/Users/andrewlamb/Software/arrow-rs/arrow-cast/src/cast/mod.rs
Line | Count | Source |
1 | | // Licensed to the Apache Software Foundation (ASF) under one |
2 | | // or more contributor license agreements. See the NOTICE file |
3 | | // distributed with this work for additional information |
4 | | // regarding copyright ownership. The ASF licenses this file |
5 | | // to you under the Apache License, Version 2.0 (the |
6 | | // "License"); you may not use this file except in compliance |
7 | | // with the License. You may obtain a copy of the License at |
8 | | // |
9 | | // http://www.apache.org/licenses/LICENSE-2.0 |
10 | | // |
11 | | // Unless required by applicable law or agreed to in writing, |
12 | | // software distributed under the License is distributed on an |
13 | | // "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY |
14 | | // KIND, either express or implied. See the License for the |
15 | | // specific language governing permissions and limitations |
16 | | // under the License. |
17 | | |
18 | | //! Cast kernels to convert [`ArrayRef`] between supported datatypes. |
19 | | //! |
20 | | //! See [`cast_with_options`] for more information on specific conversions. |
21 | | //! |
22 | | //! Example: |
23 | | //! |
24 | | //! ``` |
25 | | //! # use arrow_array::*; |
26 | | //! # use arrow_cast::cast; |
27 | | //! # use arrow_schema::DataType; |
28 | | //! # use std::sync::Arc; |
29 | | //! # use arrow_array::types::Float64Type; |
30 | | //! # use arrow_array::cast::AsArray; |
31 | | //! // int32 to float64 |
32 | | //! let a = Int32Array::from(vec![5, 6, 7]); |
33 | | //! let b = cast(&a, &DataType::Float64).unwrap(); |
34 | | //! let c = b.as_primitive::<Float64Type>(); |
35 | | //! assert_eq!(5.0, c.value(0)); |
36 | | //! assert_eq!(6.0, c.value(1)); |
37 | | //! assert_eq!(7.0, c.value(2)); |
38 | | //! ``` |
39 | | |
40 | | mod decimal; |
41 | | mod dictionary; |
42 | | mod list; |
43 | | mod map; |
44 | | mod string; |
45 | | use crate::cast::decimal::*; |
46 | | use crate::cast::dictionary::*; |
47 | | use crate::cast::list::*; |
48 | | use crate::cast::map::*; |
49 | | use crate::cast::string::*; |
50 | | |
51 | | use arrow_buffer::IntervalMonthDayNano; |
52 | | use arrow_data::ByteView; |
53 | | use chrono::{NaiveTime, Offset, TimeZone, Utc}; |
54 | | use std::cmp::Ordering; |
55 | | use std::sync::Arc; |
56 | | |
57 | | use crate::display::{ArrayFormatter, FormatOptions}; |
58 | | use crate::parse::{ |
59 | | parse_interval_day_time, parse_interval_month_day_nano, parse_interval_year_month, |
60 | | string_to_datetime, Parser, |
61 | | }; |
62 | | use arrow_array::{builder::*, cast::*, temporal_conversions::*, timezone::Tz, types::*, *}; |
63 | | use arrow_buffer::{i256, ArrowNativeType, OffsetBuffer}; |
64 | | use arrow_data::transform::MutableArrayData; |
65 | | use arrow_data::ArrayData; |
66 | | use arrow_schema::*; |
67 | | use arrow_select::take::take; |
68 | | use num::cast::AsPrimitive; |
69 | | use num::{NumCast, ToPrimitive}; |
70 | | |
71 | | /// CastOptions provides a way to override the default cast behaviors |
72 | | #[derive(Debug, Clone, PartialEq, Eq, Hash)] |
73 | | pub struct CastOptions<'a> { |
74 | | /// how to handle cast failures, either return NULL (safe=true) or return ERR (safe=false) |
75 | | pub safe: bool, |
76 | | /// Formatting options when casting from temporal types to string |
77 | | pub format_options: FormatOptions<'a>, |
78 | | } |
79 | | |
80 | | impl Default for CastOptions<'_> { |
81 | 0 | fn default() -> Self { |
82 | 0 | Self { |
83 | 0 | safe: true, |
84 | 0 | format_options: FormatOptions::default(), |
85 | 0 | } |
86 | 0 | } |
87 | | } |
88 | | |
89 | | /// Return true if a value of type `from_type` can be cast into a value of `to_type`. |
90 | | /// |
91 | | /// See [`cast_with_options`] for more information |
92 | 0 | pub fn can_cast_types(from_type: &DataType, to_type: &DataType) -> bool { |
93 | | use self::DataType::*; |
94 | | use self::IntervalUnit::*; |
95 | | use self::TimeUnit::*; |
96 | 0 | if from_type == to_type { |
97 | 0 | return true; |
98 | 0 | } |
99 | | |
100 | 0 | match (from_type, to_type) { |
101 | | ( |
102 | | Null, |
103 | | Boolean |
104 | | | Int8 |
105 | | | UInt8 |
106 | | | Int16 |
107 | | | UInt16 |
108 | | | Int32 |
109 | | | UInt32 |
110 | | | Float32 |
111 | | | Date32 |
112 | | | Time32(_) |
113 | | | Int64 |
114 | | | UInt64 |
115 | | | Float64 |
116 | | | Date64 |
117 | | | Timestamp(_, _) |
118 | | | Time64(_) |
119 | | | Duration(_) |
120 | | | Interval(_) |
121 | | | FixedSizeBinary(_) |
122 | | | Binary |
123 | | | Utf8 |
124 | | | LargeBinary |
125 | | | LargeUtf8 |
126 | | | BinaryView |
127 | | | Utf8View |
128 | | | List(_) |
129 | | | LargeList(_) |
130 | | | FixedSizeList(_, _) |
131 | | | Struct(_) |
132 | | | Map(_, _) |
133 | | | Dictionary(_, _), |
134 | 0 | ) => true, |
135 | | // Dictionary/List conditions should be put in front of others |
136 | 0 | (Dictionary(_, from_value_type), Dictionary(_, to_value_type)) => { |
137 | 0 | can_cast_types(from_value_type, to_value_type) |
138 | | } |
139 | 0 | (Dictionary(_, value_type), _) => can_cast_types(value_type, to_type), |
140 | 0 | (_, Dictionary(_, value_type)) => can_cast_types(from_type, value_type), |
141 | 0 | (List(list_from) | LargeList(list_from), List(list_to) | LargeList(list_to)) => { |
142 | 0 | can_cast_types(list_from.data_type(), list_to.data_type()) |
143 | | } |
144 | 0 | (List(list_from) | LargeList(list_from), Utf8 | LargeUtf8) => { |
145 | 0 | can_cast_types(list_from.data_type(), to_type) |
146 | | } |
147 | 0 | (List(list_from) | LargeList(list_from), FixedSizeList(list_to, _)) => { |
148 | 0 | can_cast_types(list_from.data_type(), list_to.data_type()) |
149 | | } |
150 | 0 | (List(_), _) => false, |
151 | 0 | (FixedSizeList(list_from, _), List(list_to)) |
152 | 0 | | (FixedSizeList(list_from, _), LargeList(list_to)) => { |
153 | 0 | can_cast_types(list_from.data_type(), list_to.data_type()) |
154 | | } |
155 | 0 | (FixedSizeList(inner, size), FixedSizeList(inner_to, size_to)) if size == size_to => { |
156 | 0 | can_cast_types(inner.data_type(), inner_to.data_type()) |
157 | | } |
158 | 0 | (_, List(list_to)) => can_cast_types(from_type, list_to.data_type()), |
159 | 0 | (_, LargeList(list_to)) => can_cast_types(from_type, list_to.data_type()), |
160 | 0 | (_, FixedSizeList(list_to, size)) if *size == 1 => { |
161 | 0 | can_cast_types(from_type, list_to.data_type()) |
162 | | } |
163 | 0 | (FixedSizeList(list_from, size), _) if *size == 1 => { |
164 | 0 | can_cast_types(list_from.data_type(), to_type) |
165 | | } |
166 | 0 | (Map(from_entries, ordered_from), Map(to_entries, ordered_to)) |
167 | 0 | if ordered_from == ordered_to => |
168 | | { |
169 | | match ( |
170 | 0 | key_field(from_entries), |
171 | 0 | key_field(to_entries), |
172 | 0 | value_field(from_entries), |
173 | 0 | value_field(to_entries), |
174 | | ) { |
175 | 0 | (Some(from_key), Some(to_key), Some(from_value), Some(to_value)) => { |
176 | 0 | can_cast_types(from_key.data_type(), to_key.data_type()) |
177 | 0 | && can_cast_types(from_value.data_type(), to_value.data_type()) |
178 | | } |
179 | 0 | _ => false, |
180 | | } |
181 | | } |
182 | | // cast one decimal type to another decimal type |
183 | | ( |
184 | | Decimal32(_, _) | Decimal64(_, _) | Decimal128(_, _) | Decimal256(_, _), |
185 | | Decimal32(_, _) | Decimal64(_, _) | Decimal128(_, _) | Decimal256(_, _), |
186 | 0 | ) => true, |
187 | | // unsigned integer to decimal |
188 | | ( |
189 | | UInt8 | UInt16 | UInt32 | UInt64, |
190 | | Decimal32(_, _) | Decimal64(_, _) | Decimal128(_, _) | Decimal256(_, _), |
191 | 0 | ) => true, |
192 | | // signed numeric to decimal |
193 | | ( |
194 | | Null | Int8 | Int16 | Int32 | Int64 | Float32 | Float64, |
195 | | Decimal32(_, _) | Decimal64(_, _) | Decimal128(_, _) | Decimal256(_, _), |
196 | 0 | ) => true, |
197 | | // decimal to unsigned numeric |
198 | | ( |
199 | | Decimal32(_, _) | Decimal64(_, _) | Decimal128(_, _) | Decimal256(_, _), |
200 | | UInt8 | UInt16 | UInt32 | UInt64, |
201 | 0 | ) => true, |
202 | | // decimal to signed numeric |
203 | | ( |
204 | | Decimal32(_, _) | Decimal64(_, _) | Decimal128(_, _) | Decimal256(_, _), |
205 | | Null | Int8 | Int16 | Int32 | Int64 | Float32 | Float64, |
206 | 0 | ) => true, |
207 | | // decimal to string |
208 | | ( |
209 | | Decimal32(_, _) | Decimal64(_, _) | Decimal128(_, _) | Decimal256(_, _), |
210 | | Utf8View | Utf8 | LargeUtf8, |
211 | 0 | ) => true, |
212 | | // string to decimal |
213 | | ( |
214 | | Utf8View | Utf8 | LargeUtf8, |
215 | | Decimal32(_, _) | Decimal64(_, _) | Decimal128(_, _) | Decimal256(_, _), |
216 | 0 | ) => true, |
217 | 0 | (Struct(from_fields), Struct(to_fields)) => { |
218 | 0 | from_fields.len() == to_fields.len() |
219 | 0 | && from_fields.iter().zip(to_fields.iter()).all(|(f1, f2)| { |
220 | | // Assume that nullability between two structs are compatible, if not, |
221 | | // cast kernel will return error. |
222 | 0 | can_cast_types(f1.data_type(), f2.data_type()) |
223 | 0 | }) |
224 | | } |
225 | 0 | (Struct(_), _) => false, |
226 | 0 | (_, Struct(_)) => false, |
227 | | (_, Boolean) => { |
228 | 0 | DataType::is_integer(from_type) |
229 | 0 | || DataType::is_floating(from_type) |
230 | 0 | || from_type == &Utf8View |
231 | 0 | || from_type == &Utf8 |
232 | 0 | || from_type == &LargeUtf8 |
233 | | } |
234 | | (Boolean, _) => { |
235 | 0 | DataType::is_integer(to_type) |
236 | 0 | || DataType::is_floating(to_type) |
237 | 0 | || to_type == &Utf8View |
238 | 0 | || to_type == &Utf8 |
239 | 0 | || to_type == &LargeUtf8 |
240 | | } |
241 | | |
242 | | (Binary, LargeBinary | Utf8 | LargeUtf8 | FixedSizeBinary(_) | BinaryView | Utf8View) => { |
243 | 0 | true |
244 | | } |
245 | | (LargeBinary, Binary | Utf8 | LargeUtf8 | FixedSizeBinary(_) | BinaryView | Utf8View) => { |
246 | 0 | true |
247 | | } |
248 | 0 | (FixedSizeBinary(_), Binary | LargeBinary | BinaryView) => true, |
249 | | ( |
250 | | Utf8 | LargeUtf8 | Utf8View, |
251 | | Binary |
252 | | | LargeBinary |
253 | | | Utf8 |
254 | | | LargeUtf8 |
255 | | | Date32 |
256 | | | Date64 |
257 | | | Time32(Second) |
258 | | | Time32(Millisecond) |
259 | | | Time64(Microsecond) |
260 | | | Time64(Nanosecond) |
261 | | | Timestamp(Second, _) |
262 | | | Timestamp(Millisecond, _) |
263 | | | Timestamp(Microsecond, _) |
264 | | | Timestamp(Nanosecond, _) |
265 | | | Interval(_) |
266 | | | BinaryView, |
267 | 0 | ) => true, |
268 | 0 | (Utf8 | LargeUtf8, Utf8View) => true, |
269 | 0 | (BinaryView, Binary | LargeBinary | Utf8 | LargeUtf8 | Utf8View) => true, |
270 | 0 | (Utf8View | Utf8 | LargeUtf8, _) => to_type.is_numeric() && to_type != &Float16, |
271 | 0 | (_, Utf8 | LargeUtf8) => from_type.is_primitive(), |
272 | 0 | (_, Utf8View) => from_type.is_numeric(), |
273 | | |
274 | 0 | (_, Binary | LargeBinary) => from_type.is_integer(), |
275 | | |
276 | | // start numeric casts |
277 | | ( |
278 | | UInt8 | UInt16 | UInt32 | UInt64 | Int8 | Int16 | Int32 | Int64 | Float16 | Float32 |
279 | | | Float64, |
280 | | UInt8 | UInt16 | UInt32 | UInt64 | Int8 | Int16 | Int32 | Int64 | Float16 | Float32 |
281 | | | Float64, |
282 | 0 | ) => true, |
283 | | // end numeric casts |
284 | | |
285 | | // temporal casts |
286 | 0 | (Int32, Date32 | Date64 | Time32(_)) => true, |
287 | 0 | (Date32, Int32 | Int64) => true, |
288 | 0 | (Time32(_), Int32) => true, |
289 | 0 | (Int64, Date64 | Date32 | Time64(_)) => true, |
290 | 0 | (Date64, Int64 | Int32) => true, |
291 | 0 | (Time64(_), Int64) => true, |
292 | 0 | (Date32 | Date64, Date32 | Date64) => true, |
293 | | // time casts |
294 | 0 | (Time32(_), Time32(_)) => true, |
295 | 0 | (Time32(_), Time64(_)) => true, |
296 | 0 | (Time64(_), Time64(_)) => true, |
297 | 0 | (Time64(_), Time32(to_unit)) => { |
298 | 0 | matches!(to_unit, Second | Millisecond) |
299 | | } |
300 | 0 | (Timestamp(_, _), _) if to_type.is_numeric() => true, |
301 | 0 | (_, Timestamp(_, _)) if from_type.is_numeric() => true, |
302 | 0 | (Date64, Timestamp(_, _)) => true, |
303 | 0 | (Date32, Timestamp(_, _)) => true, |
304 | | ( |
305 | | Timestamp(_, _), |
306 | | Timestamp(_, _) |
307 | | | Date32 |
308 | | | Date64 |
309 | | | Time32(Second) |
310 | | | Time32(Millisecond) |
311 | | | Time64(Microsecond) |
312 | | | Time64(Nanosecond), |
313 | 0 | ) => true, |
314 | 0 | (_, Duration(_)) if from_type.is_numeric() => true, |
315 | 0 | (Duration(_), _) if to_type.is_numeric() => true, |
316 | 0 | (Duration(_), Duration(_)) => true, |
317 | 0 | (Interval(from_type), Int64) => { |
318 | 0 | match from_type { |
319 | 0 | YearMonth => true, |
320 | 0 | DayTime => true, |
321 | 0 | MonthDayNano => false, // Native type is i128 |
322 | | } |
323 | | } |
324 | 0 | (Int32, Interval(to_type)) => match to_type { |
325 | 0 | YearMonth => true, |
326 | 0 | DayTime => false, |
327 | 0 | MonthDayNano => false, |
328 | | }, |
329 | 0 | (Duration(_), Interval(MonthDayNano)) => true, |
330 | 0 | (Interval(MonthDayNano), Duration(_)) => true, |
331 | 0 | (Interval(YearMonth), Interval(MonthDayNano)) => true, |
332 | 0 | (Interval(DayTime), Interval(MonthDayNano)) => true, |
333 | 0 | (_, _) => false, |
334 | | } |
335 | 0 | } |
336 | | |
337 | | /// Cast `array` to the provided data type and return a new Array with type `to_type`, if possible. |
338 | | /// |
339 | | /// See [`cast_with_options`] for more information |
340 | 0 | pub fn cast(array: &dyn Array, to_type: &DataType) -> Result<ArrayRef, ArrowError> { |
341 | 0 | cast_with_options(array, to_type, &CastOptions::default()) |
342 | 0 | } |
343 | | |
344 | 0 | fn cast_integer_to_decimal< |
345 | 0 | T: ArrowPrimitiveType, |
346 | 0 | D: DecimalType + ArrowPrimitiveType<Native = M>, |
347 | 0 | M, |
348 | 0 | >( |
349 | 0 | array: &PrimitiveArray<T>, |
350 | 0 | precision: u8, |
351 | 0 | scale: i8, |
352 | 0 | base: M, |
353 | 0 | cast_options: &CastOptions, |
354 | 0 | ) -> Result<ArrayRef, ArrowError> |
355 | 0 | where |
356 | 0 | <T as ArrowPrimitiveType>::Native: AsPrimitive<M>, |
357 | 0 | M: ArrowNativeTypeOp, |
358 | | { |
359 | 0 | let scale_factor = base.pow_checked(scale.unsigned_abs() as u32).map_err(|_| { |
360 | 0 | ArrowError::CastError(format!( |
361 | 0 | "Cannot cast to {:?}({}, {}). The scale causes overflow.", |
362 | 0 | D::PREFIX, |
363 | 0 | precision, |
364 | 0 | scale, |
365 | 0 | )) |
366 | 0 | })?; |
367 | | |
368 | 0 | let array = if scale < 0 { |
369 | 0 | match cast_options.safe { |
370 | 0 | true => array.unary_opt::<_, D>(|v| { |
371 | 0 | v.as_() |
372 | 0 | .div_checked(scale_factor) |
373 | 0 | .ok() |
374 | 0 | .and_then(|v| (D::is_valid_decimal_precision(v, precision)).then_some(v)) |
375 | 0 | }), |
376 | 0 | false => array.try_unary::<_, D, _>(|v| { |
377 | 0 | v.as_() |
378 | 0 | .div_checked(scale_factor) |
379 | 0 | .and_then(|v| D::validate_decimal_precision(v, precision).map(|_| v)) |
380 | 0 | })?, |
381 | | } |
382 | | } else { |
383 | 0 | match cast_options.safe { |
384 | 0 | true => array.unary_opt::<_, D>(|v| { |
385 | 0 | v.as_() |
386 | 0 | .mul_checked(scale_factor) |
387 | 0 | .ok() |
388 | 0 | .and_then(|v| (D::is_valid_decimal_precision(v, precision)).then_some(v)) |
389 | 0 | }), |
390 | 0 | false => array.try_unary::<_, D, _>(|v| { |
391 | 0 | v.as_() |
392 | 0 | .mul_checked(scale_factor) |
393 | 0 | .and_then(|v| D::validate_decimal_precision(v, precision).map(|_| v)) |
394 | 0 | })?, |
395 | | } |
396 | | }; |
397 | | |
398 | 0 | Ok(Arc::new(array.with_precision_and_scale(precision, scale)?)) |
399 | 0 | } |
400 | | |
401 | | /// Cast the array from interval year month to month day nano |
402 | 0 | fn cast_interval_year_month_to_interval_month_day_nano( |
403 | 0 | array: &dyn Array, |
404 | 0 | _cast_options: &CastOptions, |
405 | 0 | ) -> Result<ArrayRef, ArrowError> { |
406 | 0 | let array = array.as_primitive::<IntervalYearMonthType>(); |
407 | | |
408 | 0 | Ok(Arc::new(array.unary::<_, IntervalMonthDayNanoType>(|v| { |
409 | 0 | let months = IntervalYearMonthType::to_months(v); |
410 | 0 | IntervalMonthDayNanoType::make_value(months, 0, 0) |
411 | 0 | }))) |
412 | 0 | } |
413 | | |
414 | | /// Cast the array from interval day time to month day nano |
415 | 0 | fn cast_interval_day_time_to_interval_month_day_nano( |
416 | 0 | array: &dyn Array, |
417 | 0 | _cast_options: &CastOptions, |
418 | 0 | ) -> Result<ArrayRef, ArrowError> { |
419 | 0 | let array = array.as_primitive::<IntervalDayTimeType>(); |
420 | 0 | let mul = 1_000_000; |
421 | | |
422 | 0 | Ok(Arc::new(array.unary::<_, IntervalMonthDayNanoType>(|v| { |
423 | 0 | let (days, ms) = IntervalDayTimeType::to_parts(v); |
424 | 0 | IntervalMonthDayNanoType::make_value(0, days, ms as i64 * mul) |
425 | 0 | }))) |
426 | 0 | } |
427 | | |
428 | | /// Cast the array from interval to duration |
429 | 0 | fn cast_month_day_nano_to_duration<D: ArrowTemporalType<Native = i64>>( |
430 | 0 | array: &dyn Array, |
431 | 0 | cast_options: &CastOptions, |
432 | 0 | ) -> Result<ArrayRef, ArrowError> { |
433 | 0 | let array = array.as_primitive::<IntervalMonthDayNanoType>(); |
434 | 0 | let scale = match D::DATA_TYPE { |
435 | 0 | DataType::Duration(TimeUnit::Second) => 1_000_000_000, |
436 | 0 | DataType::Duration(TimeUnit::Millisecond) => 1_000_000, |
437 | 0 | DataType::Duration(TimeUnit::Microsecond) => 1_000, |
438 | 0 | DataType::Duration(TimeUnit::Nanosecond) => 1, |
439 | 0 | _ => unreachable!(), |
440 | | }; |
441 | | |
442 | 0 | if cast_options.safe { |
443 | 0 | let iter = array.iter().map(|v| { |
444 | 0 | v.and_then(|v| (v.days == 0 && v.months == 0).then_some(v.nanoseconds / scale)) |
445 | 0 | }); |
446 | 0 | Ok(Arc::new(unsafe { |
447 | 0 | PrimitiveArray::<D>::from_trusted_len_iter(iter) |
448 | 0 | })) |
449 | | } else { |
450 | 0 | let vec = array |
451 | 0 | .iter() |
452 | 0 | .map(|v| { |
453 | 0 | v.map(|v| match v.days == 0 && v.months == 0 { |
454 | 0 | true => Ok((v.nanoseconds) / scale), |
455 | 0 | _ => Err(ArrowError::ComputeError( |
456 | 0 | "Cannot convert interval containing non-zero months or days to duration" |
457 | 0 | .to_string(), |
458 | 0 | )), |
459 | 0 | }) |
460 | 0 | .transpose() |
461 | 0 | }) |
462 | 0 | .collect::<Result<Vec<_>, _>>()?; |
463 | 0 | Ok(Arc::new(unsafe { |
464 | 0 | PrimitiveArray::<D>::from_trusted_len_iter(vec.iter()) |
465 | 0 | })) |
466 | | } |
467 | 0 | } |
468 | | |
469 | | /// Cast the array from duration and interval |
470 | 0 | fn cast_duration_to_interval<D: ArrowTemporalType<Native = i64>>( |
471 | 0 | array: &dyn Array, |
472 | 0 | cast_options: &CastOptions, |
473 | 0 | ) -> Result<ArrayRef, ArrowError> { |
474 | 0 | let array = array |
475 | 0 | .as_any() |
476 | 0 | .downcast_ref::<PrimitiveArray<D>>() |
477 | 0 | .ok_or_else(|| { |
478 | 0 | ArrowError::ComputeError( |
479 | 0 | "Internal Error: Cannot cast duration to DurationArray of expected type" |
480 | 0 | .to_string(), |
481 | 0 | ) |
482 | 0 | })?; |
483 | | |
484 | 0 | let scale = match array.data_type() { |
485 | 0 | DataType::Duration(TimeUnit::Second) => 1_000_000_000, |
486 | 0 | DataType::Duration(TimeUnit::Millisecond) => 1_000_000, |
487 | 0 | DataType::Duration(TimeUnit::Microsecond) => 1_000, |
488 | 0 | DataType::Duration(TimeUnit::Nanosecond) => 1, |
489 | 0 | _ => unreachable!(), |
490 | | }; |
491 | | |
492 | 0 | if cast_options.safe { |
493 | 0 | let iter = array.iter().map(|v| { |
494 | 0 | v.and_then(|v| { |
495 | 0 | v.checked_mul(scale) |
496 | 0 | .map(|v| IntervalMonthDayNano::new(0, 0, v)) |
497 | 0 | }) |
498 | 0 | }); |
499 | 0 | Ok(Arc::new(unsafe { |
500 | 0 | PrimitiveArray::<IntervalMonthDayNanoType>::from_trusted_len_iter(iter) |
501 | 0 | })) |
502 | | } else { |
503 | 0 | let vec = array |
504 | 0 | .iter() |
505 | 0 | .map(|v| { |
506 | 0 | v.map(|v| { |
507 | 0 | if let Ok(v) = v.mul_checked(scale) { |
508 | 0 | Ok(IntervalMonthDayNano::new(0, 0, v)) |
509 | | } else { |
510 | 0 | Err(ArrowError::ComputeError(format!( |
511 | 0 | "Cannot cast to {:?}. Overflowing on {:?}", |
512 | 0 | IntervalMonthDayNanoType::DATA_TYPE, |
513 | 0 | v |
514 | 0 | ))) |
515 | | } |
516 | 0 | }) |
517 | 0 | .transpose() |
518 | 0 | }) |
519 | 0 | .collect::<Result<Vec<_>, _>>()?; |
520 | 0 | Ok(Arc::new(unsafe { |
521 | 0 | PrimitiveArray::<IntervalMonthDayNanoType>::from_trusted_len_iter(vec.iter()) |
522 | 0 | })) |
523 | | } |
524 | 0 | } |
525 | | |
526 | | /// Cast the primitive array using [`PrimitiveArray::reinterpret_cast`] |
527 | 0 | fn cast_reinterpret_arrays<I: ArrowPrimitiveType, O: ArrowPrimitiveType<Native = I::Native>>( |
528 | 0 | array: &dyn Array, |
529 | 0 | ) -> Result<ArrayRef, ArrowError> { |
530 | 0 | Ok(Arc::new(array.as_primitive::<I>().reinterpret_cast::<O>())) |
531 | 0 | } |
532 | | |
533 | 0 | fn make_timestamp_array( |
534 | 0 | array: &PrimitiveArray<Int64Type>, |
535 | 0 | unit: TimeUnit, |
536 | 0 | tz: Option<Arc<str>>, |
537 | 0 | ) -> ArrayRef { |
538 | 0 | match unit { |
539 | 0 | TimeUnit::Second => Arc::new( |
540 | 0 | array |
541 | 0 | .reinterpret_cast::<TimestampSecondType>() |
542 | 0 | .with_timezone_opt(tz), |
543 | 0 | ), |
544 | 0 | TimeUnit::Millisecond => Arc::new( |
545 | 0 | array |
546 | 0 | .reinterpret_cast::<TimestampMillisecondType>() |
547 | 0 | .with_timezone_opt(tz), |
548 | 0 | ), |
549 | 0 | TimeUnit::Microsecond => Arc::new( |
550 | 0 | array |
551 | 0 | .reinterpret_cast::<TimestampMicrosecondType>() |
552 | 0 | .with_timezone_opt(tz), |
553 | 0 | ), |
554 | 0 | TimeUnit::Nanosecond => Arc::new( |
555 | 0 | array |
556 | 0 | .reinterpret_cast::<TimestampNanosecondType>() |
557 | 0 | .with_timezone_opt(tz), |
558 | 0 | ), |
559 | | } |
560 | 0 | } |
561 | | |
562 | 0 | fn make_duration_array(array: &PrimitiveArray<Int64Type>, unit: TimeUnit) -> ArrayRef { |
563 | 0 | match unit { |
564 | 0 | TimeUnit::Second => Arc::new(array.reinterpret_cast::<DurationSecondType>()), |
565 | 0 | TimeUnit::Millisecond => Arc::new(array.reinterpret_cast::<DurationMillisecondType>()), |
566 | 0 | TimeUnit::Microsecond => Arc::new(array.reinterpret_cast::<DurationMicrosecondType>()), |
567 | 0 | TimeUnit::Nanosecond => Arc::new(array.reinterpret_cast::<DurationNanosecondType>()), |
568 | | } |
569 | 0 | } |
570 | | |
571 | 0 | fn as_time_res_with_timezone<T: ArrowPrimitiveType>( |
572 | 0 | v: i64, |
573 | 0 | tz: Option<Tz>, |
574 | 0 | ) -> Result<NaiveTime, ArrowError> { |
575 | 0 | let time = match tz { |
576 | 0 | Some(tz) => as_datetime_with_timezone::<T>(v, tz).map(|d| d.time()), |
577 | 0 | None => as_datetime::<T>(v).map(|d| d.time()), |
578 | | }; |
579 | | |
580 | 0 | time.ok_or_else(|| { |
581 | 0 | ArrowError::CastError(format!( |
582 | 0 | "Failed to create naive time with {} {}", |
583 | 0 | std::any::type_name::<T>(), |
584 | 0 | v |
585 | 0 | )) |
586 | 0 | }) |
587 | 0 | } |
588 | | |
589 | 0 | fn timestamp_to_date32<T: ArrowTimestampType>( |
590 | 0 | array: &PrimitiveArray<T>, |
591 | 0 | ) -> Result<ArrayRef, ArrowError> { |
592 | 0 | let err = |x: i64| { |
593 | 0 | ArrowError::CastError(format!( |
594 | 0 | "Cannot convert {} {x} to datetime", |
595 | 0 | std::any::type_name::<T>() |
596 | 0 | )) |
597 | 0 | }; |
598 | | |
599 | 0 | let array: Date32Array = match array.timezone() { |
600 | 0 | Some(tz) => { |
601 | 0 | let tz: Tz = tz.parse()?; |
602 | 0 | array.try_unary(|x| { |
603 | 0 | as_datetime_with_timezone::<T>(x, tz) |
604 | 0 | .ok_or_else(|| err(x)) |
605 | 0 | .map(|d| Date32Type::from_naive_date(d.date_naive())) |
606 | 0 | })? |
607 | | } |
608 | 0 | None => array.try_unary(|x| { |
609 | 0 | as_datetime::<T>(x) |
610 | 0 | .ok_or_else(|| err(x)) |
611 | 0 | .map(|d| Date32Type::from_naive_date(d.date())) |
612 | 0 | })?, |
613 | | }; |
614 | 0 | Ok(Arc::new(array)) |
615 | 0 | } |
616 | | |
617 | | /// Try to cast `array` to `to_type` if possible. |
618 | | /// |
619 | | /// Returns a new Array with type `to_type` if possible. |
620 | | /// |
621 | | /// Accepts [`CastOptions`] to specify cast behavior. See also [`cast()`]. |
622 | | /// |
623 | | /// # Behavior |
624 | | /// * `Boolean` to `Utf8`: `true` => '1', `false` => `0` |
625 | | /// * `Utf8` to `Boolean`: `true`, `yes`, `on`, `1` => `true`, `false`, `no`, `off`, `0` => `false`, |
626 | | /// short variants are accepted, other strings return null or error |
627 | | /// * `Utf8` to Numeric: strings that can't be parsed to numbers return null, float strings |
628 | | /// in integer casts return null |
629 | | /// * Numeric to `Boolean`: 0 returns `false`, any other value returns `true` |
630 | | /// * `List` to `List`: the underlying data type is cast |
631 | | /// * `List` to `FixedSizeList`: the underlying data type is cast. If safe is true and a list element |
632 | | /// has the wrong length it will be replaced with NULL, otherwise an error will be returned |
633 | | /// * Primitive to `List`: a list array with 1 value per slot is created |
634 | | /// * `Date32` and `Date64`: precision lost when going to higher interval |
635 | | /// * `Time32 and `Time64`: precision lost when going to higher interval |
636 | | /// * `Timestamp` and `Date{32|64}`: precision lost when going to higher interval |
637 | | /// * Temporal to/from backing Primitive: zero-copy with data type change |
638 | | /// * `Float32/Float64` to `Decimal(precision, scale)` rounds to the `scale` decimals |
639 | | /// (i.e. casting `6.4999` to `Decimal(10, 1)` becomes `6.5`). |
640 | | /// * `Decimal` to `Float32/Float64` is lossy and values outside the representable |
641 | | /// range become `INFINITY` or `-INFINITY` without error. |
642 | | /// |
643 | | /// Unsupported Casts (check with `can_cast_types` before calling): |
644 | | /// * To or from `StructArray` |
645 | | /// * `List` to `Primitive` |
646 | | /// * `Interval` and `Duration` |
647 | | /// |
648 | | /// # Durations and Intervals |
649 | | /// |
650 | | /// Casting integer types directly to interval types such as |
651 | | /// [`IntervalMonthDayNano`] is not supported because the meaning of the integer |
652 | | /// is ambiguous. For example, the integer could represent either nanoseconds |
653 | | /// or months. |
654 | | /// |
655 | | /// To cast an integer type to an interval type, first convert to a Duration |
656 | | /// type, and then cast that to the desired interval type. |
657 | | /// |
658 | | /// For example, to convert an `Int64` representing nanoseconds to an |
659 | | /// `IntervalMonthDayNano` you would first convert the `Int64` to a |
660 | | /// `DurationNanoseconds`, and then cast that to `IntervalMonthDayNano`. |
661 | | /// |
662 | | /// # Timestamps and Timezones |
663 | | /// |
664 | | /// Timestamps are stored with an optional timezone in Arrow. |
665 | | /// |
666 | | /// ## Casting timestamps to a timestamp without timezone / UTC |
667 | | /// ``` |
668 | | /// # use arrow_array::Int64Array; |
669 | | /// # use arrow_array::types::TimestampSecondType; |
670 | | /// # use arrow_cast::{cast, display}; |
671 | | /// # use arrow_array::cast::AsArray; |
672 | | /// # use arrow_schema::{DataType, TimeUnit}; |
673 | | /// // can use "UTC" if chrono-tz feature is enabled, here use offset based timezone |
674 | | /// let data_type = DataType::Timestamp(TimeUnit::Second, None); |
675 | | /// let a = Int64Array::from(vec![1_000_000_000, 2_000_000_000, 3_000_000_000]); |
676 | | /// let b = cast(&a, &data_type).unwrap(); |
677 | | /// let b = b.as_primitive::<TimestampSecondType>(); // downcast to result type |
678 | | /// assert_eq!(2_000_000_000, b.value(1)); // values are the same as the type has no timezone |
679 | | /// // use display to show them (note has no trailing Z) |
680 | | /// assert_eq!("2033-05-18T03:33:20", display::array_value_to_string(&b, 1).unwrap()); |
681 | | /// ``` |
682 | | /// |
683 | | /// ## Casting timestamps to a timestamp with timezone |
684 | | /// |
685 | | /// Similarly to the previous example, if you cast numeric values to a timestamp |
686 | | /// with timezone, the cast kernel will not change the underlying values |
687 | | /// but display and other functions will interpret them as being in the provided timezone. |
688 | | /// |
689 | | /// ``` |
690 | | /// # use arrow_array::Int64Array; |
691 | | /// # use arrow_array::types::TimestampSecondType; |
692 | | /// # use arrow_cast::{cast, display}; |
693 | | /// # use arrow_array::cast::AsArray; |
694 | | /// # use arrow_schema::{DataType, TimeUnit}; |
695 | | /// // can use "Americas/New_York" if chrono-tz feature is enabled, here use offset based timezone |
696 | | /// let data_type = DataType::Timestamp(TimeUnit::Second, Some("-05:00".into())); |
697 | | /// let a = Int64Array::from(vec![1_000_000_000, 2_000_000_000, 3_000_000_000]); |
698 | | /// let b = cast(&a, &data_type).unwrap(); |
699 | | /// let b = b.as_primitive::<TimestampSecondType>(); // downcast to result type |
700 | | /// assert_eq!(2_000_000_000, b.value(1)); // values are still the same |
701 | | /// // displayed in the target timezone (note the offset -05:00) |
702 | | /// assert_eq!("2033-05-17T22:33:20-05:00", display::array_value_to_string(&b, 1).unwrap()); |
703 | | /// ``` |
704 | | /// # Casting timestamps without timezone to timestamps with timezone |
705 | | /// |
706 | | /// When casting from a timestamp without timezone to a timestamp with |
707 | | /// timezone, the cast kernel interprets the timestamp values as being in |
708 | | /// the destination timezone and then adjusts the underlying value to UTC as required |
709 | | /// |
710 | | /// However, note that when casting from a timestamp with timezone BACK to a |
711 | | /// timestamp without timezone the cast kernel does not adjust the values. |
712 | | /// |
713 | | /// Thus round trip casting a timestamp without timezone to a timestamp with |
714 | | /// timezone and back to a timestamp without timezone results in different |
715 | | /// values than the starting values. |
716 | | /// |
717 | | /// ``` |
718 | | /// # use arrow_array::Int64Array; |
719 | | /// # use arrow_array::types::{TimestampSecondType}; |
720 | | /// # use arrow_cast::{cast, display}; |
721 | | /// # use arrow_array::cast::AsArray; |
722 | | /// # use arrow_schema::{DataType, TimeUnit}; |
723 | | /// let data_type = DataType::Timestamp(TimeUnit::Second, None); |
724 | | /// let data_type_tz = DataType::Timestamp(TimeUnit::Second, Some("-05:00".into())); |
725 | | /// let a = Int64Array::from(vec![1_000_000_000, 2_000_000_000, 3_000_000_000]); |
726 | | /// let b = cast(&a, &data_type).unwrap(); // cast to timestamp without timezone |
727 | | /// let b = b.as_primitive::<TimestampSecondType>(); // downcast to result type |
728 | | /// assert_eq!(2_000_000_000, b.value(1)); // values are still the same |
729 | | /// // displayed without a timezone (note lack of offset or Z) |
730 | | /// assert_eq!("2033-05-18T03:33:20", display::array_value_to_string(&b, 1).unwrap()); |
731 | | /// |
732 | | /// // Convert timestamps without a timezone to timestamps with a timezone |
733 | | /// let c = cast(&b, &data_type_tz).unwrap(); |
734 | | /// let c = c.as_primitive::<TimestampSecondType>(); // downcast to result type |
735 | | /// assert_eq!(2_000_018_000, c.value(1)); // value has been adjusted by offset |
736 | | /// // displayed with the target timezone offset (-05:00) |
737 | | /// assert_eq!("2033-05-18T03:33:20-05:00", display::array_value_to_string(&c, 1).unwrap()); |
738 | | /// |
739 | | /// // Convert from timestamp with timezone back to timestamp without timezone |
740 | | /// let d = cast(&c, &data_type).unwrap(); |
741 | | /// let d = d.as_primitive::<TimestampSecondType>(); // downcast to result type |
742 | | /// assert_eq!(2_000_018_000, d.value(1)); // value has not been adjusted |
743 | | /// // NOTE: the timestamp is adjusted (08:33:20 instead of 03:33:20 as in previous example) |
744 | | /// assert_eq!("2033-05-18T08:33:20", display::array_value_to_string(&d, 1).unwrap()); |
745 | | /// ``` |
746 | 0 | pub fn cast_with_options( |
747 | 0 | array: &dyn Array, |
748 | 0 | to_type: &DataType, |
749 | 0 | cast_options: &CastOptions, |
750 | 0 | ) -> Result<ArrayRef, ArrowError> { |
751 | | use DataType::*; |
752 | 0 | let from_type = array.data_type(); |
753 | | // clone array if types are the same |
754 | 0 | if from_type == to_type { |
755 | 0 | return Ok(make_array(array.to_data())); |
756 | 0 | } |
757 | 0 | match (from_type, to_type) { |
758 | | ( |
759 | | Null, |
760 | | Boolean |
761 | | | Int8 |
762 | | | UInt8 |
763 | | | Int16 |
764 | | | UInt16 |
765 | | | Int32 |
766 | | | UInt32 |
767 | | | Float32 |
768 | | | Date32 |
769 | | | Time32(_) |
770 | | | Int64 |
771 | | | UInt64 |
772 | | | Float64 |
773 | | | Date64 |
774 | | | Timestamp(_, _) |
775 | | | Time64(_) |
776 | | | Duration(_) |
777 | | | Interval(_) |
778 | | | FixedSizeBinary(_) |
779 | | | Binary |
780 | | | Utf8 |
781 | | | LargeBinary |
782 | | | LargeUtf8 |
783 | | | BinaryView |
784 | | | Utf8View |
785 | | | List(_) |
786 | | | LargeList(_) |
787 | | | FixedSizeList(_, _) |
788 | | | Struct(_) |
789 | | | Map(_, _) |
790 | | | Dictionary(_, _), |
791 | 0 | ) => Ok(new_null_array(to_type, array.len())), |
792 | 0 | (Dictionary(index_type, _), _) => match **index_type { |
793 | 0 | Int8 => dictionary_cast::<Int8Type>(array, to_type, cast_options), |
794 | 0 | Int16 => dictionary_cast::<Int16Type>(array, to_type, cast_options), |
795 | 0 | Int32 => dictionary_cast::<Int32Type>(array, to_type, cast_options), |
796 | 0 | Int64 => dictionary_cast::<Int64Type>(array, to_type, cast_options), |
797 | 0 | UInt8 => dictionary_cast::<UInt8Type>(array, to_type, cast_options), |
798 | 0 | UInt16 => dictionary_cast::<UInt16Type>(array, to_type, cast_options), |
799 | 0 | UInt32 => dictionary_cast::<UInt32Type>(array, to_type, cast_options), |
800 | 0 | UInt64 => dictionary_cast::<UInt64Type>(array, to_type, cast_options), |
801 | 0 | _ => Err(ArrowError::CastError(format!( |
802 | 0 | "Casting from dictionary type {from_type:?} to {to_type:?} not supported", |
803 | 0 | ))), |
804 | | }, |
805 | 0 | (_, Dictionary(index_type, value_type)) => match **index_type { |
806 | 0 | Int8 => cast_to_dictionary::<Int8Type>(array, value_type, cast_options), |
807 | 0 | Int16 => cast_to_dictionary::<Int16Type>(array, value_type, cast_options), |
808 | 0 | Int32 => cast_to_dictionary::<Int32Type>(array, value_type, cast_options), |
809 | 0 | Int64 => cast_to_dictionary::<Int64Type>(array, value_type, cast_options), |
810 | 0 | UInt8 => cast_to_dictionary::<UInt8Type>(array, value_type, cast_options), |
811 | 0 | UInt16 => cast_to_dictionary::<UInt16Type>(array, value_type, cast_options), |
812 | 0 | UInt32 => cast_to_dictionary::<UInt32Type>(array, value_type, cast_options), |
813 | 0 | UInt64 => cast_to_dictionary::<UInt64Type>(array, value_type, cast_options), |
814 | 0 | _ => Err(ArrowError::CastError(format!( |
815 | 0 | "Casting from type {from_type:?} to dictionary type {to_type:?} not supported", |
816 | 0 | ))), |
817 | | }, |
818 | 0 | (List(_), List(to)) => cast_list_values::<i32>(array, to, cast_options), |
819 | 0 | (LargeList(_), LargeList(to)) => cast_list_values::<i64>(array, to, cast_options), |
820 | 0 | (List(_), LargeList(list_to)) => cast_list::<i32, i64>(array, list_to, cast_options), |
821 | 0 | (LargeList(_), List(list_to)) => cast_list::<i64, i32>(array, list_to, cast_options), |
822 | 0 | (List(_), FixedSizeList(field, size)) => { |
823 | 0 | let array = array.as_list::<i32>(); |
824 | 0 | cast_list_to_fixed_size_list::<i32>(array, field, *size, cast_options) |
825 | | } |
826 | 0 | (LargeList(_), FixedSizeList(field, size)) => { |
827 | 0 | let array = array.as_list::<i64>(); |
828 | 0 | cast_list_to_fixed_size_list::<i64>(array, field, *size, cast_options) |
829 | | } |
830 | 0 | (List(_) | LargeList(_), _) => match to_type { |
831 | 0 | Utf8 => value_to_string::<i32>(array, cast_options), |
832 | 0 | LargeUtf8 => value_to_string::<i64>(array, cast_options), |
833 | 0 | _ => Err(ArrowError::CastError( |
834 | 0 | "Cannot cast list to non-list data types".to_string(), |
835 | 0 | )), |
836 | | }, |
837 | 0 | (FixedSizeList(list_from, size), List(list_to)) => { |
838 | 0 | if list_to.data_type() != list_from.data_type() { |
839 | | // To transform inner type, can first cast to FSL with new inner type. |
840 | 0 | let fsl_to = DataType::FixedSizeList(list_to.clone(), *size); |
841 | 0 | let array = cast_with_options(array, &fsl_to, cast_options)?; |
842 | 0 | cast_fixed_size_list_to_list::<i32>(array.as_ref()) |
843 | | } else { |
844 | 0 | cast_fixed_size_list_to_list::<i32>(array) |
845 | | } |
846 | | } |
847 | 0 | (FixedSizeList(list_from, size), LargeList(list_to)) => { |
848 | 0 | if list_to.data_type() != list_from.data_type() { |
849 | | // To transform inner type, can first cast to FSL with new inner type. |
850 | 0 | let fsl_to = DataType::FixedSizeList(list_to.clone(), *size); |
851 | 0 | let array = cast_with_options(array, &fsl_to, cast_options)?; |
852 | 0 | cast_fixed_size_list_to_list::<i64>(array.as_ref()) |
853 | | } else { |
854 | 0 | cast_fixed_size_list_to_list::<i64>(array) |
855 | | } |
856 | | } |
857 | 0 | (FixedSizeList(_, size_from), FixedSizeList(list_to, size_to)) => { |
858 | 0 | if size_from != size_to { |
859 | 0 | return Err(ArrowError::CastError( |
860 | 0 | "cannot cast fixed-size-list to fixed-size-list with different size".into(), |
861 | 0 | )); |
862 | 0 | } |
863 | 0 | let array = array.as_any().downcast_ref::<FixedSizeListArray>().unwrap(); |
864 | 0 | let values = cast_with_options(array.values(), list_to.data_type(), cast_options)?; |
865 | 0 | Ok(Arc::new(FixedSizeListArray::try_new( |
866 | 0 | list_to.clone(), |
867 | 0 | *size_from, |
868 | 0 | values, |
869 | 0 | array.nulls().cloned(), |
870 | 0 | )?)) |
871 | | } |
872 | 0 | (_, List(ref to)) => cast_values_to_list::<i32>(array, to, cast_options), |
873 | 0 | (_, LargeList(ref to)) => cast_values_to_list::<i64>(array, to, cast_options), |
874 | 0 | (_, FixedSizeList(ref to, size)) if *size == 1 => { |
875 | 0 | cast_values_to_fixed_size_list(array, to, *size, cast_options) |
876 | | } |
877 | 0 | (FixedSizeList(_, size), _) if *size == 1 => { |
878 | 0 | cast_single_element_fixed_size_list_to_values(array, to_type, cast_options) |
879 | | } |
880 | 0 | (Map(_, ordered1), Map(_, ordered2)) if ordered1 == ordered2 => { |
881 | 0 | cast_map_values(array.as_map(), to_type, cast_options, ordered1.to_owned()) |
882 | | } |
883 | | // Decimal to decimal, same width |
884 | 0 | (Decimal32(p1, s1), Decimal32(p2, s2)) => { |
885 | 0 | cast_decimal_to_decimal_same_type::<Decimal32Type>( |
886 | 0 | array.as_primitive(), |
887 | 0 | *p1, |
888 | 0 | *s1, |
889 | 0 | *p2, |
890 | 0 | *s2, |
891 | 0 | cast_options, |
892 | | ) |
893 | | } |
894 | 0 | (Decimal64(p1, s1), Decimal64(p2, s2)) => { |
895 | 0 | cast_decimal_to_decimal_same_type::<Decimal64Type>( |
896 | 0 | array.as_primitive(), |
897 | 0 | *p1, |
898 | 0 | *s1, |
899 | 0 | *p2, |
900 | 0 | *s2, |
901 | 0 | cast_options, |
902 | | ) |
903 | | } |
904 | 0 | (Decimal128(p1, s1), Decimal128(p2, s2)) => { |
905 | 0 | cast_decimal_to_decimal_same_type::<Decimal128Type>( |
906 | 0 | array.as_primitive(), |
907 | 0 | *p1, |
908 | 0 | *s1, |
909 | 0 | *p2, |
910 | 0 | *s2, |
911 | 0 | cast_options, |
912 | | ) |
913 | | } |
914 | 0 | (Decimal256(p1, s1), Decimal256(p2, s2)) => { |
915 | 0 | cast_decimal_to_decimal_same_type::<Decimal256Type>( |
916 | 0 | array.as_primitive(), |
917 | 0 | *p1, |
918 | 0 | *s1, |
919 | 0 | *p2, |
920 | 0 | *s2, |
921 | 0 | cast_options, |
922 | | ) |
923 | | } |
924 | | // Decimal to decimal, different width |
925 | 0 | (Decimal32(p1, s1), Decimal64(p2, s2)) => { |
926 | 0 | cast_decimal_to_decimal::<Decimal32Type, Decimal64Type>( |
927 | 0 | array.as_primitive(), |
928 | 0 | *p1, |
929 | 0 | *s1, |
930 | 0 | *p2, |
931 | 0 | *s2, |
932 | 0 | cast_options, |
933 | | ) |
934 | | } |
935 | 0 | (Decimal32(p1, s1), Decimal128(p2, s2)) => { |
936 | 0 | cast_decimal_to_decimal::<Decimal32Type, Decimal128Type>( |
937 | 0 | array.as_primitive(), |
938 | 0 | *p1, |
939 | 0 | *s1, |
940 | 0 | *p2, |
941 | 0 | *s2, |
942 | 0 | cast_options, |
943 | | ) |
944 | | } |
945 | 0 | (Decimal32(p1, s1), Decimal256(p2, s2)) => { |
946 | 0 | cast_decimal_to_decimal::<Decimal32Type, Decimal256Type>( |
947 | 0 | array.as_primitive(), |
948 | 0 | *p1, |
949 | 0 | *s1, |
950 | 0 | *p2, |
951 | 0 | *s2, |
952 | 0 | cast_options, |
953 | | ) |
954 | | } |
955 | 0 | (Decimal64(p1, s1), Decimal32(p2, s2)) => { |
956 | 0 | cast_decimal_to_decimal::<Decimal64Type, Decimal32Type>( |
957 | 0 | array.as_primitive(), |
958 | 0 | *p1, |
959 | 0 | *s1, |
960 | 0 | *p2, |
961 | 0 | *s2, |
962 | 0 | cast_options, |
963 | | ) |
964 | | } |
965 | 0 | (Decimal64(p1, s1), Decimal128(p2, s2)) => { |
966 | 0 | cast_decimal_to_decimal::<Decimal64Type, Decimal128Type>( |
967 | 0 | array.as_primitive(), |
968 | 0 | *p1, |
969 | 0 | *s1, |
970 | 0 | *p2, |
971 | 0 | *s2, |
972 | 0 | cast_options, |
973 | | ) |
974 | | } |
975 | 0 | (Decimal64(p1, s1), Decimal256(p2, s2)) => { |
976 | 0 | cast_decimal_to_decimal::<Decimal64Type, Decimal256Type>( |
977 | 0 | array.as_primitive(), |
978 | 0 | *p1, |
979 | 0 | *s1, |
980 | 0 | *p2, |
981 | 0 | *s2, |
982 | 0 | cast_options, |
983 | | ) |
984 | | } |
985 | 0 | (Decimal128(p1, s1), Decimal32(p2, s2)) => { |
986 | 0 | cast_decimal_to_decimal::<Decimal128Type, Decimal32Type>( |
987 | 0 | array.as_primitive(), |
988 | 0 | *p1, |
989 | 0 | *s1, |
990 | 0 | *p2, |
991 | 0 | *s2, |
992 | 0 | cast_options, |
993 | | ) |
994 | | } |
995 | 0 | (Decimal128(p1, s1), Decimal64(p2, s2)) => { |
996 | 0 | cast_decimal_to_decimal::<Decimal128Type, Decimal64Type>( |
997 | 0 | array.as_primitive(), |
998 | 0 | *p1, |
999 | 0 | *s1, |
1000 | 0 | *p2, |
1001 | 0 | *s2, |
1002 | 0 | cast_options, |
1003 | | ) |
1004 | | } |
1005 | 0 | (Decimal128(p1, s1), Decimal256(p2, s2)) => { |
1006 | 0 | cast_decimal_to_decimal::<Decimal128Type, Decimal256Type>( |
1007 | 0 | array.as_primitive(), |
1008 | 0 | *p1, |
1009 | 0 | *s1, |
1010 | 0 | *p2, |
1011 | 0 | *s2, |
1012 | 0 | cast_options, |
1013 | | ) |
1014 | | } |
1015 | 0 | (Decimal256(p1, s1), Decimal32(p2, s2)) => { |
1016 | 0 | cast_decimal_to_decimal::<Decimal256Type, Decimal32Type>( |
1017 | 0 | array.as_primitive(), |
1018 | 0 | *p1, |
1019 | 0 | *s1, |
1020 | 0 | *p2, |
1021 | 0 | *s2, |
1022 | 0 | cast_options, |
1023 | | ) |
1024 | | } |
1025 | 0 | (Decimal256(p1, s1), Decimal64(p2, s2)) => { |
1026 | 0 | cast_decimal_to_decimal::<Decimal256Type, Decimal64Type>( |
1027 | 0 | array.as_primitive(), |
1028 | 0 | *p1, |
1029 | 0 | *s1, |
1030 | 0 | *p2, |
1031 | 0 | *s2, |
1032 | 0 | cast_options, |
1033 | | ) |
1034 | | } |
1035 | 0 | (Decimal256(p1, s1), Decimal128(p2, s2)) => { |
1036 | 0 | cast_decimal_to_decimal::<Decimal256Type, Decimal128Type>( |
1037 | 0 | array.as_primitive(), |
1038 | 0 | *p1, |
1039 | 0 | *s1, |
1040 | 0 | *p2, |
1041 | 0 | *s2, |
1042 | 0 | cast_options, |
1043 | | ) |
1044 | | } |
1045 | | // Decimal to non-decimal |
1046 | 0 | (Decimal32(_, scale), _) if !to_type.is_temporal() => { |
1047 | 0 | cast_from_decimal::<Decimal32Type, _>( |
1048 | 0 | array, |
1049 | | 10_i32, |
1050 | 0 | scale, |
1051 | 0 | from_type, |
1052 | 0 | to_type, |
1053 | 0 | |x: i32| x as f64, |
1054 | 0 | cast_options, |
1055 | | ) |
1056 | | } |
1057 | 0 | (Decimal64(_, scale), _) if !to_type.is_temporal() => { |
1058 | 0 | cast_from_decimal::<Decimal64Type, _>( |
1059 | 0 | array, |
1060 | | 10_i64, |
1061 | 0 | scale, |
1062 | 0 | from_type, |
1063 | 0 | to_type, |
1064 | 0 | |x: i64| x as f64, |
1065 | 0 | cast_options, |
1066 | | ) |
1067 | | } |
1068 | 0 | (Decimal128(_, scale), _) if !to_type.is_temporal() => { |
1069 | 0 | cast_from_decimal::<Decimal128Type, _>( |
1070 | 0 | array, |
1071 | | 10_i128, |
1072 | 0 | scale, |
1073 | 0 | from_type, |
1074 | 0 | to_type, |
1075 | 0 | |x: i128| x as f64, |
1076 | 0 | cast_options, |
1077 | | ) |
1078 | | } |
1079 | 0 | (Decimal256(_, scale), _) if !to_type.is_temporal() => { |
1080 | 0 | cast_from_decimal::<Decimal256Type, _>( |
1081 | 0 | array, |
1082 | 0 | i256::from_i128(10_i128), |
1083 | 0 | scale, |
1084 | 0 | from_type, |
1085 | 0 | to_type, |
1086 | 0 | |x: i256| x.to_f64().expect("All i256 values fit in f64"), |
1087 | 0 | cast_options, |
1088 | | ) |
1089 | | } |
1090 | | // Non-decimal to decimal |
1091 | 0 | (_, Decimal32(precision, scale)) if !from_type.is_temporal() => { |
1092 | 0 | cast_to_decimal::<Decimal32Type, _>( |
1093 | 0 | array, |
1094 | | 10_i32, |
1095 | 0 | precision, |
1096 | 0 | scale, |
1097 | 0 | from_type, |
1098 | 0 | to_type, |
1099 | 0 | cast_options, |
1100 | | ) |
1101 | | } |
1102 | 0 | (_, Decimal64(precision, scale)) if !from_type.is_temporal() => { |
1103 | 0 | cast_to_decimal::<Decimal64Type, _>( |
1104 | 0 | array, |
1105 | | 10_i64, |
1106 | 0 | precision, |
1107 | 0 | scale, |
1108 | 0 | from_type, |
1109 | 0 | to_type, |
1110 | 0 | cast_options, |
1111 | | ) |
1112 | | } |
1113 | 0 | (_, Decimal128(precision, scale)) if !from_type.is_temporal() => { |
1114 | 0 | cast_to_decimal::<Decimal128Type, _>( |
1115 | 0 | array, |
1116 | | 10_i128, |
1117 | 0 | precision, |
1118 | 0 | scale, |
1119 | 0 | from_type, |
1120 | 0 | to_type, |
1121 | 0 | cast_options, |
1122 | | ) |
1123 | | } |
1124 | 0 | (_, Decimal256(precision, scale)) if !from_type.is_temporal() => { |
1125 | 0 | cast_to_decimal::<Decimal256Type, _>( |
1126 | 0 | array, |
1127 | 0 | i256::from_i128(10_i128), |
1128 | 0 | precision, |
1129 | 0 | scale, |
1130 | 0 | from_type, |
1131 | 0 | to_type, |
1132 | 0 | cast_options, |
1133 | | ) |
1134 | | } |
1135 | 0 | (Struct(_), Struct(to_fields)) => { |
1136 | 0 | let array = array.as_struct(); |
1137 | 0 | let fields = array |
1138 | 0 | .columns() |
1139 | 0 | .iter() |
1140 | 0 | .zip(to_fields.iter()) |
1141 | 0 | .map(|(l, field)| cast_with_options(l, field.data_type(), cast_options)) |
1142 | 0 | .collect::<Result<Vec<ArrayRef>, ArrowError>>()?; |
1143 | 0 | let array = StructArray::try_new(to_fields.clone(), fields, array.nulls().cloned())?; |
1144 | 0 | Ok(Arc::new(array) as ArrayRef) |
1145 | | } |
1146 | 0 | (Struct(_), _) => Err(ArrowError::CastError(format!( |
1147 | 0 | "Casting from {from_type:?} to {to_type:?} not supported" |
1148 | 0 | ))), |
1149 | 0 | (_, Struct(_)) => Err(ArrowError::CastError(format!( |
1150 | 0 | "Casting from {from_type:?} to {to_type:?} not supported" |
1151 | 0 | ))), |
1152 | 0 | (_, Boolean) => match from_type { |
1153 | 0 | UInt8 => cast_numeric_to_bool::<UInt8Type>(array), |
1154 | 0 | UInt16 => cast_numeric_to_bool::<UInt16Type>(array), |
1155 | 0 | UInt32 => cast_numeric_to_bool::<UInt32Type>(array), |
1156 | 0 | UInt64 => cast_numeric_to_bool::<UInt64Type>(array), |
1157 | 0 | Int8 => cast_numeric_to_bool::<Int8Type>(array), |
1158 | 0 | Int16 => cast_numeric_to_bool::<Int16Type>(array), |
1159 | 0 | Int32 => cast_numeric_to_bool::<Int32Type>(array), |
1160 | 0 | Int64 => cast_numeric_to_bool::<Int64Type>(array), |
1161 | 0 | Float16 => cast_numeric_to_bool::<Float16Type>(array), |
1162 | 0 | Float32 => cast_numeric_to_bool::<Float32Type>(array), |
1163 | 0 | Float64 => cast_numeric_to_bool::<Float64Type>(array), |
1164 | 0 | Utf8View => cast_utf8view_to_boolean(array, cast_options), |
1165 | 0 | Utf8 => cast_utf8_to_boolean::<i32>(array, cast_options), |
1166 | 0 | LargeUtf8 => cast_utf8_to_boolean::<i64>(array, cast_options), |
1167 | 0 | _ => Err(ArrowError::CastError(format!( |
1168 | 0 | "Casting from {from_type:?} to {to_type:?} not supported", |
1169 | 0 | ))), |
1170 | | }, |
1171 | 0 | (Boolean, _) => match to_type { |
1172 | 0 | UInt8 => cast_bool_to_numeric::<UInt8Type>(array, cast_options), |
1173 | 0 | UInt16 => cast_bool_to_numeric::<UInt16Type>(array, cast_options), |
1174 | 0 | UInt32 => cast_bool_to_numeric::<UInt32Type>(array, cast_options), |
1175 | 0 | UInt64 => cast_bool_to_numeric::<UInt64Type>(array, cast_options), |
1176 | 0 | Int8 => cast_bool_to_numeric::<Int8Type>(array, cast_options), |
1177 | 0 | Int16 => cast_bool_to_numeric::<Int16Type>(array, cast_options), |
1178 | 0 | Int32 => cast_bool_to_numeric::<Int32Type>(array, cast_options), |
1179 | 0 | Int64 => cast_bool_to_numeric::<Int64Type>(array, cast_options), |
1180 | 0 | Float16 => cast_bool_to_numeric::<Float16Type>(array, cast_options), |
1181 | 0 | Float32 => cast_bool_to_numeric::<Float32Type>(array, cast_options), |
1182 | 0 | Float64 => cast_bool_to_numeric::<Float64Type>(array, cast_options), |
1183 | 0 | Utf8View => value_to_string_view(array, cast_options), |
1184 | 0 | Utf8 => value_to_string::<i32>(array, cast_options), |
1185 | 0 | LargeUtf8 => value_to_string::<i64>(array, cast_options), |
1186 | 0 | _ => Err(ArrowError::CastError(format!( |
1187 | 0 | "Casting from {from_type:?} to {to_type:?} not supported", |
1188 | 0 | ))), |
1189 | | }, |
1190 | 0 | (Utf8, _) => match to_type { |
1191 | 0 | UInt8 => parse_string::<UInt8Type, i32>(array, cast_options), |
1192 | 0 | UInt16 => parse_string::<UInt16Type, i32>(array, cast_options), |
1193 | 0 | UInt32 => parse_string::<UInt32Type, i32>(array, cast_options), |
1194 | 0 | UInt64 => parse_string::<UInt64Type, i32>(array, cast_options), |
1195 | 0 | Int8 => parse_string::<Int8Type, i32>(array, cast_options), |
1196 | 0 | Int16 => parse_string::<Int16Type, i32>(array, cast_options), |
1197 | 0 | Int32 => parse_string::<Int32Type, i32>(array, cast_options), |
1198 | 0 | Int64 => parse_string::<Int64Type, i32>(array, cast_options), |
1199 | 0 | Float32 => parse_string::<Float32Type, i32>(array, cast_options), |
1200 | 0 | Float64 => parse_string::<Float64Type, i32>(array, cast_options), |
1201 | 0 | Date32 => parse_string::<Date32Type, i32>(array, cast_options), |
1202 | 0 | Date64 => parse_string::<Date64Type, i32>(array, cast_options), |
1203 | 0 | Binary => Ok(Arc::new(BinaryArray::from( |
1204 | 0 | array.as_string::<i32>().clone(), |
1205 | 0 | ))), |
1206 | | LargeBinary => { |
1207 | 0 | let binary = BinaryArray::from(array.as_string::<i32>().clone()); |
1208 | 0 | cast_byte_container::<BinaryType, LargeBinaryType>(&binary) |
1209 | | } |
1210 | 0 | Utf8View => Ok(Arc::new(StringViewArray::from(array.as_string::<i32>()))), |
1211 | 0 | BinaryView => Ok(Arc::new( |
1212 | 0 | StringViewArray::from(array.as_string::<i32>()).to_binary_view(), |
1213 | 0 | )), |
1214 | 0 | LargeUtf8 => cast_byte_container::<Utf8Type, LargeUtf8Type>(array), |
1215 | 0 | Time32(TimeUnit::Second) => parse_string::<Time32SecondType, i32>(array, cast_options), |
1216 | | Time32(TimeUnit::Millisecond) => { |
1217 | 0 | parse_string::<Time32MillisecondType, i32>(array, cast_options) |
1218 | | } |
1219 | | Time64(TimeUnit::Microsecond) => { |
1220 | 0 | parse_string::<Time64MicrosecondType, i32>(array, cast_options) |
1221 | | } |
1222 | | Time64(TimeUnit::Nanosecond) => { |
1223 | 0 | parse_string::<Time64NanosecondType, i32>(array, cast_options) |
1224 | | } |
1225 | 0 | Timestamp(TimeUnit::Second, to_tz) => { |
1226 | 0 | cast_string_to_timestamp::<i32, TimestampSecondType>(array, to_tz, cast_options) |
1227 | | } |
1228 | 0 | Timestamp(TimeUnit::Millisecond, to_tz) => cast_string_to_timestamp::< |
1229 | 0 | i32, |
1230 | 0 | TimestampMillisecondType, |
1231 | 0 | >(array, to_tz, cast_options), |
1232 | 0 | Timestamp(TimeUnit::Microsecond, to_tz) => cast_string_to_timestamp::< |
1233 | 0 | i32, |
1234 | 0 | TimestampMicrosecondType, |
1235 | 0 | >(array, to_tz, cast_options), |
1236 | 0 | Timestamp(TimeUnit::Nanosecond, to_tz) => { |
1237 | 0 | cast_string_to_timestamp::<i32, TimestampNanosecondType>(array, to_tz, cast_options) |
1238 | | } |
1239 | | Interval(IntervalUnit::YearMonth) => { |
1240 | 0 | cast_string_to_year_month_interval::<i32>(array, cast_options) |
1241 | | } |
1242 | | Interval(IntervalUnit::DayTime) => { |
1243 | 0 | cast_string_to_day_time_interval::<i32>(array, cast_options) |
1244 | | } |
1245 | | Interval(IntervalUnit::MonthDayNano) => { |
1246 | 0 | cast_string_to_month_day_nano_interval::<i32>(array, cast_options) |
1247 | | } |
1248 | 0 | _ => Err(ArrowError::CastError(format!( |
1249 | 0 | "Casting from {from_type:?} to {to_type:?} not supported", |
1250 | 0 | ))), |
1251 | | }, |
1252 | 0 | (Utf8View, _) => match to_type { |
1253 | 0 | UInt8 => parse_string_view::<UInt8Type>(array, cast_options), |
1254 | 0 | UInt16 => parse_string_view::<UInt16Type>(array, cast_options), |
1255 | 0 | UInt32 => parse_string_view::<UInt32Type>(array, cast_options), |
1256 | 0 | UInt64 => parse_string_view::<UInt64Type>(array, cast_options), |
1257 | 0 | Int8 => parse_string_view::<Int8Type>(array, cast_options), |
1258 | 0 | Int16 => parse_string_view::<Int16Type>(array, cast_options), |
1259 | 0 | Int32 => parse_string_view::<Int32Type>(array, cast_options), |
1260 | 0 | Int64 => parse_string_view::<Int64Type>(array, cast_options), |
1261 | 0 | Float32 => parse_string_view::<Float32Type>(array, cast_options), |
1262 | 0 | Float64 => parse_string_view::<Float64Type>(array, cast_options), |
1263 | 0 | Date32 => parse_string_view::<Date32Type>(array, cast_options), |
1264 | 0 | Date64 => parse_string_view::<Date64Type>(array, cast_options), |
1265 | 0 | Binary => cast_view_to_byte::<StringViewType, GenericBinaryType<i32>>(array), |
1266 | 0 | LargeBinary => cast_view_to_byte::<StringViewType, GenericBinaryType<i64>>(array), |
1267 | 0 | BinaryView => Ok(Arc::new(array.as_string_view().clone().to_binary_view())), |
1268 | 0 | Utf8 => cast_view_to_byte::<StringViewType, GenericStringType<i32>>(array), |
1269 | 0 | LargeUtf8 => cast_view_to_byte::<StringViewType, GenericStringType<i64>>(array), |
1270 | 0 | Time32(TimeUnit::Second) => parse_string_view::<Time32SecondType>(array, cast_options), |
1271 | | Time32(TimeUnit::Millisecond) => { |
1272 | 0 | parse_string_view::<Time32MillisecondType>(array, cast_options) |
1273 | | } |
1274 | | Time64(TimeUnit::Microsecond) => { |
1275 | 0 | parse_string_view::<Time64MicrosecondType>(array, cast_options) |
1276 | | } |
1277 | | Time64(TimeUnit::Nanosecond) => { |
1278 | 0 | parse_string_view::<Time64NanosecondType>(array, cast_options) |
1279 | | } |
1280 | 0 | Timestamp(TimeUnit::Second, to_tz) => { |
1281 | 0 | cast_view_to_timestamp::<TimestampSecondType>(array, to_tz, cast_options) |
1282 | | } |
1283 | 0 | Timestamp(TimeUnit::Millisecond, to_tz) => { |
1284 | 0 | cast_view_to_timestamp::<TimestampMillisecondType>(array, to_tz, cast_options) |
1285 | | } |
1286 | 0 | Timestamp(TimeUnit::Microsecond, to_tz) => { |
1287 | 0 | cast_view_to_timestamp::<TimestampMicrosecondType>(array, to_tz, cast_options) |
1288 | | } |
1289 | 0 | Timestamp(TimeUnit::Nanosecond, to_tz) => { |
1290 | 0 | cast_view_to_timestamp::<TimestampNanosecondType>(array, to_tz, cast_options) |
1291 | | } |
1292 | | Interval(IntervalUnit::YearMonth) => { |
1293 | 0 | cast_view_to_year_month_interval(array, cast_options) |
1294 | | } |
1295 | 0 | Interval(IntervalUnit::DayTime) => cast_view_to_day_time_interval(array, cast_options), |
1296 | | Interval(IntervalUnit::MonthDayNano) => { |
1297 | 0 | cast_view_to_month_day_nano_interval(array, cast_options) |
1298 | | } |
1299 | 0 | _ => Err(ArrowError::CastError(format!( |
1300 | 0 | "Casting from {from_type:?} to {to_type:?} not supported", |
1301 | 0 | ))), |
1302 | | }, |
1303 | 0 | (LargeUtf8, _) => match to_type { |
1304 | 0 | UInt8 => parse_string::<UInt8Type, i64>(array, cast_options), |
1305 | 0 | UInt16 => parse_string::<UInt16Type, i64>(array, cast_options), |
1306 | 0 | UInt32 => parse_string::<UInt32Type, i64>(array, cast_options), |
1307 | 0 | UInt64 => parse_string::<UInt64Type, i64>(array, cast_options), |
1308 | 0 | Int8 => parse_string::<Int8Type, i64>(array, cast_options), |
1309 | 0 | Int16 => parse_string::<Int16Type, i64>(array, cast_options), |
1310 | 0 | Int32 => parse_string::<Int32Type, i64>(array, cast_options), |
1311 | 0 | Int64 => parse_string::<Int64Type, i64>(array, cast_options), |
1312 | 0 | Float32 => parse_string::<Float32Type, i64>(array, cast_options), |
1313 | 0 | Float64 => parse_string::<Float64Type, i64>(array, cast_options), |
1314 | 0 | Date32 => parse_string::<Date32Type, i64>(array, cast_options), |
1315 | 0 | Date64 => parse_string::<Date64Type, i64>(array, cast_options), |
1316 | 0 | Utf8 => cast_byte_container::<LargeUtf8Type, Utf8Type>(array), |
1317 | | Binary => { |
1318 | 0 | let large_binary = LargeBinaryArray::from(array.as_string::<i64>().clone()); |
1319 | 0 | cast_byte_container::<LargeBinaryType, BinaryType>(&large_binary) |
1320 | | } |
1321 | 0 | LargeBinary => Ok(Arc::new(LargeBinaryArray::from( |
1322 | 0 | array.as_string::<i64>().clone(), |
1323 | 0 | ))), |
1324 | 0 | Utf8View => Ok(Arc::new(StringViewArray::from(array.as_string::<i64>()))), |
1325 | 0 | BinaryView => Ok(Arc::new(BinaryViewArray::from( |
1326 | 0 | array |
1327 | 0 | .as_string::<i64>() |
1328 | 0 | .into_iter() |
1329 | 0 | .map(|x| x.map(|x| x.as_bytes())) |
1330 | 0 | .collect::<Vec<_>>(), |
1331 | | ))), |
1332 | 0 | Time32(TimeUnit::Second) => parse_string::<Time32SecondType, i64>(array, cast_options), |
1333 | | Time32(TimeUnit::Millisecond) => { |
1334 | 0 | parse_string::<Time32MillisecondType, i64>(array, cast_options) |
1335 | | } |
1336 | | Time64(TimeUnit::Microsecond) => { |
1337 | 0 | parse_string::<Time64MicrosecondType, i64>(array, cast_options) |
1338 | | } |
1339 | | Time64(TimeUnit::Nanosecond) => { |
1340 | 0 | parse_string::<Time64NanosecondType, i64>(array, cast_options) |
1341 | | } |
1342 | 0 | Timestamp(TimeUnit::Second, to_tz) => { |
1343 | 0 | cast_string_to_timestamp::<i64, TimestampSecondType>(array, to_tz, cast_options) |
1344 | | } |
1345 | 0 | Timestamp(TimeUnit::Millisecond, to_tz) => cast_string_to_timestamp::< |
1346 | 0 | i64, |
1347 | 0 | TimestampMillisecondType, |
1348 | 0 | >(array, to_tz, cast_options), |
1349 | 0 | Timestamp(TimeUnit::Microsecond, to_tz) => cast_string_to_timestamp::< |
1350 | 0 | i64, |
1351 | 0 | TimestampMicrosecondType, |
1352 | 0 | >(array, to_tz, cast_options), |
1353 | 0 | Timestamp(TimeUnit::Nanosecond, to_tz) => { |
1354 | 0 | cast_string_to_timestamp::<i64, TimestampNanosecondType>(array, to_tz, cast_options) |
1355 | | } |
1356 | | Interval(IntervalUnit::YearMonth) => { |
1357 | 0 | cast_string_to_year_month_interval::<i64>(array, cast_options) |
1358 | | } |
1359 | | Interval(IntervalUnit::DayTime) => { |
1360 | 0 | cast_string_to_day_time_interval::<i64>(array, cast_options) |
1361 | | } |
1362 | | Interval(IntervalUnit::MonthDayNano) => { |
1363 | 0 | cast_string_to_month_day_nano_interval::<i64>(array, cast_options) |
1364 | | } |
1365 | 0 | _ => Err(ArrowError::CastError(format!( |
1366 | 0 | "Casting from {from_type:?} to {to_type:?} not supported", |
1367 | 0 | ))), |
1368 | | }, |
1369 | 0 | (Binary, _) => match to_type { |
1370 | 0 | Utf8 => cast_binary_to_string::<i32>(array, cast_options), |
1371 | | LargeUtf8 => { |
1372 | 0 | let array = cast_binary_to_string::<i32>(array, cast_options)?; |
1373 | 0 | cast_byte_container::<Utf8Type, LargeUtf8Type>(array.as_ref()) |
1374 | | } |
1375 | 0 | LargeBinary => cast_byte_container::<BinaryType, LargeBinaryType>(array), |
1376 | 0 | FixedSizeBinary(size) => { |
1377 | 0 | cast_binary_to_fixed_size_binary::<i32>(array, *size, cast_options) |
1378 | | } |
1379 | 0 | BinaryView => Ok(Arc::new(BinaryViewArray::from(array.as_binary::<i32>()))), |
1380 | 0 | Utf8View => Ok(Arc::new(StringViewArray::from( |
1381 | 0 | cast_binary_to_string::<i32>(array, cast_options)?.as_string::<i32>(), |
1382 | | ))), |
1383 | 0 | _ => Err(ArrowError::CastError(format!( |
1384 | 0 | "Casting from {from_type:?} to {to_type:?} not supported", |
1385 | 0 | ))), |
1386 | | }, |
1387 | 0 | (LargeBinary, _) => match to_type { |
1388 | | Utf8 => { |
1389 | 0 | let array = cast_binary_to_string::<i64>(array, cast_options)?; |
1390 | 0 | cast_byte_container::<LargeUtf8Type, Utf8Type>(array.as_ref()) |
1391 | | } |
1392 | 0 | LargeUtf8 => cast_binary_to_string::<i64>(array, cast_options), |
1393 | 0 | Binary => cast_byte_container::<LargeBinaryType, BinaryType>(array), |
1394 | 0 | FixedSizeBinary(size) => { |
1395 | 0 | cast_binary_to_fixed_size_binary::<i64>(array, *size, cast_options) |
1396 | | } |
1397 | 0 | BinaryView => Ok(Arc::new(BinaryViewArray::from(array.as_binary::<i64>()))), |
1398 | | Utf8View => { |
1399 | 0 | let array = cast_binary_to_string::<i64>(array, cast_options)?; |
1400 | 0 | Ok(Arc::new(StringViewArray::from(array.as_string::<i64>()))) |
1401 | | } |
1402 | 0 | _ => Err(ArrowError::CastError(format!( |
1403 | 0 | "Casting from {from_type:?} to {to_type:?} not supported", |
1404 | 0 | ))), |
1405 | | }, |
1406 | 0 | (FixedSizeBinary(size), _) => match to_type { |
1407 | 0 | Binary => cast_fixed_size_binary_to_binary::<i32>(array, *size), |
1408 | 0 | LargeBinary => cast_fixed_size_binary_to_binary::<i64>(array, *size), |
1409 | 0 | BinaryView => cast_fixed_size_binary_to_binary_view(array, *size), |
1410 | 0 | _ => Err(ArrowError::CastError(format!( |
1411 | 0 | "Casting from {from_type:?} to {to_type:?} not supported", |
1412 | 0 | ))), |
1413 | | }, |
1414 | 0 | (BinaryView, Binary) => cast_view_to_byte::<BinaryViewType, GenericBinaryType<i32>>(array), |
1415 | | (BinaryView, LargeBinary) => { |
1416 | 0 | cast_view_to_byte::<BinaryViewType, GenericBinaryType<i64>>(array) |
1417 | | } |
1418 | | (BinaryView, Utf8) => { |
1419 | 0 | let binary_arr = cast_view_to_byte::<BinaryViewType, GenericBinaryType<i32>>(array)?; |
1420 | 0 | cast_binary_to_string::<i32>(&binary_arr, cast_options) |
1421 | | } |
1422 | | (BinaryView, LargeUtf8) => { |
1423 | 0 | let binary_arr = cast_view_to_byte::<BinaryViewType, GenericBinaryType<i64>>(array)?; |
1424 | 0 | cast_binary_to_string::<i64>(&binary_arr, cast_options) |
1425 | | } |
1426 | | (BinaryView, Utf8View) => { |
1427 | 0 | Ok(Arc::new(array.as_binary_view().clone().to_string_view()?) as ArrayRef) |
1428 | | } |
1429 | 0 | (BinaryView, _) => Err(ArrowError::CastError(format!( |
1430 | 0 | "Casting from {from_type:?} to {to_type:?} not supported", |
1431 | 0 | ))), |
1432 | 0 | (from_type, Utf8View) if from_type.is_primitive() => { |
1433 | 0 | value_to_string_view(array, cast_options) |
1434 | | } |
1435 | 0 | (from_type, LargeUtf8) if from_type.is_primitive() => { |
1436 | 0 | value_to_string::<i64>(array, cast_options) |
1437 | | } |
1438 | 0 | (from_type, Utf8) if from_type.is_primitive() => { |
1439 | 0 | value_to_string::<i32>(array, cast_options) |
1440 | | } |
1441 | 0 | (from_type, Binary) if from_type.is_integer() => match from_type { |
1442 | 0 | UInt8 => cast_numeric_to_binary::<UInt8Type, i32>(array), |
1443 | 0 | UInt16 => cast_numeric_to_binary::<UInt16Type, i32>(array), |
1444 | 0 | UInt32 => cast_numeric_to_binary::<UInt32Type, i32>(array), |
1445 | 0 | UInt64 => cast_numeric_to_binary::<UInt64Type, i32>(array), |
1446 | 0 | Int8 => cast_numeric_to_binary::<Int8Type, i32>(array), |
1447 | 0 | Int16 => cast_numeric_to_binary::<Int16Type, i32>(array), |
1448 | 0 | Int32 => cast_numeric_to_binary::<Int32Type, i32>(array), |
1449 | 0 | Int64 => cast_numeric_to_binary::<Int64Type, i32>(array), |
1450 | 0 | _ => unreachable!(), |
1451 | | }, |
1452 | 0 | (from_type, LargeBinary) if from_type.is_integer() => match from_type { |
1453 | 0 | UInt8 => cast_numeric_to_binary::<UInt8Type, i64>(array), |
1454 | 0 | UInt16 => cast_numeric_to_binary::<UInt16Type, i64>(array), |
1455 | 0 | UInt32 => cast_numeric_to_binary::<UInt32Type, i64>(array), |
1456 | 0 | UInt64 => cast_numeric_to_binary::<UInt64Type, i64>(array), |
1457 | 0 | Int8 => cast_numeric_to_binary::<Int8Type, i64>(array), |
1458 | 0 | Int16 => cast_numeric_to_binary::<Int16Type, i64>(array), |
1459 | 0 | Int32 => cast_numeric_to_binary::<Int32Type, i64>(array), |
1460 | 0 | Int64 => cast_numeric_to_binary::<Int64Type, i64>(array), |
1461 | 0 | _ => unreachable!(), |
1462 | | }, |
1463 | | // start numeric casts |
1464 | 0 | (UInt8, UInt16) => cast_numeric_arrays::<UInt8Type, UInt16Type>(array, cast_options), |
1465 | 0 | (UInt8, UInt32) => cast_numeric_arrays::<UInt8Type, UInt32Type>(array, cast_options), |
1466 | 0 | (UInt8, UInt64) => cast_numeric_arrays::<UInt8Type, UInt64Type>(array, cast_options), |
1467 | 0 | (UInt8, Int8) => cast_numeric_arrays::<UInt8Type, Int8Type>(array, cast_options), |
1468 | 0 | (UInt8, Int16) => cast_numeric_arrays::<UInt8Type, Int16Type>(array, cast_options), |
1469 | 0 | (UInt8, Int32) => cast_numeric_arrays::<UInt8Type, Int32Type>(array, cast_options), |
1470 | 0 | (UInt8, Int64) => cast_numeric_arrays::<UInt8Type, Int64Type>(array, cast_options), |
1471 | 0 | (UInt8, Float16) => cast_numeric_arrays::<UInt8Type, Float16Type>(array, cast_options), |
1472 | 0 | (UInt8, Float32) => cast_numeric_arrays::<UInt8Type, Float32Type>(array, cast_options), |
1473 | 0 | (UInt8, Float64) => cast_numeric_arrays::<UInt8Type, Float64Type>(array, cast_options), |
1474 | | |
1475 | 0 | (UInt16, UInt8) => cast_numeric_arrays::<UInt16Type, UInt8Type>(array, cast_options), |
1476 | 0 | (UInt16, UInt32) => cast_numeric_arrays::<UInt16Type, UInt32Type>(array, cast_options), |
1477 | 0 | (UInt16, UInt64) => cast_numeric_arrays::<UInt16Type, UInt64Type>(array, cast_options), |
1478 | 0 | (UInt16, Int8) => cast_numeric_arrays::<UInt16Type, Int8Type>(array, cast_options), |
1479 | 0 | (UInt16, Int16) => cast_numeric_arrays::<UInt16Type, Int16Type>(array, cast_options), |
1480 | 0 | (UInt16, Int32) => cast_numeric_arrays::<UInt16Type, Int32Type>(array, cast_options), |
1481 | 0 | (UInt16, Int64) => cast_numeric_arrays::<UInt16Type, Int64Type>(array, cast_options), |
1482 | 0 | (UInt16, Float16) => cast_numeric_arrays::<UInt16Type, Float16Type>(array, cast_options), |
1483 | 0 | (UInt16, Float32) => cast_numeric_arrays::<UInt16Type, Float32Type>(array, cast_options), |
1484 | 0 | (UInt16, Float64) => cast_numeric_arrays::<UInt16Type, Float64Type>(array, cast_options), |
1485 | | |
1486 | 0 | (UInt32, UInt8) => cast_numeric_arrays::<UInt32Type, UInt8Type>(array, cast_options), |
1487 | 0 | (UInt32, UInt16) => cast_numeric_arrays::<UInt32Type, UInt16Type>(array, cast_options), |
1488 | 0 | (UInt32, UInt64) => cast_numeric_arrays::<UInt32Type, UInt64Type>(array, cast_options), |
1489 | 0 | (UInt32, Int8) => cast_numeric_arrays::<UInt32Type, Int8Type>(array, cast_options), |
1490 | 0 | (UInt32, Int16) => cast_numeric_arrays::<UInt32Type, Int16Type>(array, cast_options), |
1491 | 0 | (UInt32, Int32) => cast_numeric_arrays::<UInt32Type, Int32Type>(array, cast_options), |
1492 | 0 | (UInt32, Int64) => cast_numeric_arrays::<UInt32Type, Int64Type>(array, cast_options), |
1493 | 0 | (UInt32, Float16) => cast_numeric_arrays::<UInt32Type, Float16Type>(array, cast_options), |
1494 | 0 | (UInt32, Float32) => cast_numeric_arrays::<UInt32Type, Float32Type>(array, cast_options), |
1495 | 0 | (UInt32, Float64) => cast_numeric_arrays::<UInt32Type, Float64Type>(array, cast_options), |
1496 | | |
1497 | 0 | (UInt64, UInt8) => cast_numeric_arrays::<UInt64Type, UInt8Type>(array, cast_options), |
1498 | 0 | (UInt64, UInt16) => cast_numeric_arrays::<UInt64Type, UInt16Type>(array, cast_options), |
1499 | 0 | (UInt64, UInt32) => cast_numeric_arrays::<UInt64Type, UInt32Type>(array, cast_options), |
1500 | 0 | (UInt64, Int8) => cast_numeric_arrays::<UInt64Type, Int8Type>(array, cast_options), |
1501 | 0 | (UInt64, Int16) => cast_numeric_arrays::<UInt64Type, Int16Type>(array, cast_options), |
1502 | 0 | (UInt64, Int32) => cast_numeric_arrays::<UInt64Type, Int32Type>(array, cast_options), |
1503 | 0 | (UInt64, Int64) => cast_numeric_arrays::<UInt64Type, Int64Type>(array, cast_options), |
1504 | 0 | (UInt64, Float16) => cast_numeric_arrays::<UInt64Type, Float16Type>(array, cast_options), |
1505 | 0 | (UInt64, Float32) => cast_numeric_arrays::<UInt64Type, Float32Type>(array, cast_options), |
1506 | 0 | (UInt64, Float64) => cast_numeric_arrays::<UInt64Type, Float64Type>(array, cast_options), |
1507 | | |
1508 | 0 | (Int8, UInt8) => cast_numeric_arrays::<Int8Type, UInt8Type>(array, cast_options), |
1509 | 0 | (Int8, UInt16) => cast_numeric_arrays::<Int8Type, UInt16Type>(array, cast_options), |
1510 | 0 | (Int8, UInt32) => cast_numeric_arrays::<Int8Type, UInt32Type>(array, cast_options), |
1511 | 0 | (Int8, UInt64) => cast_numeric_arrays::<Int8Type, UInt64Type>(array, cast_options), |
1512 | 0 | (Int8, Int16) => cast_numeric_arrays::<Int8Type, Int16Type>(array, cast_options), |
1513 | 0 | (Int8, Int32) => cast_numeric_arrays::<Int8Type, Int32Type>(array, cast_options), |
1514 | 0 | (Int8, Int64) => cast_numeric_arrays::<Int8Type, Int64Type>(array, cast_options), |
1515 | 0 | (Int8, Float16) => cast_numeric_arrays::<Int8Type, Float16Type>(array, cast_options), |
1516 | 0 | (Int8, Float32) => cast_numeric_arrays::<Int8Type, Float32Type>(array, cast_options), |
1517 | 0 | (Int8, Float64) => cast_numeric_arrays::<Int8Type, Float64Type>(array, cast_options), |
1518 | | |
1519 | 0 | (Int16, UInt8) => cast_numeric_arrays::<Int16Type, UInt8Type>(array, cast_options), |
1520 | 0 | (Int16, UInt16) => cast_numeric_arrays::<Int16Type, UInt16Type>(array, cast_options), |
1521 | 0 | (Int16, UInt32) => cast_numeric_arrays::<Int16Type, UInt32Type>(array, cast_options), |
1522 | 0 | (Int16, UInt64) => cast_numeric_arrays::<Int16Type, UInt64Type>(array, cast_options), |
1523 | 0 | (Int16, Int8) => cast_numeric_arrays::<Int16Type, Int8Type>(array, cast_options), |
1524 | 0 | (Int16, Int32) => cast_numeric_arrays::<Int16Type, Int32Type>(array, cast_options), |
1525 | 0 | (Int16, Int64) => cast_numeric_arrays::<Int16Type, Int64Type>(array, cast_options), |
1526 | 0 | (Int16, Float16) => cast_numeric_arrays::<Int16Type, Float16Type>(array, cast_options), |
1527 | 0 | (Int16, Float32) => cast_numeric_arrays::<Int16Type, Float32Type>(array, cast_options), |
1528 | 0 | (Int16, Float64) => cast_numeric_arrays::<Int16Type, Float64Type>(array, cast_options), |
1529 | | |
1530 | 0 | (Int32, UInt8) => cast_numeric_arrays::<Int32Type, UInt8Type>(array, cast_options), |
1531 | 0 | (Int32, UInt16) => cast_numeric_arrays::<Int32Type, UInt16Type>(array, cast_options), |
1532 | 0 | (Int32, UInt32) => cast_numeric_arrays::<Int32Type, UInt32Type>(array, cast_options), |
1533 | 0 | (Int32, UInt64) => cast_numeric_arrays::<Int32Type, UInt64Type>(array, cast_options), |
1534 | 0 | (Int32, Int8) => cast_numeric_arrays::<Int32Type, Int8Type>(array, cast_options), |
1535 | 0 | (Int32, Int16) => cast_numeric_arrays::<Int32Type, Int16Type>(array, cast_options), |
1536 | 0 | (Int32, Int64) => cast_numeric_arrays::<Int32Type, Int64Type>(array, cast_options), |
1537 | 0 | (Int32, Float16) => cast_numeric_arrays::<Int32Type, Float16Type>(array, cast_options), |
1538 | 0 | (Int32, Float32) => cast_numeric_arrays::<Int32Type, Float32Type>(array, cast_options), |
1539 | 0 | (Int32, Float64) => cast_numeric_arrays::<Int32Type, Float64Type>(array, cast_options), |
1540 | | |
1541 | 0 | (Int64, UInt8) => cast_numeric_arrays::<Int64Type, UInt8Type>(array, cast_options), |
1542 | 0 | (Int64, UInt16) => cast_numeric_arrays::<Int64Type, UInt16Type>(array, cast_options), |
1543 | 0 | (Int64, UInt32) => cast_numeric_arrays::<Int64Type, UInt32Type>(array, cast_options), |
1544 | 0 | (Int64, UInt64) => cast_numeric_arrays::<Int64Type, UInt64Type>(array, cast_options), |
1545 | 0 | (Int64, Int8) => cast_numeric_arrays::<Int64Type, Int8Type>(array, cast_options), |
1546 | 0 | (Int64, Int16) => cast_numeric_arrays::<Int64Type, Int16Type>(array, cast_options), |
1547 | 0 | (Int64, Int32) => cast_numeric_arrays::<Int64Type, Int32Type>(array, cast_options), |
1548 | 0 | (Int64, Float16) => cast_numeric_arrays::<Int64Type, Float16Type>(array, cast_options), |
1549 | 0 | (Int64, Float32) => cast_numeric_arrays::<Int64Type, Float32Type>(array, cast_options), |
1550 | 0 | (Int64, Float64) => cast_numeric_arrays::<Int64Type, Float64Type>(array, cast_options), |
1551 | | |
1552 | 0 | (Float16, UInt8) => cast_numeric_arrays::<Float16Type, UInt8Type>(array, cast_options), |
1553 | 0 | (Float16, UInt16) => cast_numeric_arrays::<Float16Type, UInt16Type>(array, cast_options), |
1554 | 0 | (Float16, UInt32) => cast_numeric_arrays::<Float16Type, UInt32Type>(array, cast_options), |
1555 | 0 | (Float16, UInt64) => cast_numeric_arrays::<Float16Type, UInt64Type>(array, cast_options), |
1556 | 0 | (Float16, Int8) => cast_numeric_arrays::<Float16Type, Int8Type>(array, cast_options), |
1557 | 0 | (Float16, Int16) => cast_numeric_arrays::<Float16Type, Int16Type>(array, cast_options), |
1558 | 0 | (Float16, Int32) => cast_numeric_arrays::<Float16Type, Int32Type>(array, cast_options), |
1559 | 0 | (Float16, Int64) => cast_numeric_arrays::<Float16Type, Int64Type>(array, cast_options), |
1560 | 0 | (Float16, Float32) => cast_numeric_arrays::<Float16Type, Float32Type>(array, cast_options), |
1561 | 0 | (Float16, Float64) => cast_numeric_arrays::<Float16Type, Float64Type>(array, cast_options), |
1562 | | |
1563 | 0 | (Float32, UInt8) => cast_numeric_arrays::<Float32Type, UInt8Type>(array, cast_options), |
1564 | 0 | (Float32, UInt16) => cast_numeric_arrays::<Float32Type, UInt16Type>(array, cast_options), |
1565 | 0 | (Float32, UInt32) => cast_numeric_arrays::<Float32Type, UInt32Type>(array, cast_options), |
1566 | 0 | (Float32, UInt64) => cast_numeric_arrays::<Float32Type, UInt64Type>(array, cast_options), |
1567 | 0 | (Float32, Int8) => cast_numeric_arrays::<Float32Type, Int8Type>(array, cast_options), |
1568 | 0 | (Float32, Int16) => cast_numeric_arrays::<Float32Type, Int16Type>(array, cast_options), |
1569 | 0 | (Float32, Int32) => cast_numeric_arrays::<Float32Type, Int32Type>(array, cast_options), |
1570 | 0 | (Float32, Int64) => cast_numeric_arrays::<Float32Type, Int64Type>(array, cast_options), |
1571 | 0 | (Float32, Float16) => cast_numeric_arrays::<Float32Type, Float16Type>(array, cast_options), |
1572 | 0 | (Float32, Float64) => cast_numeric_arrays::<Float32Type, Float64Type>(array, cast_options), |
1573 | | |
1574 | 0 | (Float64, UInt8) => cast_numeric_arrays::<Float64Type, UInt8Type>(array, cast_options), |
1575 | 0 | (Float64, UInt16) => cast_numeric_arrays::<Float64Type, UInt16Type>(array, cast_options), |
1576 | 0 | (Float64, UInt32) => cast_numeric_arrays::<Float64Type, UInt32Type>(array, cast_options), |
1577 | 0 | (Float64, UInt64) => cast_numeric_arrays::<Float64Type, UInt64Type>(array, cast_options), |
1578 | 0 | (Float64, Int8) => cast_numeric_arrays::<Float64Type, Int8Type>(array, cast_options), |
1579 | 0 | (Float64, Int16) => cast_numeric_arrays::<Float64Type, Int16Type>(array, cast_options), |
1580 | 0 | (Float64, Int32) => cast_numeric_arrays::<Float64Type, Int32Type>(array, cast_options), |
1581 | 0 | (Float64, Int64) => cast_numeric_arrays::<Float64Type, Int64Type>(array, cast_options), |
1582 | 0 | (Float64, Float16) => cast_numeric_arrays::<Float64Type, Float16Type>(array, cast_options), |
1583 | 0 | (Float64, Float32) => cast_numeric_arrays::<Float64Type, Float32Type>(array, cast_options), |
1584 | | // end numeric casts |
1585 | | |
1586 | | // temporal casts |
1587 | 0 | (Int32, Date32) => cast_reinterpret_arrays::<Int32Type, Date32Type>(array), |
1588 | 0 | (Int32, Date64) => cast_with_options( |
1589 | 0 | &cast_with_options(array, &Date32, cast_options)?, |
1590 | 0 | &Date64, |
1591 | 0 | cast_options, |
1592 | | ), |
1593 | | (Int32, Time32(TimeUnit::Second)) => { |
1594 | 0 | cast_reinterpret_arrays::<Int32Type, Time32SecondType>(array) |
1595 | | } |
1596 | | (Int32, Time32(TimeUnit::Millisecond)) => { |
1597 | 0 | cast_reinterpret_arrays::<Int32Type, Time32MillisecondType>(array) |
1598 | | } |
1599 | | // No support for microsecond/nanosecond with i32 |
1600 | 0 | (Date32, Int32) => cast_reinterpret_arrays::<Date32Type, Int32Type>(array), |
1601 | 0 | (Date32, Int64) => cast_with_options( |
1602 | 0 | &cast_with_options(array, &Int32, cast_options)?, |
1603 | 0 | &Int64, |
1604 | 0 | cast_options, |
1605 | | ), |
1606 | | (Time32(TimeUnit::Second), Int32) => { |
1607 | 0 | cast_reinterpret_arrays::<Time32SecondType, Int32Type>(array) |
1608 | | } |
1609 | | (Time32(TimeUnit::Millisecond), Int32) => { |
1610 | 0 | cast_reinterpret_arrays::<Time32MillisecondType, Int32Type>(array) |
1611 | | } |
1612 | 0 | (Int64, Date64) => cast_reinterpret_arrays::<Int64Type, Date64Type>(array), |
1613 | 0 | (Int64, Date32) => cast_with_options( |
1614 | 0 | &cast_with_options(array, &Int32, cast_options)?, |
1615 | 0 | &Date32, |
1616 | 0 | cast_options, |
1617 | | ), |
1618 | | // No support for second/milliseconds with i64 |
1619 | | (Int64, Time64(TimeUnit::Microsecond)) => { |
1620 | 0 | cast_reinterpret_arrays::<Int64Type, Time64MicrosecondType>(array) |
1621 | | } |
1622 | | (Int64, Time64(TimeUnit::Nanosecond)) => { |
1623 | 0 | cast_reinterpret_arrays::<Int64Type, Time64NanosecondType>(array) |
1624 | | } |
1625 | | |
1626 | 0 | (Date64, Int64) => cast_reinterpret_arrays::<Date64Type, Int64Type>(array), |
1627 | 0 | (Date64, Int32) => cast_with_options( |
1628 | 0 | &cast_with_options(array, &Int64, cast_options)?, |
1629 | 0 | &Int32, |
1630 | 0 | cast_options, |
1631 | | ), |
1632 | | (Time64(TimeUnit::Microsecond), Int64) => { |
1633 | 0 | cast_reinterpret_arrays::<Time64MicrosecondType, Int64Type>(array) |
1634 | | } |
1635 | | (Time64(TimeUnit::Nanosecond), Int64) => { |
1636 | 0 | cast_reinterpret_arrays::<Time64NanosecondType, Int64Type>(array) |
1637 | | } |
1638 | 0 | (Date32, Date64) => Ok(Arc::new( |
1639 | 0 | array |
1640 | 0 | .as_primitive::<Date32Type>() |
1641 | 0 | .unary::<_, Date64Type>(|x| x as i64 * MILLISECONDS_IN_DAY), |
1642 | | )), |
1643 | 0 | (Date64, Date32) => Ok(Arc::new( |
1644 | 0 | array |
1645 | 0 | .as_primitive::<Date64Type>() |
1646 | 0 | .unary::<_, Date32Type>(|x| (x / MILLISECONDS_IN_DAY) as i32), |
1647 | | )), |
1648 | | |
1649 | 0 | (Time32(TimeUnit::Second), Time32(TimeUnit::Millisecond)) => Ok(Arc::new( |
1650 | 0 | array |
1651 | 0 | .as_primitive::<Time32SecondType>() |
1652 | 0 | .unary::<_, Time32MillisecondType>(|x| x * MILLISECONDS as i32), |
1653 | | )), |
1654 | 0 | (Time32(TimeUnit::Second), Time64(TimeUnit::Microsecond)) => Ok(Arc::new( |
1655 | 0 | array |
1656 | 0 | .as_primitive::<Time32SecondType>() |
1657 | 0 | .unary::<_, Time64MicrosecondType>(|x| x as i64 * MICROSECONDS), |
1658 | | )), |
1659 | 0 | (Time32(TimeUnit::Second), Time64(TimeUnit::Nanosecond)) => Ok(Arc::new( |
1660 | 0 | array |
1661 | 0 | .as_primitive::<Time32SecondType>() |
1662 | 0 | .unary::<_, Time64NanosecondType>(|x| x as i64 * NANOSECONDS), |
1663 | | )), |
1664 | | |
1665 | 0 | (Time32(TimeUnit::Millisecond), Time32(TimeUnit::Second)) => Ok(Arc::new( |
1666 | 0 | array |
1667 | 0 | .as_primitive::<Time32MillisecondType>() |
1668 | 0 | .unary::<_, Time32SecondType>(|x| x / MILLISECONDS as i32), |
1669 | | )), |
1670 | 0 | (Time32(TimeUnit::Millisecond), Time64(TimeUnit::Microsecond)) => Ok(Arc::new( |
1671 | 0 | array |
1672 | 0 | .as_primitive::<Time32MillisecondType>() |
1673 | 0 | .unary::<_, Time64MicrosecondType>(|x| x as i64 * (MICROSECONDS / MILLISECONDS)), |
1674 | | )), |
1675 | 0 | (Time32(TimeUnit::Millisecond), Time64(TimeUnit::Nanosecond)) => Ok(Arc::new( |
1676 | 0 | array |
1677 | 0 | .as_primitive::<Time32MillisecondType>() |
1678 | 0 | .unary::<_, Time64NanosecondType>(|x| x as i64 * (MICROSECONDS / NANOSECONDS)), |
1679 | | )), |
1680 | | |
1681 | 0 | (Time64(TimeUnit::Microsecond), Time32(TimeUnit::Second)) => Ok(Arc::new( |
1682 | 0 | array |
1683 | 0 | .as_primitive::<Time64MicrosecondType>() |
1684 | 0 | .unary::<_, Time32SecondType>(|x| (x / MICROSECONDS) as i32), |
1685 | | )), |
1686 | 0 | (Time64(TimeUnit::Microsecond), Time32(TimeUnit::Millisecond)) => Ok(Arc::new( |
1687 | 0 | array |
1688 | 0 | .as_primitive::<Time64MicrosecondType>() |
1689 | 0 | .unary::<_, Time32MillisecondType>(|x| (x / (MICROSECONDS / MILLISECONDS)) as i32), |
1690 | | )), |
1691 | 0 | (Time64(TimeUnit::Microsecond), Time64(TimeUnit::Nanosecond)) => Ok(Arc::new( |
1692 | 0 | array |
1693 | 0 | .as_primitive::<Time64MicrosecondType>() |
1694 | 0 | .unary::<_, Time64NanosecondType>(|x| x * (NANOSECONDS / MICROSECONDS)), |
1695 | | )), |
1696 | | |
1697 | 0 | (Time64(TimeUnit::Nanosecond), Time32(TimeUnit::Second)) => Ok(Arc::new( |
1698 | 0 | array |
1699 | 0 | .as_primitive::<Time64NanosecondType>() |
1700 | 0 | .unary::<_, Time32SecondType>(|x| (x / NANOSECONDS) as i32), |
1701 | | )), |
1702 | 0 | (Time64(TimeUnit::Nanosecond), Time32(TimeUnit::Millisecond)) => Ok(Arc::new( |
1703 | 0 | array |
1704 | 0 | .as_primitive::<Time64NanosecondType>() |
1705 | 0 | .unary::<_, Time32MillisecondType>(|x| (x / (NANOSECONDS / MILLISECONDS)) as i32), |
1706 | | )), |
1707 | 0 | (Time64(TimeUnit::Nanosecond), Time64(TimeUnit::Microsecond)) => Ok(Arc::new( |
1708 | 0 | array |
1709 | 0 | .as_primitive::<Time64NanosecondType>() |
1710 | 0 | .unary::<_, Time64MicrosecondType>(|x| x / (NANOSECONDS / MICROSECONDS)), |
1711 | | )), |
1712 | | |
1713 | | // Timestamp to integer/floating/decimals |
1714 | 0 | (Timestamp(TimeUnit::Second, _), _) if to_type.is_numeric() => { |
1715 | 0 | let array = cast_reinterpret_arrays::<TimestampSecondType, Int64Type>(array)?; |
1716 | 0 | cast_with_options(&array, to_type, cast_options) |
1717 | | } |
1718 | 0 | (Timestamp(TimeUnit::Millisecond, _), _) if to_type.is_numeric() => { |
1719 | 0 | let array = cast_reinterpret_arrays::<TimestampMillisecondType, Int64Type>(array)?; |
1720 | 0 | cast_with_options(&array, to_type, cast_options) |
1721 | | } |
1722 | 0 | (Timestamp(TimeUnit::Microsecond, _), _) if to_type.is_numeric() => { |
1723 | 0 | let array = cast_reinterpret_arrays::<TimestampMicrosecondType, Int64Type>(array)?; |
1724 | 0 | cast_with_options(&array, to_type, cast_options) |
1725 | | } |
1726 | 0 | (Timestamp(TimeUnit::Nanosecond, _), _) if to_type.is_numeric() => { |
1727 | 0 | let array = cast_reinterpret_arrays::<TimestampNanosecondType, Int64Type>(array)?; |
1728 | 0 | cast_with_options(&array, to_type, cast_options) |
1729 | | } |
1730 | | |
1731 | 0 | (_, Timestamp(unit, tz)) if from_type.is_numeric() => { |
1732 | 0 | let array = cast_with_options(array, &Int64, cast_options)?; |
1733 | 0 | Ok(make_timestamp_array( |
1734 | 0 | array.as_primitive(), |
1735 | 0 | *unit, |
1736 | 0 | tz.clone(), |
1737 | 0 | )) |
1738 | | } |
1739 | | |
1740 | 0 | (Timestamp(from_unit, from_tz), Timestamp(to_unit, to_tz)) => { |
1741 | 0 | let array = cast_with_options(array, &Int64, cast_options)?; |
1742 | 0 | let time_array = array.as_primitive::<Int64Type>(); |
1743 | 0 | let from_size = time_unit_multiple(from_unit); |
1744 | 0 | let to_size = time_unit_multiple(to_unit); |
1745 | | // we either divide or multiply, depending on size of each unit |
1746 | | // units are never the same when the types are the same |
1747 | 0 | let converted = match from_size.cmp(&to_size) { |
1748 | | Ordering::Greater => { |
1749 | 0 | let divisor = from_size / to_size; |
1750 | 0 | time_array.unary::<_, Int64Type>(|o| o / divisor) |
1751 | | } |
1752 | 0 | Ordering::Equal => time_array.clone(), |
1753 | | Ordering::Less => { |
1754 | 0 | let mul = to_size / from_size; |
1755 | 0 | if cast_options.safe { |
1756 | 0 | time_array.unary_opt::<_, Int64Type>(|o| o.checked_mul(mul)) |
1757 | | } else { |
1758 | 0 | time_array.try_unary::<_, Int64Type, _>(|o| o.mul_checked(mul))? |
1759 | | } |
1760 | | } |
1761 | | }; |
1762 | | // Normalize timezone |
1763 | 0 | let adjusted = match (from_tz, to_tz) { |
1764 | | // Only this case needs to be adjusted because we're casting from |
1765 | | // unknown time offset to some time offset, we want the time to be |
1766 | | // unchanged. |
1767 | | // |
1768 | | // i.e. Timestamp('2001-01-01T00:00', None) -> Timestamp('2001-01-01T00:00', '+0700') |
1769 | 0 | (None, Some(to_tz)) => { |
1770 | 0 | let to_tz: Tz = to_tz.parse()?; |
1771 | 0 | match to_unit { |
1772 | 0 | TimeUnit::Second => adjust_timestamp_to_timezone::<TimestampSecondType>( |
1773 | 0 | converted, |
1774 | 0 | &to_tz, |
1775 | 0 | cast_options, |
1776 | 0 | )?, |
1777 | 0 | TimeUnit::Millisecond => adjust_timestamp_to_timezone::< |
1778 | 0 | TimestampMillisecondType, |
1779 | 0 | >( |
1780 | 0 | converted, &to_tz, cast_options |
1781 | 0 | )?, |
1782 | 0 | TimeUnit::Microsecond => adjust_timestamp_to_timezone::< |
1783 | 0 | TimestampMicrosecondType, |
1784 | 0 | >( |
1785 | 0 | converted, &to_tz, cast_options |
1786 | 0 | )?, |
1787 | 0 | TimeUnit::Nanosecond => adjust_timestamp_to_timezone::< |
1788 | 0 | TimestampNanosecondType, |
1789 | 0 | >( |
1790 | 0 | converted, &to_tz, cast_options |
1791 | 0 | )?, |
1792 | | } |
1793 | | } |
1794 | 0 | _ => converted, |
1795 | | }; |
1796 | 0 | Ok(make_timestamp_array(&adjusted, *to_unit, to_tz.clone())) |
1797 | | } |
1798 | | (Timestamp(TimeUnit::Microsecond, _), Date32) => { |
1799 | 0 | timestamp_to_date32(array.as_primitive::<TimestampMicrosecondType>()) |
1800 | | } |
1801 | | (Timestamp(TimeUnit::Millisecond, _), Date32) => { |
1802 | 0 | timestamp_to_date32(array.as_primitive::<TimestampMillisecondType>()) |
1803 | | } |
1804 | | (Timestamp(TimeUnit::Second, _), Date32) => { |
1805 | 0 | timestamp_to_date32(array.as_primitive::<TimestampSecondType>()) |
1806 | | } |
1807 | | (Timestamp(TimeUnit::Nanosecond, _), Date32) => { |
1808 | 0 | timestamp_to_date32(array.as_primitive::<TimestampNanosecondType>()) |
1809 | | } |
1810 | 0 | (Timestamp(TimeUnit::Second, _), Date64) => Ok(Arc::new(match cast_options.safe { |
1811 | | true => { |
1812 | | // change error to None |
1813 | 0 | array |
1814 | 0 | .as_primitive::<TimestampSecondType>() |
1815 | 0 | .unary_opt::<_, Date64Type>(|x| x.checked_mul(MILLISECONDS)) |
1816 | | } |
1817 | 0 | false => array |
1818 | 0 | .as_primitive::<TimestampSecondType>() |
1819 | 0 | .try_unary::<_, Date64Type, _>(|x| x.mul_checked(MILLISECONDS))?, |
1820 | | })), |
1821 | | (Timestamp(TimeUnit::Millisecond, _), Date64) => { |
1822 | 0 | cast_reinterpret_arrays::<TimestampMillisecondType, Date64Type>(array) |
1823 | | } |
1824 | 0 | (Timestamp(TimeUnit::Microsecond, _), Date64) => Ok(Arc::new( |
1825 | 0 | array |
1826 | 0 | .as_primitive::<TimestampMicrosecondType>() |
1827 | 0 | .unary::<_, Date64Type>(|x| x / (MICROSECONDS / MILLISECONDS)), |
1828 | | )), |
1829 | 0 | (Timestamp(TimeUnit::Nanosecond, _), Date64) => Ok(Arc::new( |
1830 | 0 | array |
1831 | 0 | .as_primitive::<TimestampNanosecondType>() |
1832 | 0 | .unary::<_, Date64Type>(|x| x / (NANOSECONDS / MILLISECONDS)), |
1833 | | )), |
1834 | 0 | (Timestamp(TimeUnit::Second, tz), Time64(TimeUnit::Microsecond)) => { |
1835 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1836 | 0 | Ok(Arc::new( |
1837 | 0 | array |
1838 | 0 | .as_primitive::<TimestampSecondType>() |
1839 | 0 | .try_unary::<_, Time64MicrosecondType, ArrowError>(|x| { |
1840 | 0 | Ok(time_to_time64us(as_time_res_with_timezone::< |
1841 | 0 | TimestampSecondType, |
1842 | 0 | >(x, tz)?)) |
1843 | 0 | })?, |
1844 | | )) |
1845 | | } |
1846 | 0 | (Timestamp(TimeUnit::Second, tz), Time64(TimeUnit::Nanosecond)) => { |
1847 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1848 | 0 | Ok(Arc::new( |
1849 | 0 | array |
1850 | 0 | .as_primitive::<TimestampSecondType>() |
1851 | 0 | .try_unary::<_, Time64NanosecondType, ArrowError>(|x| { |
1852 | 0 | Ok(time_to_time64ns(as_time_res_with_timezone::< |
1853 | 0 | TimestampSecondType, |
1854 | 0 | >(x, tz)?)) |
1855 | 0 | })?, |
1856 | | )) |
1857 | | } |
1858 | 0 | (Timestamp(TimeUnit::Millisecond, tz), Time64(TimeUnit::Microsecond)) => { |
1859 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1860 | 0 | Ok(Arc::new( |
1861 | 0 | array |
1862 | 0 | .as_primitive::<TimestampMillisecondType>() |
1863 | 0 | .try_unary::<_, Time64MicrosecondType, ArrowError>(|x| { |
1864 | 0 | Ok(time_to_time64us(as_time_res_with_timezone::< |
1865 | 0 | TimestampMillisecondType, |
1866 | 0 | >(x, tz)?)) |
1867 | 0 | })?, |
1868 | | )) |
1869 | | } |
1870 | 0 | (Timestamp(TimeUnit::Millisecond, tz), Time64(TimeUnit::Nanosecond)) => { |
1871 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1872 | 0 | Ok(Arc::new( |
1873 | 0 | array |
1874 | 0 | .as_primitive::<TimestampMillisecondType>() |
1875 | 0 | .try_unary::<_, Time64NanosecondType, ArrowError>(|x| { |
1876 | 0 | Ok(time_to_time64ns(as_time_res_with_timezone::< |
1877 | 0 | TimestampMillisecondType, |
1878 | 0 | >(x, tz)?)) |
1879 | 0 | })?, |
1880 | | )) |
1881 | | } |
1882 | 0 | (Timestamp(TimeUnit::Microsecond, tz), Time64(TimeUnit::Microsecond)) => { |
1883 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1884 | 0 | Ok(Arc::new( |
1885 | 0 | array |
1886 | 0 | .as_primitive::<TimestampMicrosecondType>() |
1887 | 0 | .try_unary::<_, Time64MicrosecondType, ArrowError>(|x| { |
1888 | 0 | Ok(time_to_time64us(as_time_res_with_timezone::< |
1889 | 0 | TimestampMicrosecondType, |
1890 | 0 | >(x, tz)?)) |
1891 | 0 | })?, |
1892 | | )) |
1893 | | } |
1894 | 0 | (Timestamp(TimeUnit::Microsecond, tz), Time64(TimeUnit::Nanosecond)) => { |
1895 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1896 | 0 | Ok(Arc::new( |
1897 | 0 | array |
1898 | 0 | .as_primitive::<TimestampMicrosecondType>() |
1899 | 0 | .try_unary::<_, Time64NanosecondType, ArrowError>(|x| { |
1900 | 0 | Ok(time_to_time64ns(as_time_res_with_timezone::< |
1901 | 0 | TimestampMicrosecondType, |
1902 | 0 | >(x, tz)?)) |
1903 | 0 | })?, |
1904 | | )) |
1905 | | } |
1906 | 0 | (Timestamp(TimeUnit::Nanosecond, tz), Time64(TimeUnit::Microsecond)) => { |
1907 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1908 | 0 | Ok(Arc::new( |
1909 | 0 | array |
1910 | 0 | .as_primitive::<TimestampNanosecondType>() |
1911 | 0 | .try_unary::<_, Time64MicrosecondType, ArrowError>(|x| { |
1912 | 0 | Ok(time_to_time64us(as_time_res_with_timezone::< |
1913 | 0 | TimestampNanosecondType, |
1914 | 0 | >(x, tz)?)) |
1915 | 0 | })?, |
1916 | | )) |
1917 | | } |
1918 | 0 | (Timestamp(TimeUnit::Nanosecond, tz), Time64(TimeUnit::Nanosecond)) => { |
1919 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1920 | 0 | Ok(Arc::new( |
1921 | 0 | array |
1922 | 0 | .as_primitive::<TimestampNanosecondType>() |
1923 | 0 | .try_unary::<_, Time64NanosecondType, ArrowError>(|x| { |
1924 | 0 | Ok(time_to_time64ns(as_time_res_with_timezone::< |
1925 | 0 | TimestampNanosecondType, |
1926 | 0 | >(x, tz)?)) |
1927 | 0 | })?, |
1928 | | )) |
1929 | | } |
1930 | 0 | (Timestamp(TimeUnit::Second, tz), Time32(TimeUnit::Second)) => { |
1931 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1932 | 0 | Ok(Arc::new( |
1933 | 0 | array |
1934 | 0 | .as_primitive::<TimestampSecondType>() |
1935 | 0 | .try_unary::<_, Time32SecondType, ArrowError>(|x| { |
1936 | 0 | Ok(time_to_time32s(as_time_res_with_timezone::< |
1937 | 0 | TimestampSecondType, |
1938 | 0 | >(x, tz)?)) |
1939 | 0 | })?, |
1940 | | )) |
1941 | | } |
1942 | 0 | (Timestamp(TimeUnit::Second, tz), Time32(TimeUnit::Millisecond)) => { |
1943 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1944 | 0 | Ok(Arc::new( |
1945 | 0 | array |
1946 | 0 | .as_primitive::<TimestampSecondType>() |
1947 | 0 | .try_unary::<_, Time32MillisecondType, ArrowError>(|x| { |
1948 | 0 | Ok(time_to_time32ms(as_time_res_with_timezone::< |
1949 | 0 | TimestampSecondType, |
1950 | 0 | >(x, tz)?)) |
1951 | 0 | })?, |
1952 | | )) |
1953 | | } |
1954 | 0 | (Timestamp(TimeUnit::Millisecond, tz), Time32(TimeUnit::Second)) => { |
1955 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1956 | 0 | Ok(Arc::new( |
1957 | 0 | array |
1958 | 0 | .as_primitive::<TimestampMillisecondType>() |
1959 | 0 | .try_unary::<_, Time32SecondType, ArrowError>(|x| { |
1960 | 0 | Ok(time_to_time32s(as_time_res_with_timezone::< |
1961 | 0 | TimestampMillisecondType, |
1962 | 0 | >(x, tz)?)) |
1963 | 0 | })?, |
1964 | | )) |
1965 | | } |
1966 | 0 | (Timestamp(TimeUnit::Millisecond, tz), Time32(TimeUnit::Millisecond)) => { |
1967 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1968 | 0 | Ok(Arc::new( |
1969 | 0 | array |
1970 | 0 | .as_primitive::<TimestampMillisecondType>() |
1971 | 0 | .try_unary::<_, Time32MillisecondType, ArrowError>(|x| { |
1972 | 0 | Ok(time_to_time32ms(as_time_res_with_timezone::< |
1973 | 0 | TimestampMillisecondType, |
1974 | 0 | >(x, tz)?)) |
1975 | 0 | })?, |
1976 | | )) |
1977 | | } |
1978 | 0 | (Timestamp(TimeUnit::Microsecond, tz), Time32(TimeUnit::Second)) => { |
1979 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1980 | 0 | Ok(Arc::new( |
1981 | 0 | array |
1982 | 0 | .as_primitive::<TimestampMicrosecondType>() |
1983 | 0 | .try_unary::<_, Time32SecondType, ArrowError>(|x| { |
1984 | 0 | Ok(time_to_time32s(as_time_res_with_timezone::< |
1985 | 0 | TimestampMicrosecondType, |
1986 | 0 | >(x, tz)?)) |
1987 | 0 | })?, |
1988 | | )) |
1989 | | } |
1990 | 0 | (Timestamp(TimeUnit::Microsecond, tz), Time32(TimeUnit::Millisecond)) => { |
1991 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
1992 | 0 | Ok(Arc::new( |
1993 | 0 | array |
1994 | 0 | .as_primitive::<TimestampMicrosecondType>() |
1995 | 0 | .try_unary::<_, Time32MillisecondType, ArrowError>(|x| { |
1996 | 0 | Ok(time_to_time32ms(as_time_res_with_timezone::< |
1997 | 0 | TimestampMicrosecondType, |
1998 | 0 | >(x, tz)?)) |
1999 | 0 | })?, |
2000 | | )) |
2001 | | } |
2002 | 0 | (Timestamp(TimeUnit::Nanosecond, tz), Time32(TimeUnit::Second)) => { |
2003 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
2004 | 0 | Ok(Arc::new( |
2005 | 0 | array |
2006 | 0 | .as_primitive::<TimestampNanosecondType>() |
2007 | 0 | .try_unary::<_, Time32SecondType, ArrowError>(|x| { |
2008 | 0 | Ok(time_to_time32s(as_time_res_with_timezone::< |
2009 | 0 | TimestampNanosecondType, |
2010 | 0 | >(x, tz)?)) |
2011 | 0 | })?, |
2012 | | )) |
2013 | | } |
2014 | 0 | (Timestamp(TimeUnit::Nanosecond, tz), Time32(TimeUnit::Millisecond)) => { |
2015 | 0 | let tz = tz.as_ref().map(|tz| tz.parse()).transpose()?; |
2016 | 0 | Ok(Arc::new( |
2017 | 0 | array |
2018 | 0 | .as_primitive::<TimestampNanosecondType>() |
2019 | 0 | .try_unary::<_, Time32MillisecondType, ArrowError>(|x| { |
2020 | 0 | Ok(time_to_time32ms(as_time_res_with_timezone::< |
2021 | 0 | TimestampNanosecondType, |
2022 | 0 | >(x, tz)?)) |
2023 | 0 | })?, |
2024 | | )) |
2025 | | } |
2026 | | (Date64, Timestamp(TimeUnit::Second, _)) => { |
2027 | 0 | let array = array |
2028 | 0 | .as_primitive::<Date64Type>() |
2029 | 0 | .unary::<_, TimestampSecondType>(|x| x / MILLISECONDS); |
2030 | | |
2031 | 0 | cast_with_options(&array, to_type, cast_options) |
2032 | | } |
2033 | | (Date64, Timestamp(TimeUnit::Millisecond, _)) => { |
2034 | 0 | let array = array |
2035 | 0 | .as_primitive::<Date64Type>() |
2036 | 0 | .reinterpret_cast::<TimestampMillisecondType>(); |
2037 | | |
2038 | 0 | cast_with_options(&array, to_type, cast_options) |
2039 | | } |
2040 | | |
2041 | | (Date64, Timestamp(TimeUnit::Microsecond, _)) => { |
2042 | 0 | let array = array |
2043 | 0 | .as_primitive::<Date64Type>() |
2044 | 0 | .unary::<_, TimestampMicrosecondType>(|x| x * (MICROSECONDS / MILLISECONDS)); |
2045 | | |
2046 | 0 | cast_with_options(&array, to_type, cast_options) |
2047 | | } |
2048 | | (Date64, Timestamp(TimeUnit::Nanosecond, _)) => { |
2049 | 0 | let array = array |
2050 | 0 | .as_primitive::<Date64Type>() |
2051 | 0 | .unary::<_, TimestampNanosecondType>(|x| x * (NANOSECONDS / MILLISECONDS)); |
2052 | | |
2053 | 0 | cast_with_options(&array, to_type, cast_options) |
2054 | | } |
2055 | | (Date32, Timestamp(TimeUnit::Second, _)) => { |
2056 | 0 | let array = array |
2057 | 0 | .as_primitive::<Date32Type>() |
2058 | 0 | .unary::<_, TimestampSecondType>(|x| (x as i64) * SECONDS_IN_DAY); |
2059 | | |
2060 | 0 | cast_with_options(&array, to_type, cast_options) |
2061 | | } |
2062 | | (Date32, Timestamp(TimeUnit::Millisecond, _)) => { |
2063 | 0 | let array = array |
2064 | 0 | .as_primitive::<Date32Type>() |
2065 | 0 | .unary::<_, TimestampMillisecondType>(|x| (x as i64) * MILLISECONDS_IN_DAY); |
2066 | | |
2067 | 0 | cast_with_options(&array, to_type, cast_options) |
2068 | | } |
2069 | | (Date32, Timestamp(TimeUnit::Microsecond, _)) => { |
2070 | 0 | let array = array |
2071 | 0 | .as_primitive::<Date32Type>() |
2072 | 0 | .unary::<_, TimestampMicrosecondType>(|x| (x as i64) * MICROSECONDS_IN_DAY); |
2073 | | |
2074 | 0 | cast_with_options(&array, to_type, cast_options) |
2075 | | } |
2076 | | (Date32, Timestamp(TimeUnit::Nanosecond, _)) => { |
2077 | 0 | let array = array |
2078 | 0 | .as_primitive::<Date32Type>() |
2079 | 0 | .unary::<_, TimestampNanosecondType>(|x| (x as i64) * NANOSECONDS_IN_DAY); |
2080 | | |
2081 | 0 | cast_with_options(&array, to_type, cast_options) |
2082 | | } |
2083 | | |
2084 | 0 | (_, Duration(unit)) if from_type.is_numeric() => { |
2085 | 0 | let array = cast_with_options(array, &Int64, cast_options)?; |
2086 | 0 | Ok(make_duration_array(array.as_primitive(), *unit)) |
2087 | | } |
2088 | 0 | (Duration(TimeUnit::Second), _) if to_type.is_numeric() => { |
2089 | 0 | let array = cast_reinterpret_arrays::<DurationSecondType, Int64Type>(array)?; |
2090 | 0 | cast_with_options(&array, to_type, cast_options) |
2091 | | } |
2092 | 0 | (Duration(TimeUnit::Millisecond), _) if to_type.is_numeric() => { |
2093 | 0 | let array = cast_reinterpret_arrays::<DurationMillisecondType, Int64Type>(array)?; |
2094 | 0 | cast_with_options(&array, to_type, cast_options) |
2095 | | } |
2096 | 0 | (Duration(TimeUnit::Microsecond), _) if to_type.is_numeric() => { |
2097 | 0 | let array = cast_reinterpret_arrays::<DurationMicrosecondType, Int64Type>(array)?; |
2098 | 0 | cast_with_options(&array, to_type, cast_options) |
2099 | | } |
2100 | 0 | (Duration(TimeUnit::Nanosecond), _) if to_type.is_numeric() => { |
2101 | 0 | let array = cast_reinterpret_arrays::<DurationNanosecondType, Int64Type>(array)?; |
2102 | 0 | cast_with_options(&array, to_type, cast_options) |
2103 | | } |
2104 | | |
2105 | 0 | (Duration(from_unit), Duration(to_unit)) => { |
2106 | 0 | let array = cast_with_options(array, &Int64, cast_options)?; |
2107 | 0 | let time_array = array.as_primitive::<Int64Type>(); |
2108 | 0 | let from_size = time_unit_multiple(from_unit); |
2109 | 0 | let to_size = time_unit_multiple(to_unit); |
2110 | | // we either divide or multiply, depending on size of each unit |
2111 | | // units are never the same when the types are the same |
2112 | 0 | let converted = match from_size.cmp(&to_size) { |
2113 | | Ordering::Greater => { |
2114 | 0 | let divisor = from_size / to_size; |
2115 | 0 | time_array.unary::<_, Int64Type>(|o| o / divisor) |
2116 | | } |
2117 | 0 | Ordering::Equal => time_array.clone(), |
2118 | | Ordering::Less => { |
2119 | 0 | let mul = to_size / from_size; |
2120 | 0 | if cast_options.safe { |
2121 | 0 | time_array.unary_opt::<_, Int64Type>(|o| o.checked_mul(mul)) |
2122 | | } else { |
2123 | 0 | time_array.try_unary::<_, Int64Type, _>(|o| o.mul_checked(mul))? |
2124 | | } |
2125 | | } |
2126 | | }; |
2127 | 0 | Ok(make_duration_array(&converted, *to_unit)) |
2128 | | } |
2129 | | |
2130 | | (Duration(TimeUnit::Second), Interval(IntervalUnit::MonthDayNano)) => { |
2131 | 0 | cast_duration_to_interval::<DurationSecondType>(array, cast_options) |
2132 | | } |
2133 | | (Duration(TimeUnit::Millisecond), Interval(IntervalUnit::MonthDayNano)) => { |
2134 | 0 | cast_duration_to_interval::<DurationMillisecondType>(array, cast_options) |
2135 | | } |
2136 | | (Duration(TimeUnit::Microsecond), Interval(IntervalUnit::MonthDayNano)) => { |
2137 | 0 | cast_duration_to_interval::<DurationMicrosecondType>(array, cast_options) |
2138 | | } |
2139 | | (Duration(TimeUnit::Nanosecond), Interval(IntervalUnit::MonthDayNano)) => { |
2140 | 0 | cast_duration_to_interval::<DurationNanosecondType>(array, cast_options) |
2141 | | } |
2142 | | (Interval(IntervalUnit::MonthDayNano), Duration(TimeUnit::Second)) => { |
2143 | 0 | cast_month_day_nano_to_duration::<DurationSecondType>(array, cast_options) |
2144 | | } |
2145 | | (Interval(IntervalUnit::MonthDayNano), Duration(TimeUnit::Millisecond)) => { |
2146 | 0 | cast_month_day_nano_to_duration::<DurationMillisecondType>(array, cast_options) |
2147 | | } |
2148 | | (Interval(IntervalUnit::MonthDayNano), Duration(TimeUnit::Microsecond)) => { |
2149 | 0 | cast_month_day_nano_to_duration::<DurationMicrosecondType>(array, cast_options) |
2150 | | } |
2151 | | (Interval(IntervalUnit::MonthDayNano), Duration(TimeUnit::Nanosecond)) => { |
2152 | 0 | cast_month_day_nano_to_duration::<DurationNanosecondType>(array, cast_options) |
2153 | | } |
2154 | | (Interval(IntervalUnit::YearMonth), Interval(IntervalUnit::MonthDayNano)) => { |
2155 | 0 | cast_interval_year_month_to_interval_month_day_nano(array, cast_options) |
2156 | | } |
2157 | | (Interval(IntervalUnit::DayTime), Interval(IntervalUnit::MonthDayNano)) => { |
2158 | 0 | cast_interval_day_time_to_interval_month_day_nano(array, cast_options) |
2159 | | } |
2160 | | (Int32, Interval(IntervalUnit::YearMonth)) => { |
2161 | 0 | cast_reinterpret_arrays::<Int32Type, IntervalYearMonthType>(array) |
2162 | | } |
2163 | 0 | (_, _) => Err(ArrowError::CastError(format!( |
2164 | 0 | "Casting from {from_type:?} to {to_type:?} not supported", |
2165 | 0 | ))), |
2166 | | } |
2167 | 0 | } |
2168 | | |
2169 | 0 | fn cast_from_decimal<D, F>( |
2170 | 0 | array: &dyn Array, |
2171 | 0 | base: D::Native, |
2172 | 0 | scale: &i8, |
2173 | 0 | from_type: &DataType, |
2174 | 0 | to_type: &DataType, |
2175 | 0 | as_float: F, |
2176 | 0 | cast_options: &CastOptions, |
2177 | 0 | ) -> Result<ArrayRef, ArrowError> |
2178 | 0 | where |
2179 | 0 | D: DecimalType + ArrowPrimitiveType, |
2180 | 0 | <D as ArrowPrimitiveType>::Native: ArrowNativeTypeOp + ToPrimitive, |
2181 | 0 | F: Fn(D::Native) -> f64, |
2182 | | { |
2183 | | use DataType::*; |
2184 | | // cast decimal to other type |
2185 | 0 | match to_type { |
2186 | 0 | UInt8 => cast_decimal_to_integer::<D, UInt8Type>(array, base, *scale, cast_options), |
2187 | 0 | UInt16 => cast_decimal_to_integer::<D, UInt16Type>(array, base, *scale, cast_options), |
2188 | 0 | UInt32 => cast_decimal_to_integer::<D, UInt32Type>(array, base, *scale, cast_options), |
2189 | 0 | UInt64 => cast_decimal_to_integer::<D, UInt64Type>(array, base, *scale, cast_options), |
2190 | 0 | Int8 => cast_decimal_to_integer::<D, Int8Type>(array, base, *scale, cast_options), |
2191 | 0 | Int16 => cast_decimal_to_integer::<D, Int16Type>(array, base, *scale, cast_options), |
2192 | 0 | Int32 => cast_decimal_to_integer::<D, Int32Type>(array, base, *scale, cast_options), |
2193 | 0 | Int64 => cast_decimal_to_integer::<D, Int64Type>(array, base, *scale, cast_options), |
2194 | 0 | Float32 => cast_decimal_to_float::<D, Float32Type, _>(array, |x| { |
2195 | 0 | (as_float(x) / 10_f64.powi(*scale as i32)) as f32 |
2196 | 0 | }), |
2197 | 0 | Float64 => cast_decimal_to_float::<D, Float64Type, _>(array, |x| { |
2198 | 0 | as_float(x) / 10_f64.powi(*scale as i32) |
2199 | 0 | }), |
2200 | 0 | Utf8View => value_to_string_view(array, cast_options), |
2201 | 0 | Utf8 => value_to_string::<i32>(array, cast_options), |
2202 | 0 | LargeUtf8 => value_to_string::<i64>(array, cast_options), |
2203 | 0 | Null => Ok(new_null_array(to_type, array.len())), |
2204 | 0 | _ => Err(ArrowError::CastError(format!( |
2205 | 0 | "Casting from {from_type:?} to {to_type:?} not supported" |
2206 | 0 | ))), |
2207 | | } |
2208 | 0 | } |
2209 | | |
2210 | 0 | fn cast_to_decimal<D, M>( |
2211 | 0 | array: &dyn Array, |
2212 | 0 | base: M, |
2213 | 0 | precision: &u8, |
2214 | 0 | scale: &i8, |
2215 | 0 | from_type: &DataType, |
2216 | 0 | to_type: &DataType, |
2217 | 0 | cast_options: &CastOptions, |
2218 | 0 | ) -> Result<ArrayRef, ArrowError> |
2219 | 0 | where |
2220 | 0 | D: DecimalType + ArrowPrimitiveType<Native = M>, |
2221 | 0 | M: ArrowNativeTypeOp + DecimalCast, |
2222 | 0 | u8: num::traits::AsPrimitive<M>, |
2223 | 0 | u16: num::traits::AsPrimitive<M>, |
2224 | 0 | u32: num::traits::AsPrimitive<M>, |
2225 | 0 | u64: num::traits::AsPrimitive<M>, |
2226 | 0 | i8: num::traits::AsPrimitive<M>, |
2227 | 0 | i16: num::traits::AsPrimitive<M>, |
2228 | 0 | i32: num::traits::AsPrimitive<M>, |
2229 | 0 | i64: num::traits::AsPrimitive<M>, |
2230 | | { |
2231 | | use DataType::*; |
2232 | | // cast data to decimal |
2233 | 0 | match from_type { |
2234 | 0 | UInt8 => cast_integer_to_decimal::<_, D, M>( |
2235 | 0 | array.as_primitive::<UInt8Type>(), |
2236 | 0 | *precision, |
2237 | 0 | *scale, |
2238 | 0 | base, |
2239 | 0 | cast_options, |
2240 | | ), |
2241 | 0 | UInt16 => cast_integer_to_decimal::<_, D, _>( |
2242 | 0 | array.as_primitive::<UInt16Type>(), |
2243 | 0 | *precision, |
2244 | 0 | *scale, |
2245 | 0 | base, |
2246 | 0 | cast_options, |
2247 | | ), |
2248 | 0 | UInt32 => cast_integer_to_decimal::<_, D, _>( |
2249 | 0 | array.as_primitive::<UInt32Type>(), |
2250 | 0 | *precision, |
2251 | 0 | *scale, |
2252 | 0 | base, |
2253 | 0 | cast_options, |
2254 | | ), |
2255 | 0 | UInt64 => cast_integer_to_decimal::<_, D, _>( |
2256 | 0 | array.as_primitive::<UInt64Type>(), |
2257 | 0 | *precision, |
2258 | 0 | *scale, |
2259 | 0 | base, |
2260 | 0 | cast_options, |
2261 | | ), |
2262 | 0 | Int8 => cast_integer_to_decimal::<_, D, _>( |
2263 | 0 | array.as_primitive::<Int8Type>(), |
2264 | 0 | *precision, |
2265 | 0 | *scale, |
2266 | 0 | base, |
2267 | 0 | cast_options, |
2268 | | ), |
2269 | 0 | Int16 => cast_integer_to_decimal::<_, D, _>( |
2270 | 0 | array.as_primitive::<Int16Type>(), |
2271 | 0 | *precision, |
2272 | 0 | *scale, |
2273 | 0 | base, |
2274 | 0 | cast_options, |
2275 | | ), |
2276 | 0 | Int32 => cast_integer_to_decimal::<_, D, _>( |
2277 | 0 | array.as_primitive::<Int32Type>(), |
2278 | 0 | *precision, |
2279 | 0 | *scale, |
2280 | 0 | base, |
2281 | 0 | cast_options, |
2282 | | ), |
2283 | 0 | Int64 => cast_integer_to_decimal::<_, D, _>( |
2284 | 0 | array.as_primitive::<Int64Type>(), |
2285 | 0 | *precision, |
2286 | 0 | *scale, |
2287 | 0 | base, |
2288 | 0 | cast_options, |
2289 | | ), |
2290 | 0 | Float32 => cast_floating_point_to_decimal::<_, D>( |
2291 | 0 | array.as_primitive::<Float32Type>(), |
2292 | 0 | *precision, |
2293 | 0 | *scale, |
2294 | 0 | cast_options, |
2295 | | ), |
2296 | 0 | Float64 => cast_floating_point_to_decimal::<_, D>( |
2297 | 0 | array.as_primitive::<Float64Type>(), |
2298 | 0 | *precision, |
2299 | 0 | *scale, |
2300 | 0 | cast_options, |
2301 | | ), |
2302 | | Utf8View | Utf8 => { |
2303 | 0 | cast_string_to_decimal::<D, i32>(array, *precision, *scale, cast_options) |
2304 | | } |
2305 | 0 | LargeUtf8 => cast_string_to_decimal::<D, i64>(array, *precision, *scale, cast_options), |
2306 | 0 | Null => Ok(new_null_array(to_type, array.len())), |
2307 | 0 | _ => Err(ArrowError::CastError(format!( |
2308 | 0 | "Casting from {from_type:?} to {to_type:?} not supported" |
2309 | 0 | ))), |
2310 | | } |
2311 | 0 | } |
2312 | | |
2313 | | /// Get the time unit as a multiple of a second |
2314 | 0 | const fn time_unit_multiple(unit: &TimeUnit) -> i64 { |
2315 | 0 | match unit { |
2316 | 0 | TimeUnit::Second => 1, |
2317 | 0 | TimeUnit::Millisecond => MILLISECONDS, |
2318 | 0 | TimeUnit::Microsecond => MICROSECONDS, |
2319 | 0 | TimeUnit::Nanosecond => NANOSECONDS, |
2320 | | } |
2321 | 0 | } |
2322 | | |
2323 | | /// Convert Array into a PrimitiveArray of type, and apply numeric cast |
2324 | 0 | fn cast_numeric_arrays<FROM, TO>( |
2325 | 0 | from: &dyn Array, |
2326 | 0 | cast_options: &CastOptions, |
2327 | 0 | ) -> Result<ArrayRef, ArrowError> |
2328 | 0 | where |
2329 | 0 | FROM: ArrowPrimitiveType, |
2330 | 0 | TO: ArrowPrimitiveType, |
2331 | 0 | FROM::Native: NumCast, |
2332 | 0 | TO::Native: NumCast, |
2333 | | { |
2334 | 0 | if cast_options.safe { |
2335 | | // If the value can't be casted to the `TO::Native`, return null |
2336 | 0 | Ok(Arc::new(numeric_cast::<FROM, TO>( |
2337 | 0 | from.as_primitive::<FROM>(), |
2338 | 0 | ))) |
2339 | | } else { |
2340 | | // If the value can't be casted to the `TO::Native`, return error |
2341 | 0 | Ok(Arc::new(try_numeric_cast::<FROM, TO>( |
2342 | 0 | from.as_primitive::<FROM>(), |
2343 | 0 | )?)) |
2344 | | } |
2345 | 0 | } |
2346 | | |
2347 | | // Natural cast between numeric types |
2348 | | // If the value of T can't be casted to R, will throw error |
2349 | 0 | fn try_numeric_cast<T, R>(from: &PrimitiveArray<T>) -> Result<PrimitiveArray<R>, ArrowError> |
2350 | 0 | where |
2351 | 0 | T: ArrowPrimitiveType, |
2352 | 0 | R: ArrowPrimitiveType, |
2353 | 0 | T::Native: NumCast, |
2354 | 0 | R::Native: NumCast, |
2355 | | { |
2356 | 0 | from.try_unary(|value| { |
2357 | 0 | num::cast::cast::<T::Native, R::Native>(value).ok_or_else(|| { |
2358 | 0 | ArrowError::CastError(format!( |
2359 | 0 | "Can't cast value {:?} to type {}", |
2360 | 0 | value, |
2361 | 0 | R::DATA_TYPE |
2362 | 0 | )) |
2363 | 0 | }) |
2364 | 0 | }) |
2365 | 0 | } |
2366 | | |
2367 | | // Natural cast between numeric types |
2368 | | // If the value of T can't be casted to R, it will be converted to null |
2369 | 0 | fn numeric_cast<T, R>(from: &PrimitiveArray<T>) -> PrimitiveArray<R> |
2370 | 0 | where |
2371 | 0 | T: ArrowPrimitiveType, |
2372 | 0 | R: ArrowPrimitiveType, |
2373 | 0 | T::Native: NumCast, |
2374 | 0 | R::Native: NumCast, |
2375 | | { |
2376 | 0 | from.unary_opt::<_, R>(num::cast::cast::<T::Native, R::Native>) |
2377 | 0 | } |
2378 | | |
2379 | 0 | fn cast_numeric_to_binary<FROM: ArrowPrimitiveType, O: OffsetSizeTrait>( |
2380 | 0 | array: &dyn Array, |
2381 | 0 | ) -> Result<ArrayRef, ArrowError> { |
2382 | 0 | let array = array.as_primitive::<FROM>(); |
2383 | 0 | let size = std::mem::size_of::<FROM::Native>(); |
2384 | 0 | let offsets = OffsetBuffer::from_lengths(std::iter::repeat_n(size, array.len())); |
2385 | 0 | Ok(Arc::new(GenericBinaryArray::<O>::new( |
2386 | 0 | offsets, |
2387 | 0 | array.values().inner().clone(), |
2388 | 0 | array.nulls().cloned(), |
2389 | 0 | ))) |
2390 | 0 | } |
2391 | | |
2392 | 0 | fn adjust_timestamp_to_timezone<T: ArrowTimestampType>( |
2393 | 0 | array: PrimitiveArray<Int64Type>, |
2394 | 0 | to_tz: &Tz, |
2395 | 0 | cast_options: &CastOptions, |
2396 | 0 | ) -> Result<PrimitiveArray<Int64Type>, ArrowError> { |
2397 | 0 | let adjust = |o| { |
2398 | 0 | let local = as_datetime::<T>(o)?; |
2399 | 0 | let offset = to_tz.offset_from_local_datetime(&local).single()?; |
2400 | 0 | T::make_value(local - offset.fix()) |
2401 | 0 | }; |
2402 | 0 | let adjusted = if cast_options.safe { |
2403 | 0 | array.unary_opt::<_, Int64Type>(adjust) |
2404 | | } else { |
2405 | 0 | array.try_unary::<_, Int64Type, _>(|o| { |
2406 | 0 | adjust(o).ok_or_else(|| { |
2407 | 0 | ArrowError::CastError("Cannot cast timezone to different timezone".to_string()) |
2408 | 0 | }) |
2409 | 0 | })? |
2410 | | }; |
2411 | 0 | Ok(adjusted) |
2412 | 0 | } |
2413 | | |
2414 | | /// Cast numeric types to Boolean |
2415 | | /// |
2416 | | /// Any zero value returns `false` while non-zero returns `true` |
2417 | 0 | fn cast_numeric_to_bool<FROM>(from: &dyn Array) -> Result<ArrayRef, ArrowError> |
2418 | 0 | where |
2419 | 0 | FROM: ArrowPrimitiveType, |
2420 | | { |
2421 | 0 | numeric_to_bool_cast::<FROM>(from.as_primitive::<FROM>()).map(|to| Arc::new(to) as ArrayRef) |
2422 | 0 | } |
2423 | | |
2424 | 0 | fn numeric_to_bool_cast<T>(from: &PrimitiveArray<T>) -> Result<BooleanArray, ArrowError> |
2425 | 0 | where |
2426 | 0 | T: ArrowPrimitiveType + ArrowPrimitiveType, |
2427 | | { |
2428 | 0 | let mut b = BooleanBuilder::with_capacity(from.len()); |
2429 | | |
2430 | 0 | for i in 0..from.len() { |
2431 | 0 | if from.is_null(i) { |
2432 | 0 | b.append_null(); |
2433 | 0 | } else if from.value(i) != T::default_value() { |
2434 | 0 | b.append_value(true); |
2435 | 0 | } else { |
2436 | 0 | b.append_value(false); |
2437 | 0 | } |
2438 | | } |
2439 | | |
2440 | 0 | Ok(b.finish()) |
2441 | 0 | } |
2442 | | |
2443 | | /// Cast Boolean types to numeric |
2444 | | /// |
2445 | | /// `false` returns 0 while `true` returns 1 |
2446 | 0 | fn cast_bool_to_numeric<TO>( |
2447 | 0 | from: &dyn Array, |
2448 | 0 | cast_options: &CastOptions, |
2449 | 0 | ) -> Result<ArrayRef, ArrowError> |
2450 | 0 | where |
2451 | 0 | TO: ArrowPrimitiveType, |
2452 | 0 | TO::Native: num::cast::NumCast, |
2453 | | { |
2454 | 0 | Ok(Arc::new(bool_to_numeric_cast::<TO>( |
2455 | 0 | from.as_any().downcast_ref::<BooleanArray>().unwrap(), |
2456 | 0 | cast_options, |
2457 | 0 | ))) |
2458 | 0 | } |
2459 | | |
2460 | 0 | fn bool_to_numeric_cast<T>(from: &BooleanArray, _cast_options: &CastOptions) -> PrimitiveArray<T> |
2461 | 0 | where |
2462 | 0 | T: ArrowPrimitiveType, |
2463 | 0 | T::Native: num::NumCast, |
2464 | | { |
2465 | 0 | let iter = (0..from.len()).map(|i| { |
2466 | 0 | if from.is_null(i) { |
2467 | 0 | None |
2468 | 0 | } else if from.value(i) { |
2469 | | // a workaround to cast a primitive to T::Native, infallible |
2470 | 0 | num::cast::cast(1) |
2471 | | } else { |
2472 | 0 | Some(T::default_value()) |
2473 | | } |
2474 | 0 | }); |
2475 | | // Benefit: |
2476 | | // 20% performance improvement |
2477 | | // Soundness: |
2478 | | // The iterator is trustedLen because it comes from a Range |
2479 | 0 | unsafe { PrimitiveArray::<T>::from_trusted_len_iter(iter) } |
2480 | 0 | } |
2481 | | |
2482 | | /// Helper function to cast from one `BinaryArray` or 'LargeBinaryArray' to 'FixedSizeBinaryArray'. |
2483 | 0 | fn cast_binary_to_fixed_size_binary<O: OffsetSizeTrait>( |
2484 | 0 | array: &dyn Array, |
2485 | 0 | byte_width: i32, |
2486 | 0 | cast_options: &CastOptions, |
2487 | 0 | ) -> Result<ArrayRef, ArrowError> { |
2488 | 0 | let array = array.as_binary::<O>(); |
2489 | 0 | let mut builder = FixedSizeBinaryBuilder::with_capacity(array.len(), byte_width); |
2490 | | |
2491 | 0 | for i in 0..array.len() { |
2492 | 0 | if array.is_null(i) { |
2493 | 0 | builder.append_null(); |
2494 | 0 | } else { |
2495 | 0 | match builder.append_value(array.value(i)) { |
2496 | 0 | Ok(_) => {} |
2497 | 0 | Err(e) => match cast_options.safe { |
2498 | 0 | true => builder.append_null(), |
2499 | 0 | false => return Err(e), |
2500 | | }, |
2501 | | } |
2502 | | } |
2503 | | } |
2504 | | |
2505 | 0 | Ok(Arc::new(builder.finish())) |
2506 | 0 | } |
2507 | | |
2508 | | /// Helper function to cast from 'FixedSizeBinaryArray' to one `BinaryArray` or 'LargeBinaryArray'. |
2509 | | /// If the target one is too large for the source array it will return an Error. |
2510 | 0 | fn cast_fixed_size_binary_to_binary<O: OffsetSizeTrait>( |
2511 | 0 | array: &dyn Array, |
2512 | 0 | byte_width: i32, |
2513 | 0 | ) -> Result<ArrayRef, ArrowError> { |
2514 | 0 | let array = array |
2515 | 0 | .as_any() |
2516 | 0 | .downcast_ref::<FixedSizeBinaryArray>() |
2517 | 0 | .unwrap(); |
2518 | | |
2519 | 0 | let offsets: i128 = byte_width as i128 * array.len() as i128; |
2520 | | |
2521 | 0 | let is_binary = matches!(GenericBinaryType::<O>::DATA_TYPE, DataType::Binary); |
2522 | 0 | if is_binary && offsets > i32::MAX as i128 { |
2523 | 0 | return Err(ArrowError::ComputeError( |
2524 | 0 | "FixedSizeBinary array too large to cast to Binary array".to_string(), |
2525 | 0 | )); |
2526 | 0 | } else if !is_binary && offsets > i64::MAX as i128 { |
2527 | 0 | return Err(ArrowError::ComputeError( |
2528 | 0 | "FixedSizeBinary array too large to cast to LargeBinary array".to_string(), |
2529 | 0 | )); |
2530 | 0 | } |
2531 | | |
2532 | 0 | let mut builder = GenericBinaryBuilder::<O>::with_capacity(array.len(), array.len()); |
2533 | | |
2534 | 0 | for i in 0..array.len() { |
2535 | 0 | if array.is_null(i) { |
2536 | 0 | builder.append_null(); |
2537 | 0 | } else { |
2538 | 0 | builder.append_value(array.value(i)); |
2539 | 0 | } |
2540 | | } |
2541 | | |
2542 | 0 | Ok(Arc::new(builder.finish())) |
2543 | 0 | } |
2544 | | |
2545 | 0 | fn cast_fixed_size_binary_to_binary_view( |
2546 | 0 | array: &dyn Array, |
2547 | 0 | _byte_width: i32, |
2548 | 0 | ) -> Result<ArrayRef, ArrowError> { |
2549 | 0 | let array = array |
2550 | 0 | .as_any() |
2551 | 0 | .downcast_ref::<FixedSizeBinaryArray>() |
2552 | 0 | .unwrap(); |
2553 | | |
2554 | 0 | let mut builder = BinaryViewBuilder::with_capacity(array.len()); |
2555 | 0 | for i in 0..array.len() { |
2556 | 0 | if array.is_null(i) { |
2557 | 0 | builder.append_null(); |
2558 | 0 | } else { |
2559 | 0 | builder.append_value(array.value(i)); |
2560 | 0 | } |
2561 | | } |
2562 | | |
2563 | 0 | Ok(Arc::new(builder.finish())) |
2564 | 0 | } |
2565 | | |
2566 | | /// Helper function to cast from one `ByteArrayType` to another and vice versa. |
2567 | | /// If the target one (e.g., `LargeUtf8`) is too large for the source array it will return an Error. |
2568 | 0 | fn cast_byte_container<FROM, TO>(array: &dyn Array) -> Result<ArrayRef, ArrowError> |
2569 | 0 | where |
2570 | 0 | FROM: ByteArrayType, |
2571 | 0 | TO: ByteArrayType<Native = FROM::Native>, |
2572 | 0 | FROM::Offset: OffsetSizeTrait + ToPrimitive, |
2573 | 0 | TO::Offset: OffsetSizeTrait + NumCast, |
2574 | | { |
2575 | 0 | let data = array.to_data(); |
2576 | 0 | assert_eq!(data.data_type(), &FROM::DATA_TYPE); |
2577 | 0 | let str_values_buf = data.buffers()[1].clone(); |
2578 | 0 | let offsets = data.buffers()[0].typed_data::<FROM::Offset>(); |
2579 | | |
2580 | 0 | let mut offset_builder = BufferBuilder::<TO::Offset>::new(offsets.len()); |
2581 | 0 | offsets |
2582 | 0 | .iter() |
2583 | 0 | .try_for_each::<_, Result<_, ArrowError>>(|offset| { |
2584 | 0 | let offset = |
2585 | 0 | <<TO as ByteArrayType>::Offset as NumCast>::from(*offset).ok_or_else(|| { |
2586 | 0 | ArrowError::ComputeError(format!( |
2587 | 0 | "{}{} array too large to cast to {}{} array", |
2588 | 0 | FROM::Offset::PREFIX, |
2589 | 0 | FROM::PREFIX, |
2590 | 0 | TO::Offset::PREFIX, |
2591 | 0 | TO::PREFIX |
2592 | 0 | )) |
2593 | 0 | })?; |
2594 | 0 | offset_builder.append(offset); |
2595 | 0 | Ok(()) |
2596 | 0 | })?; |
2597 | | |
2598 | 0 | let offset_buffer = offset_builder.finish(); |
2599 | | |
2600 | 0 | let dtype = TO::DATA_TYPE; |
2601 | | |
2602 | 0 | let builder = ArrayData::builder(dtype) |
2603 | 0 | .offset(array.offset()) |
2604 | 0 | .len(array.len()) |
2605 | 0 | .add_buffer(offset_buffer) |
2606 | 0 | .add_buffer(str_values_buf) |
2607 | 0 | .nulls(data.nulls().cloned()); |
2608 | | |
2609 | 0 | let array_data = unsafe { builder.build_unchecked() }; |
2610 | | |
2611 | 0 | Ok(Arc::new(GenericByteArray::<TO>::from(array_data))) |
2612 | 0 | } |
2613 | | |
2614 | | /// Helper function to cast from one `ByteViewType` array to `ByteArrayType` array. |
2615 | 0 | fn cast_view_to_byte<FROM, TO>(array: &dyn Array) -> Result<ArrayRef, ArrowError> |
2616 | 0 | where |
2617 | 0 | FROM: ByteViewType, |
2618 | 0 | TO: ByteArrayType, |
2619 | 0 | FROM::Native: AsRef<TO::Native>, |
2620 | | { |
2621 | 0 | let data = array.to_data(); |
2622 | 0 | let view_array = GenericByteViewArray::<FROM>::from(data); |
2623 | | |
2624 | 0 | let len = view_array.len(); |
2625 | 0 | let bytes = view_array |
2626 | 0 | .views() |
2627 | 0 | .iter() |
2628 | 0 | .map(|v| ByteView::from(*v).length as usize) |
2629 | 0 | .sum::<usize>(); |
2630 | | |
2631 | 0 | let mut byte_array_builder = GenericByteBuilder::<TO>::with_capacity(len, bytes); |
2632 | | |
2633 | 0 | for val in view_array.iter() { |
2634 | 0 | byte_array_builder.append_option(val); |
2635 | 0 | } |
2636 | | |
2637 | 0 | Ok(Arc::new(byte_array_builder.finish())) |
2638 | 0 | } |
2639 | | |
2640 | | #[cfg(test)] |
2641 | | mod tests { |
2642 | | use super::*; |
2643 | | use arrow_buffer::i256; |
2644 | | use arrow_buffer::{Buffer, IntervalDayTime, NullBuffer}; |
2645 | | use chrono::NaiveDate; |
2646 | | use half::f16; |
2647 | | |
2648 | | #[derive(Clone)] |
2649 | | struct DecimalCastTestConfig { |
2650 | | input_prec: u8, |
2651 | | input_scale: i8, |
2652 | | input_repr: i128, |
2653 | | output_prec: u8, |
2654 | | output_scale: i8, |
2655 | | expected_output_repr: Result<i128, String>, // the error variant can contain a string |
2656 | | // template where the "{}" will be |
2657 | | // replaced with the decimal type name |
2658 | | // (e.g. Decimal128) |
2659 | | } |
2660 | | |
2661 | | macro_rules! generate_cast_test_case { |
2662 | | ($INPUT_ARRAY: expr, $OUTPUT_TYPE_ARRAY: ident, $OUTPUT_TYPE: expr, $OUTPUT_VALUES: expr) => { |
2663 | | let output = |
2664 | | $OUTPUT_TYPE_ARRAY::from($OUTPUT_VALUES).with_data_type($OUTPUT_TYPE.clone()); |
2665 | | |
2666 | | // assert cast type |
2667 | | let input_array_type = $INPUT_ARRAY.data_type(); |
2668 | | assert!(can_cast_types(input_array_type, $OUTPUT_TYPE)); |
2669 | | let result = cast($INPUT_ARRAY, $OUTPUT_TYPE).unwrap(); |
2670 | | assert_eq!($OUTPUT_TYPE, result.data_type()); |
2671 | | assert_eq!(result.as_ref(), &output); |
2672 | | |
2673 | | let cast_option = CastOptions { |
2674 | | safe: false, |
2675 | | format_options: FormatOptions::default(), |
2676 | | }; |
2677 | | let result = cast_with_options($INPUT_ARRAY, $OUTPUT_TYPE, &cast_option).unwrap(); |
2678 | | assert_eq!($OUTPUT_TYPE, result.data_type()); |
2679 | | assert_eq!(result.as_ref(), &output); |
2680 | | }; |
2681 | | } |
2682 | | |
2683 | | fn run_decimal_cast_test_case<I, O>(t: DecimalCastTestConfig) |
2684 | | where |
2685 | | I: DecimalType, |
2686 | | O: DecimalType, |
2687 | | I::Native: DecimalCast, |
2688 | | O::Native: DecimalCast, |
2689 | | { |
2690 | | let array = vec![I::Native::from_decimal(t.input_repr)]; |
2691 | | let array = array |
2692 | | .into_iter() |
2693 | | .collect::<PrimitiveArray<I>>() |
2694 | | .with_precision_and_scale(t.input_prec, t.input_scale) |
2695 | | .unwrap(); |
2696 | | let input_type = array.data_type(); |
2697 | | let output_type = O::TYPE_CONSTRUCTOR(t.output_prec, t.output_scale); |
2698 | | assert!(can_cast_types(input_type, &output_type)); |
2699 | | |
2700 | | let options = CastOptions { |
2701 | | safe: false, |
2702 | | ..Default::default() |
2703 | | }; |
2704 | | let result = cast_with_options(&array, &output_type, &options); |
2705 | | |
2706 | | match t.expected_output_repr { |
2707 | | Ok(v) => { |
2708 | | let expected_array = vec![O::Native::from_decimal(v)]; |
2709 | | let expected_array = expected_array |
2710 | | .into_iter() |
2711 | | .collect::<PrimitiveArray<O>>() |
2712 | | .with_precision_and_scale(t.output_prec, t.output_scale) |
2713 | | .unwrap(); |
2714 | | assert_eq!(*result.unwrap(), expected_array); |
2715 | | } |
2716 | | Err(expected_output_message_template) => { |
2717 | | assert!(result.is_err()); |
2718 | | let expected_error_message = |
2719 | | expected_output_message_template.replace("{}", O::PREFIX); |
2720 | | assert_eq!(result.unwrap_err().to_string(), expected_error_message); |
2721 | | } |
2722 | | } |
2723 | | } |
2724 | | |
2725 | | fn create_decimal32_array( |
2726 | | array: Vec<Option<i32>>, |
2727 | | precision: u8, |
2728 | | scale: i8, |
2729 | | ) -> Result<Decimal32Array, ArrowError> { |
2730 | | array |
2731 | | .into_iter() |
2732 | | .collect::<Decimal32Array>() |
2733 | | .with_precision_and_scale(precision, scale) |
2734 | | } |
2735 | | |
2736 | | fn create_decimal64_array( |
2737 | | array: Vec<Option<i64>>, |
2738 | | precision: u8, |
2739 | | scale: i8, |
2740 | | ) -> Result<Decimal64Array, ArrowError> { |
2741 | | array |
2742 | | .into_iter() |
2743 | | .collect::<Decimal64Array>() |
2744 | | .with_precision_and_scale(precision, scale) |
2745 | | } |
2746 | | |
2747 | | fn create_decimal128_array( |
2748 | | array: Vec<Option<i128>>, |
2749 | | precision: u8, |
2750 | | scale: i8, |
2751 | | ) -> Result<Decimal128Array, ArrowError> { |
2752 | | array |
2753 | | .into_iter() |
2754 | | .collect::<Decimal128Array>() |
2755 | | .with_precision_and_scale(precision, scale) |
2756 | | } |
2757 | | |
2758 | | fn create_decimal256_array( |
2759 | | array: Vec<Option<i256>>, |
2760 | | precision: u8, |
2761 | | scale: i8, |
2762 | | ) -> Result<Decimal256Array, ArrowError> { |
2763 | | array |
2764 | | .into_iter() |
2765 | | .collect::<Decimal256Array>() |
2766 | | .with_precision_and_scale(precision, scale) |
2767 | | } |
2768 | | |
2769 | | #[test] |
2770 | | #[cfg(not(feature = "force_validate"))] |
2771 | | #[should_panic( |
2772 | | expected = "Cannot cast to Decimal128(20, 3). Overflowing on 57896044618658097711785492504343953926634992332820282019728792003956564819967" |
2773 | | )] |
2774 | | fn test_cast_decimal_to_decimal_round_with_error() { |
2775 | | // decimal256 to decimal128 overflow |
2776 | | let array = vec![ |
2777 | | Some(i256::from_i128(1123454)), |
2778 | | Some(i256::from_i128(2123456)), |
2779 | | Some(i256::from_i128(-3123453)), |
2780 | | Some(i256::from_i128(-3123456)), |
2781 | | None, |
2782 | | Some(i256::MAX), |
2783 | | Some(i256::MIN), |
2784 | | ]; |
2785 | | let input_decimal_array = create_decimal256_array(array, 76, 4).unwrap(); |
2786 | | let array = Arc::new(input_decimal_array) as ArrayRef; |
2787 | | let input_type = DataType::Decimal256(76, 4); |
2788 | | let output_type = DataType::Decimal128(20, 3); |
2789 | | assert!(can_cast_types(&input_type, &output_type)); |
2790 | | generate_cast_test_case!( |
2791 | | &array, |
2792 | | Decimal128Array, |
2793 | | &output_type, |
2794 | | vec![ |
2795 | | Some(112345_i128), |
2796 | | Some(212346_i128), |
2797 | | Some(-312345_i128), |
2798 | | Some(-312346_i128), |
2799 | | None, |
2800 | | None, |
2801 | | None, |
2802 | | ] |
2803 | | ); |
2804 | | } |
2805 | | |
2806 | | #[test] |
2807 | | #[cfg(not(feature = "force_validate"))] |
2808 | | fn test_cast_decimal_to_decimal_round() { |
2809 | | let array = vec![ |
2810 | | Some(1123454), |
2811 | | Some(2123456), |
2812 | | Some(-3123453), |
2813 | | Some(-3123456), |
2814 | | None, |
2815 | | ]; |
2816 | | let array = create_decimal128_array(array, 20, 4).unwrap(); |
2817 | | // decimal128 to decimal128 |
2818 | | let input_type = DataType::Decimal128(20, 4); |
2819 | | let output_type = DataType::Decimal128(20, 3); |
2820 | | assert!(can_cast_types(&input_type, &output_type)); |
2821 | | generate_cast_test_case!( |
2822 | | &array, |
2823 | | Decimal128Array, |
2824 | | &output_type, |
2825 | | vec![ |
2826 | | Some(112345_i128), |
2827 | | Some(212346_i128), |
2828 | | Some(-312345_i128), |
2829 | | Some(-312346_i128), |
2830 | | None |
2831 | | ] |
2832 | | ); |
2833 | | |
2834 | | // decimal128 to decimal256 |
2835 | | let input_type = DataType::Decimal128(20, 4); |
2836 | | let output_type = DataType::Decimal256(20, 3); |
2837 | | assert!(can_cast_types(&input_type, &output_type)); |
2838 | | generate_cast_test_case!( |
2839 | | &array, |
2840 | | Decimal256Array, |
2841 | | &output_type, |
2842 | | vec![ |
2843 | | Some(i256::from_i128(112345_i128)), |
2844 | | Some(i256::from_i128(212346_i128)), |
2845 | | Some(i256::from_i128(-312345_i128)), |
2846 | | Some(i256::from_i128(-312346_i128)), |
2847 | | None |
2848 | | ] |
2849 | | ); |
2850 | | |
2851 | | // decimal256 |
2852 | | let array = vec![ |
2853 | | Some(i256::from_i128(1123454)), |
2854 | | Some(i256::from_i128(2123456)), |
2855 | | Some(i256::from_i128(-3123453)), |
2856 | | Some(i256::from_i128(-3123456)), |
2857 | | None, |
2858 | | ]; |
2859 | | let array = create_decimal256_array(array, 20, 4).unwrap(); |
2860 | | |
2861 | | // decimal256 to decimal256 |
2862 | | let input_type = DataType::Decimal256(20, 4); |
2863 | | let output_type = DataType::Decimal256(20, 3); |
2864 | | assert!(can_cast_types(&input_type, &output_type)); |
2865 | | generate_cast_test_case!( |
2866 | | &array, |
2867 | | Decimal256Array, |
2868 | | &output_type, |
2869 | | vec![ |
2870 | | Some(i256::from_i128(112345_i128)), |
2871 | | Some(i256::from_i128(212346_i128)), |
2872 | | Some(i256::from_i128(-312345_i128)), |
2873 | | Some(i256::from_i128(-312346_i128)), |
2874 | | None |
2875 | | ] |
2876 | | ); |
2877 | | // decimal256 to decimal128 |
2878 | | let input_type = DataType::Decimal256(20, 4); |
2879 | | let output_type = DataType::Decimal128(20, 3); |
2880 | | assert!(can_cast_types(&input_type, &output_type)); |
2881 | | generate_cast_test_case!( |
2882 | | &array, |
2883 | | Decimal128Array, |
2884 | | &output_type, |
2885 | | vec![ |
2886 | | Some(112345_i128), |
2887 | | Some(212346_i128), |
2888 | | Some(-312345_i128), |
2889 | | Some(-312346_i128), |
2890 | | None |
2891 | | ] |
2892 | | ); |
2893 | | } |
2894 | | |
2895 | | #[test] |
2896 | | fn test_cast_decimal32_to_decimal32() { |
2897 | | // test changing precision |
2898 | | let input_type = DataType::Decimal32(9, 3); |
2899 | | let output_type = DataType::Decimal32(9, 4); |
2900 | | assert!(can_cast_types(&input_type, &output_type)); |
2901 | | let array = vec![Some(1123456), Some(2123456), Some(3123456), None]; |
2902 | | let array = create_decimal32_array(array, 9, 3).unwrap(); |
2903 | | generate_cast_test_case!( |
2904 | | &array, |
2905 | | Decimal32Array, |
2906 | | &output_type, |
2907 | | vec![ |
2908 | | Some(11234560_i32), |
2909 | | Some(21234560_i32), |
2910 | | Some(31234560_i32), |
2911 | | None |
2912 | | ] |
2913 | | ); |
2914 | | // negative test |
2915 | | let array = vec![Some(123456), None]; |
2916 | | let array = create_decimal32_array(array, 9, 0).unwrap(); |
2917 | | let result_safe = cast(&array, &DataType::Decimal32(2, 2)); |
2918 | | assert!(result_safe.is_ok()); |
2919 | | let options = CastOptions { |
2920 | | safe: false, |
2921 | | ..Default::default() |
2922 | | }; |
2923 | | |
2924 | | let result_unsafe = cast_with_options(&array, &DataType::Decimal32(2, 2), &options); |
2925 | | assert_eq!("Invalid argument error: 12345600 is too large to store in a Decimal32 of precision 2. Max is 99", |
2926 | | result_unsafe.unwrap_err().to_string()); |
2927 | | } |
2928 | | |
2929 | | #[test] |
2930 | | fn test_cast_decimal64_to_decimal64() { |
2931 | | // test changing precision |
2932 | | let input_type = DataType::Decimal64(17, 3); |
2933 | | let output_type = DataType::Decimal64(17, 4); |
2934 | | assert!(can_cast_types(&input_type, &output_type)); |
2935 | | let array = vec![Some(1123456), Some(2123456), Some(3123456), None]; |
2936 | | let array = create_decimal64_array(array, 17, 3).unwrap(); |
2937 | | generate_cast_test_case!( |
2938 | | &array, |
2939 | | Decimal64Array, |
2940 | | &output_type, |
2941 | | vec![ |
2942 | | Some(11234560_i64), |
2943 | | Some(21234560_i64), |
2944 | | Some(31234560_i64), |
2945 | | None |
2946 | | ] |
2947 | | ); |
2948 | | // negative test |
2949 | | let array = vec![Some(123456), None]; |
2950 | | let array = create_decimal64_array(array, 9, 0).unwrap(); |
2951 | | let result_safe = cast(&array, &DataType::Decimal64(2, 2)); |
2952 | | assert!(result_safe.is_ok()); |
2953 | | let options = CastOptions { |
2954 | | safe: false, |
2955 | | ..Default::default() |
2956 | | }; |
2957 | | |
2958 | | let result_unsafe = cast_with_options(&array, &DataType::Decimal64(2, 2), &options); |
2959 | | assert_eq!("Invalid argument error: 12345600 is too large to store in a Decimal64 of precision 2. Max is 99", |
2960 | | result_unsafe.unwrap_err().to_string()); |
2961 | | } |
2962 | | |
2963 | | #[test] |
2964 | | fn test_cast_decimal128_to_decimal128() { |
2965 | | // test changing precision |
2966 | | let input_type = DataType::Decimal128(20, 3); |
2967 | | let output_type = DataType::Decimal128(20, 4); |
2968 | | assert!(can_cast_types(&input_type, &output_type)); |
2969 | | let array = vec![Some(1123456), Some(2123456), Some(3123456), None]; |
2970 | | let array = create_decimal128_array(array, 20, 3).unwrap(); |
2971 | | generate_cast_test_case!( |
2972 | | &array, |
2973 | | Decimal128Array, |
2974 | | &output_type, |
2975 | | vec![ |
2976 | | Some(11234560_i128), |
2977 | | Some(21234560_i128), |
2978 | | Some(31234560_i128), |
2979 | | None |
2980 | | ] |
2981 | | ); |
2982 | | // negative test |
2983 | | let array = vec![Some(123456), None]; |
2984 | | let array = create_decimal128_array(array, 10, 0).unwrap(); |
2985 | | let result_safe = cast(&array, &DataType::Decimal128(2, 2)); |
2986 | | assert!(result_safe.is_ok()); |
2987 | | let options = CastOptions { |
2988 | | safe: false, |
2989 | | ..Default::default() |
2990 | | }; |
2991 | | |
2992 | | let result_unsafe = cast_with_options(&array, &DataType::Decimal128(2, 2), &options); |
2993 | | assert_eq!("Invalid argument error: 12345600 is too large to store in a Decimal128 of precision 2. Max is 99", |
2994 | | result_unsafe.unwrap_err().to_string()); |
2995 | | } |
2996 | | |
2997 | | #[test] |
2998 | | fn test_cast_decimal32_to_decimal32_dict() { |
2999 | | let p = 9; |
3000 | | let s = 3; |
3001 | | let input_type = DataType::Decimal32(p, s); |
3002 | | let output_type = DataType::Dictionary( |
3003 | | Box::new(DataType::Int32), |
3004 | | Box::new(DataType::Decimal32(p, s)), |
3005 | | ); |
3006 | | assert!(can_cast_types(&input_type, &output_type)); |
3007 | | let array = vec![Some(1123456), Some(2123456), Some(3123456), None]; |
3008 | | let array = create_decimal32_array(array, p, s).unwrap(); |
3009 | | let cast_array = cast_with_options(&array, &output_type, &CastOptions::default()).unwrap(); |
3010 | | assert_eq!(cast_array.data_type(), &output_type); |
3011 | | } |
3012 | | |
3013 | | #[test] |
3014 | | fn test_cast_decimal64_to_decimal64_dict() { |
3015 | | let p = 15; |
3016 | | let s = 3; |
3017 | | let input_type = DataType::Decimal64(p, s); |
3018 | | let output_type = DataType::Dictionary( |
3019 | | Box::new(DataType::Int32), |
3020 | | Box::new(DataType::Decimal64(p, s)), |
3021 | | ); |
3022 | | assert!(can_cast_types(&input_type, &output_type)); |
3023 | | let array = vec![Some(1123456), Some(2123456), Some(3123456), None]; |
3024 | | let array = create_decimal64_array(array, p, s).unwrap(); |
3025 | | let cast_array = cast_with_options(&array, &output_type, &CastOptions::default()).unwrap(); |
3026 | | assert_eq!(cast_array.data_type(), &output_type); |
3027 | | } |
3028 | | |
3029 | | #[test] |
3030 | | fn test_cast_decimal128_to_decimal128_dict() { |
3031 | | let p = 20; |
3032 | | let s = 3; |
3033 | | let input_type = DataType::Decimal128(p, s); |
3034 | | let output_type = DataType::Dictionary( |
3035 | | Box::new(DataType::Int32), |
3036 | | Box::new(DataType::Decimal128(p, s)), |
3037 | | ); |
3038 | | assert!(can_cast_types(&input_type, &output_type)); |
3039 | | let array = vec![Some(1123456), Some(2123456), Some(3123456), None]; |
3040 | | let array = create_decimal128_array(array, p, s).unwrap(); |
3041 | | let cast_array = cast_with_options(&array, &output_type, &CastOptions::default()).unwrap(); |
3042 | | assert_eq!(cast_array.data_type(), &output_type); |
3043 | | } |
3044 | | |
3045 | | #[test] |
3046 | | fn test_cast_decimal256_to_decimal256_dict() { |
3047 | | let p = 20; |
3048 | | let s = 3; |
3049 | | let input_type = DataType::Decimal256(p, s); |
3050 | | let output_type = DataType::Dictionary( |
3051 | | Box::new(DataType::Int32), |
3052 | | Box::new(DataType::Decimal256(p, s)), |
3053 | | ); |
3054 | | assert!(can_cast_types(&input_type, &output_type)); |
3055 | | let array = vec![Some(1123456), Some(2123456), Some(3123456), None]; |
3056 | | let array = create_decimal128_array(array, p, s).unwrap(); |
3057 | | let cast_array = cast_with_options(&array, &output_type, &CastOptions::default()).unwrap(); |
3058 | | assert_eq!(cast_array.data_type(), &output_type); |
3059 | | } |
3060 | | |
3061 | | #[test] |
3062 | | fn test_cast_decimal32_to_decimal32_overflow() { |
3063 | | let input_type = DataType::Decimal32(9, 3); |
3064 | | let output_type = DataType::Decimal32(9, 9); |
3065 | | assert!(can_cast_types(&input_type, &output_type)); |
3066 | | |
3067 | | let array = vec![Some(i32::MAX)]; |
3068 | | let array = create_decimal32_array(array, 9, 3).unwrap(); |
3069 | | let result = cast_with_options( |
3070 | | &array, |
3071 | | &output_type, |
3072 | | &CastOptions { |
3073 | | safe: false, |
3074 | | format_options: FormatOptions::default(), |
3075 | | }, |
3076 | | ); |
3077 | | assert_eq!( |
3078 | | "Cast error: Cannot cast to Decimal32(9, 9). Overflowing on 2147483647", |
3079 | | result.unwrap_err().to_string() |
3080 | | ); |
3081 | | } |
3082 | | |
3083 | | #[test] |
3084 | | fn test_cast_decimal64_to_decimal64_overflow() { |
3085 | | let input_type = DataType::Decimal64(18, 3); |
3086 | | let output_type = DataType::Decimal64(18, 18); |
3087 | | assert!(can_cast_types(&input_type, &output_type)); |
3088 | | |
3089 | | let array = vec![Some(i64::MAX)]; |
3090 | | let array = create_decimal64_array(array, 18, 3).unwrap(); |
3091 | | let result = cast_with_options( |
3092 | | &array, |
3093 | | &output_type, |
3094 | | &CastOptions { |
3095 | | safe: false, |
3096 | | format_options: FormatOptions::default(), |
3097 | | }, |
3098 | | ); |
3099 | | assert_eq!( |
3100 | | "Cast error: Cannot cast to Decimal64(18, 18). Overflowing on 9223372036854775807", |
3101 | | result.unwrap_err().to_string() |
3102 | | ); |
3103 | | } |
3104 | | |
3105 | | #[test] |
3106 | | fn test_cast_decimal128_to_decimal128_overflow() { |
3107 | | let input_type = DataType::Decimal128(38, 3); |
3108 | | let output_type = DataType::Decimal128(38, 38); |
3109 | | assert!(can_cast_types(&input_type, &output_type)); |
3110 | | |
3111 | | let array = vec![Some(i128::MAX)]; |
3112 | | let array = create_decimal128_array(array, 38, 3).unwrap(); |
3113 | | let result = cast_with_options( |
3114 | | &array, |
3115 | | &output_type, |
3116 | | &CastOptions { |
3117 | | safe: false, |
3118 | | format_options: FormatOptions::default(), |
3119 | | }, |
3120 | | ); |
3121 | | assert_eq!("Cast error: Cannot cast to Decimal128(38, 38). Overflowing on 170141183460469231731687303715884105727", |
3122 | | result.unwrap_err().to_string()); |
3123 | | } |
3124 | | |
3125 | | #[test] |
3126 | | fn test_cast_decimal128_to_decimal256_overflow() { |
3127 | | let input_type = DataType::Decimal128(38, 3); |
3128 | | let output_type = DataType::Decimal256(76, 76); |
3129 | | assert!(can_cast_types(&input_type, &output_type)); |
3130 | | |
3131 | | let array = vec![Some(i128::MAX)]; |
3132 | | let array = create_decimal128_array(array, 38, 3).unwrap(); |
3133 | | let result = cast_with_options( |
3134 | | &array, |
3135 | | &output_type, |
3136 | | &CastOptions { |
3137 | | safe: false, |
3138 | | format_options: FormatOptions::default(), |
3139 | | }, |
3140 | | ); |
3141 | | assert_eq!("Cast error: Cannot cast to Decimal256(76, 76). Overflowing on 170141183460469231731687303715884105727", |
3142 | | result.unwrap_err().to_string()); |
3143 | | } |
3144 | | |
3145 | | #[test] |
3146 | | fn test_cast_decimal32_to_decimal256() { |
3147 | | let input_type = DataType::Decimal32(8, 3); |
3148 | | let output_type = DataType::Decimal256(20, 4); |
3149 | | assert!(can_cast_types(&input_type, &output_type)); |
3150 | | let array = vec![Some(1123456), Some(2123456), Some(3123456), None]; |
3151 | | let array = create_decimal32_array(array, 8, 3).unwrap(); |
3152 | | generate_cast_test_case!( |
3153 | | &array, |
3154 | | Decimal256Array, |
3155 | | &output_type, |
3156 | | vec![ |
3157 | | Some(i256::from_i128(11234560_i128)), |
3158 | | Some(i256::from_i128(21234560_i128)), |
3159 | | Some(i256::from_i128(31234560_i128)), |
3160 | | None |
3161 | | ] |
3162 | | ); |
3163 | | } |
3164 | | #[test] |
3165 | | fn test_cast_decimal64_to_decimal256() { |
3166 | | let input_type = DataType::Decimal64(12, 3); |
3167 | | let output_type = DataType::Decimal256(20, 4); |
3168 | | assert!(can_cast_types(&input_type, &output_type)); |
3169 | | let array = vec![Some(1123456), Some(2123456), Some(3123456), None]; |
3170 | | let array = create_decimal64_array(array, 12, 3).unwrap(); |
3171 | | generate_cast_test_case!( |
3172 | | &array, |
3173 | | Decimal256Array, |
3174 | | &output_type, |
3175 | | vec![ |
3176 | | Some(i256::from_i128(11234560_i128)), |
3177 | | Some(i256::from_i128(21234560_i128)), |
3178 | | Some(i256::from_i128(31234560_i128)), |
3179 | | None |
3180 | | ] |
3181 | | ); |
3182 | | } |
3183 | | #[test] |
3184 | | fn test_cast_decimal128_to_decimal256() { |
3185 | | let input_type = DataType::Decimal128(20, 3); |
3186 | | let output_type = DataType::Decimal256(20, 4); |
3187 | | assert!(can_cast_types(&input_type, &output_type)); |
3188 | | let array = vec![Some(1123456), Some(2123456), Some(3123456), None]; |
3189 | | let array = create_decimal128_array(array, 20, 3).unwrap(); |
3190 | | generate_cast_test_case!( |
3191 | | &array, |
3192 | | Decimal256Array, |
3193 | | &output_type, |
3194 | | vec![ |
3195 | | Some(i256::from_i128(11234560_i128)), |
3196 | | Some(i256::from_i128(21234560_i128)), |
3197 | | Some(i256::from_i128(31234560_i128)), |
3198 | | None |
3199 | | ] |
3200 | | ); |
3201 | | } |
3202 | | |
3203 | | #[test] |
3204 | | fn test_cast_decimal256_to_decimal128_overflow() { |
3205 | | let input_type = DataType::Decimal256(76, 5); |
3206 | | let output_type = DataType::Decimal128(38, 7); |
3207 | | assert!(can_cast_types(&input_type, &output_type)); |
3208 | | let array = vec![Some(i256::from_i128(i128::MAX))]; |
3209 | | let array = create_decimal256_array(array, 76, 5).unwrap(); |
3210 | | let result = cast_with_options( |
3211 | | &array, |
3212 | | &output_type, |
3213 | | &CastOptions { |
3214 | | safe: false, |
3215 | | format_options: FormatOptions::default(), |
3216 | | }, |
3217 | | ); |
3218 | | assert_eq!("Cast error: Cannot cast to Decimal128(38, 7). Overflowing on 170141183460469231731687303715884105727", |
3219 | | result.unwrap_err().to_string()); |
3220 | | } |
3221 | | |
3222 | | #[test] |
3223 | | fn test_cast_decimal256_to_decimal256_overflow() { |
3224 | | let input_type = DataType::Decimal256(76, 5); |
3225 | | let output_type = DataType::Decimal256(76, 55); |
3226 | | assert!(can_cast_types(&input_type, &output_type)); |
3227 | | let array = vec![Some(i256::from_i128(i128::MAX))]; |
3228 | | let array = create_decimal256_array(array, 76, 5).unwrap(); |
3229 | | let result = cast_with_options( |
3230 | | &array, |
3231 | | &output_type, |
3232 | | &CastOptions { |
3233 | | safe: false, |
3234 | | format_options: FormatOptions::default(), |
3235 | | }, |
3236 | | ); |
3237 | | assert_eq!("Cast error: Cannot cast to Decimal256(76, 55). Overflowing on 170141183460469231731687303715884105727", |
3238 | | result.unwrap_err().to_string()); |
3239 | | } |
3240 | | |
3241 | | #[test] |
3242 | | fn test_cast_decimal256_to_decimal128() { |
3243 | | let input_type = DataType::Decimal256(20, 3); |
3244 | | let output_type = DataType::Decimal128(20, 4); |
3245 | | assert!(can_cast_types(&input_type, &output_type)); |
3246 | | let array = vec![ |
3247 | | Some(i256::from_i128(1123456)), |
3248 | | Some(i256::from_i128(2123456)), |
3249 | | Some(i256::from_i128(3123456)), |
3250 | | None, |
3251 | | ]; |
3252 | | let array = create_decimal256_array(array, 20, 3).unwrap(); |
3253 | | generate_cast_test_case!( |
3254 | | &array, |
3255 | | Decimal128Array, |
3256 | | &output_type, |
3257 | | vec![ |
3258 | | Some(11234560_i128), |
3259 | | Some(21234560_i128), |
3260 | | Some(31234560_i128), |
3261 | | None |
3262 | | ] |
3263 | | ); |
3264 | | } |
3265 | | |
3266 | | #[test] |
3267 | | fn test_cast_decimal256_to_decimal256() { |
3268 | | let input_type = DataType::Decimal256(20, 3); |
3269 | | let output_type = DataType::Decimal256(20, 4); |
3270 | | assert!(can_cast_types(&input_type, &output_type)); |
3271 | | let array = vec![ |
3272 | | Some(i256::from_i128(1123456)), |
3273 | | Some(i256::from_i128(2123456)), |
3274 | | Some(i256::from_i128(3123456)), |
3275 | | None, |
3276 | | ]; |
3277 | | let array = create_decimal256_array(array, 20, 3).unwrap(); |
3278 | | generate_cast_test_case!( |
3279 | | &array, |
3280 | | Decimal256Array, |
3281 | | &output_type, |
3282 | | vec![ |
3283 | | Some(i256::from_i128(11234560_i128)), |
3284 | | Some(i256::from_i128(21234560_i128)), |
3285 | | Some(i256::from_i128(31234560_i128)), |
3286 | | None |
3287 | | ] |
3288 | | ); |
3289 | | } |
3290 | | |
3291 | | fn generate_decimal_to_numeric_cast_test_case<T>(array: &PrimitiveArray<T>) |
3292 | | where |
3293 | | T: ArrowPrimitiveType + DecimalType, |
3294 | | { |
3295 | | // u8 |
3296 | | generate_cast_test_case!( |
3297 | | array, |
3298 | | UInt8Array, |
3299 | | &DataType::UInt8, |
3300 | | vec![Some(1_u8), Some(2_u8), Some(3_u8), None, Some(5_u8)] |
3301 | | ); |
3302 | | // u16 |
3303 | | generate_cast_test_case!( |
3304 | | array, |
3305 | | UInt16Array, |
3306 | | &DataType::UInt16, |
3307 | | vec![Some(1_u16), Some(2_u16), Some(3_u16), None, Some(5_u16)] |
3308 | | ); |
3309 | | // u32 |
3310 | | generate_cast_test_case!( |
3311 | | array, |
3312 | | UInt32Array, |
3313 | | &DataType::UInt32, |
3314 | | vec![Some(1_u32), Some(2_u32), Some(3_u32), None, Some(5_u32)] |
3315 | | ); |
3316 | | // u64 |
3317 | | generate_cast_test_case!( |
3318 | | array, |
3319 | | UInt64Array, |
3320 | | &DataType::UInt64, |
3321 | | vec![Some(1_u64), Some(2_u64), Some(3_u64), None, Some(5_u64)] |
3322 | | ); |
3323 | | // i8 |
3324 | | generate_cast_test_case!( |
3325 | | array, |
3326 | | Int8Array, |
3327 | | &DataType::Int8, |
3328 | | vec![Some(1_i8), Some(2_i8), Some(3_i8), None, Some(5_i8)] |
3329 | | ); |
3330 | | // i16 |
3331 | | generate_cast_test_case!( |
3332 | | array, |
3333 | | Int16Array, |
3334 | | &DataType::Int16, |
3335 | | vec![Some(1_i16), Some(2_i16), Some(3_i16), None, Some(5_i16)] |
3336 | | ); |
3337 | | // i32 |
3338 | | generate_cast_test_case!( |
3339 | | array, |
3340 | | Int32Array, |
3341 | | &DataType::Int32, |
3342 | | vec![Some(1_i32), Some(2_i32), Some(3_i32), None, Some(5_i32)] |
3343 | | ); |
3344 | | // i64 |
3345 | | generate_cast_test_case!( |
3346 | | array, |
3347 | | Int64Array, |
3348 | | &DataType::Int64, |
3349 | | vec![Some(1_i64), Some(2_i64), Some(3_i64), None, Some(5_i64)] |
3350 | | ); |
3351 | | // f32 |
3352 | | generate_cast_test_case!( |
3353 | | array, |
3354 | | Float32Array, |
3355 | | &DataType::Float32, |
3356 | | vec![ |
3357 | | Some(1.25_f32), |
3358 | | Some(2.25_f32), |
3359 | | Some(3.25_f32), |
3360 | | None, |
3361 | | Some(5.25_f32) |
3362 | | ] |
3363 | | ); |
3364 | | // f64 |
3365 | | generate_cast_test_case!( |
3366 | | array, |
3367 | | Float64Array, |
3368 | | &DataType::Float64, |
3369 | | vec![ |
3370 | | Some(1.25_f64), |
3371 | | Some(2.25_f64), |
3372 | | Some(3.25_f64), |
3373 | | None, |
3374 | | Some(5.25_f64) |
3375 | | ] |
3376 | | ); |
3377 | | } |
3378 | | |
3379 | | #[test] |
3380 | | fn test_cast_decimal32_to_numeric() { |
3381 | | let value_array: Vec<Option<i32>> = vec![Some(125), Some(225), Some(325), None, Some(525)]; |
3382 | | let array = create_decimal32_array(value_array, 8, 2).unwrap(); |
3383 | | |
3384 | | generate_decimal_to_numeric_cast_test_case(&array); |
3385 | | } |
3386 | | |
3387 | | #[test] |
3388 | | fn test_cast_decimal64_to_numeric() { |
3389 | | let value_array: Vec<Option<i64>> = vec![Some(125), Some(225), Some(325), None, Some(525)]; |
3390 | | let array = create_decimal64_array(value_array, 8, 2).unwrap(); |
3391 | | |
3392 | | generate_decimal_to_numeric_cast_test_case(&array); |
3393 | | } |
3394 | | |
3395 | | #[test] |
3396 | | fn test_cast_decimal128_to_numeric() { |
3397 | | let value_array: Vec<Option<i128>> = vec![Some(125), Some(225), Some(325), None, Some(525)]; |
3398 | | let array = create_decimal128_array(value_array, 38, 2).unwrap(); |
3399 | | |
3400 | | generate_decimal_to_numeric_cast_test_case(&array); |
3401 | | |
3402 | | // overflow test: out of range of max u8 |
3403 | | let value_array: Vec<Option<i128>> = vec![Some(51300)]; |
3404 | | let array = create_decimal128_array(value_array, 38, 2).unwrap(); |
3405 | | let casted_array = cast_with_options( |
3406 | | &array, |
3407 | | &DataType::UInt8, |
3408 | | &CastOptions { |
3409 | | safe: false, |
3410 | | format_options: FormatOptions::default(), |
3411 | | }, |
3412 | | ); |
3413 | | assert_eq!( |
3414 | | "Cast error: value of 513 is out of range UInt8".to_string(), |
3415 | | casted_array.unwrap_err().to_string() |
3416 | | ); |
3417 | | |
3418 | | let casted_array = cast_with_options( |
3419 | | &array, |
3420 | | &DataType::UInt8, |
3421 | | &CastOptions { |
3422 | | safe: true, |
3423 | | format_options: FormatOptions::default(), |
3424 | | }, |
3425 | | ); |
3426 | | assert!(casted_array.is_ok()); |
3427 | | assert!(casted_array.unwrap().is_null(0)); |
3428 | | |
3429 | | // overflow test: out of range of max i8 |
3430 | | let value_array: Vec<Option<i128>> = vec![Some(24400)]; |
3431 | | let array = create_decimal128_array(value_array, 38, 2).unwrap(); |
3432 | | let casted_array = cast_with_options( |
3433 | | &array, |
3434 | | &DataType::Int8, |
3435 | | &CastOptions { |
3436 | | safe: false, |
3437 | | format_options: FormatOptions::default(), |
3438 | | }, |
3439 | | ); |
3440 | | assert_eq!( |
3441 | | "Cast error: value of 244 is out of range Int8".to_string(), |
3442 | | casted_array.unwrap_err().to_string() |
3443 | | ); |
3444 | | |
3445 | | let casted_array = cast_with_options( |
3446 | | &array, |
3447 | | &DataType::Int8, |
3448 | | &CastOptions { |
3449 | | safe: true, |
3450 | | format_options: FormatOptions::default(), |
3451 | | }, |
3452 | | ); |
3453 | | assert!(casted_array.is_ok()); |
3454 | | assert!(casted_array.unwrap().is_null(0)); |
3455 | | |
3456 | | // loss the precision: convert decimal to f32、f64 |
3457 | | // f32 |
3458 | | // 112345678_f32 and 112345679_f32 are same, so the 112345679_f32 will lose precision. |
3459 | | let value_array: Vec<Option<i128>> = vec![ |
3460 | | Some(125), |
3461 | | Some(225), |
3462 | | Some(325), |
3463 | | None, |
3464 | | Some(525), |
3465 | | Some(112345678), |
3466 | | Some(112345679), |
3467 | | ]; |
3468 | | let array = create_decimal128_array(value_array, 38, 2).unwrap(); |
3469 | | generate_cast_test_case!( |
3470 | | &array, |
3471 | | Float32Array, |
3472 | | &DataType::Float32, |
3473 | | vec![ |
3474 | | Some(1.25_f32), |
3475 | | Some(2.25_f32), |
3476 | | Some(3.25_f32), |
3477 | | None, |
3478 | | Some(5.25_f32), |
3479 | | Some(1_123_456.7_f32), |
3480 | | Some(1_123_456.7_f32) |
3481 | | ] |
3482 | | ); |
3483 | | |
3484 | | // f64 |
3485 | | // 112345678901234568_f64 and 112345678901234560_f64 are same, so the 112345678901234568_f64 will lose precision. |
3486 | | let value_array: Vec<Option<i128>> = vec![ |
3487 | | Some(125), |
3488 | | Some(225), |
3489 | | Some(325), |
3490 | | None, |
3491 | | Some(525), |
3492 | | Some(112345678901234568), |
3493 | | Some(112345678901234560), |
3494 | | ]; |
3495 | | let array = create_decimal128_array(value_array, 38, 2).unwrap(); |
3496 | | generate_cast_test_case!( |
3497 | | &array, |
3498 | | Float64Array, |
3499 | | &DataType::Float64, |
3500 | | vec![ |
3501 | | Some(1.25_f64), |
3502 | | Some(2.25_f64), |
3503 | | Some(3.25_f64), |
3504 | | None, |
3505 | | Some(5.25_f64), |
3506 | | Some(1_123_456_789_012_345.6_f64), |
3507 | | Some(1_123_456_789_012_345.6_f64), |
3508 | | ] |
3509 | | ); |
3510 | | } |
3511 | | |
3512 | | #[test] |
3513 | | fn test_cast_decimal256_to_numeric() { |
3514 | | let value_array: Vec<Option<i256>> = vec![ |
3515 | | Some(i256::from_i128(125)), |
3516 | | Some(i256::from_i128(225)), |
3517 | | Some(i256::from_i128(325)), |
3518 | | None, |
3519 | | Some(i256::from_i128(525)), |
3520 | | ]; |
3521 | | let array = create_decimal256_array(value_array, 38, 2).unwrap(); |
3522 | | // u8 |
3523 | | generate_cast_test_case!( |
3524 | | &array, |
3525 | | UInt8Array, |
3526 | | &DataType::UInt8, |
3527 | | vec![Some(1_u8), Some(2_u8), Some(3_u8), None, Some(5_u8)] |
3528 | | ); |
3529 | | // u16 |
3530 | | generate_cast_test_case!( |
3531 | | &array, |
3532 | | UInt16Array, |
3533 | | &DataType::UInt16, |
3534 | | vec![Some(1_u16), Some(2_u16), Some(3_u16), None, Some(5_u16)] |
3535 | | ); |
3536 | | // u32 |
3537 | | generate_cast_test_case!( |
3538 | | &array, |
3539 | | UInt32Array, |
3540 | | &DataType::UInt32, |
3541 | | vec![Some(1_u32), Some(2_u32), Some(3_u32), None, Some(5_u32)] |
3542 | | ); |
3543 | | // u64 |
3544 | | generate_cast_test_case!( |
3545 | | &array, |
3546 | | UInt64Array, |
3547 | | &DataType::UInt64, |
3548 | | vec![Some(1_u64), Some(2_u64), Some(3_u64), None, Some(5_u64)] |
3549 | | ); |
3550 | | // i8 |
3551 | | generate_cast_test_case!( |
3552 | | &array, |
3553 | | Int8Array, |
3554 | | &DataType::Int8, |
3555 | | vec![Some(1_i8), Some(2_i8), Some(3_i8), None, Some(5_i8)] |
3556 | | ); |
3557 | | // i16 |
3558 | | generate_cast_test_case!( |
3559 | | &array, |
3560 | | Int16Array, |
3561 | | &DataType::Int16, |
3562 | | vec![Some(1_i16), Some(2_i16), Some(3_i16), None, Some(5_i16)] |
3563 | | ); |
3564 | | // i32 |
3565 | | generate_cast_test_case!( |
3566 | | &array, |
3567 | | Int32Array, |
3568 | | &DataType::Int32, |
3569 | | vec![Some(1_i32), Some(2_i32), Some(3_i32), None, Some(5_i32)] |
3570 | | ); |
3571 | | // i64 |
3572 | | generate_cast_test_case!( |
3573 | | &array, |
3574 | | Int64Array, |
3575 | | &DataType::Int64, |
3576 | | vec![Some(1_i64), Some(2_i64), Some(3_i64), None, Some(5_i64)] |
3577 | | ); |
3578 | | // f32 |
3579 | | generate_cast_test_case!( |
3580 | | &array, |
3581 | | Float32Array, |
3582 | | &DataType::Float32, |
3583 | | vec![ |
3584 | | Some(1.25_f32), |
3585 | | Some(2.25_f32), |
3586 | | Some(3.25_f32), |
3587 | | None, |
3588 | | Some(5.25_f32) |
3589 | | ] |
3590 | | ); |
3591 | | // f64 |
3592 | | generate_cast_test_case!( |
3593 | | &array, |
3594 | | Float64Array, |
3595 | | &DataType::Float64, |
3596 | | vec![ |
3597 | | Some(1.25_f64), |
3598 | | Some(2.25_f64), |
3599 | | Some(3.25_f64), |
3600 | | None, |
3601 | | Some(5.25_f64) |
3602 | | ] |
3603 | | ); |
3604 | | |
3605 | | // overflow test: out of range of max i8 |
3606 | | let value_array: Vec<Option<i256>> = vec![Some(i256::from_i128(24400))]; |
3607 | | let array = create_decimal256_array(value_array, 38, 2).unwrap(); |
3608 | | let casted_array = cast_with_options( |
3609 | | &array, |
3610 | | &DataType::Int8, |
3611 | | &CastOptions { |
3612 | | safe: false, |
3613 | | format_options: FormatOptions::default(), |
3614 | | }, |
3615 | | ); |
3616 | | assert_eq!( |
3617 | | "Cast error: value of 244 is out of range Int8".to_string(), |
3618 | | casted_array.unwrap_err().to_string() |
3619 | | ); |
3620 | | |
3621 | | let casted_array = cast_with_options( |
3622 | | &array, |
3623 | | &DataType::Int8, |
3624 | | &CastOptions { |
3625 | | safe: true, |
3626 | | format_options: FormatOptions::default(), |
3627 | | }, |
3628 | | ); |
3629 | | assert!(casted_array.is_ok()); |
3630 | | assert!(casted_array.unwrap().is_null(0)); |
3631 | | |
3632 | | // loss the precision: convert decimal to f32、f64 |
3633 | | // f32 |
3634 | | // 112345678_f32 and 112345679_f32 are same, so the 112345679_f32 will lose precision. |
3635 | | let value_array: Vec<Option<i256>> = vec![ |
3636 | | Some(i256::from_i128(125)), |
3637 | | Some(i256::from_i128(225)), |
3638 | | Some(i256::from_i128(325)), |
3639 | | None, |
3640 | | Some(i256::from_i128(525)), |
3641 | | Some(i256::from_i128(112345678)), |
3642 | | Some(i256::from_i128(112345679)), |
3643 | | ]; |
3644 | | let array = create_decimal256_array(value_array, 76, 2).unwrap(); |
3645 | | generate_cast_test_case!( |
3646 | | &array, |
3647 | | Float32Array, |
3648 | | &DataType::Float32, |
3649 | | vec![ |
3650 | | Some(1.25_f32), |
3651 | | Some(2.25_f32), |
3652 | | Some(3.25_f32), |
3653 | | None, |
3654 | | Some(5.25_f32), |
3655 | | Some(1_123_456.7_f32), |
3656 | | Some(1_123_456.7_f32) |
3657 | | ] |
3658 | | ); |
3659 | | |
3660 | | // f64 |
3661 | | // 112345678901234568_f64 and 112345678901234560_f64 are same, so the 112345678901234568_f64 will lose precision. |
3662 | | let value_array: Vec<Option<i256>> = vec![ |
3663 | | Some(i256::from_i128(125)), |
3664 | | Some(i256::from_i128(225)), |
3665 | | Some(i256::from_i128(325)), |
3666 | | None, |
3667 | | Some(i256::from_i128(525)), |
3668 | | Some(i256::from_i128(112345678901234568)), |
3669 | | Some(i256::from_i128(112345678901234560)), |
3670 | | ]; |
3671 | | let array = create_decimal256_array(value_array, 76, 2).unwrap(); |
3672 | | generate_cast_test_case!( |
3673 | | &array, |
3674 | | Float64Array, |
3675 | | &DataType::Float64, |
3676 | | vec![ |
3677 | | Some(1.25_f64), |
3678 | | Some(2.25_f64), |
3679 | | Some(3.25_f64), |
3680 | | None, |
3681 | | Some(5.25_f64), |
3682 | | Some(1_123_456_789_012_345.6_f64), |
3683 | | Some(1_123_456_789_012_345.6_f64), |
3684 | | ] |
3685 | | ); |
3686 | | } |
3687 | | |
3688 | | #[test] |
3689 | | fn test_cast_numeric_to_decimal128() { |
3690 | | let decimal_type = DataType::Decimal128(38, 6); |
3691 | | // u8, u16, u32, u64 |
3692 | | let input_datas = vec![ |
3693 | | Arc::new(UInt8Array::from(vec![ |
3694 | | Some(1), |
3695 | | Some(2), |
3696 | | Some(3), |
3697 | | None, |
3698 | | Some(5), |
3699 | | ])) as ArrayRef, // u8 |
3700 | | Arc::new(UInt16Array::from(vec![ |
3701 | | Some(1), |
3702 | | Some(2), |
3703 | | Some(3), |
3704 | | None, |
3705 | | Some(5), |
3706 | | ])) as ArrayRef, // u16 |
3707 | | Arc::new(UInt32Array::from(vec![ |
3708 | | Some(1), |
3709 | | Some(2), |
3710 | | Some(3), |
3711 | | None, |
3712 | | Some(5), |
3713 | | ])) as ArrayRef, // u32 |
3714 | | Arc::new(UInt64Array::from(vec![ |
3715 | | Some(1), |
3716 | | Some(2), |
3717 | | Some(3), |
3718 | | None, |
3719 | | Some(5), |
3720 | | ])) as ArrayRef, // u64 |
3721 | | ]; |
3722 | | |
3723 | | for array in input_datas { |
3724 | | generate_cast_test_case!( |
3725 | | &array, |
3726 | | Decimal128Array, |
3727 | | &decimal_type, |
3728 | | vec![ |
3729 | | Some(1000000_i128), |
3730 | | Some(2000000_i128), |
3731 | | Some(3000000_i128), |
3732 | | None, |
3733 | | Some(5000000_i128) |
3734 | | ] |
3735 | | ); |
3736 | | } |
3737 | | |
3738 | | // i8, i16, i32, i64 |
3739 | | let input_datas = vec![ |
3740 | | Arc::new(Int8Array::from(vec![ |
3741 | | Some(1), |
3742 | | Some(2), |
3743 | | Some(3), |
3744 | | None, |
3745 | | Some(5), |
3746 | | ])) as ArrayRef, // i8 |
3747 | | Arc::new(Int16Array::from(vec![ |
3748 | | Some(1), |
3749 | | Some(2), |
3750 | | Some(3), |
3751 | | None, |
3752 | | Some(5), |
3753 | | ])) as ArrayRef, // i16 |
3754 | | Arc::new(Int32Array::from(vec![ |
3755 | | Some(1), |
3756 | | Some(2), |
3757 | | Some(3), |
3758 | | None, |
3759 | | Some(5), |
3760 | | ])) as ArrayRef, // i32 |
3761 | | Arc::new(Int64Array::from(vec![ |
3762 | | Some(1), |
3763 | | Some(2), |
3764 | | Some(3), |
3765 | | None, |
3766 | | Some(5), |
3767 | | ])) as ArrayRef, // i64 |
3768 | | ]; |
3769 | | for array in input_datas { |
3770 | | generate_cast_test_case!( |
3771 | | &array, |
3772 | | Decimal128Array, |
3773 | | &decimal_type, |
3774 | | vec![ |
3775 | | Some(1000000_i128), |
3776 | | Some(2000000_i128), |
3777 | | Some(3000000_i128), |
3778 | | None, |
3779 | | Some(5000000_i128) |
3780 | | ] |
3781 | | ); |
3782 | | } |
3783 | | |
3784 | | // test u8 to decimal type with overflow the result type |
3785 | | // the 100 will be converted to 1000_i128, but it is out of range for max value in the precision 3. |
3786 | | let array = UInt8Array::from(vec![1, 2, 3, 4, 100]); |
3787 | | let casted_array = cast(&array, &DataType::Decimal128(3, 1)); |
3788 | | assert!(casted_array.is_ok()); |
3789 | | let array = casted_array.unwrap(); |
3790 | | let array: &Decimal128Array = array.as_primitive(); |
3791 | | assert!(array.is_null(4)); |
3792 | | |
3793 | | // test i8 to decimal type with overflow the result type |
3794 | | // the 100 will be converted to 1000_i128, but it is out of range for max value in the precision 3. |
3795 | | let array = Int8Array::from(vec![1, 2, 3, 4, 100]); |
3796 | | let casted_array = cast(&array, &DataType::Decimal128(3, 1)); |
3797 | | assert!(casted_array.is_ok()); |
3798 | | let array = casted_array.unwrap(); |
3799 | | let array: &Decimal128Array = array.as_primitive(); |
3800 | | assert!(array.is_null(4)); |
3801 | | |
3802 | | // test f32 to decimal type |
3803 | | let array = Float32Array::from(vec![ |
3804 | | Some(1.1), |
3805 | | Some(2.2), |
3806 | | Some(4.4), |
3807 | | None, |
3808 | | Some(1.123_456_4), // round down |
3809 | | Some(1.123_456_7), // round up |
3810 | | ]); |
3811 | | let array = Arc::new(array) as ArrayRef; |
3812 | | generate_cast_test_case!( |
3813 | | &array, |
3814 | | Decimal128Array, |
3815 | | &decimal_type, |
3816 | | vec![ |
3817 | | Some(1100000_i128), |
3818 | | Some(2200000_i128), |
3819 | | Some(4400000_i128), |
3820 | | None, |
3821 | | Some(1123456_i128), // round down |
3822 | | Some(1123457_i128), // round up |
3823 | | ] |
3824 | | ); |
3825 | | |
3826 | | // test f64 to decimal type |
3827 | | let array = Float64Array::from(vec![ |
3828 | | Some(1.1), |
3829 | | Some(2.2), |
3830 | | Some(4.4), |
3831 | | None, |
3832 | | Some(1.123_456_489_123_4), // round up |
3833 | | Some(1.123_456_789_123_4), // round up |
3834 | | Some(1.123_456_489_012_345_6), // round down |
3835 | | Some(1.123_456_789_012_345_6), // round up |
3836 | | ]); |
3837 | | generate_cast_test_case!( |
3838 | | &array, |
3839 | | Decimal128Array, |
3840 | | &decimal_type, |
3841 | | vec![ |
3842 | | Some(1100000_i128), |
3843 | | Some(2200000_i128), |
3844 | | Some(4400000_i128), |
3845 | | None, |
3846 | | Some(1123456_i128), // round down |
3847 | | Some(1123457_i128), // round up |
3848 | | Some(1123456_i128), // round down |
3849 | | Some(1123457_i128), // round up |
3850 | | ] |
3851 | | ); |
3852 | | } |
3853 | | |
3854 | | #[test] |
3855 | | fn test_cast_numeric_to_decimal256() { |
3856 | | let decimal_type = DataType::Decimal256(76, 6); |
3857 | | // u8, u16, u32, u64 |
3858 | | let input_datas = vec![ |
3859 | | Arc::new(UInt8Array::from(vec![ |
3860 | | Some(1), |
3861 | | Some(2), |
3862 | | Some(3), |
3863 | | None, |
3864 | | Some(5), |
3865 | | ])) as ArrayRef, // u8 |
3866 | | Arc::new(UInt16Array::from(vec![ |
3867 | | Some(1), |
3868 | | Some(2), |
3869 | | Some(3), |
3870 | | None, |
3871 | | Some(5), |
3872 | | ])) as ArrayRef, // u16 |
3873 | | Arc::new(UInt32Array::from(vec![ |
3874 | | Some(1), |
3875 | | Some(2), |
3876 | | Some(3), |
3877 | | None, |
3878 | | Some(5), |
3879 | | ])) as ArrayRef, // u32 |
3880 | | Arc::new(UInt64Array::from(vec![ |
3881 | | Some(1), |
3882 | | Some(2), |
3883 | | Some(3), |
3884 | | None, |
3885 | | Some(5), |
3886 | | ])) as ArrayRef, // u64 |
3887 | | ]; |
3888 | | |
3889 | | for array in input_datas { |
3890 | | generate_cast_test_case!( |
3891 | | &array, |
3892 | | Decimal256Array, |
3893 | | &decimal_type, |
3894 | | vec![ |
3895 | | Some(i256::from_i128(1000000_i128)), |
3896 | | Some(i256::from_i128(2000000_i128)), |
3897 | | Some(i256::from_i128(3000000_i128)), |
3898 | | None, |
3899 | | Some(i256::from_i128(5000000_i128)) |
3900 | | ] |
3901 | | ); |
3902 | | } |
3903 | | |
3904 | | // i8, i16, i32, i64 |
3905 | | let input_datas = vec![ |
3906 | | Arc::new(Int8Array::from(vec![ |
3907 | | Some(1), |
3908 | | Some(2), |
3909 | | Some(3), |
3910 | | None, |
3911 | | Some(5), |
3912 | | ])) as ArrayRef, // i8 |
3913 | | Arc::new(Int16Array::from(vec![ |
3914 | | Some(1), |
3915 | | Some(2), |
3916 | | Some(3), |
3917 | | None, |
3918 | | Some(5), |
3919 | | ])) as ArrayRef, // i16 |
3920 | | Arc::new(Int32Array::from(vec![ |
3921 | | Some(1), |
3922 | | Some(2), |
3923 | | Some(3), |
3924 | | None, |
3925 | | Some(5), |
3926 | | ])) as ArrayRef, // i32 |
3927 | | Arc::new(Int64Array::from(vec![ |
3928 | | Some(1), |
3929 | | Some(2), |
3930 | | Some(3), |
3931 | | None, |
3932 | | Some(5), |
3933 | | ])) as ArrayRef, // i64 |
3934 | | ]; |
3935 | | for array in input_datas { |
3936 | | generate_cast_test_case!( |
3937 | | &array, |
3938 | | Decimal256Array, |
3939 | | &decimal_type, |
3940 | | vec![ |
3941 | | Some(i256::from_i128(1000000_i128)), |
3942 | | Some(i256::from_i128(2000000_i128)), |
3943 | | Some(i256::from_i128(3000000_i128)), |
3944 | | None, |
3945 | | Some(i256::from_i128(5000000_i128)) |
3946 | | ] |
3947 | | ); |
3948 | | } |
3949 | | |
3950 | | // test i8 to decimal type with overflow the result type |
3951 | | // the 100 will be converted to 1000_i128, but it is out of range for max value in the precision 3. |
3952 | | let array = Int8Array::from(vec![1, 2, 3, 4, 100]); |
3953 | | let array = Arc::new(array) as ArrayRef; |
3954 | | let casted_array = cast(&array, &DataType::Decimal256(3, 1)); |
3955 | | assert!(casted_array.is_ok()); |
3956 | | let array = casted_array.unwrap(); |
3957 | | let array: &Decimal256Array = array.as_primitive(); |
3958 | | assert!(array.is_null(4)); |
3959 | | |
3960 | | // test f32 to decimal type |
3961 | | let array = Float32Array::from(vec![ |
3962 | | Some(1.1), |
3963 | | Some(2.2), |
3964 | | Some(4.4), |
3965 | | None, |
3966 | | Some(1.123_456_4), // round down |
3967 | | Some(1.123_456_7), // round up |
3968 | | ]); |
3969 | | generate_cast_test_case!( |
3970 | | &array, |
3971 | | Decimal256Array, |
3972 | | &decimal_type, |
3973 | | vec![ |
3974 | | Some(i256::from_i128(1100000_i128)), |
3975 | | Some(i256::from_i128(2200000_i128)), |
3976 | | Some(i256::from_i128(4400000_i128)), |
3977 | | None, |
3978 | | Some(i256::from_i128(1123456_i128)), // round down |
3979 | | Some(i256::from_i128(1123457_i128)), // round up |
3980 | | ] |
3981 | | ); |
3982 | | |
3983 | | // test f64 to decimal type |
3984 | | let array = Float64Array::from(vec![ |
3985 | | Some(1.1), |
3986 | | Some(2.2), |
3987 | | Some(4.4), |
3988 | | None, |
3989 | | Some(1.123_456_489_123_4), // round down |
3990 | | Some(1.123_456_789_123_4), // round up |
3991 | | Some(1.123_456_489_012_345_6), // round down |
3992 | | Some(1.123_456_789_012_345_6), // round up |
3993 | | ]); |
3994 | | generate_cast_test_case!( |
3995 | | &array, |
3996 | | Decimal256Array, |
3997 | | &decimal_type, |
3998 | | vec![ |
3999 | | Some(i256::from_i128(1100000_i128)), |
4000 | | Some(i256::from_i128(2200000_i128)), |
4001 | | Some(i256::from_i128(4400000_i128)), |
4002 | | None, |
4003 | | Some(i256::from_i128(1123456_i128)), // round down |
4004 | | Some(i256::from_i128(1123457_i128)), // round up |
4005 | | Some(i256::from_i128(1123456_i128)), // round down |
4006 | | Some(i256::from_i128(1123457_i128)), // round up |
4007 | | ] |
4008 | | ); |
4009 | | } |
4010 | | |
4011 | | #[test] |
4012 | | fn test_cast_i32_to_f64() { |
4013 | | let array = Int32Array::from(vec![5, 6, 7, 8, 9]); |
4014 | | let b = cast(&array, &DataType::Float64).unwrap(); |
4015 | | let c = b.as_primitive::<Float64Type>(); |
4016 | | assert_eq!(5.0, c.value(0)); |
4017 | | assert_eq!(6.0, c.value(1)); |
4018 | | assert_eq!(7.0, c.value(2)); |
4019 | | assert_eq!(8.0, c.value(3)); |
4020 | | assert_eq!(9.0, c.value(4)); |
4021 | | } |
4022 | | |
4023 | | #[test] |
4024 | | fn test_cast_i32_to_u8() { |
4025 | | let array = Int32Array::from(vec![-5, 6, -7, 8, 100000000]); |
4026 | | let b = cast(&array, &DataType::UInt8).unwrap(); |
4027 | | let c = b.as_primitive::<UInt8Type>(); |
4028 | | assert!(!c.is_valid(0)); |
4029 | | assert_eq!(6, c.value(1)); |
4030 | | assert!(!c.is_valid(2)); |
4031 | | assert_eq!(8, c.value(3)); |
4032 | | // overflows return None |
4033 | | assert!(!c.is_valid(4)); |
4034 | | } |
4035 | | |
4036 | | #[test] |
4037 | | #[should_panic(expected = "Can't cast value -5 to type UInt8")] |
4038 | | fn test_cast_int32_to_u8_with_error() { |
4039 | | let array = Int32Array::from(vec![-5, 6, -7, 8, 100000000]); |
4040 | | // overflow with the error |
4041 | | let cast_option = CastOptions { |
4042 | | safe: false, |
4043 | | format_options: FormatOptions::default(), |
4044 | | }; |
4045 | | let result = cast_with_options(&array, &DataType::UInt8, &cast_option); |
4046 | | assert!(result.is_err()); |
4047 | | result.unwrap(); |
4048 | | } |
4049 | | |
4050 | | #[test] |
4051 | | fn test_cast_i32_to_u8_sliced() { |
4052 | | let array = Int32Array::from(vec![-5, 6, -7, 8, 100000000]); |
4053 | | assert_eq!(0, array.offset()); |
4054 | | let array = array.slice(2, 3); |
4055 | | let b = cast(&array, &DataType::UInt8).unwrap(); |
4056 | | assert_eq!(3, b.len()); |
4057 | | let c = b.as_primitive::<UInt8Type>(); |
4058 | | assert!(!c.is_valid(0)); |
4059 | | assert_eq!(8, c.value(1)); |
4060 | | // overflows return None |
4061 | | assert!(!c.is_valid(2)); |
4062 | | } |
4063 | | |
4064 | | #[test] |
4065 | | fn test_cast_i32_to_i32() { |
4066 | | let array = Int32Array::from(vec![5, 6, 7, 8, 9]); |
4067 | | let b = cast(&array, &DataType::Int32).unwrap(); |
4068 | | let c = b.as_primitive::<Int32Type>(); |
4069 | | assert_eq!(5, c.value(0)); |
4070 | | assert_eq!(6, c.value(1)); |
4071 | | assert_eq!(7, c.value(2)); |
4072 | | assert_eq!(8, c.value(3)); |
4073 | | assert_eq!(9, c.value(4)); |
4074 | | } |
4075 | | |
4076 | | #[test] |
4077 | | fn test_cast_i32_to_list_i32() { |
4078 | | let array = Int32Array::from(vec![5, 6, 7, 8, 9]); |
4079 | | let b = cast( |
4080 | | &array, |
4081 | | &DataType::List(Arc::new(Field::new_list_field(DataType::Int32, true))), |
4082 | | ) |
4083 | | .unwrap(); |
4084 | | assert_eq!(5, b.len()); |
4085 | | let arr = b.as_list::<i32>(); |
4086 | | assert_eq!(&[0, 1, 2, 3, 4, 5], arr.value_offsets()); |
4087 | | assert_eq!(1, arr.value_length(0)); |
4088 | | assert_eq!(1, arr.value_length(1)); |
4089 | | assert_eq!(1, arr.value_length(2)); |
4090 | | assert_eq!(1, arr.value_length(3)); |
4091 | | assert_eq!(1, arr.value_length(4)); |
4092 | | let c = arr.values().as_primitive::<Int32Type>(); |
4093 | | assert_eq!(5, c.value(0)); |
4094 | | assert_eq!(6, c.value(1)); |
4095 | | assert_eq!(7, c.value(2)); |
4096 | | assert_eq!(8, c.value(3)); |
4097 | | assert_eq!(9, c.value(4)); |
4098 | | } |
4099 | | |
4100 | | #[test] |
4101 | | fn test_cast_i32_to_list_i32_nullable() { |
4102 | | let array = Int32Array::from(vec![Some(5), None, Some(7), Some(8), Some(9)]); |
4103 | | let b = cast( |
4104 | | &array, |
4105 | | &DataType::List(Arc::new(Field::new_list_field(DataType::Int32, true))), |
4106 | | ) |
4107 | | .unwrap(); |
4108 | | assert_eq!(5, b.len()); |
4109 | | assert_eq!(0, b.null_count()); |
4110 | | let arr = b.as_list::<i32>(); |
4111 | | assert_eq!(&[0, 1, 2, 3, 4, 5], arr.value_offsets()); |
4112 | | assert_eq!(1, arr.value_length(0)); |
4113 | | assert_eq!(1, arr.value_length(1)); |
4114 | | assert_eq!(1, arr.value_length(2)); |
4115 | | assert_eq!(1, arr.value_length(3)); |
4116 | | assert_eq!(1, arr.value_length(4)); |
4117 | | |
4118 | | let c = arr.values().as_primitive::<Int32Type>(); |
4119 | | assert_eq!(1, c.null_count()); |
4120 | | assert_eq!(5, c.value(0)); |
4121 | | assert!(!c.is_valid(1)); |
4122 | | assert_eq!(7, c.value(2)); |
4123 | | assert_eq!(8, c.value(3)); |
4124 | | assert_eq!(9, c.value(4)); |
4125 | | } |
4126 | | |
4127 | | #[test] |
4128 | | fn test_cast_i32_to_list_f64_nullable_sliced() { |
4129 | | let array = Int32Array::from(vec![Some(5), None, Some(7), Some(8), None, Some(10)]); |
4130 | | let array = array.slice(2, 4); |
4131 | | let b = cast( |
4132 | | &array, |
4133 | | &DataType::List(Arc::new(Field::new_list_field(DataType::Float64, true))), |
4134 | | ) |
4135 | | .unwrap(); |
4136 | | assert_eq!(4, b.len()); |
4137 | | assert_eq!(0, b.null_count()); |
4138 | | let arr = b.as_list::<i32>(); |
4139 | | assert_eq!(&[0, 1, 2, 3, 4], arr.value_offsets()); |
4140 | | assert_eq!(1, arr.value_length(0)); |
4141 | | assert_eq!(1, arr.value_length(1)); |
4142 | | assert_eq!(1, arr.value_length(2)); |
4143 | | assert_eq!(1, arr.value_length(3)); |
4144 | | let c = arr.values().as_primitive::<Float64Type>(); |
4145 | | assert_eq!(1, c.null_count()); |
4146 | | assert_eq!(7.0, c.value(0)); |
4147 | | assert_eq!(8.0, c.value(1)); |
4148 | | assert!(!c.is_valid(2)); |
4149 | | assert_eq!(10.0, c.value(3)); |
4150 | | } |
4151 | | |
4152 | | #[test] |
4153 | | fn test_cast_int_to_utf8view() { |
4154 | | let inputs = vec![ |
4155 | | Arc::new(Int8Array::from(vec![None, Some(8), Some(9), Some(10)])) as ArrayRef, |
4156 | | Arc::new(Int16Array::from(vec![None, Some(8), Some(9), Some(10)])) as ArrayRef, |
4157 | | Arc::new(Int32Array::from(vec![None, Some(8), Some(9), Some(10)])) as ArrayRef, |
4158 | | Arc::new(Int64Array::from(vec![None, Some(8), Some(9), Some(10)])) as ArrayRef, |
4159 | | Arc::new(UInt8Array::from(vec![None, Some(8), Some(9), Some(10)])) as ArrayRef, |
4160 | | Arc::new(UInt16Array::from(vec![None, Some(8), Some(9), Some(10)])) as ArrayRef, |
4161 | | Arc::new(UInt32Array::from(vec![None, Some(8), Some(9), Some(10)])) as ArrayRef, |
4162 | | Arc::new(UInt64Array::from(vec![None, Some(8), Some(9), Some(10)])) as ArrayRef, |
4163 | | ]; |
4164 | | let expected: ArrayRef = Arc::new(StringViewArray::from(vec![ |
4165 | | None, |
4166 | | Some("8"), |
4167 | | Some("9"), |
4168 | | Some("10"), |
4169 | | ])); |
4170 | | |
4171 | | for array in inputs { |
4172 | | assert!(can_cast_types(array.data_type(), &DataType::Utf8View)); |
4173 | | let arr = cast(&array, &DataType::Utf8View).unwrap(); |
4174 | | assert_eq!(expected.as_ref(), arr.as_ref()); |
4175 | | } |
4176 | | } |
4177 | | |
4178 | | #[test] |
4179 | | fn test_cast_float_to_utf8view() { |
4180 | | let inputs = vec![ |
4181 | | Arc::new(Float16Array::from(vec![ |
4182 | | Some(f16::from_f64(1.5)), |
4183 | | Some(f16::from_f64(2.5)), |
4184 | | None, |
4185 | | ])) as ArrayRef, |
4186 | | Arc::new(Float32Array::from(vec![Some(1.5), Some(2.5), None])) as ArrayRef, |
4187 | | Arc::new(Float64Array::from(vec![Some(1.5), Some(2.5), None])) as ArrayRef, |
4188 | | ]; |
4189 | | |
4190 | | let expected: ArrayRef = |
4191 | | Arc::new(StringViewArray::from(vec![Some("1.5"), Some("2.5"), None])); |
4192 | | |
4193 | | for array in inputs { |
4194 | | assert!(can_cast_types(array.data_type(), &DataType::Utf8View)); |
4195 | | let arr = cast(&array, &DataType::Utf8View).unwrap(); |
4196 | | assert_eq!(expected.as_ref(), arr.as_ref()); |
4197 | | } |
4198 | | } |
4199 | | |
4200 | | #[test] |
4201 | | fn test_cast_utf8_to_i32() { |
4202 | | let array = StringArray::from(vec!["5", "6", "seven", "8", "9.1"]); |
4203 | | let b = cast(&array, &DataType::Int32).unwrap(); |
4204 | | let c = b.as_primitive::<Int32Type>(); |
4205 | | assert_eq!(5, c.value(0)); |
4206 | | assert_eq!(6, c.value(1)); |
4207 | | assert!(!c.is_valid(2)); |
4208 | | assert_eq!(8, c.value(3)); |
4209 | | assert!(!c.is_valid(4)); |
4210 | | } |
4211 | | |
4212 | | #[test] |
4213 | | fn test_cast_utf8view_to_i32() { |
4214 | | let array = StringViewArray::from(vec!["5", "6", "seven", "8", "9.1"]); |
4215 | | let b = cast(&array, &DataType::Int32).unwrap(); |
4216 | | let c = b.as_primitive::<Int32Type>(); |
4217 | | assert_eq!(5, c.value(0)); |
4218 | | assert_eq!(6, c.value(1)); |
4219 | | assert!(!c.is_valid(2)); |
4220 | | assert_eq!(8, c.value(3)); |
4221 | | assert!(!c.is_valid(4)); |
4222 | | } |
4223 | | |
4224 | | #[test] |
4225 | | fn test_cast_utf8view_to_f32() { |
4226 | | let array = StringViewArray::from(vec!["3", "4.56", "seven", "8.9"]); |
4227 | | let b = cast(&array, &DataType::Float32).unwrap(); |
4228 | | let c = b.as_primitive::<Float32Type>(); |
4229 | | assert_eq!(3.0, c.value(0)); |
4230 | | assert_eq!(4.56, c.value(1)); |
4231 | | assert!(!c.is_valid(2)); |
4232 | | assert_eq!(8.9, c.value(3)); |
4233 | | } |
4234 | | |
4235 | | #[test] |
4236 | | fn test_cast_utf8view_to_decimal128() { |
4237 | | let array = StringViewArray::from(vec![None, Some("4"), Some("5.6"), Some("7.89")]); |
4238 | | let arr = Arc::new(array) as ArrayRef; |
4239 | | generate_cast_test_case!( |
4240 | | &arr, |
4241 | | Decimal128Array, |
4242 | | &DataType::Decimal128(4, 2), |
4243 | | vec![None, Some(400_i128), Some(560_i128), Some(789_i128)] |
4244 | | ); |
4245 | | } |
4246 | | |
4247 | | #[test] |
4248 | | fn test_cast_with_options_utf8_to_i32() { |
4249 | | let array = StringArray::from(vec!["5", "6", "seven", "8", "9.1"]); |
4250 | | let result = cast_with_options( |
4251 | | &array, |
4252 | | &DataType::Int32, |
4253 | | &CastOptions { |
4254 | | safe: false, |
4255 | | format_options: FormatOptions::default(), |
4256 | | }, |
4257 | | ); |
4258 | | match result { |
4259 | | Ok(_) => panic!("expected error"), |
4260 | | Err(e) => { |
4261 | | assert!( |
4262 | | e.to_string() |
4263 | | .contains("Cast error: Cannot cast string 'seven' to value of Int32 type",), |
4264 | | "Error: {e}" |
4265 | | ) |
4266 | | } |
4267 | | } |
4268 | | } |
4269 | | |
4270 | | #[test] |
4271 | | fn test_cast_utf8_to_bool() { |
4272 | | let strings = StringArray::from(vec!["true", "false", "invalid", " Y ", ""]); |
4273 | | let casted = cast(&strings, &DataType::Boolean).unwrap(); |
4274 | | let expected = BooleanArray::from(vec![Some(true), Some(false), None, Some(true), None]); |
4275 | | assert_eq!(*as_boolean_array(&casted), expected); |
4276 | | } |
4277 | | |
4278 | | #[test] |
4279 | | fn test_cast_utf8view_to_bool() { |
4280 | | let strings = StringViewArray::from(vec!["true", "false", "invalid", " Y ", ""]); |
4281 | | let casted = cast(&strings, &DataType::Boolean).unwrap(); |
4282 | | let expected = BooleanArray::from(vec![Some(true), Some(false), None, Some(true), None]); |
4283 | | assert_eq!(*as_boolean_array(&casted), expected); |
4284 | | } |
4285 | | |
4286 | | #[test] |
4287 | | fn test_cast_with_options_utf8_to_bool() { |
4288 | | let strings = StringArray::from(vec!["true", "false", "invalid", " Y ", ""]); |
4289 | | let casted = cast_with_options( |
4290 | | &strings, |
4291 | | &DataType::Boolean, |
4292 | | &CastOptions { |
4293 | | safe: false, |
4294 | | format_options: FormatOptions::default(), |
4295 | | }, |
4296 | | ); |
4297 | | match casted { |
4298 | | Ok(_) => panic!("expected error"), |
4299 | | Err(e) => { |
4300 | | assert!(e |
4301 | | .to_string() |
4302 | | .contains("Cast error: Cannot cast value 'invalid' to value of Boolean type")) |
4303 | | } |
4304 | | } |
4305 | | } |
4306 | | |
4307 | | #[test] |
4308 | | fn test_cast_bool_to_i32() { |
4309 | | let array = BooleanArray::from(vec![Some(true), Some(false), None]); |
4310 | | let b = cast(&array, &DataType::Int32).unwrap(); |
4311 | | let c = b.as_primitive::<Int32Type>(); |
4312 | | assert_eq!(1, c.value(0)); |
4313 | | assert_eq!(0, c.value(1)); |
4314 | | assert!(!c.is_valid(2)); |
4315 | | } |
4316 | | |
4317 | | #[test] |
4318 | | fn test_cast_bool_to_utf8view() { |
4319 | | let array = BooleanArray::from(vec![Some(true), Some(false), None]); |
4320 | | let b = cast(&array, &DataType::Utf8View).unwrap(); |
4321 | | let c = b.as_any().downcast_ref::<StringViewArray>().unwrap(); |
4322 | | assert_eq!("true", c.value(0)); |
4323 | | assert_eq!("false", c.value(1)); |
4324 | | assert!(!c.is_valid(2)); |
4325 | | } |
4326 | | |
4327 | | #[test] |
4328 | | fn test_cast_bool_to_utf8() { |
4329 | | let array = BooleanArray::from(vec![Some(true), Some(false), None]); |
4330 | | let b = cast(&array, &DataType::Utf8).unwrap(); |
4331 | | let c = b.as_any().downcast_ref::<StringArray>().unwrap(); |
4332 | | assert_eq!("true", c.value(0)); |
4333 | | assert_eq!("false", c.value(1)); |
4334 | | assert!(!c.is_valid(2)); |
4335 | | } |
4336 | | |
4337 | | #[test] |
4338 | | fn test_cast_bool_to_large_utf8() { |
4339 | | let array = BooleanArray::from(vec![Some(true), Some(false), None]); |
4340 | | let b = cast(&array, &DataType::LargeUtf8).unwrap(); |
4341 | | let c = b.as_any().downcast_ref::<LargeStringArray>().unwrap(); |
4342 | | assert_eq!("true", c.value(0)); |
4343 | | assert_eq!("false", c.value(1)); |
4344 | | assert!(!c.is_valid(2)); |
4345 | | } |
4346 | | |
4347 | | #[test] |
4348 | | fn test_cast_bool_to_f64() { |
4349 | | let array = BooleanArray::from(vec![Some(true), Some(false), None]); |
4350 | | let b = cast(&array, &DataType::Float64).unwrap(); |
4351 | | let c = b.as_primitive::<Float64Type>(); |
4352 | | assert_eq!(1.0, c.value(0)); |
4353 | | assert_eq!(0.0, c.value(1)); |
4354 | | assert!(!c.is_valid(2)); |
4355 | | } |
4356 | | |
4357 | | #[test] |
4358 | | fn test_cast_integer_to_timestamp() { |
4359 | | let array = Int64Array::from(vec![Some(2), Some(10), None]); |
4360 | | let expected = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4361 | | |
4362 | | let array = Int8Array::from(vec![Some(2), Some(10), None]); |
4363 | | let actual = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4364 | | |
4365 | | assert_eq!(&actual, &expected); |
4366 | | |
4367 | | let array = Int16Array::from(vec![Some(2), Some(10), None]); |
4368 | | let actual = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4369 | | |
4370 | | assert_eq!(&actual, &expected); |
4371 | | |
4372 | | let array = Int32Array::from(vec![Some(2), Some(10), None]); |
4373 | | let actual = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4374 | | |
4375 | | assert_eq!(&actual, &expected); |
4376 | | |
4377 | | let array = UInt8Array::from(vec![Some(2), Some(10), None]); |
4378 | | let actual = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4379 | | |
4380 | | assert_eq!(&actual, &expected); |
4381 | | |
4382 | | let array = UInt16Array::from(vec![Some(2), Some(10), None]); |
4383 | | let actual = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4384 | | |
4385 | | assert_eq!(&actual, &expected); |
4386 | | |
4387 | | let array = UInt32Array::from(vec![Some(2), Some(10), None]); |
4388 | | let actual = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4389 | | |
4390 | | assert_eq!(&actual, &expected); |
4391 | | |
4392 | | let array = UInt64Array::from(vec![Some(2), Some(10), None]); |
4393 | | let actual = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4394 | | |
4395 | | assert_eq!(&actual, &expected); |
4396 | | } |
4397 | | |
4398 | | #[test] |
4399 | | fn test_cast_timestamp_to_integer() { |
4400 | | let array = TimestampMillisecondArray::from(vec![Some(5), Some(1), None]) |
4401 | | .with_timezone("UTC".to_string()); |
4402 | | let expected = cast(&array, &DataType::Int64).unwrap(); |
4403 | | |
4404 | | let actual = cast(&cast(&array, &DataType::Int8).unwrap(), &DataType::Int64).unwrap(); |
4405 | | assert_eq!(&actual, &expected); |
4406 | | |
4407 | | let actual = cast(&cast(&array, &DataType::Int16).unwrap(), &DataType::Int64).unwrap(); |
4408 | | assert_eq!(&actual, &expected); |
4409 | | |
4410 | | let actual = cast(&cast(&array, &DataType::Int32).unwrap(), &DataType::Int64).unwrap(); |
4411 | | assert_eq!(&actual, &expected); |
4412 | | |
4413 | | let actual = cast(&cast(&array, &DataType::UInt8).unwrap(), &DataType::Int64).unwrap(); |
4414 | | assert_eq!(&actual, &expected); |
4415 | | |
4416 | | let actual = cast(&cast(&array, &DataType::UInt16).unwrap(), &DataType::Int64).unwrap(); |
4417 | | assert_eq!(&actual, &expected); |
4418 | | |
4419 | | let actual = cast(&cast(&array, &DataType::UInt32).unwrap(), &DataType::Int64).unwrap(); |
4420 | | assert_eq!(&actual, &expected); |
4421 | | |
4422 | | let actual = cast(&cast(&array, &DataType::UInt64).unwrap(), &DataType::Int64).unwrap(); |
4423 | | assert_eq!(&actual, &expected); |
4424 | | } |
4425 | | |
4426 | | #[test] |
4427 | | fn test_cast_floating_to_timestamp() { |
4428 | | let array = Int64Array::from(vec![Some(2), Some(10), None]); |
4429 | | let expected = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4430 | | |
4431 | | let array = Float16Array::from(vec![ |
4432 | | Some(f16::from_f32(2.0)), |
4433 | | Some(f16::from_f32(10.6)), |
4434 | | None, |
4435 | | ]); |
4436 | | let actual = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4437 | | |
4438 | | assert_eq!(&actual, &expected); |
4439 | | |
4440 | | let array = Float32Array::from(vec![Some(2.0), Some(10.6), None]); |
4441 | | let actual = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4442 | | |
4443 | | assert_eq!(&actual, &expected); |
4444 | | |
4445 | | let array = Float64Array::from(vec![Some(2.1), Some(10.2), None]); |
4446 | | let actual = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4447 | | |
4448 | | assert_eq!(&actual, &expected); |
4449 | | } |
4450 | | |
4451 | | #[test] |
4452 | | fn test_cast_timestamp_to_floating() { |
4453 | | let array = TimestampMillisecondArray::from(vec![Some(5), Some(1), None]) |
4454 | | .with_timezone("UTC".to_string()); |
4455 | | let expected = cast(&array, &DataType::Int64).unwrap(); |
4456 | | |
4457 | | let actual = cast(&cast(&array, &DataType::Float16).unwrap(), &DataType::Int64).unwrap(); |
4458 | | assert_eq!(&actual, &expected); |
4459 | | |
4460 | | let actual = cast(&cast(&array, &DataType::Float32).unwrap(), &DataType::Int64).unwrap(); |
4461 | | assert_eq!(&actual, &expected); |
4462 | | |
4463 | | let actual = cast(&cast(&array, &DataType::Float64).unwrap(), &DataType::Int64).unwrap(); |
4464 | | assert_eq!(&actual, &expected); |
4465 | | } |
4466 | | |
4467 | | #[test] |
4468 | | fn test_cast_decimal_to_timestamp() { |
4469 | | let array = Int64Array::from(vec![Some(2), Some(10), None]); |
4470 | | let expected = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4471 | | |
4472 | | let array = Decimal128Array::from(vec![Some(200), Some(1000), None]) |
4473 | | .with_precision_and_scale(4, 2) |
4474 | | .unwrap(); |
4475 | | let actual = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4476 | | |
4477 | | assert_eq!(&actual, &expected); |
4478 | | |
4479 | | let array = Decimal256Array::from(vec![ |
4480 | | Some(i256::from_i128(2000)), |
4481 | | Some(i256::from_i128(10000)), |
4482 | | None, |
4483 | | ]) |
4484 | | .with_precision_and_scale(5, 3) |
4485 | | .unwrap(); |
4486 | | let actual = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
4487 | | |
4488 | | assert_eq!(&actual, &expected); |
4489 | | } |
4490 | | |
4491 | | #[test] |
4492 | | fn test_cast_timestamp_to_decimal() { |
4493 | | let array = TimestampMillisecondArray::from(vec![Some(5), Some(1), None]) |
4494 | | .with_timezone("UTC".to_string()); |
4495 | | let expected = cast(&array, &DataType::Int64).unwrap(); |
4496 | | |
4497 | | let actual = cast( |
4498 | | &cast(&array, &DataType::Decimal128(5, 2)).unwrap(), |
4499 | | &DataType::Int64, |
4500 | | ) |
4501 | | .unwrap(); |
4502 | | assert_eq!(&actual, &expected); |
4503 | | |
4504 | | let actual = cast( |
4505 | | &cast(&array, &DataType::Decimal256(10, 5)).unwrap(), |
4506 | | &DataType::Int64, |
4507 | | ) |
4508 | | .unwrap(); |
4509 | | assert_eq!(&actual, &expected); |
4510 | | } |
4511 | | |
4512 | | #[test] |
4513 | | fn test_cast_list_i32_to_list_u16() { |
4514 | | let value_data = Int32Array::from(vec![0, 0, 0, -1, -2, -1, 2, 100000000]).into_data(); |
4515 | | |
4516 | | let value_offsets = Buffer::from_slice_ref([0, 3, 6, 8]); |
4517 | | |
4518 | | // Construct a list array from the above two |
4519 | | // [[0,0,0], [-1, -2, -1], [2, 100000000]] |
4520 | | let list_data_type = DataType::List(Arc::new(Field::new_list_field(DataType::Int32, true))); |
4521 | | let list_data = ArrayData::builder(list_data_type) |
4522 | | .len(3) |
4523 | | .add_buffer(value_offsets) |
4524 | | .add_child_data(value_data) |
4525 | | .build() |
4526 | | .unwrap(); |
4527 | | let list_array = ListArray::from(list_data); |
4528 | | |
4529 | | let cast_array = cast( |
4530 | | &list_array, |
4531 | | &DataType::List(Arc::new(Field::new_list_field(DataType::UInt16, true))), |
4532 | | ) |
4533 | | .unwrap(); |
4534 | | |
4535 | | // For the ListArray itself, there are no null values (as there were no nulls when they went in) |
4536 | | // |
4537 | | // 3 negative values should get lost when casting to unsigned, |
4538 | | // 1 value should overflow |
4539 | | assert_eq!(0, cast_array.null_count()); |
4540 | | |
4541 | | // offsets should be the same |
4542 | | let array = cast_array.as_list::<i32>(); |
4543 | | assert_eq!(list_array.value_offsets(), array.value_offsets()); |
4544 | | |
4545 | | assert_eq!(DataType::UInt16, array.value_type()); |
4546 | | assert_eq!(3, array.value_length(0)); |
4547 | | assert_eq!(3, array.value_length(1)); |
4548 | | assert_eq!(2, array.value_length(2)); |
4549 | | |
4550 | | // expect 4 nulls: negative numbers and overflow |
4551 | | let u16arr = array.values().as_primitive::<UInt16Type>(); |
4552 | | assert_eq!(4, u16arr.null_count()); |
4553 | | |
4554 | | // expect 4 nulls: negative numbers and overflow |
4555 | | let expected: UInt16Array = |
4556 | | vec![Some(0), Some(0), Some(0), None, None, None, Some(2), None] |
4557 | | .into_iter() |
4558 | | .collect(); |
4559 | | |
4560 | | assert_eq!(u16arr, &expected); |
4561 | | } |
4562 | | |
4563 | | #[test] |
4564 | | fn test_cast_list_i32_to_list_timestamp() { |
4565 | | // Construct a value array |
4566 | | let value_data = Int32Array::from(vec![0, 0, 0, -1, -2, -1, 2, 8, 100000000]).into_data(); |
4567 | | |
4568 | | let value_offsets = Buffer::from_slice_ref([0, 3, 6, 9]); |
4569 | | |
4570 | | // Construct a list array from the above two |
4571 | | let list_data_type = DataType::List(Arc::new(Field::new_list_field(DataType::Int32, true))); |
4572 | | let list_data = ArrayData::builder(list_data_type) |
4573 | | .len(3) |
4574 | | .add_buffer(value_offsets) |
4575 | | .add_child_data(value_data) |
4576 | | .build() |
4577 | | .unwrap(); |
4578 | | let list_array = Arc::new(ListArray::from(list_data)) as ArrayRef; |
4579 | | |
4580 | | let actual = cast( |
4581 | | &list_array, |
4582 | | &DataType::List(Arc::new(Field::new_list_field( |
4583 | | DataType::Timestamp(TimeUnit::Microsecond, None), |
4584 | | true, |
4585 | | ))), |
4586 | | ) |
4587 | | .unwrap(); |
4588 | | |
4589 | | let expected = cast( |
4590 | | &cast( |
4591 | | &list_array, |
4592 | | &DataType::List(Arc::new(Field::new_list_field(DataType::Int64, true))), |
4593 | | ) |
4594 | | .unwrap(), |
4595 | | &DataType::List(Arc::new(Field::new_list_field( |
4596 | | DataType::Timestamp(TimeUnit::Microsecond, None), |
4597 | | true, |
4598 | | ))), |
4599 | | ) |
4600 | | .unwrap(); |
4601 | | |
4602 | | assert_eq!(&actual, &expected); |
4603 | | } |
4604 | | |
4605 | | #[test] |
4606 | | fn test_cast_date32_to_date64() { |
4607 | | let a = Date32Array::from(vec![10000, 17890]); |
4608 | | let array = Arc::new(a) as ArrayRef; |
4609 | | let b = cast(&array, &DataType::Date64).unwrap(); |
4610 | | let c = b.as_primitive::<Date64Type>(); |
4611 | | assert_eq!(864000000000, c.value(0)); |
4612 | | assert_eq!(1545696000000, c.value(1)); |
4613 | | } |
4614 | | |
4615 | | #[test] |
4616 | | fn test_cast_date64_to_date32() { |
4617 | | let a = Date64Array::from(vec![Some(864000000005), Some(1545696000001), None]); |
4618 | | let array = Arc::new(a) as ArrayRef; |
4619 | | let b = cast(&array, &DataType::Date32).unwrap(); |
4620 | | let c = b.as_primitive::<Date32Type>(); |
4621 | | assert_eq!(10000, c.value(0)); |
4622 | | assert_eq!(17890, c.value(1)); |
4623 | | assert!(c.is_null(2)); |
4624 | | } |
4625 | | |
4626 | | #[test] |
4627 | | fn test_cast_string_to_integral_overflow() { |
4628 | | let str = Arc::new(StringArray::from(vec![ |
4629 | | Some("123"), |
4630 | | Some("-123"), |
4631 | | Some("86374"), |
4632 | | None, |
4633 | | ])) as ArrayRef; |
4634 | | |
4635 | | let options = CastOptions { |
4636 | | safe: true, |
4637 | | format_options: FormatOptions::default(), |
4638 | | }; |
4639 | | let res = cast_with_options(&str, &DataType::Int16, &options).expect("should cast to i16"); |
4640 | | let expected = |
4641 | | Arc::new(Int16Array::from(vec![Some(123), Some(-123), None, None])) as ArrayRef; |
4642 | | assert_eq!(&res, &expected); |
4643 | | } |
4644 | | |
4645 | | #[test] |
4646 | | fn test_cast_string_to_timestamp() { |
4647 | | let a0 = Arc::new(StringViewArray::from(vec![ |
4648 | | Some("2020-09-08T12:00:00.123456789+00:00"), |
4649 | | Some("Not a valid date"), |
4650 | | None, |
4651 | | ])) as ArrayRef; |
4652 | | let a1 = Arc::new(StringArray::from(vec![ |
4653 | | Some("2020-09-08T12:00:00.123456789+00:00"), |
4654 | | Some("Not a valid date"), |
4655 | | None, |
4656 | | ])) as ArrayRef; |
4657 | | let a2 = Arc::new(LargeStringArray::from(vec![ |
4658 | | Some("2020-09-08T12:00:00.123456789+00:00"), |
4659 | | Some("Not a valid date"), |
4660 | | None, |
4661 | | ])) as ArrayRef; |
4662 | | for array in &[a0, a1, a2] { |
4663 | | for time_unit in &[ |
4664 | | TimeUnit::Second, |
4665 | | TimeUnit::Millisecond, |
4666 | | TimeUnit::Microsecond, |
4667 | | TimeUnit::Nanosecond, |
4668 | | ] { |
4669 | | let to_type = DataType::Timestamp(*time_unit, None); |
4670 | | let b = cast(array, &to_type).unwrap(); |
4671 | | |
4672 | | match time_unit { |
4673 | | TimeUnit::Second => { |
4674 | | let c = b.as_primitive::<TimestampSecondType>(); |
4675 | | assert_eq!(1599566400, c.value(0)); |
4676 | | assert!(c.is_null(1)); |
4677 | | assert!(c.is_null(2)); |
4678 | | } |
4679 | | TimeUnit::Millisecond => { |
4680 | | let c = b |
4681 | | .as_any() |
4682 | | .downcast_ref::<TimestampMillisecondArray>() |
4683 | | .unwrap(); |
4684 | | assert_eq!(1599566400123, c.value(0)); |
4685 | | assert!(c.is_null(1)); |
4686 | | assert!(c.is_null(2)); |
4687 | | } |
4688 | | TimeUnit::Microsecond => { |
4689 | | let c = b |
4690 | | .as_any() |
4691 | | .downcast_ref::<TimestampMicrosecondArray>() |
4692 | | .unwrap(); |
4693 | | assert_eq!(1599566400123456, c.value(0)); |
4694 | | assert!(c.is_null(1)); |
4695 | | assert!(c.is_null(2)); |
4696 | | } |
4697 | | TimeUnit::Nanosecond => { |
4698 | | let c = b |
4699 | | .as_any() |
4700 | | .downcast_ref::<TimestampNanosecondArray>() |
4701 | | .unwrap(); |
4702 | | assert_eq!(1599566400123456789, c.value(0)); |
4703 | | assert!(c.is_null(1)); |
4704 | | assert!(c.is_null(2)); |
4705 | | } |
4706 | | } |
4707 | | |
4708 | | let options = CastOptions { |
4709 | | safe: false, |
4710 | | format_options: FormatOptions::default(), |
4711 | | }; |
4712 | | let err = cast_with_options(array, &to_type, &options).unwrap_err(); |
4713 | | assert_eq!( |
4714 | | err.to_string(), |
4715 | | "Parser error: Error parsing timestamp from 'Not a valid date': error parsing date" |
4716 | | ); |
4717 | | } |
4718 | | } |
4719 | | } |
4720 | | |
4721 | | #[test] |
4722 | | fn test_cast_string_to_timestamp_overflow() { |
4723 | | let array = StringArray::from(vec!["9800-09-08T12:00:00.123456789"]); |
4724 | | let result = cast(&array, &DataType::Timestamp(TimeUnit::Second, None)).unwrap(); |
4725 | | let result = result.as_primitive::<TimestampSecondType>(); |
4726 | | assert_eq!(result.values(), &[247112596800]); |
4727 | | } |
4728 | | |
4729 | | #[test] |
4730 | | fn test_cast_string_to_date32() { |
4731 | | let a0 = Arc::new(StringViewArray::from(vec![ |
4732 | | Some("2018-12-25"), |
4733 | | Some("Not a valid date"), |
4734 | | None, |
4735 | | ])) as ArrayRef; |
4736 | | let a1 = Arc::new(StringArray::from(vec![ |
4737 | | Some("2018-12-25"), |
4738 | | Some("Not a valid date"), |
4739 | | None, |
4740 | | ])) as ArrayRef; |
4741 | | let a2 = Arc::new(LargeStringArray::from(vec![ |
4742 | | Some("2018-12-25"), |
4743 | | Some("Not a valid date"), |
4744 | | None, |
4745 | | ])) as ArrayRef; |
4746 | | for array in &[a0, a1, a2] { |
4747 | | let to_type = DataType::Date32; |
4748 | | let b = cast(array, &to_type).unwrap(); |
4749 | | let c = b.as_primitive::<Date32Type>(); |
4750 | | assert_eq!(17890, c.value(0)); |
4751 | | assert!(c.is_null(1)); |
4752 | | assert!(c.is_null(2)); |
4753 | | |
4754 | | let options = CastOptions { |
4755 | | safe: false, |
4756 | | format_options: FormatOptions::default(), |
4757 | | }; |
4758 | | let err = cast_with_options(array, &to_type, &options).unwrap_err(); |
4759 | | assert_eq!( |
4760 | | err.to_string(), |
4761 | | "Cast error: Cannot cast string 'Not a valid date' to value of Date32 type" |
4762 | | ); |
4763 | | } |
4764 | | } |
4765 | | |
4766 | | #[test] |
4767 | | fn test_cast_string_with_large_date_to_date32() { |
4768 | | let array = Arc::new(StringArray::from(vec![ |
4769 | | Some("+10999-12-31"), |
4770 | | Some("-0010-02-28"), |
4771 | | Some("0010-02-28"), |
4772 | | Some("0000-01-01"), |
4773 | | Some("-0000-01-01"), |
4774 | | Some("-0001-01-01"), |
4775 | | ])) as ArrayRef; |
4776 | | let to_type = DataType::Date32; |
4777 | | let options = CastOptions { |
4778 | | safe: false, |
4779 | | format_options: FormatOptions::default(), |
4780 | | }; |
4781 | | let b = cast_with_options(&array, &to_type, &options).unwrap(); |
4782 | | let c = b.as_primitive::<Date32Type>(); |
4783 | | assert_eq!(3298139, c.value(0)); // 10999-12-31 |
4784 | | assert_eq!(-723122, c.value(1)); // -0010-02-28 |
4785 | | assert_eq!(-715817, c.value(2)); // 0010-02-28 |
4786 | | assert_eq!(c.value(3), c.value(4)); // Expect 0000-01-01 and -0000-01-01 to be parsed the same |
4787 | | assert_eq!(-719528, c.value(3)); // 0000-01-01 |
4788 | | assert_eq!(-719528, c.value(4)); // -0000-01-01 |
4789 | | assert_eq!(-719893, c.value(5)); // -0001-01-01 |
4790 | | } |
4791 | | |
4792 | | #[test] |
4793 | | fn test_cast_invalid_string_with_large_date_to_date32() { |
4794 | | // Large dates need to be prefixed with a + or - sign, otherwise they are not parsed correctly |
4795 | | let array = Arc::new(StringArray::from(vec![Some("10999-12-31")])) as ArrayRef; |
4796 | | let to_type = DataType::Date32; |
4797 | | let options = CastOptions { |
4798 | | safe: false, |
4799 | | format_options: FormatOptions::default(), |
4800 | | }; |
4801 | | let err = cast_with_options(&array, &to_type, &options).unwrap_err(); |
4802 | | assert_eq!( |
4803 | | err.to_string(), |
4804 | | "Cast error: Cannot cast string '10999-12-31' to value of Date32 type" |
4805 | | ); |
4806 | | } |
4807 | | |
4808 | | #[test] |
4809 | | fn test_cast_string_format_yyyymmdd_to_date32() { |
4810 | | let a0 = Arc::new(StringViewArray::from(vec![ |
4811 | | Some("2020-12-25"), |
4812 | | Some("20201117"), |
4813 | | ])) as ArrayRef; |
4814 | | let a1 = Arc::new(StringArray::from(vec![ |
4815 | | Some("2020-12-25"), |
4816 | | Some("20201117"), |
4817 | | ])) as ArrayRef; |
4818 | | let a2 = Arc::new(LargeStringArray::from(vec![ |
4819 | | Some("2020-12-25"), |
4820 | | Some("20201117"), |
4821 | | ])) as ArrayRef; |
4822 | | |
4823 | | for array in &[a0, a1, a2] { |
4824 | | let to_type = DataType::Date32; |
4825 | | let options = CastOptions { |
4826 | | safe: false, |
4827 | | format_options: FormatOptions::default(), |
4828 | | }; |
4829 | | let result = cast_with_options(&array, &to_type, &options).unwrap(); |
4830 | | let c = result.as_primitive::<Date32Type>(); |
4831 | | assert_eq!( |
4832 | | chrono::NaiveDate::from_ymd_opt(2020, 12, 25), |
4833 | | c.value_as_date(0) |
4834 | | ); |
4835 | | assert_eq!( |
4836 | | chrono::NaiveDate::from_ymd_opt(2020, 11, 17), |
4837 | | c.value_as_date(1) |
4838 | | ); |
4839 | | } |
4840 | | } |
4841 | | |
4842 | | #[test] |
4843 | | fn test_cast_string_to_time32second() { |
4844 | | let a0 = Arc::new(StringViewArray::from(vec![ |
4845 | | Some("08:08:35.091323414"), |
4846 | | Some("08:08:60.091323414"), // leap second |
4847 | | Some("08:08:61.091323414"), // not valid |
4848 | | Some("Not a valid time"), |
4849 | | None, |
4850 | | ])) as ArrayRef; |
4851 | | let a1 = Arc::new(StringArray::from(vec![ |
4852 | | Some("08:08:35.091323414"), |
4853 | | Some("08:08:60.091323414"), // leap second |
4854 | | Some("08:08:61.091323414"), // not valid |
4855 | | Some("Not a valid time"), |
4856 | | None, |
4857 | | ])) as ArrayRef; |
4858 | | let a2 = Arc::new(LargeStringArray::from(vec![ |
4859 | | Some("08:08:35.091323414"), |
4860 | | Some("08:08:60.091323414"), // leap second |
4861 | | Some("08:08:61.091323414"), // not valid |
4862 | | Some("Not a valid time"), |
4863 | | None, |
4864 | | ])) as ArrayRef; |
4865 | | for array in &[a0, a1, a2] { |
4866 | | let to_type = DataType::Time32(TimeUnit::Second); |
4867 | | let b = cast(array, &to_type).unwrap(); |
4868 | | let c = b.as_primitive::<Time32SecondType>(); |
4869 | | assert_eq!(29315, c.value(0)); |
4870 | | assert_eq!(29340, c.value(1)); |
4871 | | assert!(c.is_null(2)); |
4872 | | assert!(c.is_null(3)); |
4873 | | assert!(c.is_null(4)); |
4874 | | |
4875 | | let options = CastOptions { |
4876 | | safe: false, |
4877 | | format_options: FormatOptions::default(), |
4878 | | }; |
4879 | | let err = cast_with_options(array, &to_type, &options).unwrap_err(); |
4880 | | assert_eq!(err.to_string(), "Cast error: Cannot cast string '08:08:61.091323414' to value of Time32(Second) type"); |
4881 | | } |
4882 | | } |
4883 | | |
4884 | | #[test] |
4885 | | fn test_cast_string_to_time32millisecond() { |
4886 | | let a0 = Arc::new(StringViewArray::from(vec![ |
4887 | | Some("08:08:35.091323414"), |
4888 | | Some("08:08:60.091323414"), // leap second |
4889 | | Some("08:08:61.091323414"), // not valid |
4890 | | Some("Not a valid time"), |
4891 | | None, |
4892 | | ])) as ArrayRef; |
4893 | | let a1 = Arc::new(StringArray::from(vec![ |
4894 | | Some("08:08:35.091323414"), |
4895 | | Some("08:08:60.091323414"), // leap second |
4896 | | Some("08:08:61.091323414"), // not valid |
4897 | | Some("Not a valid time"), |
4898 | | None, |
4899 | | ])) as ArrayRef; |
4900 | | let a2 = Arc::new(LargeStringArray::from(vec![ |
4901 | | Some("08:08:35.091323414"), |
4902 | | Some("08:08:60.091323414"), // leap second |
4903 | | Some("08:08:61.091323414"), // not valid |
4904 | | Some("Not a valid time"), |
4905 | | None, |
4906 | | ])) as ArrayRef; |
4907 | | for array in &[a0, a1, a2] { |
4908 | | let to_type = DataType::Time32(TimeUnit::Millisecond); |
4909 | | let b = cast(array, &to_type).unwrap(); |
4910 | | let c = b.as_primitive::<Time32MillisecondType>(); |
4911 | | assert_eq!(29315091, c.value(0)); |
4912 | | assert_eq!(29340091, c.value(1)); |
4913 | | assert!(c.is_null(2)); |
4914 | | assert!(c.is_null(3)); |
4915 | | assert!(c.is_null(4)); |
4916 | | |
4917 | | let options = CastOptions { |
4918 | | safe: false, |
4919 | | format_options: FormatOptions::default(), |
4920 | | }; |
4921 | | let err = cast_with_options(array, &to_type, &options).unwrap_err(); |
4922 | | assert_eq!(err.to_string(), "Cast error: Cannot cast string '08:08:61.091323414' to value of Time32(Millisecond) type"); |
4923 | | } |
4924 | | } |
4925 | | |
4926 | | #[test] |
4927 | | fn test_cast_string_to_time64microsecond() { |
4928 | | let a0 = Arc::new(StringViewArray::from(vec![ |
4929 | | Some("08:08:35.091323414"), |
4930 | | Some("Not a valid time"), |
4931 | | None, |
4932 | | ])) as ArrayRef; |
4933 | | let a1 = Arc::new(StringArray::from(vec![ |
4934 | | Some("08:08:35.091323414"), |
4935 | | Some("Not a valid time"), |
4936 | | None, |
4937 | | ])) as ArrayRef; |
4938 | | let a2 = Arc::new(LargeStringArray::from(vec![ |
4939 | | Some("08:08:35.091323414"), |
4940 | | Some("Not a valid time"), |
4941 | | None, |
4942 | | ])) as ArrayRef; |
4943 | | for array in &[a0, a1, a2] { |
4944 | | let to_type = DataType::Time64(TimeUnit::Microsecond); |
4945 | | let b = cast(array, &to_type).unwrap(); |
4946 | | let c = b.as_primitive::<Time64MicrosecondType>(); |
4947 | | assert_eq!(29315091323, c.value(0)); |
4948 | | assert!(c.is_null(1)); |
4949 | | assert!(c.is_null(2)); |
4950 | | |
4951 | | let options = CastOptions { |
4952 | | safe: false, |
4953 | | format_options: FormatOptions::default(), |
4954 | | }; |
4955 | | let err = cast_with_options(array, &to_type, &options).unwrap_err(); |
4956 | | assert_eq!(err.to_string(), "Cast error: Cannot cast string 'Not a valid time' to value of Time64(Microsecond) type"); |
4957 | | } |
4958 | | } |
4959 | | |
4960 | | #[test] |
4961 | | fn test_cast_string_to_time64nanosecond() { |
4962 | | let a0 = Arc::new(StringViewArray::from(vec![ |
4963 | | Some("08:08:35.091323414"), |
4964 | | Some("Not a valid time"), |
4965 | | None, |
4966 | | ])) as ArrayRef; |
4967 | | let a1 = Arc::new(StringArray::from(vec![ |
4968 | | Some("08:08:35.091323414"), |
4969 | | Some("Not a valid time"), |
4970 | | None, |
4971 | | ])) as ArrayRef; |
4972 | | let a2 = Arc::new(LargeStringArray::from(vec![ |
4973 | | Some("08:08:35.091323414"), |
4974 | | Some("Not a valid time"), |
4975 | | None, |
4976 | | ])) as ArrayRef; |
4977 | | for array in &[a0, a1, a2] { |
4978 | | let to_type = DataType::Time64(TimeUnit::Nanosecond); |
4979 | | let b = cast(array, &to_type).unwrap(); |
4980 | | let c = b.as_primitive::<Time64NanosecondType>(); |
4981 | | assert_eq!(29315091323414, c.value(0)); |
4982 | | assert!(c.is_null(1)); |
4983 | | assert!(c.is_null(2)); |
4984 | | |
4985 | | let options = CastOptions { |
4986 | | safe: false, |
4987 | | format_options: FormatOptions::default(), |
4988 | | }; |
4989 | | let err = cast_with_options(array, &to_type, &options).unwrap_err(); |
4990 | | assert_eq!(err.to_string(), "Cast error: Cannot cast string 'Not a valid time' to value of Time64(Nanosecond) type"); |
4991 | | } |
4992 | | } |
4993 | | |
4994 | | #[test] |
4995 | | fn test_cast_string_to_date64() { |
4996 | | let a0 = Arc::new(StringViewArray::from(vec![ |
4997 | | Some("2020-09-08T12:00:00"), |
4998 | | Some("Not a valid date"), |
4999 | | None, |
5000 | | ])) as ArrayRef; |
5001 | | let a1 = Arc::new(StringArray::from(vec![ |
5002 | | Some("2020-09-08T12:00:00"), |
5003 | | Some("Not a valid date"), |
5004 | | None, |
5005 | | ])) as ArrayRef; |
5006 | | let a2 = Arc::new(LargeStringArray::from(vec![ |
5007 | | Some("2020-09-08T12:00:00"), |
5008 | | Some("Not a valid date"), |
5009 | | None, |
5010 | | ])) as ArrayRef; |
5011 | | for array in &[a0, a1, a2] { |
5012 | | let to_type = DataType::Date64; |
5013 | | let b = cast(array, &to_type).unwrap(); |
5014 | | let c = b.as_primitive::<Date64Type>(); |
5015 | | assert_eq!(1599566400000, c.value(0)); |
5016 | | assert!(c.is_null(1)); |
5017 | | assert!(c.is_null(2)); |
5018 | | |
5019 | | let options = CastOptions { |
5020 | | safe: false, |
5021 | | format_options: FormatOptions::default(), |
5022 | | }; |
5023 | | let err = cast_with_options(array, &to_type, &options).unwrap_err(); |
5024 | | assert_eq!( |
5025 | | err.to_string(), |
5026 | | "Cast error: Cannot cast string 'Not a valid date' to value of Date64 type" |
5027 | | ); |
5028 | | } |
5029 | | } |
5030 | | |
5031 | | macro_rules! test_safe_string_to_interval { |
5032 | | ($data_vec:expr, $interval_unit:expr, $array_ty:ty, $expect_vec:expr) => { |
5033 | | let source_string_array = Arc::new(StringArray::from($data_vec.clone())) as ArrayRef; |
5034 | | |
5035 | | let options = CastOptions { |
5036 | | safe: true, |
5037 | | format_options: FormatOptions::default(), |
5038 | | }; |
5039 | | |
5040 | | let target_interval_array = cast_with_options( |
5041 | | &source_string_array.clone(), |
5042 | | &DataType::Interval($interval_unit), |
5043 | | &options, |
5044 | | ) |
5045 | | .unwrap() |
5046 | | .as_any() |
5047 | | .downcast_ref::<$array_ty>() |
5048 | | .unwrap() |
5049 | | .clone() as $array_ty; |
5050 | | |
5051 | | let target_string_array = |
5052 | | cast_with_options(&target_interval_array, &DataType::Utf8, &options) |
5053 | | .unwrap() |
5054 | | .as_any() |
5055 | | .downcast_ref::<StringArray>() |
5056 | | .unwrap() |
5057 | | .clone(); |
5058 | | |
5059 | | let expect_string_array = StringArray::from($expect_vec); |
5060 | | |
5061 | | assert_eq!(target_string_array, expect_string_array); |
5062 | | |
5063 | | let target_large_string_array = |
5064 | | cast_with_options(&target_interval_array, &DataType::LargeUtf8, &options) |
5065 | | .unwrap() |
5066 | | .as_any() |
5067 | | .downcast_ref::<LargeStringArray>() |
5068 | | .unwrap() |
5069 | | .clone(); |
5070 | | |
5071 | | let expect_large_string_array = LargeStringArray::from($expect_vec); |
5072 | | |
5073 | | assert_eq!(target_large_string_array, expect_large_string_array); |
5074 | | }; |
5075 | | } |
5076 | | |
5077 | | #[test] |
5078 | | fn test_cast_string_to_interval_year_month() { |
5079 | | test_safe_string_to_interval!( |
5080 | | vec![ |
5081 | | Some("1 year 1 month"), |
5082 | | Some("1.5 years 13 month"), |
5083 | | Some("30 days"), |
5084 | | Some("31 days"), |
5085 | | Some("2 months 31 days"), |
5086 | | Some("2 months 31 days 1 second"), |
5087 | | Some("foobar"), |
5088 | | ], |
5089 | | IntervalUnit::YearMonth, |
5090 | | IntervalYearMonthArray, |
5091 | | vec![ |
5092 | | Some("1 years 1 mons"), |
5093 | | Some("2 years 7 mons"), |
5094 | | None, |
5095 | | None, |
5096 | | None, |
5097 | | None, |
5098 | | None, |
5099 | | ] |
5100 | | ); |
5101 | | } |
5102 | | |
5103 | | #[test] |
5104 | | fn test_cast_string_to_interval_day_time() { |
5105 | | test_safe_string_to_interval!( |
5106 | | vec![ |
5107 | | Some("1 year 1 month"), |
5108 | | Some("1.5 years 13 month"), |
5109 | | Some("30 days"), |
5110 | | Some("1 day 2 second 3.5 milliseconds"), |
5111 | | Some("foobar"), |
5112 | | ], |
5113 | | IntervalUnit::DayTime, |
5114 | | IntervalDayTimeArray, |
5115 | | vec![ |
5116 | | Some("390 days"), |
5117 | | Some("930 days"), |
5118 | | Some("30 days"), |
5119 | | None, |
5120 | | None, |
5121 | | ] |
5122 | | ); |
5123 | | } |
5124 | | |
5125 | | #[test] |
5126 | | fn test_cast_string_to_interval_month_day_nano() { |
5127 | | test_safe_string_to_interval!( |
5128 | | vec![ |
5129 | | Some("1 year 1 month 1 day"), |
5130 | | None, |
5131 | | Some("1.5 years 13 month 35 days 1.4 milliseconds"), |
5132 | | Some("3 days"), |
5133 | | Some("8 seconds"), |
5134 | | None, |
5135 | | Some("1 day 29800 milliseconds"), |
5136 | | Some("3 months 1 second"), |
5137 | | Some("6 minutes 120 second"), |
5138 | | Some("2 years 39 months 9 days 19 hours 1 minute 83 seconds 399222 milliseconds"), |
5139 | | Some("foobar"), |
5140 | | ], |
5141 | | IntervalUnit::MonthDayNano, |
5142 | | IntervalMonthDayNanoArray, |
5143 | | vec![ |
5144 | | Some("13 mons 1 days"), |
5145 | | None, |
5146 | | Some("31 mons 35 days 0.001400000 secs"), |
5147 | | Some("3 days"), |
5148 | | Some("8.000000000 secs"), |
5149 | | None, |
5150 | | Some("1 days 29.800000000 secs"), |
5151 | | Some("3 mons 1.000000000 secs"), |
5152 | | Some("8 mins"), |
5153 | | Some("63 mons 9 days 19 hours 9 mins 2.222000000 secs"), |
5154 | | None, |
5155 | | ] |
5156 | | ); |
5157 | | } |
5158 | | |
5159 | | macro_rules! test_unsafe_string_to_interval_err { |
5160 | | ($data_vec:expr, $interval_unit:expr, $error_msg:expr) => { |
5161 | | let string_array = Arc::new(StringArray::from($data_vec.clone())) as ArrayRef; |
5162 | | let options = CastOptions { |
5163 | | safe: false, |
5164 | | format_options: FormatOptions::default(), |
5165 | | }; |
5166 | | let arrow_err = cast_with_options( |
5167 | | &string_array.clone(), |
5168 | | &DataType::Interval($interval_unit), |
5169 | | &options, |
5170 | | ) |
5171 | | .unwrap_err(); |
5172 | | assert_eq!($error_msg, arrow_err.to_string()); |
5173 | | }; |
5174 | | } |
5175 | | |
5176 | | #[test] |
5177 | | fn test_cast_string_to_interval_err() { |
5178 | | test_unsafe_string_to_interval_err!( |
5179 | | vec![Some("foobar")], |
5180 | | IntervalUnit::YearMonth, |
5181 | | r#"Parser error: Invalid input syntax for type interval: "foobar""# |
5182 | | ); |
5183 | | test_unsafe_string_to_interval_err!( |
5184 | | vec![Some("foobar")], |
5185 | | IntervalUnit::DayTime, |
5186 | | r#"Parser error: Invalid input syntax for type interval: "foobar""# |
5187 | | ); |
5188 | | test_unsafe_string_to_interval_err!( |
5189 | | vec![Some("foobar")], |
5190 | | IntervalUnit::MonthDayNano, |
5191 | | r#"Parser error: Invalid input syntax for type interval: "foobar""# |
5192 | | ); |
5193 | | test_unsafe_string_to_interval_err!( |
5194 | | vec![Some("2 months 31 days 1 second")], |
5195 | | IntervalUnit::YearMonth, |
5196 | | r#"Cast error: Cannot cast 2 months 31 days 1 second to IntervalYearMonth. Only year and month fields are allowed."# |
5197 | | ); |
5198 | | test_unsafe_string_to_interval_err!( |
5199 | | vec![Some("1 day 1.5 milliseconds")], |
5200 | | IntervalUnit::DayTime, |
5201 | | r#"Cast error: Cannot cast 1 day 1.5 milliseconds to IntervalDayTime because the nanos part isn't multiple of milliseconds"# |
5202 | | ); |
5203 | | |
5204 | | // overflow |
5205 | | test_unsafe_string_to_interval_err!( |
5206 | | vec![Some(format!( |
5207 | | "{} century {} year {} month", |
5208 | | i64::MAX - 2, |
5209 | | i64::MAX - 2, |
5210 | | i64::MAX - 2 |
5211 | | ))], |
5212 | | IntervalUnit::DayTime, |
5213 | | format!( |
5214 | | "Arithmetic overflow: Overflow happened on: {} * 100", |
5215 | | i64::MAX - 2 |
5216 | | ) |
5217 | | ); |
5218 | | test_unsafe_string_to_interval_err!( |
5219 | | vec![Some(format!( |
5220 | | "{} year {} month {} day", |
5221 | | i64::MAX - 2, |
5222 | | i64::MAX - 2, |
5223 | | i64::MAX - 2 |
5224 | | ))], |
5225 | | IntervalUnit::MonthDayNano, |
5226 | | format!( |
5227 | | "Arithmetic overflow: Overflow happened on: {} * 12", |
5228 | | i64::MAX - 2 |
5229 | | ) |
5230 | | ); |
5231 | | } |
5232 | | |
5233 | | #[test] |
5234 | | fn test_cast_binary_to_fixed_size_binary() { |
5235 | | let bytes_1 = "Hiiii".as_bytes(); |
5236 | | let bytes_2 = "Hello".as_bytes(); |
5237 | | |
5238 | | let binary_data = vec![Some(bytes_1), Some(bytes_2), None]; |
5239 | | let a1 = Arc::new(BinaryArray::from(binary_data.clone())) as ArrayRef; |
5240 | | let a2 = Arc::new(LargeBinaryArray::from(binary_data)) as ArrayRef; |
5241 | | |
5242 | | let array_ref = cast(&a1, &DataType::FixedSizeBinary(5)).unwrap(); |
5243 | | let down_cast = array_ref |
5244 | | .as_any() |
5245 | | .downcast_ref::<FixedSizeBinaryArray>() |
5246 | | .unwrap(); |
5247 | | assert_eq!(bytes_1, down_cast.value(0)); |
5248 | | assert_eq!(bytes_2, down_cast.value(1)); |
5249 | | assert!(down_cast.is_null(2)); |
5250 | | |
5251 | | let array_ref = cast(&a2, &DataType::FixedSizeBinary(5)).unwrap(); |
5252 | | let down_cast = array_ref |
5253 | | .as_any() |
5254 | | .downcast_ref::<FixedSizeBinaryArray>() |
5255 | | .unwrap(); |
5256 | | assert_eq!(bytes_1, down_cast.value(0)); |
5257 | | assert_eq!(bytes_2, down_cast.value(1)); |
5258 | | assert!(down_cast.is_null(2)); |
5259 | | |
5260 | | // test error cases when the length of binary are not same |
5261 | | let bytes_1 = "Hi".as_bytes(); |
5262 | | let bytes_2 = "Hello".as_bytes(); |
5263 | | |
5264 | | let binary_data = vec![Some(bytes_1), Some(bytes_2), None]; |
5265 | | let a1 = Arc::new(BinaryArray::from(binary_data.clone())) as ArrayRef; |
5266 | | let a2 = Arc::new(LargeBinaryArray::from(binary_data)) as ArrayRef; |
5267 | | |
5268 | | let array_ref = cast_with_options( |
5269 | | &a1, |
5270 | | &DataType::FixedSizeBinary(5), |
5271 | | &CastOptions { |
5272 | | safe: false, |
5273 | | format_options: FormatOptions::default(), |
5274 | | }, |
5275 | | ); |
5276 | | assert!(array_ref.is_err()); |
5277 | | |
5278 | | let array_ref = cast_with_options( |
5279 | | &a2, |
5280 | | &DataType::FixedSizeBinary(5), |
5281 | | &CastOptions { |
5282 | | safe: false, |
5283 | | format_options: FormatOptions::default(), |
5284 | | }, |
5285 | | ); |
5286 | | assert!(array_ref.is_err()); |
5287 | | } |
5288 | | |
5289 | | #[test] |
5290 | | fn test_fixed_size_binary_to_binary() { |
5291 | | let bytes_1 = "Hiiii".as_bytes(); |
5292 | | let bytes_2 = "Hello".as_bytes(); |
5293 | | |
5294 | | let binary_data = vec![Some(bytes_1), Some(bytes_2), None]; |
5295 | | let a1 = Arc::new(FixedSizeBinaryArray::from(binary_data.clone())) as ArrayRef; |
5296 | | |
5297 | | let array_ref = cast(&a1, &DataType::Binary).unwrap(); |
5298 | | let down_cast = array_ref.as_binary::<i32>(); |
5299 | | assert_eq!(bytes_1, down_cast.value(0)); |
5300 | | assert_eq!(bytes_2, down_cast.value(1)); |
5301 | | assert!(down_cast.is_null(2)); |
5302 | | |
5303 | | let array_ref = cast(&a1, &DataType::LargeBinary).unwrap(); |
5304 | | let down_cast = array_ref.as_binary::<i64>(); |
5305 | | assert_eq!(bytes_1, down_cast.value(0)); |
5306 | | assert_eq!(bytes_2, down_cast.value(1)); |
5307 | | assert!(down_cast.is_null(2)); |
5308 | | |
5309 | | let array_ref = cast(&a1, &DataType::BinaryView).unwrap(); |
5310 | | let down_cast = array_ref.as_binary_view(); |
5311 | | assert_eq!(bytes_1, down_cast.value(0)); |
5312 | | assert_eq!(bytes_2, down_cast.value(1)); |
5313 | | assert!(down_cast.is_null(2)); |
5314 | | } |
5315 | | |
5316 | | #[test] |
5317 | | fn test_fixed_size_binary_to_dictionary() { |
5318 | | let bytes_1 = "Hiiii".as_bytes(); |
5319 | | let bytes_2 = "Hello".as_bytes(); |
5320 | | |
5321 | | let binary_data = vec![Some(bytes_1), Some(bytes_2), Some(bytes_1), None]; |
5322 | | let a1 = Arc::new(FixedSizeBinaryArray::from(binary_data.clone())) as ArrayRef; |
5323 | | |
5324 | | let cast_type = DataType::Dictionary( |
5325 | | Box::new(DataType::Int8), |
5326 | | Box::new(DataType::FixedSizeBinary(5)), |
5327 | | ); |
5328 | | let cast_array = cast(&a1, &cast_type).unwrap(); |
5329 | | assert_eq!(cast_array.data_type(), &cast_type); |
5330 | | assert_eq!( |
5331 | | array_to_strings(&cast_array), |
5332 | | vec!["4869696969", "48656c6c6f", "4869696969", "null"] |
5333 | | ); |
5334 | | // dictionary should only have two distinct values |
5335 | | let dict_array = cast_array |
5336 | | .as_any() |
5337 | | .downcast_ref::<DictionaryArray<Int8Type>>() |
5338 | | .unwrap(); |
5339 | | assert_eq!(dict_array.values().len(), 2); |
5340 | | } |
5341 | | |
5342 | | #[test] |
5343 | | fn test_binary_to_dictionary() { |
5344 | | let mut builder = GenericBinaryBuilder::<i32>::new(); |
5345 | | builder.append_value(b"hello"); |
5346 | | builder.append_value(b"hiiii"); |
5347 | | builder.append_value(b"hiiii"); // duplicate |
5348 | | builder.append_null(); |
5349 | | builder.append_value(b"rustt"); |
5350 | | |
5351 | | let a1 = builder.finish(); |
5352 | | |
5353 | | let cast_type = DataType::Dictionary( |
5354 | | Box::new(DataType::Int8), |
5355 | | Box::new(DataType::FixedSizeBinary(5)), |
5356 | | ); |
5357 | | let cast_array = cast(&a1, &cast_type).unwrap(); |
5358 | | assert_eq!(cast_array.data_type(), &cast_type); |
5359 | | assert_eq!( |
5360 | | array_to_strings(&cast_array), |
5361 | | vec![ |
5362 | | "68656c6c6f", |
5363 | | "6869696969", |
5364 | | "6869696969", |
5365 | | "null", |
5366 | | "7275737474" |
5367 | | ] |
5368 | | ); |
5369 | | // dictionary should only have three distinct values |
5370 | | let dict_array = cast_array |
5371 | | .as_any() |
5372 | | .downcast_ref::<DictionaryArray<Int8Type>>() |
5373 | | .unwrap(); |
5374 | | assert_eq!(dict_array.values().len(), 3); |
5375 | | } |
5376 | | |
5377 | | #[test] |
5378 | | fn test_numeric_to_binary() { |
5379 | | let a = Int16Array::from(vec![Some(1), Some(511), None]); |
5380 | | |
5381 | | let array_ref = cast(&a, &DataType::Binary).unwrap(); |
5382 | | let down_cast = array_ref.as_binary::<i32>(); |
5383 | | assert_eq!(&1_i16.to_le_bytes(), down_cast.value(0)); |
5384 | | assert_eq!(&511_i16.to_le_bytes(), down_cast.value(1)); |
5385 | | assert!(down_cast.is_null(2)); |
5386 | | |
5387 | | let a = Int64Array::from(vec![Some(-1), Some(123456789), None]); |
5388 | | |
5389 | | let array_ref = cast(&a, &DataType::Binary).unwrap(); |
5390 | | let down_cast = array_ref.as_binary::<i32>(); |
5391 | | assert_eq!(&(-1_i64).to_le_bytes(), down_cast.value(0)); |
5392 | | assert_eq!(&123456789_i64.to_le_bytes(), down_cast.value(1)); |
5393 | | assert!(down_cast.is_null(2)); |
5394 | | } |
5395 | | |
5396 | | #[test] |
5397 | | fn test_numeric_to_large_binary() { |
5398 | | let a = Int16Array::from(vec![Some(1), Some(511), None]); |
5399 | | |
5400 | | let array_ref = cast(&a, &DataType::LargeBinary).unwrap(); |
5401 | | let down_cast = array_ref.as_binary::<i64>(); |
5402 | | assert_eq!(&1_i16.to_le_bytes(), down_cast.value(0)); |
5403 | | assert_eq!(&511_i16.to_le_bytes(), down_cast.value(1)); |
5404 | | assert!(down_cast.is_null(2)); |
5405 | | |
5406 | | let a = Int64Array::from(vec![Some(-1), Some(123456789), None]); |
5407 | | |
5408 | | let array_ref = cast(&a, &DataType::LargeBinary).unwrap(); |
5409 | | let down_cast = array_ref.as_binary::<i64>(); |
5410 | | assert_eq!(&(-1_i64).to_le_bytes(), down_cast.value(0)); |
5411 | | assert_eq!(&123456789_i64.to_le_bytes(), down_cast.value(1)); |
5412 | | assert!(down_cast.is_null(2)); |
5413 | | } |
5414 | | |
5415 | | #[test] |
5416 | | fn test_cast_date32_to_int32() { |
5417 | | let array = Date32Array::from(vec![10000, 17890]); |
5418 | | let b = cast(&array, &DataType::Int32).unwrap(); |
5419 | | let c = b.as_primitive::<Int32Type>(); |
5420 | | assert_eq!(10000, c.value(0)); |
5421 | | assert_eq!(17890, c.value(1)); |
5422 | | } |
5423 | | |
5424 | | #[test] |
5425 | | fn test_cast_int32_to_date32() { |
5426 | | let array = Int32Array::from(vec![10000, 17890]); |
5427 | | let b = cast(&array, &DataType::Date32).unwrap(); |
5428 | | let c = b.as_primitive::<Date32Type>(); |
5429 | | assert_eq!(10000, c.value(0)); |
5430 | | assert_eq!(17890, c.value(1)); |
5431 | | } |
5432 | | |
5433 | | #[test] |
5434 | | fn test_cast_timestamp_to_date32() { |
5435 | | let array = |
5436 | | TimestampMillisecondArray::from(vec![Some(864000000005), Some(1545696000001), None]) |
5437 | | .with_timezone("+00:00".to_string()); |
5438 | | let b = cast(&array, &DataType::Date32).unwrap(); |
5439 | | let c = b.as_primitive::<Date32Type>(); |
5440 | | assert_eq!(10000, c.value(0)); |
5441 | | assert_eq!(17890, c.value(1)); |
5442 | | assert!(c.is_null(2)); |
5443 | | } |
5444 | | #[test] |
5445 | | fn test_cast_timestamp_to_date32_zone() { |
5446 | | let strings = StringArray::from_iter([ |
5447 | | Some("1970-01-01T00:00:01"), |
5448 | | Some("1970-01-01T23:59:59"), |
5449 | | None, |
5450 | | Some("2020-03-01T02:00:23+00:00"), |
5451 | | ]); |
5452 | | let dt = DataType::Timestamp(TimeUnit::Millisecond, Some("-07:00".into())); |
5453 | | let timestamps = cast(&strings, &dt).unwrap(); |
5454 | | let dates = cast(timestamps.as_ref(), &DataType::Date32).unwrap(); |
5455 | | |
5456 | | let c = dates.as_primitive::<Date32Type>(); |
5457 | | let expected = NaiveDate::from_ymd_opt(1970, 1, 1).unwrap(); |
5458 | | assert_eq!(c.value_as_date(0).unwrap(), expected); |
5459 | | assert_eq!(c.value_as_date(1).unwrap(), expected); |
5460 | | assert!(c.is_null(2)); |
5461 | | let expected = NaiveDate::from_ymd_opt(2020, 2, 29).unwrap(); |
5462 | | assert_eq!(c.value_as_date(3).unwrap(), expected); |
5463 | | } |
5464 | | #[test] |
5465 | | fn test_cast_timestamp_to_date64() { |
5466 | | let array = |
5467 | | TimestampMillisecondArray::from(vec![Some(864000000005), Some(1545696000001), None]); |
5468 | | let b = cast(&array, &DataType::Date64).unwrap(); |
5469 | | let c = b.as_primitive::<Date64Type>(); |
5470 | | assert_eq!(864000000005, c.value(0)); |
5471 | | assert_eq!(1545696000001, c.value(1)); |
5472 | | assert!(c.is_null(2)); |
5473 | | |
5474 | | let array = TimestampSecondArray::from(vec![Some(864000000005), Some(1545696000001)]); |
5475 | | let b = cast(&array, &DataType::Date64).unwrap(); |
5476 | | let c = b.as_primitive::<Date64Type>(); |
5477 | | assert_eq!(864000000005000, c.value(0)); |
5478 | | assert_eq!(1545696000001000, c.value(1)); |
5479 | | |
5480 | | // test overflow, safe cast |
5481 | | let array = TimestampSecondArray::from(vec![Some(i64::MAX)]); |
5482 | | let b = cast(&array, &DataType::Date64).unwrap(); |
5483 | | assert!(b.is_null(0)); |
5484 | | // test overflow, unsafe cast |
5485 | | let array = TimestampSecondArray::from(vec![Some(i64::MAX)]); |
5486 | | let options = CastOptions { |
5487 | | safe: false, |
5488 | | format_options: FormatOptions::default(), |
5489 | | }; |
5490 | | let b = cast_with_options(&array, &DataType::Date64, &options); |
5491 | | assert!(b.is_err()); |
5492 | | } |
5493 | | |
5494 | | #[test] |
5495 | | fn test_cast_timestamp_to_time64() { |
5496 | | // test timestamp secs |
5497 | | let array = TimestampSecondArray::from(vec![Some(86405), Some(1), None]) |
5498 | | .with_timezone("+01:00".to_string()); |
5499 | | let b = cast(&array, &DataType::Time64(TimeUnit::Microsecond)).unwrap(); |
5500 | | let c = b.as_primitive::<Time64MicrosecondType>(); |
5501 | | assert_eq!(3605000000, c.value(0)); |
5502 | | assert_eq!(3601000000, c.value(1)); |
5503 | | assert!(c.is_null(2)); |
5504 | | let b = cast(&array, &DataType::Time64(TimeUnit::Nanosecond)).unwrap(); |
5505 | | let c = b.as_primitive::<Time64NanosecondType>(); |
5506 | | assert_eq!(3605000000000, c.value(0)); |
5507 | | assert_eq!(3601000000000, c.value(1)); |
5508 | | assert!(c.is_null(2)); |
5509 | | |
5510 | | // test timestamp milliseconds |
5511 | | let a = TimestampMillisecondArray::from(vec![Some(86405000), Some(1000), None]) |
5512 | | .with_timezone("+01:00".to_string()); |
5513 | | let array = Arc::new(a) as ArrayRef; |
5514 | | let b = cast(&array, &DataType::Time64(TimeUnit::Microsecond)).unwrap(); |
5515 | | let c = b.as_primitive::<Time64MicrosecondType>(); |
5516 | | assert_eq!(3605000000, c.value(0)); |
5517 | | assert_eq!(3601000000, c.value(1)); |
5518 | | assert!(c.is_null(2)); |
5519 | | let b = cast(&array, &DataType::Time64(TimeUnit::Nanosecond)).unwrap(); |
5520 | | let c = b.as_primitive::<Time64NanosecondType>(); |
5521 | | assert_eq!(3605000000000, c.value(0)); |
5522 | | assert_eq!(3601000000000, c.value(1)); |
5523 | | assert!(c.is_null(2)); |
5524 | | |
5525 | | // test timestamp microseconds |
5526 | | let a = TimestampMicrosecondArray::from(vec![Some(86405000000), Some(1000000), None]) |
5527 | | .with_timezone("+01:00".to_string()); |
5528 | | let array = Arc::new(a) as ArrayRef; |
5529 | | let b = cast(&array, &DataType::Time64(TimeUnit::Microsecond)).unwrap(); |
5530 | | let c = b.as_primitive::<Time64MicrosecondType>(); |
5531 | | assert_eq!(3605000000, c.value(0)); |
5532 | | assert_eq!(3601000000, c.value(1)); |
5533 | | assert!(c.is_null(2)); |
5534 | | let b = cast(&array, &DataType::Time64(TimeUnit::Nanosecond)).unwrap(); |
5535 | | let c = b.as_primitive::<Time64NanosecondType>(); |
5536 | | assert_eq!(3605000000000, c.value(0)); |
5537 | | assert_eq!(3601000000000, c.value(1)); |
5538 | | assert!(c.is_null(2)); |
5539 | | |
5540 | | // test timestamp nanoseconds |
5541 | | let a = TimestampNanosecondArray::from(vec![Some(86405000000000), Some(1000000000), None]) |
5542 | | .with_timezone("+01:00".to_string()); |
5543 | | let array = Arc::new(a) as ArrayRef; |
5544 | | let b = cast(&array, &DataType::Time64(TimeUnit::Microsecond)).unwrap(); |
5545 | | let c = b.as_primitive::<Time64MicrosecondType>(); |
5546 | | assert_eq!(3605000000, c.value(0)); |
5547 | | assert_eq!(3601000000, c.value(1)); |
5548 | | assert!(c.is_null(2)); |
5549 | | let b = cast(&array, &DataType::Time64(TimeUnit::Nanosecond)).unwrap(); |
5550 | | let c = b.as_primitive::<Time64NanosecondType>(); |
5551 | | assert_eq!(3605000000000, c.value(0)); |
5552 | | assert_eq!(3601000000000, c.value(1)); |
5553 | | assert!(c.is_null(2)); |
5554 | | |
5555 | | // test overflow |
5556 | | let a = |
5557 | | TimestampSecondArray::from(vec![Some(i64::MAX)]).with_timezone("+01:00".to_string()); |
5558 | | let array = Arc::new(a) as ArrayRef; |
5559 | | let b = cast(&array, &DataType::Time64(TimeUnit::Microsecond)); |
5560 | | assert!(b.is_err()); |
5561 | | let b = cast(&array, &DataType::Time64(TimeUnit::Nanosecond)); |
5562 | | assert!(b.is_err()); |
5563 | | let b = cast(&array, &DataType::Time64(TimeUnit::Millisecond)); |
5564 | | assert!(b.is_err()); |
5565 | | } |
5566 | | |
5567 | | #[test] |
5568 | | fn test_cast_timestamp_to_time32() { |
5569 | | // test timestamp secs |
5570 | | let a = TimestampSecondArray::from(vec![Some(86405), Some(1), None]) |
5571 | | .with_timezone("+01:00".to_string()); |
5572 | | let array = Arc::new(a) as ArrayRef; |
5573 | | let b = cast(&array, &DataType::Time32(TimeUnit::Second)).unwrap(); |
5574 | | let c = b.as_primitive::<Time32SecondType>(); |
5575 | | assert_eq!(3605, c.value(0)); |
5576 | | assert_eq!(3601, c.value(1)); |
5577 | | assert!(c.is_null(2)); |
5578 | | let b = cast(&array, &DataType::Time32(TimeUnit::Millisecond)).unwrap(); |
5579 | | let c = b.as_primitive::<Time32MillisecondType>(); |
5580 | | assert_eq!(3605000, c.value(0)); |
5581 | | assert_eq!(3601000, c.value(1)); |
5582 | | assert!(c.is_null(2)); |
5583 | | |
5584 | | // test timestamp milliseconds |
5585 | | let a = TimestampMillisecondArray::from(vec![Some(86405000), Some(1000), None]) |
5586 | | .with_timezone("+01:00".to_string()); |
5587 | | let array = Arc::new(a) as ArrayRef; |
5588 | | let b = cast(&array, &DataType::Time32(TimeUnit::Second)).unwrap(); |
5589 | | let c = b.as_primitive::<Time32SecondType>(); |
5590 | | assert_eq!(3605, c.value(0)); |
5591 | | assert_eq!(3601, c.value(1)); |
5592 | | assert!(c.is_null(2)); |
5593 | | let b = cast(&array, &DataType::Time32(TimeUnit::Millisecond)).unwrap(); |
5594 | | let c = b.as_primitive::<Time32MillisecondType>(); |
5595 | | assert_eq!(3605000, c.value(0)); |
5596 | | assert_eq!(3601000, c.value(1)); |
5597 | | assert!(c.is_null(2)); |
5598 | | |
5599 | | // test timestamp microseconds |
5600 | | let a = TimestampMicrosecondArray::from(vec![Some(86405000000), Some(1000000), None]) |
5601 | | .with_timezone("+01:00".to_string()); |
5602 | | let array = Arc::new(a) as ArrayRef; |
5603 | | let b = cast(&array, &DataType::Time32(TimeUnit::Second)).unwrap(); |
5604 | | let c = b.as_primitive::<Time32SecondType>(); |
5605 | | assert_eq!(3605, c.value(0)); |
5606 | | assert_eq!(3601, c.value(1)); |
5607 | | assert!(c.is_null(2)); |
5608 | | let b = cast(&array, &DataType::Time32(TimeUnit::Millisecond)).unwrap(); |
5609 | | let c = b.as_primitive::<Time32MillisecondType>(); |
5610 | | assert_eq!(3605000, c.value(0)); |
5611 | | assert_eq!(3601000, c.value(1)); |
5612 | | assert!(c.is_null(2)); |
5613 | | |
5614 | | // test timestamp nanoseconds |
5615 | | let a = TimestampNanosecondArray::from(vec![Some(86405000000000), Some(1000000000), None]) |
5616 | | .with_timezone("+01:00".to_string()); |
5617 | | let array = Arc::new(a) as ArrayRef; |
5618 | | let b = cast(&array, &DataType::Time32(TimeUnit::Second)).unwrap(); |
5619 | | let c = b.as_primitive::<Time32SecondType>(); |
5620 | | assert_eq!(3605, c.value(0)); |
5621 | | assert_eq!(3601, c.value(1)); |
5622 | | assert!(c.is_null(2)); |
5623 | | let b = cast(&array, &DataType::Time32(TimeUnit::Millisecond)).unwrap(); |
5624 | | let c = b.as_primitive::<Time32MillisecondType>(); |
5625 | | assert_eq!(3605000, c.value(0)); |
5626 | | assert_eq!(3601000, c.value(1)); |
5627 | | assert!(c.is_null(2)); |
5628 | | |
5629 | | // test overflow |
5630 | | let a = |
5631 | | TimestampSecondArray::from(vec![Some(i64::MAX)]).with_timezone("+01:00".to_string()); |
5632 | | let array = Arc::new(a) as ArrayRef; |
5633 | | let b = cast(&array, &DataType::Time32(TimeUnit::Second)); |
5634 | | assert!(b.is_err()); |
5635 | | let b = cast(&array, &DataType::Time32(TimeUnit::Millisecond)); |
5636 | | assert!(b.is_err()); |
5637 | | } |
5638 | | |
5639 | | // Cast Timestamp(_, None) -> Timestamp(_, Some(timezone)) |
5640 | | #[test] |
5641 | | fn test_cast_timestamp_with_timezone_1() { |
5642 | | let string_array: Arc<dyn Array> = Arc::new(StringArray::from(vec![ |
5643 | | Some("2000-01-01T00:00:00.123456789"), |
5644 | | Some("2010-01-01T00:00:00.123456789"), |
5645 | | None, |
5646 | | ])); |
5647 | | let to_type = DataType::Timestamp(TimeUnit::Nanosecond, None); |
5648 | | let timestamp_array = cast(&string_array, &to_type).unwrap(); |
5649 | | |
5650 | | let to_type = DataType::Timestamp(TimeUnit::Microsecond, Some("+0700".into())); |
5651 | | let timestamp_array = cast(×tamp_array, &to_type).unwrap(); |
5652 | | |
5653 | | let string_array = cast(×tamp_array, &DataType::Utf8).unwrap(); |
5654 | | let result = string_array.as_string::<i32>(); |
5655 | | assert_eq!("2000-01-01T00:00:00.123456+07:00", result.value(0)); |
5656 | | assert_eq!("2010-01-01T00:00:00.123456+07:00", result.value(1)); |
5657 | | assert!(result.is_null(2)); |
5658 | | } |
5659 | | |
5660 | | // Cast Timestamp(_, Some(timezone)) -> Timestamp(_, None) |
5661 | | #[test] |
5662 | | fn test_cast_timestamp_with_timezone_2() { |
5663 | | let string_array: Arc<dyn Array> = Arc::new(StringArray::from(vec![ |
5664 | | Some("2000-01-01T07:00:00.123456789"), |
5665 | | Some("2010-01-01T07:00:00.123456789"), |
5666 | | None, |
5667 | | ])); |
5668 | | let to_type = DataType::Timestamp(TimeUnit::Millisecond, Some("+0700".into())); |
5669 | | let timestamp_array = cast(&string_array, &to_type).unwrap(); |
5670 | | |
5671 | | // Check intermediate representation is correct |
5672 | | let string_array = cast(×tamp_array, &DataType::Utf8).unwrap(); |
5673 | | let result = string_array.as_string::<i32>(); |
5674 | | assert_eq!("2000-01-01T07:00:00.123+07:00", result.value(0)); |
5675 | | assert_eq!("2010-01-01T07:00:00.123+07:00", result.value(1)); |
5676 | | assert!(result.is_null(2)); |
5677 | | |
5678 | | let to_type = DataType::Timestamp(TimeUnit::Nanosecond, None); |
5679 | | let timestamp_array = cast(×tamp_array, &to_type).unwrap(); |
5680 | | |
5681 | | let string_array = cast(×tamp_array, &DataType::Utf8).unwrap(); |
5682 | | let result = string_array.as_string::<i32>(); |
5683 | | assert_eq!("2000-01-01T00:00:00.123", result.value(0)); |
5684 | | assert_eq!("2010-01-01T00:00:00.123", result.value(1)); |
5685 | | assert!(result.is_null(2)); |
5686 | | } |
5687 | | |
5688 | | // Cast Timestamp(_, Some(timezone)) -> Timestamp(_, Some(timezone)) |
5689 | | #[test] |
5690 | | fn test_cast_timestamp_with_timezone_3() { |
5691 | | let string_array: Arc<dyn Array> = Arc::new(StringArray::from(vec![ |
5692 | | Some("2000-01-01T07:00:00.123456789"), |
5693 | | Some("2010-01-01T07:00:00.123456789"), |
5694 | | None, |
5695 | | ])); |
5696 | | let to_type = DataType::Timestamp(TimeUnit::Microsecond, Some("+0700".into())); |
5697 | | let timestamp_array = cast(&string_array, &to_type).unwrap(); |
5698 | | |
5699 | | // Check intermediate representation is correct |
5700 | | let string_array = cast(×tamp_array, &DataType::Utf8).unwrap(); |
5701 | | let result = string_array.as_string::<i32>(); |
5702 | | assert_eq!("2000-01-01T07:00:00.123456+07:00", result.value(0)); |
5703 | | assert_eq!("2010-01-01T07:00:00.123456+07:00", result.value(1)); |
5704 | | assert!(result.is_null(2)); |
5705 | | |
5706 | | let to_type = DataType::Timestamp(TimeUnit::Second, Some("-08:00".into())); |
5707 | | let timestamp_array = cast(×tamp_array, &to_type).unwrap(); |
5708 | | |
5709 | | let string_array = cast(×tamp_array, &DataType::Utf8).unwrap(); |
5710 | | let result = string_array.as_string::<i32>(); |
5711 | | assert_eq!("1999-12-31T16:00:00-08:00", result.value(0)); |
5712 | | assert_eq!("2009-12-31T16:00:00-08:00", result.value(1)); |
5713 | | assert!(result.is_null(2)); |
5714 | | } |
5715 | | |
5716 | | #[test] |
5717 | | fn test_cast_date64_to_timestamp() { |
5718 | | let array = Date64Array::from(vec![Some(864000000005), Some(1545696000001), None]); |
5719 | | let b = cast(&array, &DataType::Timestamp(TimeUnit::Second, None)).unwrap(); |
5720 | | let c = b.as_primitive::<TimestampSecondType>(); |
5721 | | assert_eq!(864000000, c.value(0)); |
5722 | | assert_eq!(1545696000, c.value(1)); |
5723 | | assert!(c.is_null(2)); |
5724 | | } |
5725 | | |
5726 | | #[test] |
5727 | | fn test_cast_date64_to_timestamp_ms() { |
5728 | | let array = Date64Array::from(vec![Some(864000000005), Some(1545696000001), None]); |
5729 | | let b = cast(&array, &DataType::Timestamp(TimeUnit::Millisecond, None)).unwrap(); |
5730 | | let c = b |
5731 | | .as_any() |
5732 | | .downcast_ref::<TimestampMillisecondArray>() |
5733 | | .unwrap(); |
5734 | | assert_eq!(864000000005, c.value(0)); |
5735 | | assert_eq!(1545696000001, c.value(1)); |
5736 | | assert!(c.is_null(2)); |
5737 | | } |
5738 | | |
5739 | | #[test] |
5740 | | fn test_cast_date64_to_timestamp_us() { |
5741 | | let array = Date64Array::from(vec![Some(864000000005), Some(1545696000001), None]); |
5742 | | let b = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
5743 | | let c = b |
5744 | | .as_any() |
5745 | | .downcast_ref::<TimestampMicrosecondArray>() |
5746 | | .unwrap(); |
5747 | | assert_eq!(864000000005000, c.value(0)); |
5748 | | assert_eq!(1545696000001000, c.value(1)); |
5749 | | assert!(c.is_null(2)); |
5750 | | } |
5751 | | |
5752 | | #[test] |
5753 | | fn test_cast_date64_to_timestamp_ns() { |
5754 | | let array = Date64Array::from(vec![Some(864000000005), Some(1545696000001), None]); |
5755 | | let b = cast(&array, &DataType::Timestamp(TimeUnit::Nanosecond, None)).unwrap(); |
5756 | | let c = b |
5757 | | .as_any() |
5758 | | .downcast_ref::<TimestampNanosecondArray>() |
5759 | | .unwrap(); |
5760 | | assert_eq!(864000000005000000, c.value(0)); |
5761 | | assert_eq!(1545696000001000000, c.value(1)); |
5762 | | assert!(c.is_null(2)); |
5763 | | } |
5764 | | |
5765 | | #[test] |
5766 | | fn test_cast_timestamp_to_i64() { |
5767 | | let array = |
5768 | | TimestampMillisecondArray::from(vec![Some(864000000005), Some(1545696000001), None]) |
5769 | | .with_timezone("UTC".to_string()); |
5770 | | let b = cast(&array, &DataType::Int64).unwrap(); |
5771 | | let c = b.as_primitive::<Int64Type>(); |
5772 | | assert_eq!(&DataType::Int64, c.data_type()); |
5773 | | assert_eq!(864000000005, c.value(0)); |
5774 | | assert_eq!(1545696000001, c.value(1)); |
5775 | | assert!(c.is_null(2)); |
5776 | | } |
5777 | | |
5778 | | #[test] |
5779 | | fn test_cast_date32_to_string() { |
5780 | | let array = Date32Array::from(vec![10000, 17890]); |
5781 | | let b = cast(&array, &DataType::Utf8).unwrap(); |
5782 | | let c = b.as_any().downcast_ref::<StringArray>().unwrap(); |
5783 | | assert_eq!(&DataType::Utf8, c.data_type()); |
5784 | | assert_eq!("1997-05-19", c.value(0)); |
5785 | | assert_eq!("2018-12-25", c.value(1)); |
5786 | | } |
5787 | | |
5788 | | #[test] |
5789 | | fn test_cast_date64_to_string() { |
5790 | | let array = Date64Array::from(vec![10000 * 86400000, 17890 * 86400000]); |
5791 | | let b = cast(&array, &DataType::Utf8).unwrap(); |
5792 | | let c = b.as_any().downcast_ref::<StringArray>().unwrap(); |
5793 | | assert_eq!(&DataType::Utf8, c.data_type()); |
5794 | | assert_eq!("1997-05-19T00:00:00", c.value(0)); |
5795 | | assert_eq!("2018-12-25T00:00:00", c.value(1)); |
5796 | | } |
5797 | | |
5798 | | macro_rules! assert_cast_timestamp_to_string { |
5799 | | ($array:expr, $datatype:expr, $output_array_type: ty, $expected:expr) => {{ |
5800 | | let out = cast(&$array, &$datatype).unwrap(); |
5801 | | let actual = out |
5802 | | .as_any() |
5803 | | .downcast_ref::<$output_array_type>() |
5804 | | .unwrap() |
5805 | | .into_iter() |
5806 | | .collect::<Vec<_>>(); |
5807 | | assert_eq!(actual, $expected); |
5808 | | }}; |
5809 | | ($array:expr, $datatype:expr, $output_array_type: ty, $options:expr, $expected:expr) => {{ |
5810 | | let out = cast_with_options(&$array, &$datatype, &$options).unwrap(); |
5811 | | let actual = out |
5812 | | .as_any() |
5813 | | .downcast_ref::<$output_array_type>() |
5814 | | .unwrap() |
5815 | | .into_iter() |
5816 | | .collect::<Vec<_>>(); |
5817 | | assert_eq!(actual, $expected); |
5818 | | }}; |
5819 | | } |
5820 | | |
5821 | | #[test] |
5822 | | fn test_cast_date32_to_timestamp_and_timestamp_with_timezone() { |
5823 | | let tz = "+0545"; // UTC + 0545 is Asia/Kathmandu |
5824 | | let a = Date32Array::from(vec![Some(18628), None, None]); // 2021-1-1, 2022-1-1 |
5825 | | let array = Arc::new(a) as ArrayRef; |
5826 | | |
5827 | | let b = cast( |
5828 | | &array, |
5829 | | &DataType::Timestamp(TimeUnit::Second, Some(tz.into())), |
5830 | | ) |
5831 | | .unwrap(); |
5832 | | let c = b.as_primitive::<TimestampSecondType>(); |
5833 | | let string_array = cast(&c, &DataType::Utf8).unwrap(); |
5834 | | let result = string_array.as_string::<i32>(); |
5835 | | assert_eq!("2021-01-01T00:00:00+05:45", result.value(0)); |
5836 | | |
5837 | | let b = cast(&array, &DataType::Timestamp(TimeUnit::Second, None)).unwrap(); |
5838 | | let c = b.as_primitive::<TimestampSecondType>(); |
5839 | | let string_array = cast(&c, &DataType::Utf8).unwrap(); |
5840 | | let result = string_array.as_string::<i32>(); |
5841 | | assert_eq!("2021-01-01T00:00:00", result.value(0)); |
5842 | | } |
5843 | | |
5844 | | #[test] |
5845 | | fn test_cast_date32_to_timestamp_with_timezone() { |
5846 | | let tz = "+0545"; // UTC + 0545 is Asia/Kathmandu |
5847 | | let a = Date32Array::from(vec![Some(18628), Some(18993), None]); // 2021-1-1, 2022-1-1 |
5848 | | let array = Arc::new(a) as ArrayRef; |
5849 | | let b = cast( |
5850 | | &array, |
5851 | | &DataType::Timestamp(TimeUnit::Second, Some(tz.into())), |
5852 | | ) |
5853 | | .unwrap(); |
5854 | | let c = b.as_primitive::<TimestampSecondType>(); |
5855 | | assert_eq!(1609438500, c.value(0)); |
5856 | | assert_eq!(1640974500, c.value(1)); |
5857 | | assert!(c.is_null(2)); |
5858 | | |
5859 | | let string_array = cast(&c, &DataType::Utf8).unwrap(); |
5860 | | let result = string_array.as_string::<i32>(); |
5861 | | assert_eq!("2021-01-01T00:00:00+05:45", result.value(0)); |
5862 | | assert_eq!("2022-01-01T00:00:00+05:45", result.value(1)); |
5863 | | } |
5864 | | |
5865 | | #[test] |
5866 | | fn test_cast_date32_to_timestamp_with_timezone_ms() { |
5867 | | let tz = "+0545"; // UTC + 0545 is Asia/Kathmandu |
5868 | | let a = Date32Array::from(vec![Some(18628), Some(18993), None]); // 2021-1-1, 2022-1-1 |
5869 | | let array = Arc::new(a) as ArrayRef; |
5870 | | let b = cast( |
5871 | | &array, |
5872 | | &DataType::Timestamp(TimeUnit::Millisecond, Some(tz.into())), |
5873 | | ) |
5874 | | .unwrap(); |
5875 | | let c = b.as_primitive::<TimestampMillisecondType>(); |
5876 | | assert_eq!(1609438500000, c.value(0)); |
5877 | | assert_eq!(1640974500000, c.value(1)); |
5878 | | assert!(c.is_null(2)); |
5879 | | |
5880 | | let string_array = cast(&c, &DataType::Utf8).unwrap(); |
5881 | | let result = string_array.as_string::<i32>(); |
5882 | | assert_eq!("2021-01-01T00:00:00+05:45", result.value(0)); |
5883 | | assert_eq!("2022-01-01T00:00:00+05:45", result.value(1)); |
5884 | | } |
5885 | | |
5886 | | #[test] |
5887 | | fn test_cast_date32_to_timestamp_with_timezone_us() { |
5888 | | let tz = "+0545"; // UTC + 0545 is Asia/Kathmandu |
5889 | | let a = Date32Array::from(vec![Some(18628), Some(18993), None]); // 2021-1-1, 2022-1-1 |
5890 | | let array = Arc::new(a) as ArrayRef; |
5891 | | let b = cast( |
5892 | | &array, |
5893 | | &DataType::Timestamp(TimeUnit::Microsecond, Some(tz.into())), |
5894 | | ) |
5895 | | .unwrap(); |
5896 | | let c = b.as_primitive::<TimestampMicrosecondType>(); |
5897 | | assert_eq!(1609438500000000, c.value(0)); |
5898 | | assert_eq!(1640974500000000, c.value(1)); |
5899 | | assert!(c.is_null(2)); |
5900 | | |
5901 | | let string_array = cast(&c, &DataType::Utf8).unwrap(); |
5902 | | let result = string_array.as_string::<i32>(); |
5903 | | assert_eq!("2021-01-01T00:00:00+05:45", result.value(0)); |
5904 | | assert_eq!("2022-01-01T00:00:00+05:45", result.value(1)); |
5905 | | } |
5906 | | |
5907 | | #[test] |
5908 | | fn test_cast_date32_to_timestamp_with_timezone_ns() { |
5909 | | let tz = "+0545"; // UTC + 0545 is Asia/Kathmandu |
5910 | | let a = Date32Array::from(vec![Some(18628), Some(18993), None]); // 2021-1-1, 2022-1-1 |
5911 | | let array = Arc::new(a) as ArrayRef; |
5912 | | let b = cast( |
5913 | | &array, |
5914 | | &DataType::Timestamp(TimeUnit::Nanosecond, Some(tz.into())), |
5915 | | ) |
5916 | | .unwrap(); |
5917 | | let c = b.as_primitive::<TimestampNanosecondType>(); |
5918 | | assert_eq!(1609438500000000000, c.value(0)); |
5919 | | assert_eq!(1640974500000000000, c.value(1)); |
5920 | | assert!(c.is_null(2)); |
5921 | | |
5922 | | let string_array = cast(&c, &DataType::Utf8).unwrap(); |
5923 | | let result = string_array.as_string::<i32>(); |
5924 | | assert_eq!("2021-01-01T00:00:00+05:45", result.value(0)); |
5925 | | assert_eq!("2022-01-01T00:00:00+05:45", result.value(1)); |
5926 | | } |
5927 | | |
5928 | | #[test] |
5929 | | fn test_cast_date64_to_timestamp_with_timezone() { |
5930 | | let array = Date64Array::from(vec![Some(864000000005), Some(1545696000001), None]); |
5931 | | let tz = "+0545"; // UTC + 0545 is Asia/Kathmandu |
5932 | | let b = cast( |
5933 | | &array, |
5934 | | &DataType::Timestamp(TimeUnit::Second, Some(tz.into())), |
5935 | | ) |
5936 | | .unwrap(); |
5937 | | |
5938 | | let c = b.as_primitive::<TimestampSecondType>(); |
5939 | | assert_eq!(863979300, c.value(0)); |
5940 | | assert_eq!(1545675300, c.value(1)); |
5941 | | assert!(c.is_null(2)); |
5942 | | |
5943 | | let string_array = cast(&c, &DataType::Utf8).unwrap(); |
5944 | | let result = string_array.as_string::<i32>(); |
5945 | | assert_eq!("1997-05-19T00:00:00+05:45", result.value(0)); |
5946 | | assert_eq!("2018-12-25T00:00:00+05:45", result.value(1)); |
5947 | | } |
5948 | | |
5949 | | #[test] |
5950 | | fn test_cast_date64_to_timestamp_with_timezone_ms() { |
5951 | | let array = Date64Array::from(vec![Some(864000000005), Some(1545696000001), None]); |
5952 | | let tz = "+0545"; // UTC + 0545 is Asia/Kathmandu |
5953 | | let b = cast( |
5954 | | &array, |
5955 | | &DataType::Timestamp(TimeUnit::Millisecond, Some(tz.into())), |
5956 | | ) |
5957 | | .unwrap(); |
5958 | | |
5959 | | let c = b.as_primitive::<TimestampMillisecondType>(); |
5960 | | assert_eq!(863979300005, c.value(0)); |
5961 | | assert_eq!(1545675300001, c.value(1)); |
5962 | | assert!(c.is_null(2)); |
5963 | | |
5964 | | let string_array = cast(&c, &DataType::Utf8).unwrap(); |
5965 | | let result = string_array.as_string::<i32>(); |
5966 | | assert_eq!("1997-05-19T00:00:00.005+05:45", result.value(0)); |
5967 | | assert_eq!("2018-12-25T00:00:00.001+05:45", result.value(1)); |
5968 | | } |
5969 | | |
5970 | | #[test] |
5971 | | fn test_cast_date64_to_timestamp_with_timezone_us() { |
5972 | | let array = Date64Array::from(vec![Some(864000000005), Some(1545696000001), None]); |
5973 | | let tz = "+0545"; // UTC + 0545 is Asia/Kathmandu |
5974 | | let b = cast( |
5975 | | &array, |
5976 | | &DataType::Timestamp(TimeUnit::Microsecond, Some(tz.into())), |
5977 | | ) |
5978 | | .unwrap(); |
5979 | | |
5980 | | let c = b.as_primitive::<TimestampMicrosecondType>(); |
5981 | | assert_eq!(863979300005000, c.value(0)); |
5982 | | assert_eq!(1545675300001000, c.value(1)); |
5983 | | assert!(c.is_null(2)); |
5984 | | |
5985 | | let string_array = cast(&c, &DataType::Utf8).unwrap(); |
5986 | | let result = string_array.as_string::<i32>(); |
5987 | | assert_eq!("1997-05-19T00:00:00.005+05:45", result.value(0)); |
5988 | | assert_eq!("2018-12-25T00:00:00.001+05:45", result.value(1)); |
5989 | | } |
5990 | | |
5991 | | #[test] |
5992 | | fn test_cast_date64_to_timestamp_with_timezone_ns() { |
5993 | | let array = Date64Array::from(vec![Some(864000000005), Some(1545696000001), None]); |
5994 | | let tz = "+0545"; // UTC + 0545 is Asia/Kathmandu |
5995 | | let b = cast( |
5996 | | &array, |
5997 | | &DataType::Timestamp(TimeUnit::Nanosecond, Some(tz.into())), |
5998 | | ) |
5999 | | .unwrap(); |
6000 | | |
6001 | | let c = b.as_primitive::<TimestampNanosecondType>(); |
6002 | | assert_eq!(863979300005000000, c.value(0)); |
6003 | | assert_eq!(1545675300001000000, c.value(1)); |
6004 | | assert!(c.is_null(2)); |
6005 | | |
6006 | | let string_array = cast(&c, &DataType::Utf8).unwrap(); |
6007 | | let result = string_array.as_string::<i32>(); |
6008 | | assert_eq!("1997-05-19T00:00:00.005+05:45", result.value(0)); |
6009 | | assert_eq!("2018-12-25T00:00:00.001+05:45", result.value(1)); |
6010 | | } |
6011 | | |
6012 | | #[test] |
6013 | | fn test_cast_timestamp_to_strings() { |
6014 | | // "2018-12-25T00:00:02.001", "1997-05-19T00:00:03.005", None |
6015 | | let array = |
6016 | | TimestampMillisecondArray::from(vec![Some(864000003005), Some(1545696002001), None]); |
6017 | | let expected = vec![ |
6018 | | Some("1997-05-19T00:00:03.005"), |
6019 | | Some("2018-12-25T00:00:02.001"), |
6020 | | None, |
6021 | | ]; |
6022 | | |
6023 | | assert_cast_timestamp_to_string!(array, DataType::Utf8View, StringViewArray, expected); |
6024 | | assert_cast_timestamp_to_string!(array, DataType::Utf8, StringArray, expected); |
6025 | | assert_cast_timestamp_to_string!(array, DataType::LargeUtf8, LargeStringArray, expected); |
6026 | | } |
6027 | | |
6028 | | #[test] |
6029 | | fn test_cast_timestamp_to_strings_opt() { |
6030 | | let ts_format = "%Y-%m-%d %H:%M:%S%.6f"; |
6031 | | let tz = "+0545"; // UTC + 0545 is Asia/Kathmandu |
6032 | | let cast_options = CastOptions { |
6033 | | safe: true, |
6034 | | format_options: FormatOptions::default() |
6035 | | .with_timestamp_format(Some(ts_format)) |
6036 | | .with_timestamp_tz_format(Some(ts_format)), |
6037 | | }; |
6038 | | |
6039 | | // "2018-12-25T00:00:02.001", "1997-05-19T00:00:03.005", None |
6040 | | let array_without_tz = |
6041 | | TimestampMillisecondArray::from(vec![Some(864000003005), Some(1545696002001), None]); |
6042 | | let expected = vec![ |
6043 | | Some("1997-05-19 00:00:03.005000"), |
6044 | | Some("2018-12-25 00:00:02.001000"), |
6045 | | None, |
6046 | | ]; |
6047 | | assert_cast_timestamp_to_string!( |
6048 | | array_without_tz, |
6049 | | DataType::Utf8View, |
6050 | | StringViewArray, |
6051 | | cast_options, |
6052 | | expected |
6053 | | ); |
6054 | | assert_cast_timestamp_to_string!( |
6055 | | array_without_tz, |
6056 | | DataType::Utf8, |
6057 | | StringArray, |
6058 | | cast_options, |
6059 | | expected |
6060 | | ); |
6061 | | assert_cast_timestamp_to_string!( |
6062 | | array_without_tz, |
6063 | | DataType::LargeUtf8, |
6064 | | LargeStringArray, |
6065 | | cast_options, |
6066 | | expected |
6067 | | ); |
6068 | | |
6069 | | let array_with_tz = |
6070 | | TimestampMillisecondArray::from(vec![Some(864000003005), Some(1545696002001), None]) |
6071 | | .with_timezone(tz.to_string()); |
6072 | | let expected = vec![ |
6073 | | Some("1997-05-19 05:45:03.005000"), |
6074 | | Some("2018-12-25 05:45:02.001000"), |
6075 | | None, |
6076 | | ]; |
6077 | | assert_cast_timestamp_to_string!( |
6078 | | array_with_tz, |
6079 | | DataType::Utf8View, |
6080 | | StringViewArray, |
6081 | | cast_options, |
6082 | | expected |
6083 | | ); |
6084 | | assert_cast_timestamp_to_string!( |
6085 | | array_with_tz, |
6086 | | DataType::Utf8, |
6087 | | StringArray, |
6088 | | cast_options, |
6089 | | expected |
6090 | | ); |
6091 | | assert_cast_timestamp_to_string!( |
6092 | | array_with_tz, |
6093 | | DataType::LargeUtf8, |
6094 | | LargeStringArray, |
6095 | | cast_options, |
6096 | | expected |
6097 | | ); |
6098 | | } |
6099 | | |
6100 | | #[test] |
6101 | | fn test_cast_between_timestamps() { |
6102 | | let array = |
6103 | | TimestampMillisecondArray::from(vec![Some(864000003005), Some(1545696002001), None]); |
6104 | | let b = cast(&array, &DataType::Timestamp(TimeUnit::Second, None)).unwrap(); |
6105 | | let c = b.as_primitive::<TimestampSecondType>(); |
6106 | | assert_eq!(864000003, c.value(0)); |
6107 | | assert_eq!(1545696002, c.value(1)); |
6108 | | assert!(c.is_null(2)); |
6109 | | } |
6110 | | |
6111 | | #[test] |
6112 | | fn test_cast_duration_to_i64() { |
6113 | | let base = vec![5, 6, 7, 8, 100000000]; |
6114 | | |
6115 | | let duration_arrays = vec![ |
6116 | | Arc::new(DurationNanosecondArray::from(base.clone())) as ArrayRef, |
6117 | | Arc::new(DurationMicrosecondArray::from(base.clone())) as ArrayRef, |
6118 | | Arc::new(DurationMillisecondArray::from(base.clone())) as ArrayRef, |
6119 | | Arc::new(DurationSecondArray::from(base.clone())) as ArrayRef, |
6120 | | ]; |
6121 | | |
6122 | | for arr in duration_arrays { |
6123 | | assert!(can_cast_types(arr.data_type(), &DataType::Int64)); |
6124 | | let result = cast(&arr, &DataType::Int64).unwrap(); |
6125 | | let result = result.as_primitive::<Int64Type>(); |
6126 | | assert_eq!(base.as_slice(), result.values()); |
6127 | | } |
6128 | | } |
6129 | | |
6130 | | #[test] |
6131 | | fn test_cast_between_durations_and_numerics() { |
6132 | | fn test_cast_between_durations<FromType, ToType>() |
6133 | | where |
6134 | | FromType: ArrowPrimitiveType<Native = i64>, |
6135 | | ToType: ArrowPrimitiveType<Native = i64>, |
6136 | | PrimitiveArray<FromType>: From<Vec<Option<i64>>>, |
6137 | | { |
6138 | | let from_unit = match FromType::DATA_TYPE { |
6139 | | DataType::Duration(unit) => unit, |
6140 | | _ => panic!("Expected a duration type"), |
6141 | | }; |
6142 | | let to_unit = match ToType::DATA_TYPE { |
6143 | | DataType::Duration(unit) => unit, |
6144 | | _ => panic!("Expected a duration type"), |
6145 | | }; |
6146 | | let from_size = time_unit_multiple(&from_unit); |
6147 | | let to_size = time_unit_multiple(&to_unit); |
6148 | | |
6149 | | let (v1_before, v2_before) = (8640003005, 1696002001); |
6150 | | let (v1_after, v2_after) = if from_size >= to_size { |
6151 | | ( |
6152 | | v1_before / (from_size / to_size), |
6153 | | v2_before / (from_size / to_size), |
6154 | | ) |
6155 | | } else { |
6156 | | ( |
6157 | | v1_before * (to_size / from_size), |
6158 | | v2_before * (to_size / from_size), |
6159 | | ) |
6160 | | }; |
6161 | | |
6162 | | let array = |
6163 | | PrimitiveArray::<FromType>::from(vec![Some(v1_before), Some(v2_before), None]); |
6164 | | let b = cast(&array, &ToType::DATA_TYPE).unwrap(); |
6165 | | let c = b.as_primitive::<ToType>(); |
6166 | | assert_eq!(v1_after, c.value(0)); |
6167 | | assert_eq!(v2_after, c.value(1)); |
6168 | | assert!(c.is_null(2)); |
6169 | | } |
6170 | | |
6171 | | // between each individual duration type |
6172 | | test_cast_between_durations::<DurationSecondType, DurationMillisecondType>(); |
6173 | | test_cast_between_durations::<DurationSecondType, DurationMicrosecondType>(); |
6174 | | test_cast_between_durations::<DurationSecondType, DurationNanosecondType>(); |
6175 | | test_cast_between_durations::<DurationMillisecondType, DurationSecondType>(); |
6176 | | test_cast_between_durations::<DurationMillisecondType, DurationMicrosecondType>(); |
6177 | | test_cast_between_durations::<DurationMillisecondType, DurationNanosecondType>(); |
6178 | | test_cast_between_durations::<DurationMicrosecondType, DurationSecondType>(); |
6179 | | test_cast_between_durations::<DurationMicrosecondType, DurationMillisecondType>(); |
6180 | | test_cast_between_durations::<DurationMicrosecondType, DurationNanosecondType>(); |
6181 | | test_cast_between_durations::<DurationNanosecondType, DurationSecondType>(); |
6182 | | test_cast_between_durations::<DurationNanosecondType, DurationMillisecondType>(); |
6183 | | test_cast_between_durations::<DurationNanosecondType, DurationMicrosecondType>(); |
6184 | | |
6185 | | // cast failed |
6186 | | let array = DurationSecondArray::from(vec![ |
6187 | | Some(i64::MAX), |
6188 | | Some(8640203410378005), |
6189 | | Some(10241096), |
6190 | | None, |
6191 | | ]); |
6192 | | let b = cast(&array, &DataType::Duration(TimeUnit::Nanosecond)).unwrap(); |
6193 | | let c = b.as_primitive::<DurationNanosecondType>(); |
6194 | | assert!(c.is_null(0)); |
6195 | | assert!(c.is_null(1)); |
6196 | | assert_eq!(10241096000000000, c.value(2)); |
6197 | | assert!(c.is_null(3)); |
6198 | | |
6199 | | // durations to numerics |
6200 | | let array = DurationSecondArray::from(vec![ |
6201 | | Some(i64::MAX), |
6202 | | Some(8640203410378005), |
6203 | | Some(10241096), |
6204 | | None, |
6205 | | ]); |
6206 | | let b = cast(&array, &DataType::Int64).unwrap(); |
6207 | | let c = b.as_primitive::<Int64Type>(); |
6208 | | assert_eq!(i64::MAX, c.value(0)); |
6209 | | assert_eq!(8640203410378005, c.value(1)); |
6210 | | assert_eq!(10241096, c.value(2)); |
6211 | | assert!(c.is_null(3)); |
6212 | | |
6213 | | let b = cast(&array, &DataType::Int32).unwrap(); |
6214 | | let c = b.as_primitive::<Int32Type>(); |
6215 | | assert_eq!(0, c.value(0)); |
6216 | | assert_eq!(0, c.value(1)); |
6217 | | assert_eq!(10241096, c.value(2)); |
6218 | | assert!(c.is_null(3)); |
6219 | | |
6220 | | // numerics to durations |
6221 | | let array = Int32Array::from(vec![Some(i32::MAX), Some(802034103), Some(10241096), None]); |
6222 | | let b = cast(&array, &DataType::Duration(TimeUnit::Second)).unwrap(); |
6223 | | let c = b.as_any().downcast_ref::<DurationSecondArray>().unwrap(); |
6224 | | assert_eq!(i32::MAX as i64, c.value(0)); |
6225 | | assert_eq!(802034103, c.value(1)); |
6226 | | assert_eq!(10241096, c.value(2)); |
6227 | | assert!(c.is_null(3)); |
6228 | | } |
6229 | | |
6230 | | #[test] |
6231 | | fn test_cast_to_strings() { |
6232 | | let a = Int32Array::from(vec![1, 2, 3]); |
6233 | | let out = cast(&a, &DataType::Utf8).unwrap(); |
6234 | | let out = out |
6235 | | .as_any() |
6236 | | .downcast_ref::<StringArray>() |
6237 | | .unwrap() |
6238 | | .into_iter() |
6239 | | .collect::<Vec<_>>(); |
6240 | | assert_eq!(out, vec![Some("1"), Some("2"), Some("3")]); |
6241 | | let out = cast(&a, &DataType::LargeUtf8).unwrap(); |
6242 | | let out = out |
6243 | | .as_any() |
6244 | | .downcast_ref::<LargeStringArray>() |
6245 | | .unwrap() |
6246 | | .into_iter() |
6247 | | .collect::<Vec<_>>(); |
6248 | | assert_eq!(out, vec![Some("1"), Some("2"), Some("3")]); |
6249 | | } |
6250 | | |
6251 | | #[test] |
6252 | | fn test_str_to_str_casts() { |
6253 | | for data in [ |
6254 | | vec![Some("foo"), Some("bar"), Some("ham")], |
6255 | | vec![Some("foo"), None, Some("bar")], |
6256 | | ] { |
6257 | | let a = LargeStringArray::from(data.clone()); |
6258 | | let to = cast(&a, &DataType::Utf8).unwrap(); |
6259 | | let expect = a |
6260 | | .as_any() |
6261 | | .downcast_ref::<LargeStringArray>() |
6262 | | .unwrap() |
6263 | | .into_iter() |
6264 | | .collect::<Vec<_>>(); |
6265 | | let out = to |
6266 | | .as_any() |
6267 | | .downcast_ref::<StringArray>() |
6268 | | .unwrap() |
6269 | | .into_iter() |
6270 | | .collect::<Vec<_>>(); |
6271 | | assert_eq!(expect, out); |
6272 | | |
6273 | | let a = StringArray::from(data); |
6274 | | let to = cast(&a, &DataType::LargeUtf8).unwrap(); |
6275 | | let expect = a |
6276 | | .as_any() |
6277 | | .downcast_ref::<StringArray>() |
6278 | | .unwrap() |
6279 | | .into_iter() |
6280 | | .collect::<Vec<_>>(); |
6281 | | let out = to |
6282 | | .as_any() |
6283 | | .downcast_ref::<LargeStringArray>() |
6284 | | .unwrap() |
6285 | | .into_iter() |
6286 | | .collect::<Vec<_>>(); |
6287 | | assert_eq!(expect, out); |
6288 | | } |
6289 | | } |
6290 | | |
6291 | | const VIEW_TEST_DATA: [Option<&str>; 5] = [ |
6292 | | Some("hello"), |
6293 | | Some("repeated"), |
6294 | | None, |
6295 | | Some("large payload over 12 bytes"), |
6296 | | Some("repeated"), |
6297 | | ]; |
6298 | | |
6299 | | #[test] |
6300 | | fn test_string_view_to_binary_view() { |
6301 | | let string_view_array = StringViewArray::from_iter(VIEW_TEST_DATA); |
6302 | | |
6303 | | assert!(can_cast_types( |
6304 | | string_view_array.data_type(), |
6305 | | &DataType::BinaryView |
6306 | | )); |
6307 | | |
6308 | | let binary_view_array = cast(&string_view_array, &DataType::BinaryView).unwrap(); |
6309 | | assert_eq!(binary_view_array.data_type(), &DataType::BinaryView); |
6310 | | |
6311 | | let expect_binary_view_array = BinaryViewArray::from_iter(VIEW_TEST_DATA); |
6312 | | assert_eq!(binary_view_array.as_ref(), &expect_binary_view_array); |
6313 | | } |
6314 | | |
6315 | | #[test] |
6316 | | fn test_binary_view_to_string_view() { |
6317 | | let binary_view_array = BinaryViewArray::from_iter(VIEW_TEST_DATA); |
6318 | | |
6319 | | assert!(can_cast_types( |
6320 | | binary_view_array.data_type(), |
6321 | | &DataType::Utf8View |
6322 | | )); |
6323 | | |
6324 | | let string_view_array = cast(&binary_view_array, &DataType::Utf8View).unwrap(); |
6325 | | assert_eq!(string_view_array.data_type(), &DataType::Utf8View); |
6326 | | |
6327 | | let expect_string_view_array = StringViewArray::from_iter(VIEW_TEST_DATA); |
6328 | | assert_eq!(string_view_array.as_ref(), &expect_string_view_array); |
6329 | | } |
6330 | | |
6331 | | #[test] |
6332 | | fn test_string_to_view() { |
6333 | | _test_string_to_view::<i32>(); |
6334 | | _test_string_to_view::<i64>(); |
6335 | | } |
6336 | | |
6337 | | fn _test_string_to_view<O>() |
6338 | | where |
6339 | | O: OffsetSizeTrait, |
6340 | | { |
6341 | | let string_array = GenericStringArray::<O>::from_iter(VIEW_TEST_DATA); |
6342 | | |
6343 | | assert!(can_cast_types( |
6344 | | string_array.data_type(), |
6345 | | &DataType::Utf8View |
6346 | | )); |
6347 | | |
6348 | | assert!(can_cast_types( |
6349 | | string_array.data_type(), |
6350 | | &DataType::BinaryView |
6351 | | )); |
6352 | | |
6353 | | let string_view_array = cast(&string_array, &DataType::Utf8View).unwrap(); |
6354 | | assert_eq!(string_view_array.data_type(), &DataType::Utf8View); |
6355 | | |
6356 | | let binary_view_array = cast(&string_array, &DataType::BinaryView).unwrap(); |
6357 | | assert_eq!(binary_view_array.data_type(), &DataType::BinaryView); |
6358 | | |
6359 | | let expect_string_view_array = StringViewArray::from_iter(VIEW_TEST_DATA); |
6360 | | assert_eq!(string_view_array.as_ref(), &expect_string_view_array); |
6361 | | |
6362 | | let expect_binary_view_array = BinaryViewArray::from_iter(VIEW_TEST_DATA); |
6363 | | assert_eq!(binary_view_array.as_ref(), &expect_binary_view_array); |
6364 | | } |
6365 | | |
6366 | | #[test] |
6367 | | fn test_bianry_to_view() { |
6368 | | _test_binary_to_view::<i32>(); |
6369 | | _test_binary_to_view::<i64>(); |
6370 | | } |
6371 | | |
6372 | | fn _test_binary_to_view<O>() |
6373 | | where |
6374 | | O: OffsetSizeTrait, |
6375 | | { |
6376 | | let binary_array = GenericBinaryArray::<O>::from_iter(VIEW_TEST_DATA); |
6377 | | |
6378 | | assert!(can_cast_types( |
6379 | | binary_array.data_type(), |
6380 | | &DataType::Utf8View |
6381 | | )); |
6382 | | |
6383 | | assert!(can_cast_types( |
6384 | | binary_array.data_type(), |
6385 | | &DataType::BinaryView |
6386 | | )); |
6387 | | |
6388 | | let string_view_array = cast(&binary_array, &DataType::Utf8View).unwrap(); |
6389 | | assert_eq!(string_view_array.data_type(), &DataType::Utf8View); |
6390 | | |
6391 | | let binary_view_array = cast(&binary_array, &DataType::BinaryView).unwrap(); |
6392 | | assert_eq!(binary_view_array.data_type(), &DataType::BinaryView); |
6393 | | |
6394 | | let expect_string_view_array = StringViewArray::from_iter(VIEW_TEST_DATA); |
6395 | | assert_eq!(string_view_array.as_ref(), &expect_string_view_array); |
6396 | | |
6397 | | let expect_binary_view_array = BinaryViewArray::from_iter(VIEW_TEST_DATA); |
6398 | | assert_eq!(binary_view_array.as_ref(), &expect_binary_view_array); |
6399 | | } |
6400 | | |
6401 | | #[test] |
6402 | | fn test_dict_to_view() { |
6403 | | let values = StringArray::from_iter(VIEW_TEST_DATA); |
6404 | | let keys = Int8Array::from_iter([Some(1), Some(0), None, Some(3), None, Some(1), Some(4)]); |
6405 | | let string_dict_array = |
6406 | | DictionaryArray::<Int8Type>::try_new(keys, Arc::new(values)).unwrap(); |
6407 | | let typed_dict = string_dict_array.downcast_dict::<StringArray>().unwrap(); |
6408 | | |
6409 | | let string_view_array = { |
6410 | | let mut builder = StringViewBuilder::new().with_fixed_block_size(8); // multiple buffers. |
6411 | | for v in typed_dict.into_iter() { |
6412 | | builder.append_option(v); |
6413 | | } |
6414 | | builder.finish() |
6415 | | }; |
6416 | | let expected_string_array_type = string_view_array.data_type(); |
6417 | | let casted_string_array = cast(&string_dict_array, expected_string_array_type).unwrap(); |
6418 | | assert_eq!(casted_string_array.data_type(), expected_string_array_type); |
6419 | | assert_eq!(casted_string_array.as_ref(), &string_view_array); |
6420 | | |
6421 | | let binary_buffer = cast(&typed_dict.values(), &DataType::Binary).unwrap(); |
6422 | | let binary_dict_array = |
6423 | | DictionaryArray::<Int8Type>::new(typed_dict.keys().clone(), binary_buffer); |
6424 | | let typed_binary_dict = binary_dict_array.downcast_dict::<BinaryArray>().unwrap(); |
6425 | | |
6426 | | let binary_view_array = { |
6427 | | let mut builder = BinaryViewBuilder::new().with_fixed_block_size(8); // multiple buffers. |
6428 | | for v in typed_binary_dict.into_iter() { |
6429 | | builder.append_option(v); |
6430 | | } |
6431 | | builder.finish() |
6432 | | }; |
6433 | | let expected_binary_array_type = binary_view_array.data_type(); |
6434 | | let casted_binary_array = cast(&binary_dict_array, expected_binary_array_type).unwrap(); |
6435 | | assert_eq!(casted_binary_array.data_type(), expected_binary_array_type); |
6436 | | assert_eq!(casted_binary_array.as_ref(), &binary_view_array); |
6437 | | } |
6438 | | |
6439 | | #[test] |
6440 | | fn test_view_to_dict() { |
6441 | | let string_view_array = StringViewArray::from_iter(VIEW_TEST_DATA); |
6442 | | let string_dict_array: DictionaryArray<Int8Type> = VIEW_TEST_DATA.into_iter().collect(); |
6443 | | let casted_type = string_dict_array.data_type(); |
6444 | | let casted_dict_array = cast(&string_view_array, casted_type).unwrap(); |
6445 | | assert_eq!(casted_dict_array.data_type(), casted_type); |
6446 | | assert_eq!(casted_dict_array.as_ref(), &string_dict_array); |
6447 | | |
6448 | | let binary_view_array = BinaryViewArray::from_iter(VIEW_TEST_DATA); |
6449 | | let binary_dict_array = string_dict_array.downcast_dict::<StringArray>().unwrap(); |
6450 | | let binary_buffer = cast(&binary_dict_array.values(), &DataType::Binary).unwrap(); |
6451 | | let binary_dict_array = |
6452 | | DictionaryArray::<Int8Type>::new(binary_dict_array.keys().clone(), binary_buffer); |
6453 | | let casted_type = binary_dict_array.data_type(); |
6454 | | let casted_binary_array = cast(&binary_view_array, casted_type).unwrap(); |
6455 | | assert_eq!(casted_binary_array.data_type(), casted_type); |
6456 | | assert_eq!(casted_binary_array.as_ref(), &binary_dict_array); |
6457 | | } |
6458 | | |
6459 | | #[test] |
6460 | | fn test_view_to_string() { |
6461 | | _test_view_to_string::<i32>(); |
6462 | | _test_view_to_string::<i64>(); |
6463 | | } |
6464 | | |
6465 | | fn _test_view_to_string<O>() |
6466 | | where |
6467 | | O: OffsetSizeTrait, |
6468 | | { |
6469 | | let string_view_array = { |
6470 | | let mut builder = StringViewBuilder::new().with_fixed_block_size(8); // multiple buffers. |
6471 | | for s in VIEW_TEST_DATA.iter() { |
6472 | | builder.append_option(*s); |
6473 | | } |
6474 | | builder.finish() |
6475 | | }; |
6476 | | |
6477 | | let binary_view_array = BinaryViewArray::from_iter(VIEW_TEST_DATA); |
6478 | | |
6479 | | let expected_string_array = GenericStringArray::<O>::from_iter(VIEW_TEST_DATA); |
6480 | | let expected_type = expected_string_array.data_type(); |
6481 | | |
6482 | | assert!(can_cast_types(string_view_array.data_type(), expected_type)); |
6483 | | assert!(can_cast_types(binary_view_array.data_type(), expected_type)); |
6484 | | |
6485 | | let string_view_casted_array = cast(&string_view_array, expected_type).unwrap(); |
6486 | | assert_eq!(string_view_casted_array.data_type(), expected_type); |
6487 | | assert_eq!(string_view_casted_array.as_ref(), &expected_string_array); |
6488 | | |
6489 | | let binary_view_casted_array = cast(&binary_view_array, expected_type).unwrap(); |
6490 | | assert_eq!(binary_view_casted_array.data_type(), expected_type); |
6491 | | assert_eq!(binary_view_casted_array.as_ref(), &expected_string_array); |
6492 | | } |
6493 | | |
6494 | | #[test] |
6495 | | fn test_view_to_binary() { |
6496 | | _test_view_to_binary::<i32>(); |
6497 | | _test_view_to_binary::<i64>(); |
6498 | | } |
6499 | | |
6500 | | fn _test_view_to_binary<O>() |
6501 | | where |
6502 | | O: OffsetSizeTrait, |
6503 | | { |
6504 | | let view_array = { |
6505 | | let mut builder = BinaryViewBuilder::new().with_fixed_block_size(8); // multiple buffers. |
6506 | | for s in VIEW_TEST_DATA.iter() { |
6507 | | builder.append_option(*s); |
6508 | | } |
6509 | | builder.finish() |
6510 | | }; |
6511 | | |
6512 | | let expected_binary_array = GenericBinaryArray::<O>::from_iter(VIEW_TEST_DATA); |
6513 | | let expected_type = expected_binary_array.data_type(); |
6514 | | |
6515 | | assert!(can_cast_types(view_array.data_type(), expected_type)); |
6516 | | |
6517 | | let binary_array = cast(&view_array, expected_type).unwrap(); |
6518 | | assert_eq!(binary_array.data_type(), expected_type); |
6519 | | |
6520 | | assert_eq!(binary_array.as_ref(), &expected_binary_array); |
6521 | | } |
6522 | | |
6523 | | #[test] |
6524 | | fn test_cast_from_f64() { |
6525 | | let f64_values: Vec<f64> = vec![ |
6526 | | i64::MIN as f64, |
6527 | | i32::MIN as f64, |
6528 | | i16::MIN as f64, |
6529 | | i8::MIN as f64, |
6530 | | 0_f64, |
6531 | | u8::MAX as f64, |
6532 | | u16::MAX as f64, |
6533 | | u32::MAX as f64, |
6534 | | u64::MAX as f64, |
6535 | | ]; |
6536 | | let f64_array: ArrayRef = Arc::new(Float64Array::from(f64_values)); |
6537 | | |
6538 | | let f64_expected = vec![ |
6539 | | -9223372036854776000.0, |
6540 | | -2147483648.0, |
6541 | | -32768.0, |
6542 | | -128.0, |
6543 | | 0.0, |
6544 | | 255.0, |
6545 | | 65535.0, |
6546 | | 4294967295.0, |
6547 | | 18446744073709552000.0, |
6548 | | ]; |
6549 | | assert_eq!( |
6550 | | f64_expected, |
6551 | | get_cast_values::<Float64Type>(&f64_array, &DataType::Float64) |
6552 | | .iter() |
6553 | | .map(|i| i.parse::<f64>().unwrap()) |
6554 | | .collect::<Vec<f64>>() |
6555 | | ); |
6556 | | |
6557 | | let f32_expected = vec![ |
6558 | | -9223372000000000000.0, |
6559 | | -2147483600.0, |
6560 | | -32768.0, |
6561 | | -128.0, |
6562 | | 0.0, |
6563 | | 255.0, |
6564 | | 65535.0, |
6565 | | 4294967300.0, |
6566 | | 18446744000000000000.0, |
6567 | | ]; |
6568 | | assert_eq!( |
6569 | | f32_expected, |
6570 | | get_cast_values::<Float32Type>(&f64_array, &DataType::Float32) |
6571 | | .iter() |
6572 | | .map(|i| i.parse::<f32>().unwrap()) |
6573 | | .collect::<Vec<f32>>() |
6574 | | ); |
6575 | | |
6576 | | let f16_expected = vec![ |
6577 | | f16::from_f64(-9223372000000000000.0), |
6578 | | f16::from_f64(-2147483600.0), |
6579 | | f16::from_f64(-32768.0), |
6580 | | f16::from_f64(-128.0), |
6581 | | f16::from_f64(0.0), |
6582 | | f16::from_f64(255.0), |
6583 | | f16::from_f64(65535.0), |
6584 | | f16::from_f64(4294967300.0), |
6585 | | f16::from_f64(18446744000000000000.0), |
6586 | | ]; |
6587 | | assert_eq!( |
6588 | | f16_expected, |
6589 | | get_cast_values::<Float16Type>(&f64_array, &DataType::Float16) |
6590 | | .iter() |
6591 | | .map(|i| i.parse::<f16>().unwrap()) |
6592 | | .collect::<Vec<f16>>() |
6593 | | ); |
6594 | | |
6595 | | let i64_expected = vec![ |
6596 | | "-9223372036854775808", |
6597 | | "-2147483648", |
6598 | | "-32768", |
6599 | | "-128", |
6600 | | "0", |
6601 | | "255", |
6602 | | "65535", |
6603 | | "4294967295", |
6604 | | "null", |
6605 | | ]; |
6606 | | assert_eq!( |
6607 | | i64_expected, |
6608 | | get_cast_values::<Int64Type>(&f64_array, &DataType::Int64) |
6609 | | ); |
6610 | | |
6611 | | let i32_expected = vec![ |
6612 | | "null", |
6613 | | "-2147483648", |
6614 | | "-32768", |
6615 | | "-128", |
6616 | | "0", |
6617 | | "255", |
6618 | | "65535", |
6619 | | "null", |
6620 | | "null", |
6621 | | ]; |
6622 | | assert_eq!( |
6623 | | i32_expected, |
6624 | | get_cast_values::<Int32Type>(&f64_array, &DataType::Int32) |
6625 | | ); |
6626 | | |
6627 | | let i16_expected = vec![ |
6628 | | "null", "null", "-32768", "-128", "0", "255", "null", "null", "null", |
6629 | | ]; |
6630 | | assert_eq!( |
6631 | | i16_expected, |
6632 | | get_cast_values::<Int16Type>(&f64_array, &DataType::Int16) |
6633 | | ); |
6634 | | |
6635 | | let i8_expected = vec![ |
6636 | | "null", "null", "null", "-128", "0", "null", "null", "null", "null", |
6637 | | ]; |
6638 | | assert_eq!( |
6639 | | i8_expected, |
6640 | | get_cast_values::<Int8Type>(&f64_array, &DataType::Int8) |
6641 | | ); |
6642 | | |
6643 | | let u64_expected = vec![ |
6644 | | "null", |
6645 | | "null", |
6646 | | "null", |
6647 | | "null", |
6648 | | "0", |
6649 | | "255", |
6650 | | "65535", |
6651 | | "4294967295", |
6652 | | "null", |
6653 | | ]; |
6654 | | assert_eq!( |
6655 | | u64_expected, |
6656 | | get_cast_values::<UInt64Type>(&f64_array, &DataType::UInt64) |
6657 | | ); |
6658 | | |
6659 | | let u32_expected = vec![ |
6660 | | "null", |
6661 | | "null", |
6662 | | "null", |
6663 | | "null", |
6664 | | "0", |
6665 | | "255", |
6666 | | "65535", |
6667 | | "4294967295", |
6668 | | "null", |
6669 | | ]; |
6670 | | assert_eq!( |
6671 | | u32_expected, |
6672 | | get_cast_values::<UInt32Type>(&f64_array, &DataType::UInt32) |
6673 | | ); |
6674 | | |
6675 | | let u16_expected = vec![ |
6676 | | "null", "null", "null", "null", "0", "255", "65535", "null", "null", |
6677 | | ]; |
6678 | | assert_eq!( |
6679 | | u16_expected, |
6680 | | get_cast_values::<UInt16Type>(&f64_array, &DataType::UInt16) |
6681 | | ); |
6682 | | |
6683 | | let u8_expected = vec![ |
6684 | | "null", "null", "null", "null", "0", "255", "null", "null", "null", |
6685 | | ]; |
6686 | | assert_eq!( |
6687 | | u8_expected, |
6688 | | get_cast_values::<UInt8Type>(&f64_array, &DataType::UInt8) |
6689 | | ); |
6690 | | } |
6691 | | |
6692 | | #[test] |
6693 | | fn test_cast_from_f32() { |
6694 | | let f32_values: Vec<f32> = vec![ |
6695 | | i32::MIN as f32, |
6696 | | i32::MIN as f32, |
6697 | | i16::MIN as f32, |
6698 | | i8::MIN as f32, |
6699 | | 0_f32, |
6700 | | u8::MAX as f32, |
6701 | | u16::MAX as f32, |
6702 | | u32::MAX as f32, |
6703 | | u32::MAX as f32, |
6704 | | ]; |
6705 | | let f32_array: ArrayRef = Arc::new(Float32Array::from(f32_values)); |
6706 | | |
6707 | | let f64_expected = vec![ |
6708 | | "-2147483648.0", |
6709 | | "-2147483648.0", |
6710 | | "-32768.0", |
6711 | | "-128.0", |
6712 | | "0.0", |
6713 | | "255.0", |
6714 | | "65535.0", |
6715 | | "4294967296.0", |
6716 | | "4294967296.0", |
6717 | | ]; |
6718 | | assert_eq!( |
6719 | | f64_expected, |
6720 | | get_cast_values::<Float64Type>(&f32_array, &DataType::Float64) |
6721 | | ); |
6722 | | |
6723 | | let f32_expected = vec![ |
6724 | | "-2147483600.0", |
6725 | | "-2147483600.0", |
6726 | | "-32768.0", |
6727 | | "-128.0", |
6728 | | "0.0", |
6729 | | "255.0", |
6730 | | "65535.0", |
6731 | | "4294967300.0", |
6732 | | "4294967300.0", |
6733 | | ]; |
6734 | | assert_eq!( |
6735 | | f32_expected, |
6736 | | get_cast_values::<Float32Type>(&f32_array, &DataType::Float32) |
6737 | | ); |
6738 | | |
6739 | | let f16_expected = vec![ |
6740 | | "-inf", "-inf", "-32768.0", "-128.0", "0.0", "255.0", "inf", "inf", "inf", |
6741 | | ]; |
6742 | | assert_eq!( |
6743 | | f16_expected, |
6744 | | get_cast_values::<Float16Type>(&f32_array, &DataType::Float16) |
6745 | | ); |
6746 | | |
6747 | | let i64_expected = vec![ |
6748 | | "-2147483648", |
6749 | | "-2147483648", |
6750 | | "-32768", |
6751 | | "-128", |
6752 | | "0", |
6753 | | "255", |
6754 | | "65535", |
6755 | | "4294967296", |
6756 | | "4294967296", |
6757 | | ]; |
6758 | | assert_eq!( |
6759 | | i64_expected, |
6760 | | get_cast_values::<Int64Type>(&f32_array, &DataType::Int64) |
6761 | | ); |
6762 | | |
6763 | | let i32_expected = vec![ |
6764 | | "-2147483648", |
6765 | | "-2147483648", |
6766 | | "-32768", |
6767 | | "-128", |
6768 | | "0", |
6769 | | "255", |
6770 | | "65535", |
6771 | | "null", |
6772 | | "null", |
6773 | | ]; |
6774 | | assert_eq!( |
6775 | | i32_expected, |
6776 | | get_cast_values::<Int32Type>(&f32_array, &DataType::Int32) |
6777 | | ); |
6778 | | |
6779 | | let i16_expected = vec![ |
6780 | | "null", "null", "-32768", "-128", "0", "255", "null", "null", "null", |
6781 | | ]; |
6782 | | assert_eq!( |
6783 | | i16_expected, |
6784 | | get_cast_values::<Int16Type>(&f32_array, &DataType::Int16) |
6785 | | ); |
6786 | | |
6787 | | let i8_expected = vec![ |
6788 | | "null", "null", "null", "-128", "0", "null", "null", "null", "null", |
6789 | | ]; |
6790 | | assert_eq!( |
6791 | | i8_expected, |
6792 | | get_cast_values::<Int8Type>(&f32_array, &DataType::Int8) |
6793 | | ); |
6794 | | |
6795 | | let u64_expected = vec![ |
6796 | | "null", |
6797 | | "null", |
6798 | | "null", |
6799 | | "null", |
6800 | | "0", |
6801 | | "255", |
6802 | | "65535", |
6803 | | "4294967296", |
6804 | | "4294967296", |
6805 | | ]; |
6806 | | assert_eq!( |
6807 | | u64_expected, |
6808 | | get_cast_values::<UInt64Type>(&f32_array, &DataType::UInt64) |
6809 | | ); |
6810 | | |
6811 | | let u32_expected = vec![ |
6812 | | "null", "null", "null", "null", "0", "255", "65535", "null", "null", |
6813 | | ]; |
6814 | | assert_eq!( |
6815 | | u32_expected, |
6816 | | get_cast_values::<UInt32Type>(&f32_array, &DataType::UInt32) |
6817 | | ); |
6818 | | |
6819 | | let u16_expected = vec![ |
6820 | | "null", "null", "null", "null", "0", "255", "65535", "null", "null", |
6821 | | ]; |
6822 | | assert_eq!( |
6823 | | u16_expected, |
6824 | | get_cast_values::<UInt16Type>(&f32_array, &DataType::UInt16) |
6825 | | ); |
6826 | | |
6827 | | let u8_expected = vec![ |
6828 | | "null", "null", "null", "null", "0", "255", "null", "null", "null", |
6829 | | ]; |
6830 | | assert_eq!( |
6831 | | u8_expected, |
6832 | | get_cast_values::<UInt8Type>(&f32_array, &DataType::UInt8) |
6833 | | ); |
6834 | | } |
6835 | | |
6836 | | #[test] |
6837 | | fn test_cast_from_uint64() { |
6838 | | let u64_values: Vec<u64> = vec![ |
6839 | | 0, |
6840 | | u8::MAX as u64, |
6841 | | u16::MAX as u64, |
6842 | | u32::MAX as u64, |
6843 | | u64::MAX, |
6844 | | ]; |
6845 | | let u64_array: ArrayRef = Arc::new(UInt64Array::from(u64_values)); |
6846 | | |
6847 | | let f64_expected = vec![0.0, 255.0, 65535.0, 4294967295.0, 18446744073709552000.0]; |
6848 | | assert_eq!( |
6849 | | f64_expected, |
6850 | | get_cast_values::<Float64Type>(&u64_array, &DataType::Float64) |
6851 | | .iter() |
6852 | | .map(|i| i.parse::<f64>().unwrap()) |
6853 | | .collect::<Vec<f64>>() |
6854 | | ); |
6855 | | |
6856 | | let f32_expected = vec![0.0, 255.0, 65535.0, 4294967300.0, 18446744000000000000.0]; |
6857 | | assert_eq!( |
6858 | | f32_expected, |
6859 | | get_cast_values::<Float32Type>(&u64_array, &DataType::Float32) |
6860 | | .iter() |
6861 | | .map(|i| i.parse::<f32>().unwrap()) |
6862 | | .collect::<Vec<f32>>() |
6863 | | ); |
6864 | | |
6865 | | let f16_expected = vec![ |
6866 | | f16::from_f64(0.0), |
6867 | | f16::from_f64(255.0), |
6868 | | f16::from_f64(65535.0), |
6869 | | f16::from_f64(4294967300.0), |
6870 | | f16::from_f64(18446744000000000000.0), |
6871 | | ]; |
6872 | | assert_eq!( |
6873 | | f16_expected, |
6874 | | get_cast_values::<Float16Type>(&u64_array, &DataType::Float16) |
6875 | | .iter() |
6876 | | .map(|i| i.parse::<f16>().unwrap()) |
6877 | | .collect::<Vec<f16>>() |
6878 | | ); |
6879 | | |
6880 | | let i64_expected = vec!["0", "255", "65535", "4294967295", "null"]; |
6881 | | assert_eq!( |
6882 | | i64_expected, |
6883 | | get_cast_values::<Int64Type>(&u64_array, &DataType::Int64) |
6884 | | ); |
6885 | | |
6886 | | let i32_expected = vec!["0", "255", "65535", "null", "null"]; |
6887 | | assert_eq!( |
6888 | | i32_expected, |
6889 | | get_cast_values::<Int32Type>(&u64_array, &DataType::Int32) |
6890 | | ); |
6891 | | |
6892 | | let i16_expected = vec!["0", "255", "null", "null", "null"]; |
6893 | | assert_eq!( |
6894 | | i16_expected, |
6895 | | get_cast_values::<Int16Type>(&u64_array, &DataType::Int16) |
6896 | | ); |
6897 | | |
6898 | | let i8_expected = vec!["0", "null", "null", "null", "null"]; |
6899 | | assert_eq!( |
6900 | | i8_expected, |
6901 | | get_cast_values::<Int8Type>(&u64_array, &DataType::Int8) |
6902 | | ); |
6903 | | |
6904 | | let u64_expected = vec!["0", "255", "65535", "4294967295", "18446744073709551615"]; |
6905 | | assert_eq!( |
6906 | | u64_expected, |
6907 | | get_cast_values::<UInt64Type>(&u64_array, &DataType::UInt64) |
6908 | | ); |
6909 | | |
6910 | | let u32_expected = vec!["0", "255", "65535", "4294967295", "null"]; |
6911 | | assert_eq!( |
6912 | | u32_expected, |
6913 | | get_cast_values::<UInt32Type>(&u64_array, &DataType::UInt32) |
6914 | | ); |
6915 | | |
6916 | | let u16_expected = vec!["0", "255", "65535", "null", "null"]; |
6917 | | assert_eq!( |
6918 | | u16_expected, |
6919 | | get_cast_values::<UInt16Type>(&u64_array, &DataType::UInt16) |
6920 | | ); |
6921 | | |
6922 | | let u8_expected = vec!["0", "255", "null", "null", "null"]; |
6923 | | assert_eq!( |
6924 | | u8_expected, |
6925 | | get_cast_values::<UInt8Type>(&u64_array, &DataType::UInt8) |
6926 | | ); |
6927 | | } |
6928 | | |
6929 | | #[test] |
6930 | | fn test_cast_from_uint32() { |
6931 | | let u32_values: Vec<u32> = vec![0, u8::MAX as u32, u16::MAX as u32, u32::MAX]; |
6932 | | let u32_array: ArrayRef = Arc::new(UInt32Array::from(u32_values)); |
6933 | | |
6934 | | let f64_expected = vec!["0.0", "255.0", "65535.0", "4294967295.0"]; |
6935 | | assert_eq!( |
6936 | | f64_expected, |
6937 | | get_cast_values::<Float64Type>(&u32_array, &DataType::Float64) |
6938 | | ); |
6939 | | |
6940 | | let f32_expected = vec!["0.0", "255.0", "65535.0", "4294967300.0"]; |
6941 | | assert_eq!( |
6942 | | f32_expected, |
6943 | | get_cast_values::<Float32Type>(&u32_array, &DataType::Float32) |
6944 | | ); |
6945 | | |
6946 | | let f16_expected = vec!["0.0", "255.0", "inf", "inf"]; |
6947 | | assert_eq!( |
6948 | | f16_expected, |
6949 | | get_cast_values::<Float16Type>(&u32_array, &DataType::Float16) |
6950 | | ); |
6951 | | |
6952 | | let i64_expected = vec!["0", "255", "65535", "4294967295"]; |
6953 | | assert_eq!( |
6954 | | i64_expected, |
6955 | | get_cast_values::<Int64Type>(&u32_array, &DataType::Int64) |
6956 | | ); |
6957 | | |
6958 | | let i32_expected = vec!["0", "255", "65535", "null"]; |
6959 | | assert_eq!( |
6960 | | i32_expected, |
6961 | | get_cast_values::<Int32Type>(&u32_array, &DataType::Int32) |
6962 | | ); |
6963 | | |
6964 | | let i16_expected = vec!["0", "255", "null", "null"]; |
6965 | | assert_eq!( |
6966 | | i16_expected, |
6967 | | get_cast_values::<Int16Type>(&u32_array, &DataType::Int16) |
6968 | | ); |
6969 | | |
6970 | | let i8_expected = vec!["0", "null", "null", "null"]; |
6971 | | assert_eq!( |
6972 | | i8_expected, |
6973 | | get_cast_values::<Int8Type>(&u32_array, &DataType::Int8) |
6974 | | ); |
6975 | | |
6976 | | let u64_expected = vec!["0", "255", "65535", "4294967295"]; |
6977 | | assert_eq!( |
6978 | | u64_expected, |
6979 | | get_cast_values::<UInt64Type>(&u32_array, &DataType::UInt64) |
6980 | | ); |
6981 | | |
6982 | | let u32_expected = vec!["0", "255", "65535", "4294967295"]; |
6983 | | assert_eq!( |
6984 | | u32_expected, |
6985 | | get_cast_values::<UInt32Type>(&u32_array, &DataType::UInt32) |
6986 | | ); |
6987 | | |
6988 | | let u16_expected = vec!["0", "255", "65535", "null"]; |
6989 | | assert_eq!( |
6990 | | u16_expected, |
6991 | | get_cast_values::<UInt16Type>(&u32_array, &DataType::UInt16) |
6992 | | ); |
6993 | | |
6994 | | let u8_expected = vec!["0", "255", "null", "null"]; |
6995 | | assert_eq!( |
6996 | | u8_expected, |
6997 | | get_cast_values::<UInt8Type>(&u32_array, &DataType::UInt8) |
6998 | | ); |
6999 | | } |
7000 | | |
7001 | | #[test] |
7002 | | fn test_cast_from_uint16() { |
7003 | | let u16_values: Vec<u16> = vec![0, u8::MAX as u16, u16::MAX]; |
7004 | | let u16_array: ArrayRef = Arc::new(UInt16Array::from(u16_values)); |
7005 | | |
7006 | | let f64_expected = vec!["0.0", "255.0", "65535.0"]; |
7007 | | assert_eq!( |
7008 | | f64_expected, |
7009 | | get_cast_values::<Float64Type>(&u16_array, &DataType::Float64) |
7010 | | ); |
7011 | | |
7012 | | let f32_expected = vec!["0.0", "255.0", "65535.0"]; |
7013 | | assert_eq!( |
7014 | | f32_expected, |
7015 | | get_cast_values::<Float32Type>(&u16_array, &DataType::Float32) |
7016 | | ); |
7017 | | |
7018 | | let f16_expected = vec!["0.0", "255.0", "inf"]; |
7019 | | assert_eq!( |
7020 | | f16_expected, |
7021 | | get_cast_values::<Float16Type>(&u16_array, &DataType::Float16) |
7022 | | ); |
7023 | | |
7024 | | let i64_expected = vec!["0", "255", "65535"]; |
7025 | | assert_eq!( |
7026 | | i64_expected, |
7027 | | get_cast_values::<Int64Type>(&u16_array, &DataType::Int64) |
7028 | | ); |
7029 | | |
7030 | | let i32_expected = vec!["0", "255", "65535"]; |
7031 | | assert_eq!( |
7032 | | i32_expected, |
7033 | | get_cast_values::<Int32Type>(&u16_array, &DataType::Int32) |
7034 | | ); |
7035 | | |
7036 | | let i16_expected = vec!["0", "255", "null"]; |
7037 | | assert_eq!( |
7038 | | i16_expected, |
7039 | | get_cast_values::<Int16Type>(&u16_array, &DataType::Int16) |
7040 | | ); |
7041 | | |
7042 | | let i8_expected = vec!["0", "null", "null"]; |
7043 | | assert_eq!( |
7044 | | i8_expected, |
7045 | | get_cast_values::<Int8Type>(&u16_array, &DataType::Int8) |
7046 | | ); |
7047 | | |
7048 | | let u64_expected = vec!["0", "255", "65535"]; |
7049 | | assert_eq!( |
7050 | | u64_expected, |
7051 | | get_cast_values::<UInt64Type>(&u16_array, &DataType::UInt64) |
7052 | | ); |
7053 | | |
7054 | | let u32_expected = vec!["0", "255", "65535"]; |
7055 | | assert_eq!( |
7056 | | u32_expected, |
7057 | | get_cast_values::<UInt32Type>(&u16_array, &DataType::UInt32) |
7058 | | ); |
7059 | | |
7060 | | let u16_expected = vec!["0", "255", "65535"]; |
7061 | | assert_eq!( |
7062 | | u16_expected, |
7063 | | get_cast_values::<UInt16Type>(&u16_array, &DataType::UInt16) |
7064 | | ); |
7065 | | |
7066 | | let u8_expected = vec!["0", "255", "null"]; |
7067 | | assert_eq!( |
7068 | | u8_expected, |
7069 | | get_cast_values::<UInt8Type>(&u16_array, &DataType::UInt8) |
7070 | | ); |
7071 | | } |
7072 | | |
7073 | | #[test] |
7074 | | fn test_cast_from_uint8() { |
7075 | | let u8_values: Vec<u8> = vec![0, u8::MAX]; |
7076 | | let u8_array: ArrayRef = Arc::new(UInt8Array::from(u8_values)); |
7077 | | |
7078 | | let f64_expected = vec!["0.0", "255.0"]; |
7079 | | assert_eq!( |
7080 | | f64_expected, |
7081 | | get_cast_values::<Float64Type>(&u8_array, &DataType::Float64) |
7082 | | ); |
7083 | | |
7084 | | let f32_expected = vec!["0.0", "255.0"]; |
7085 | | assert_eq!( |
7086 | | f32_expected, |
7087 | | get_cast_values::<Float32Type>(&u8_array, &DataType::Float32) |
7088 | | ); |
7089 | | |
7090 | | let f16_expected = vec!["0.0", "255.0"]; |
7091 | | assert_eq!( |
7092 | | f16_expected, |
7093 | | get_cast_values::<Float16Type>(&u8_array, &DataType::Float16) |
7094 | | ); |
7095 | | |
7096 | | let i64_expected = vec!["0", "255"]; |
7097 | | assert_eq!( |
7098 | | i64_expected, |
7099 | | get_cast_values::<Int64Type>(&u8_array, &DataType::Int64) |
7100 | | ); |
7101 | | |
7102 | | let i32_expected = vec!["0", "255"]; |
7103 | | assert_eq!( |
7104 | | i32_expected, |
7105 | | get_cast_values::<Int32Type>(&u8_array, &DataType::Int32) |
7106 | | ); |
7107 | | |
7108 | | let i16_expected = vec!["0", "255"]; |
7109 | | assert_eq!( |
7110 | | i16_expected, |
7111 | | get_cast_values::<Int16Type>(&u8_array, &DataType::Int16) |
7112 | | ); |
7113 | | |
7114 | | let i8_expected = vec!["0", "null"]; |
7115 | | assert_eq!( |
7116 | | i8_expected, |
7117 | | get_cast_values::<Int8Type>(&u8_array, &DataType::Int8) |
7118 | | ); |
7119 | | |
7120 | | let u64_expected = vec!["0", "255"]; |
7121 | | assert_eq!( |
7122 | | u64_expected, |
7123 | | get_cast_values::<UInt64Type>(&u8_array, &DataType::UInt64) |
7124 | | ); |
7125 | | |
7126 | | let u32_expected = vec!["0", "255"]; |
7127 | | assert_eq!( |
7128 | | u32_expected, |
7129 | | get_cast_values::<UInt32Type>(&u8_array, &DataType::UInt32) |
7130 | | ); |
7131 | | |
7132 | | let u16_expected = vec!["0", "255"]; |
7133 | | assert_eq!( |
7134 | | u16_expected, |
7135 | | get_cast_values::<UInt16Type>(&u8_array, &DataType::UInt16) |
7136 | | ); |
7137 | | |
7138 | | let u8_expected = vec!["0", "255"]; |
7139 | | assert_eq!( |
7140 | | u8_expected, |
7141 | | get_cast_values::<UInt8Type>(&u8_array, &DataType::UInt8) |
7142 | | ); |
7143 | | } |
7144 | | |
7145 | | #[test] |
7146 | | fn test_cast_from_int64() { |
7147 | | let i64_values: Vec<i64> = vec![ |
7148 | | i64::MIN, |
7149 | | i32::MIN as i64, |
7150 | | i16::MIN as i64, |
7151 | | i8::MIN as i64, |
7152 | | 0, |
7153 | | i8::MAX as i64, |
7154 | | i16::MAX as i64, |
7155 | | i32::MAX as i64, |
7156 | | i64::MAX, |
7157 | | ]; |
7158 | | let i64_array: ArrayRef = Arc::new(Int64Array::from(i64_values)); |
7159 | | |
7160 | | let f64_expected = vec![ |
7161 | | -9223372036854776000.0, |
7162 | | -2147483648.0, |
7163 | | -32768.0, |
7164 | | -128.0, |
7165 | | 0.0, |
7166 | | 127.0, |
7167 | | 32767.0, |
7168 | | 2147483647.0, |
7169 | | 9223372036854776000.0, |
7170 | | ]; |
7171 | | assert_eq!( |
7172 | | f64_expected, |
7173 | | get_cast_values::<Float64Type>(&i64_array, &DataType::Float64) |
7174 | | .iter() |
7175 | | .map(|i| i.parse::<f64>().unwrap()) |
7176 | | .collect::<Vec<f64>>() |
7177 | | ); |
7178 | | |
7179 | | let f32_expected = vec![ |
7180 | | -9223372000000000000.0, |
7181 | | -2147483600.0, |
7182 | | -32768.0, |
7183 | | -128.0, |
7184 | | 0.0, |
7185 | | 127.0, |
7186 | | 32767.0, |
7187 | | 2147483600.0, |
7188 | | 9223372000000000000.0, |
7189 | | ]; |
7190 | | assert_eq!( |
7191 | | f32_expected, |
7192 | | get_cast_values::<Float32Type>(&i64_array, &DataType::Float32) |
7193 | | .iter() |
7194 | | .map(|i| i.parse::<f32>().unwrap()) |
7195 | | .collect::<Vec<f32>>() |
7196 | | ); |
7197 | | |
7198 | | let f16_expected = vec![ |
7199 | | f16::from_f64(-9223372000000000000.0), |
7200 | | f16::from_f64(-2147483600.0), |
7201 | | f16::from_f64(-32768.0), |
7202 | | f16::from_f64(-128.0), |
7203 | | f16::from_f64(0.0), |
7204 | | f16::from_f64(127.0), |
7205 | | f16::from_f64(32767.0), |
7206 | | f16::from_f64(2147483600.0), |
7207 | | f16::from_f64(9223372000000000000.0), |
7208 | | ]; |
7209 | | assert_eq!( |
7210 | | f16_expected, |
7211 | | get_cast_values::<Float16Type>(&i64_array, &DataType::Float16) |
7212 | | .iter() |
7213 | | .map(|i| i.parse::<f16>().unwrap()) |
7214 | | .collect::<Vec<f16>>() |
7215 | | ); |
7216 | | |
7217 | | let i64_expected = vec![ |
7218 | | "-9223372036854775808", |
7219 | | "-2147483648", |
7220 | | "-32768", |
7221 | | "-128", |
7222 | | "0", |
7223 | | "127", |
7224 | | "32767", |
7225 | | "2147483647", |
7226 | | "9223372036854775807", |
7227 | | ]; |
7228 | | assert_eq!( |
7229 | | i64_expected, |
7230 | | get_cast_values::<Int64Type>(&i64_array, &DataType::Int64) |
7231 | | ); |
7232 | | |
7233 | | let i32_expected = vec![ |
7234 | | "null", |
7235 | | "-2147483648", |
7236 | | "-32768", |
7237 | | "-128", |
7238 | | "0", |
7239 | | "127", |
7240 | | "32767", |
7241 | | "2147483647", |
7242 | | "null", |
7243 | | ]; |
7244 | | assert_eq!( |
7245 | | i32_expected, |
7246 | | get_cast_values::<Int32Type>(&i64_array, &DataType::Int32) |
7247 | | ); |
7248 | | |
7249 | | assert_eq!( |
7250 | | i32_expected, |
7251 | | get_cast_values::<Date32Type>(&i64_array, &DataType::Date32) |
7252 | | ); |
7253 | | |
7254 | | let i16_expected = vec![ |
7255 | | "null", "null", "-32768", "-128", "0", "127", "32767", "null", "null", |
7256 | | ]; |
7257 | | assert_eq!( |
7258 | | i16_expected, |
7259 | | get_cast_values::<Int16Type>(&i64_array, &DataType::Int16) |
7260 | | ); |
7261 | | |
7262 | | let i8_expected = vec![ |
7263 | | "null", "null", "null", "-128", "0", "127", "null", "null", "null", |
7264 | | ]; |
7265 | | assert_eq!( |
7266 | | i8_expected, |
7267 | | get_cast_values::<Int8Type>(&i64_array, &DataType::Int8) |
7268 | | ); |
7269 | | |
7270 | | let u64_expected = vec![ |
7271 | | "null", |
7272 | | "null", |
7273 | | "null", |
7274 | | "null", |
7275 | | "0", |
7276 | | "127", |
7277 | | "32767", |
7278 | | "2147483647", |
7279 | | "9223372036854775807", |
7280 | | ]; |
7281 | | assert_eq!( |
7282 | | u64_expected, |
7283 | | get_cast_values::<UInt64Type>(&i64_array, &DataType::UInt64) |
7284 | | ); |
7285 | | |
7286 | | let u32_expected = vec![ |
7287 | | "null", |
7288 | | "null", |
7289 | | "null", |
7290 | | "null", |
7291 | | "0", |
7292 | | "127", |
7293 | | "32767", |
7294 | | "2147483647", |
7295 | | "null", |
7296 | | ]; |
7297 | | assert_eq!( |
7298 | | u32_expected, |
7299 | | get_cast_values::<UInt32Type>(&i64_array, &DataType::UInt32) |
7300 | | ); |
7301 | | |
7302 | | let u16_expected = vec![ |
7303 | | "null", "null", "null", "null", "0", "127", "32767", "null", "null", |
7304 | | ]; |
7305 | | assert_eq!( |
7306 | | u16_expected, |
7307 | | get_cast_values::<UInt16Type>(&i64_array, &DataType::UInt16) |
7308 | | ); |
7309 | | |
7310 | | let u8_expected = vec![ |
7311 | | "null", "null", "null", "null", "0", "127", "null", "null", "null", |
7312 | | ]; |
7313 | | assert_eq!( |
7314 | | u8_expected, |
7315 | | get_cast_values::<UInt8Type>(&i64_array, &DataType::UInt8) |
7316 | | ); |
7317 | | } |
7318 | | |
7319 | | #[test] |
7320 | | fn test_cast_from_int32() { |
7321 | | let i32_values: Vec<i32> = vec![ |
7322 | | i32::MIN, |
7323 | | i16::MIN as i32, |
7324 | | i8::MIN as i32, |
7325 | | 0, |
7326 | | i8::MAX as i32, |
7327 | | i16::MAX as i32, |
7328 | | i32::MAX, |
7329 | | ]; |
7330 | | let i32_array: ArrayRef = Arc::new(Int32Array::from(i32_values)); |
7331 | | |
7332 | | let f64_expected = vec![ |
7333 | | "-2147483648.0", |
7334 | | "-32768.0", |
7335 | | "-128.0", |
7336 | | "0.0", |
7337 | | "127.0", |
7338 | | "32767.0", |
7339 | | "2147483647.0", |
7340 | | ]; |
7341 | | assert_eq!( |
7342 | | f64_expected, |
7343 | | get_cast_values::<Float64Type>(&i32_array, &DataType::Float64) |
7344 | | ); |
7345 | | |
7346 | | let f32_expected = vec![ |
7347 | | "-2147483600.0", |
7348 | | "-32768.0", |
7349 | | "-128.0", |
7350 | | "0.0", |
7351 | | "127.0", |
7352 | | "32767.0", |
7353 | | "2147483600.0", |
7354 | | ]; |
7355 | | assert_eq!( |
7356 | | f32_expected, |
7357 | | get_cast_values::<Float32Type>(&i32_array, &DataType::Float32) |
7358 | | ); |
7359 | | |
7360 | | let f16_expected = vec![ |
7361 | | f16::from_f64(-2147483600.0), |
7362 | | f16::from_f64(-32768.0), |
7363 | | f16::from_f64(-128.0), |
7364 | | f16::from_f64(0.0), |
7365 | | f16::from_f64(127.0), |
7366 | | f16::from_f64(32767.0), |
7367 | | f16::from_f64(2147483600.0), |
7368 | | ]; |
7369 | | assert_eq!( |
7370 | | f16_expected, |
7371 | | get_cast_values::<Float16Type>(&i32_array, &DataType::Float16) |
7372 | | .iter() |
7373 | | .map(|i| i.parse::<f16>().unwrap()) |
7374 | | .collect::<Vec<f16>>() |
7375 | | ); |
7376 | | |
7377 | | let i16_expected = vec!["null", "-32768", "-128", "0", "127", "32767", "null"]; |
7378 | | assert_eq!( |
7379 | | i16_expected, |
7380 | | get_cast_values::<Int16Type>(&i32_array, &DataType::Int16) |
7381 | | ); |
7382 | | |
7383 | | let i8_expected = vec!["null", "null", "-128", "0", "127", "null", "null"]; |
7384 | | assert_eq!( |
7385 | | i8_expected, |
7386 | | get_cast_values::<Int8Type>(&i32_array, &DataType::Int8) |
7387 | | ); |
7388 | | |
7389 | | let u64_expected = vec!["null", "null", "null", "0", "127", "32767", "2147483647"]; |
7390 | | assert_eq!( |
7391 | | u64_expected, |
7392 | | get_cast_values::<UInt64Type>(&i32_array, &DataType::UInt64) |
7393 | | ); |
7394 | | |
7395 | | let u32_expected = vec!["null", "null", "null", "0", "127", "32767", "2147483647"]; |
7396 | | assert_eq!( |
7397 | | u32_expected, |
7398 | | get_cast_values::<UInt32Type>(&i32_array, &DataType::UInt32) |
7399 | | ); |
7400 | | |
7401 | | let u16_expected = vec!["null", "null", "null", "0", "127", "32767", "null"]; |
7402 | | assert_eq!( |
7403 | | u16_expected, |
7404 | | get_cast_values::<UInt16Type>(&i32_array, &DataType::UInt16) |
7405 | | ); |
7406 | | |
7407 | | let u8_expected = vec!["null", "null", "null", "0", "127", "null", "null"]; |
7408 | | assert_eq!( |
7409 | | u8_expected, |
7410 | | get_cast_values::<UInt8Type>(&i32_array, &DataType::UInt8) |
7411 | | ); |
7412 | | |
7413 | | // The date32 to date64 cast increases the numerical values in order to keep the same dates. |
7414 | | let i64_expected = vec![ |
7415 | | "-185542587187200000", |
7416 | | "-2831155200000", |
7417 | | "-11059200000", |
7418 | | "0", |
7419 | | "10972800000", |
7420 | | "2831068800000", |
7421 | | "185542587100800000", |
7422 | | ]; |
7423 | | assert_eq!( |
7424 | | i64_expected, |
7425 | | get_cast_values::<Date64Type>(&i32_array, &DataType::Date64) |
7426 | | ); |
7427 | | } |
7428 | | |
7429 | | #[test] |
7430 | | fn test_cast_from_int16() { |
7431 | | let i16_values: Vec<i16> = vec![i16::MIN, i8::MIN as i16, 0, i8::MAX as i16, i16::MAX]; |
7432 | | let i16_array: ArrayRef = Arc::new(Int16Array::from(i16_values)); |
7433 | | |
7434 | | let f64_expected = vec!["-32768.0", "-128.0", "0.0", "127.0", "32767.0"]; |
7435 | | assert_eq!( |
7436 | | f64_expected, |
7437 | | get_cast_values::<Float64Type>(&i16_array, &DataType::Float64) |
7438 | | ); |
7439 | | |
7440 | | let f32_expected = vec!["-32768.0", "-128.0", "0.0", "127.0", "32767.0"]; |
7441 | | assert_eq!( |
7442 | | f32_expected, |
7443 | | get_cast_values::<Float32Type>(&i16_array, &DataType::Float32) |
7444 | | ); |
7445 | | |
7446 | | let f16_expected = vec![ |
7447 | | f16::from_f64(-32768.0), |
7448 | | f16::from_f64(-128.0), |
7449 | | f16::from_f64(0.0), |
7450 | | f16::from_f64(127.0), |
7451 | | f16::from_f64(32767.0), |
7452 | | ]; |
7453 | | assert_eq!( |
7454 | | f16_expected, |
7455 | | get_cast_values::<Float16Type>(&i16_array, &DataType::Float16) |
7456 | | .iter() |
7457 | | .map(|i| i.parse::<f16>().unwrap()) |
7458 | | .collect::<Vec<f16>>() |
7459 | | ); |
7460 | | |
7461 | | let i64_expected = vec!["-32768", "-128", "0", "127", "32767"]; |
7462 | | assert_eq!( |
7463 | | i64_expected, |
7464 | | get_cast_values::<Int64Type>(&i16_array, &DataType::Int64) |
7465 | | ); |
7466 | | |
7467 | | let i32_expected = vec!["-32768", "-128", "0", "127", "32767"]; |
7468 | | assert_eq!( |
7469 | | i32_expected, |
7470 | | get_cast_values::<Int32Type>(&i16_array, &DataType::Int32) |
7471 | | ); |
7472 | | |
7473 | | let i16_expected = vec!["-32768", "-128", "0", "127", "32767"]; |
7474 | | assert_eq!( |
7475 | | i16_expected, |
7476 | | get_cast_values::<Int16Type>(&i16_array, &DataType::Int16) |
7477 | | ); |
7478 | | |
7479 | | let i8_expected = vec!["null", "-128", "0", "127", "null"]; |
7480 | | assert_eq!( |
7481 | | i8_expected, |
7482 | | get_cast_values::<Int8Type>(&i16_array, &DataType::Int8) |
7483 | | ); |
7484 | | |
7485 | | let u64_expected = vec!["null", "null", "0", "127", "32767"]; |
7486 | | assert_eq!( |
7487 | | u64_expected, |
7488 | | get_cast_values::<UInt64Type>(&i16_array, &DataType::UInt64) |
7489 | | ); |
7490 | | |
7491 | | let u32_expected = vec!["null", "null", "0", "127", "32767"]; |
7492 | | assert_eq!( |
7493 | | u32_expected, |
7494 | | get_cast_values::<UInt32Type>(&i16_array, &DataType::UInt32) |
7495 | | ); |
7496 | | |
7497 | | let u16_expected = vec!["null", "null", "0", "127", "32767"]; |
7498 | | assert_eq!( |
7499 | | u16_expected, |
7500 | | get_cast_values::<UInt16Type>(&i16_array, &DataType::UInt16) |
7501 | | ); |
7502 | | |
7503 | | let u8_expected = vec!["null", "null", "0", "127", "null"]; |
7504 | | assert_eq!( |
7505 | | u8_expected, |
7506 | | get_cast_values::<UInt8Type>(&i16_array, &DataType::UInt8) |
7507 | | ); |
7508 | | } |
7509 | | |
7510 | | #[test] |
7511 | | fn test_cast_from_date32() { |
7512 | | let i32_values: Vec<i32> = vec![ |
7513 | | i32::MIN, |
7514 | | i16::MIN as i32, |
7515 | | i8::MIN as i32, |
7516 | | 0, |
7517 | | i8::MAX as i32, |
7518 | | i16::MAX as i32, |
7519 | | i32::MAX, |
7520 | | ]; |
7521 | | let date32_array: ArrayRef = Arc::new(Date32Array::from(i32_values)); |
7522 | | |
7523 | | let i64_expected = vec![ |
7524 | | "-2147483648", |
7525 | | "-32768", |
7526 | | "-128", |
7527 | | "0", |
7528 | | "127", |
7529 | | "32767", |
7530 | | "2147483647", |
7531 | | ]; |
7532 | | assert_eq!( |
7533 | | i64_expected, |
7534 | | get_cast_values::<Int64Type>(&date32_array, &DataType::Int64) |
7535 | | ); |
7536 | | } |
7537 | | |
7538 | | #[test] |
7539 | | fn test_cast_from_int8() { |
7540 | | let i8_values: Vec<i8> = vec![i8::MIN, 0, i8::MAX]; |
7541 | | let i8_array = Int8Array::from(i8_values); |
7542 | | |
7543 | | let f64_expected = vec!["-128.0", "0.0", "127.0"]; |
7544 | | assert_eq!( |
7545 | | f64_expected, |
7546 | | get_cast_values::<Float64Type>(&i8_array, &DataType::Float64) |
7547 | | ); |
7548 | | |
7549 | | let f32_expected = vec!["-128.0", "0.0", "127.0"]; |
7550 | | assert_eq!( |
7551 | | f32_expected, |
7552 | | get_cast_values::<Float32Type>(&i8_array, &DataType::Float32) |
7553 | | ); |
7554 | | |
7555 | | let f16_expected = vec!["-128.0", "0.0", "127.0"]; |
7556 | | assert_eq!( |
7557 | | f16_expected, |
7558 | | get_cast_values::<Float16Type>(&i8_array, &DataType::Float16) |
7559 | | ); |
7560 | | |
7561 | | let i64_expected = vec!["-128", "0", "127"]; |
7562 | | assert_eq!( |
7563 | | i64_expected, |
7564 | | get_cast_values::<Int64Type>(&i8_array, &DataType::Int64) |
7565 | | ); |
7566 | | |
7567 | | let i32_expected = vec!["-128", "0", "127"]; |
7568 | | assert_eq!( |
7569 | | i32_expected, |
7570 | | get_cast_values::<Int32Type>(&i8_array, &DataType::Int32) |
7571 | | ); |
7572 | | |
7573 | | let i16_expected = vec!["-128", "0", "127"]; |
7574 | | assert_eq!( |
7575 | | i16_expected, |
7576 | | get_cast_values::<Int16Type>(&i8_array, &DataType::Int16) |
7577 | | ); |
7578 | | |
7579 | | let i8_expected = vec!["-128", "0", "127"]; |
7580 | | assert_eq!( |
7581 | | i8_expected, |
7582 | | get_cast_values::<Int8Type>(&i8_array, &DataType::Int8) |
7583 | | ); |
7584 | | |
7585 | | let u64_expected = vec!["null", "0", "127"]; |
7586 | | assert_eq!( |
7587 | | u64_expected, |
7588 | | get_cast_values::<UInt64Type>(&i8_array, &DataType::UInt64) |
7589 | | ); |
7590 | | |
7591 | | let u32_expected = vec!["null", "0", "127"]; |
7592 | | assert_eq!( |
7593 | | u32_expected, |
7594 | | get_cast_values::<UInt32Type>(&i8_array, &DataType::UInt32) |
7595 | | ); |
7596 | | |
7597 | | let u16_expected = vec!["null", "0", "127"]; |
7598 | | assert_eq!( |
7599 | | u16_expected, |
7600 | | get_cast_values::<UInt16Type>(&i8_array, &DataType::UInt16) |
7601 | | ); |
7602 | | |
7603 | | let u8_expected = vec!["null", "0", "127"]; |
7604 | | assert_eq!( |
7605 | | u8_expected, |
7606 | | get_cast_values::<UInt8Type>(&i8_array, &DataType::UInt8) |
7607 | | ); |
7608 | | } |
7609 | | |
7610 | | /// Convert `array` into a vector of strings by casting to data type dt |
7611 | | fn get_cast_values<T>(array: &dyn Array, dt: &DataType) -> Vec<String> |
7612 | | where |
7613 | | T: ArrowPrimitiveType, |
7614 | | { |
7615 | | let c = cast(array, dt).unwrap(); |
7616 | | let a = c.as_primitive::<T>(); |
7617 | | let mut v: Vec<String> = vec![]; |
7618 | | for i in 0..array.len() { |
7619 | | if a.is_null(i) { |
7620 | | v.push("null".to_string()) |
7621 | | } else { |
7622 | | v.push(format!("{:?}", a.value(i))); |
7623 | | } |
7624 | | } |
7625 | | v |
7626 | | } |
7627 | | |
7628 | | #[test] |
7629 | | fn test_cast_utf8_dict() { |
7630 | | // FROM a dictionary with of Utf8 values |
7631 | | use DataType::*; |
7632 | | |
7633 | | let mut builder = StringDictionaryBuilder::<Int8Type>::new(); |
7634 | | builder.append("one").unwrap(); |
7635 | | builder.append_null(); |
7636 | | builder.append("three").unwrap(); |
7637 | | let array: ArrayRef = Arc::new(builder.finish()); |
7638 | | |
7639 | | let expected = vec!["one", "null", "three"]; |
7640 | | |
7641 | | // Test casting TO StringArray |
7642 | | let cast_type = Utf8; |
7643 | | let cast_array = cast(&array, &cast_type).expect("cast to UTF-8 failed"); |
7644 | | assert_eq!(cast_array.data_type(), &cast_type); |
7645 | | assert_eq!(array_to_strings(&cast_array), expected); |
7646 | | |
7647 | | // Test casting TO Dictionary (with different index sizes) |
7648 | | |
7649 | | let cast_type = Dictionary(Box::new(Int16), Box::new(Utf8)); |
7650 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7651 | | assert_eq!(cast_array.data_type(), &cast_type); |
7652 | | assert_eq!(array_to_strings(&cast_array), expected); |
7653 | | |
7654 | | let cast_type = Dictionary(Box::new(Int32), Box::new(Utf8)); |
7655 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7656 | | assert_eq!(cast_array.data_type(), &cast_type); |
7657 | | assert_eq!(array_to_strings(&cast_array), expected); |
7658 | | |
7659 | | let cast_type = Dictionary(Box::new(Int64), Box::new(Utf8)); |
7660 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7661 | | assert_eq!(cast_array.data_type(), &cast_type); |
7662 | | assert_eq!(array_to_strings(&cast_array), expected); |
7663 | | |
7664 | | let cast_type = Dictionary(Box::new(UInt8), Box::new(Utf8)); |
7665 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7666 | | assert_eq!(cast_array.data_type(), &cast_type); |
7667 | | assert_eq!(array_to_strings(&cast_array), expected); |
7668 | | |
7669 | | let cast_type = Dictionary(Box::new(UInt16), Box::new(Utf8)); |
7670 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7671 | | assert_eq!(cast_array.data_type(), &cast_type); |
7672 | | assert_eq!(array_to_strings(&cast_array), expected); |
7673 | | |
7674 | | let cast_type = Dictionary(Box::new(UInt32), Box::new(Utf8)); |
7675 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7676 | | assert_eq!(cast_array.data_type(), &cast_type); |
7677 | | assert_eq!(array_to_strings(&cast_array), expected); |
7678 | | |
7679 | | let cast_type = Dictionary(Box::new(UInt64), Box::new(Utf8)); |
7680 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7681 | | assert_eq!(cast_array.data_type(), &cast_type); |
7682 | | assert_eq!(array_to_strings(&cast_array), expected); |
7683 | | } |
7684 | | |
7685 | | #[test] |
7686 | | fn test_cast_dict_to_dict_bad_index_value_primitive() { |
7687 | | use DataType::*; |
7688 | | // test converting from an array that has indexes of a type |
7689 | | // that are out of bounds for a particular other kind of |
7690 | | // index. |
7691 | | |
7692 | | let mut builder = PrimitiveDictionaryBuilder::<Int32Type, Int64Type>::new(); |
7693 | | |
7694 | | // add 200 distinct values (which can be stored by a |
7695 | | // dictionary indexed by int32, but not a dictionary indexed |
7696 | | // with int8) |
7697 | | for i in 0..200 { |
7698 | | builder.append(i).unwrap(); |
7699 | | } |
7700 | | let array: ArrayRef = Arc::new(builder.finish()); |
7701 | | |
7702 | | let cast_type = Dictionary(Box::new(Int8), Box::new(Utf8)); |
7703 | | let res = cast(&array, &cast_type); |
7704 | | assert!(res.is_err()); |
7705 | | let actual_error = format!("{res:?}"); |
7706 | | let expected_error = "Could not convert 72 dictionary indexes from Int32 to Int8"; |
7707 | | assert!( |
7708 | | actual_error.contains(expected_error), |
7709 | | "did not find expected error '{actual_error}' in actual error '{expected_error}'" |
7710 | | ); |
7711 | | } |
7712 | | |
7713 | | #[test] |
7714 | | fn test_cast_dict_to_dict_bad_index_value_utf8() { |
7715 | | use DataType::*; |
7716 | | // Same test as test_cast_dict_to_dict_bad_index_value but use |
7717 | | // string values (and encode the expected behavior here); |
7718 | | |
7719 | | let mut builder = StringDictionaryBuilder::<Int32Type>::new(); |
7720 | | |
7721 | | // add 200 distinct values (which can be stored by a |
7722 | | // dictionary indexed by int32, but not a dictionary indexed |
7723 | | // with int8) |
7724 | | for i in 0..200 { |
7725 | | let val = format!("val{i}"); |
7726 | | builder.append(&val).unwrap(); |
7727 | | } |
7728 | | let array = builder.finish(); |
7729 | | |
7730 | | let cast_type = Dictionary(Box::new(Int8), Box::new(Utf8)); |
7731 | | let res = cast(&array, &cast_type); |
7732 | | assert!(res.is_err()); |
7733 | | let actual_error = format!("{res:?}"); |
7734 | | let expected_error = "Could not convert 72 dictionary indexes from Int32 to Int8"; |
7735 | | assert!( |
7736 | | actual_error.contains(expected_error), |
7737 | | "did not find expected error '{actual_error}' in actual error '{expected_error}'" |
7738 | | ); |
7739 | | } |
7740 | | |
7741 | | #[test] |
7742 | | fn test_cast_primitive_dict() { |
7743 | | // FROM a dictionary with of INT32 values |
7744 | | use DataType::*; |
7745 | | |
7746 | | let mut builder = PrimitiveDictionaryBuilder::<Int8Type, Int32Type>::new(); |
7747 | | builder.append(1).unwrap(); |
7748 | | builder.append_null(); |
7749 | | builder.append(3).unwrap(); |
7750 | | let array: ArrayRef = Arc::new(builder.finish()); |
7751 | | |
7752 | | let expected = vec!["1", "null", "3"]; |
7753 | | |
7754 | | // Test casting TO PrimitiveArray, different dictionary type |
7755 | | let cast_array = cast(&array, &Utf8).expect("cast to UTF-8 failed"); |
7756 | | assert_eq!(array_to_strings(&cast_array), expected); |
7757 | | assert_eq!(cast_array.data_type(), &Utf8); |
7758 | | |
7759 | | let cast_array = cast(&array, &Int64).expect("cast to int64 failed"); |
7760 | | assert_eq!(array_to_strings(&cast_array), expected); |
7761 | | assert_eq!(cast_array.data_type(), &Int64); |
7762 | | } |
7763 | | |
7764 | | #[test] |
7765 | | fn test_cast_primitive_array_to_dict() { |
7766 | | use DataType::*; |
7767 | | |
7768 | | let mut builder = PrimitiveBuilder::<Int32Type>::new(); |
7769 | | builder.append_value(1); |
7770 | | builder.append_null(); |
7771 | | builder.append_value(3); |
7772 | | let array: ArrayRef = Arc::new(builder.finish()); |
7773 | | |
7774 | | let expected = vec!["1", "null", "3"]; |
7775 | | |
7776 | | // Cast to a dictionary (same value type, Int32) |
7777 | | let cast_type = Dictionary(Box::new(UInt8), Box::new(Int32)); |
7778 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7779 | | assert_eq!(cast_array.data_type(), &cast_type); |
7780 | | assert_eq!(array_to_strings(&cast_array), expected); |
7781 | | |
7782 | | // Cast to a dictionary (different value type, Int8) |
7783 | | let cast_type = Dictionary(Box::new(UInt8), Box::new(Int8)); |
7784 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7785 | | assert_eq!(cast_array.data_type(), &cast_type); |
7786 | | assert_eq!(array_to_strings(&cast_array), expected); |
7787 | | } |
7788 | | |
7789 | | #[test] |
7790 | | fn test_cast_time_array_to_dict() { |
7791 | | use DataType::*; |
7792 | | |
7793 | | let array = Arc::new(Date32Array::from(vec![Some(1000), None, Some(2000)])) as ArrayRef; |
7794 | | |
7795 | | let expected = vec!["1972-09-27", "null", "1975-06-24"]; |
7796 | | |
7797 | | let cast_type = Dictionary(Box::new(UInt8), Box::new(Date32)); |
7798 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7799 | | assert_eq!(cast_array.data_type(), &cast_type); |
7800 | | assert_eq!(array_to_strings(&cast_array), expected); |
7801 | | } |
7802 | | |
7803 | | #[test] |
7804 | | fn test_cast_timestamp_array_to_dict() { |
7805 | | use DataType::*; |
7806 | | |
7807 | | let array = Arc::new( |
7808 | | TimestampSecondArray::from(vec![Some(1000), None, Some(2000)]).with_timezone_utc(), |
7809 | | ) as ArrayRef; |
7810 | | |
7811 | | let expected = vec!["1970-01-01T00:16:40", "null", "1970-01-01T00:33:20"]; |
7812 | | |
7813 | | let cast_type = Dictionary(Box::new(UInt8), Box::new(Timestamp(TimeUnit::Second, None))); |
7814 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7815 | | assert_eq!(cast_array.data_type(), &cast_type); |
7816 | | assert_eq!(array_to_strings(&cast_array), expected); |
7817 | | } |
7818 | | |
7819 | | #[test] |
7820 | | fn test_cast_string_array_to_dict() { |
7821 | | use DataType::*; |
7822 | | |
7823 | | let array = Arc::new(StringArray::from(vec![Some("one"), None, Some("three")])) as ArrayRef; |
7824 | | |
7825 | | let expected = vec!["one", "null", "three"]; |
7826 | | |
7827 | | // Cast to a dictionary (same value type, Utf8) |
7828 | | let cast_type = Dictionary(Box::new(UInt8), Box::new(Utf8)); |
7829 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7830 | | assert_eq!(cast_array.data_type(), &cast_type); |
7831 | | assert_eq!(array_to_strings(&cast_array), expected); |
7832 | | } |
7833 | | |
7834 | | #[test] |
7835 | | fn test_cast_null_array_to_from_decimal_array() { |
7836 | | let data_type = DataType::Decimal128(12, 4); |
7837 | | let array = new_null_array(&DataType::Null, 4); |
7838 | | assert_eq!(array.data_type(), &DataType::Null); |
7839 | | let cast_array = cast(&array, &data_type).expect("cast failed"); |
7840 | | assert_eq!(cast_array.data_type(), &data_type); |
7841 | | for i in 0..4 { |
7842 | | assert!(cast_array.is_null(i)); |
7843 | | } |
7844 | | |
7845 | | let array = new_null_array(&data_type, 4); |
7846 | | assert_eq!(array.data_type(), &data_type); |
7847 | | let cast_array = cast(&array, &DataType::Null).expect("cast failed"); |
7848 | | assert_eq!(cast_array.data_type(), &DataType::Null); |
7849 | | assert_eq!(cast_array.len(), 4); |
7850 | | assert_eq!(cast_array.logical_nulls().unwrap().null_count(), 4); |
7851 | | } |
7852 | | |
7853 | | #[test] |
7854 | | fn test_cast_null_array_from_and_to_primitive_array() { |
7855 | | macro_rules! typed_test { |
7856 | | ($ARR_TYPE:ident, $DATATYPE:ident, $TYPE:tt) => {{ |
7857 | | { |
7858 | | let array = Arc::new(NullArray::new(6)) as ArrayRef; |
7859 | | let expected = $ARR_TYPE::from(vec![None; 6]); |
7860 | | let cast_type = DataType::$DATATYPE; |
7861 | | let cast_array = cast(&array, &cast_type).expect("cast failed"); |
7862 | | let cast_array = cast_array.as_primitive::<$TYPE>(); |
7863 | | assert_eq!(cast_array.data_type(), &cast_type); |
7864 | | assert_eq!(cast_array, &expected); |
7865 | | } |
7866 | | }}; |
7867 | | } |
7868 | | |
7869 | | typed_test!(Int16Array, Int16, Int16Type); |
7870 | | typed_test!(Int32Array, Int32, Int32Type); |
7871 | | typed_test!(Int64Array, Int64, Int64Type); |
7872 | | |
7873 | | typed_test!(UInt16Array, UInt16, UInt16Type); |
7874 | | typed_test!(UInt32Array, UInt32, UInt32Type); |
7875 | | typed_test!(UInt64Array, UInt64, UInt64Type); |
7876 | | |
7877 | | typed_test!(Float32Array, Float32, Float32Type); |
7878 | | typed_test!(Float64Array, Float64, Float64Type); |
7879 | | |
7880 | | typed_test!(Date32Array, Date32, Date32Type); |
7881 | | typed_test!(Date64Array, Date64, Date64Type); |
7882 | | } |
7883 | | |
7884 | | fn cast_from_null_to_other(data_type: &DataType) { |
7885 | | // Cast from null to data_type |
7886 | | { |
7887 | | let array = new_null_array(&DataType::Null, 4); |
7888 | | assert_eq!(array.data_type(), &DataType::Null); |
7889 | | let cast_array = cast(&array, data_type).expect("cast failed"); |
7890 | | assert_eq!(cast_array.data_type(), data_type); |
7891 | | for i in 0..4 { |
7892 | | assert!(cast_array.is_null(i)); |
7893 | | } |
7894 | | } |
7895 | | } |
7896 | | |
7897 | | #[test] |
7898 | | fn test_cast_null_from_and_to_variable_sized() { |
7899 | | cast_from_null_to_other(&DataType::Utf8); |
7900 | | cast_from_null_to_other(&DataType::LargeUtf8); |
7901 | | cast_from_null_to_other(&DataType::Binary); |
7902 | | cast_from_null_to_other(&DataType::LargeBinary); |
7903 | | } |
7904 | | |
7905 | | #[test] |
7906 | | fn test_cast_null_from_and_to_nested_type() { |
7907 | | // Cast null from and to map |
7908 | | let data_type = DataType::Map( |
7909 | | Arc::new(Field::new_struct( |
7910 | | "entry", |
7911 | | vec![ |
7912 | | Field::new("key", DataType::Utf8, false), |
7913 | | Field::new("value", DataType::Int32, true), |
7914 | | ], |
7915 | | false, |
7916 | | )), |
7917 | | false, |
7918 | | ); |
7919 | | cast_from_null_to_other(&data_type); |
7920 | | |
7921 | | // Cast null from and to list |
7922 | | let data_type = DataType::List(Arc::new(Field::new_list_field(DataType::Int32, true))); |
7923 | | cast_from_null_to_other(&data_type); |
7924 | | let data_type = DataType::LargeList(Arc::new(Field::new_list_field(DataType::Int32, true))); |
7925 | | cast_from_null_to_other(&data_type); |
7926 | | let data_type = |
7927 | | DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int32, true)), 4); |
7928 | | cast_from_null_to_other(&data_type); |
7929 | | |
7930 | | // Cast null from and to dictionary |
7931 | | let values = vec![None, None, None, None] as Vec<Option<&str>>; |
7932 | | let array: DictionaryArray<Int8Type> = values.into_iter().collect(); |
7933 | | let array = Arc::new(array) as ArrayRef; |
7934 | | let data_type = array.data_type().to_owned(); |
7935 | | cast_from_null_to_other(&data_type); |
7936 | | |
7937 | | // Cast null from and to struct |
7938 | | let data_type = DataType::Struct(vec![Field::new("data", DataType::Int64, false)].into()); |
7939 | | cast_from_null_to_other(&data_type); |
7940 | | } |
7941 | | |
7942 | | /// Print the `DictionaryArray` `array` as a vector of strings |
7943 | | fn array_to_strings(array: &ArrayRef) -> Vec<String> { |
7944 | | let options = FormatOptions::new().with_null("null"); |
7945 | | let formatter = ArrayFormatter::try_new(array.as_ref(), &options).unwrap(); |
7946 | | (0..array.len()) |
7947 | | .map(|i| formatter.value(i).to_string()) |
7948 | | .collect() |
7949 | | } |
7950 | | |
7951 | | #[test] |
7952 | | fn test_cast_utf8_to_date32() { |
7953 | | use chrono::NaiveDate; |
7954 | | let from_ymd = chrono::NaiveDate::from_ymd_opt; |
7955 | | let since = chrono::NaiveDate::signed_duration_since; |
7956 | | |
7957 | | let a = StringArray::from(vec![ |
7958 | | "2000-01-01", // valid date with leading 0s |
7959 | | "2000-01-01T12:00:00", // valid datetime, will throw away the time part |
7960 | | "2000-2-2", // valid date without leading 0s |
7961 | | "2000-00-00", // invalid month and day |
7962 | | "2000", // just a year is invalid |
7963 | | ]); |
7964 | | let array = Arc::new(a) as ArrayRef; |
7965 | | let b = cast(&array, &DataType::Date32).unwrap(); |
7966 | | let c = b.as_primitive::<Date32Type>(); |
7967 | | |
7968 | | // test valid inputs |
7969 | | let date_value = since( |
7970 | | NaiveDate::from_ymd_opt(2000, 1, 1).unwrap(), |
7971 | | from_ymd(1970, 1, 1).unwrap(), |
7972 | | ) |
7973 | | .num_days() as i32; |
7974 | | assert!(c.is_valid(0)); // "2000-01-01" |
7975 | | assert_eq!(date_value, c.value(0)); |
7976 | | |
7977 | | assert!(c.is_valid(1)); // "2000-01-01T12:00:00" |
7978 | | assert_eq!(date_value, c.value(1)); |
7979 | | |
7980 | | let date_value = since( |
7981 | | NaiveDate::from_ymd_opt(2000, 2, 2).unwrap(), |
7982 | | from_ymd(1970, 1, 1).unwrap(), |
7983 | | ) |
7984 | | .num_days() as i32; |
7985 | | assert!(c.is_valid(2)); // "2000-2-2" |
7986 | | assert_eq!(date_value, c.value(2)); |
7987 | | |
7988 | | // test invalid inputs |
7989 | | assert!(!c.is_valid(3)); // "2000-00-00" |
7990 | | assert!(!c.is_valid(4)); // "2000" |
7991 | | } |
7992 | | |
7993 | | #[test] |
7994 | | fn test_cast_utf8_to_date64() { |
7995 | | let a = StringArray::from(vec![ |
7996 | | "2000-01-01T12:00:00", // date + time valid |
7997 | | "2020-12-15T12:34:56", // date + time valid |
7998 | | "2020-2-2T12:34:56", // valid date time without leading 0s |
7999 | | "2000-00-00T12:00:00", // invalid month and day |
8000 | | "2000-01-01 12:00:00", // missing the 'T' |
8001 | | "2000-01-01", // just a date is invalid |
8002 | | ]); |
8003 | | let array = Arc::new(a) as ArrayRef; |
8004 | | let b = cast(&array, &DataType::Date64).unwrap(); |
8005 | | let c = b.as_primitive::<Date64Type>(); |
8006 | | |
8007 | | // test valid inputs |
8008 | | assert!(c.is_valid(0)); // "2000-01-01T12:00:00" |
8009 | | assert_eq!(946728000000, c.value(0)); |
8010 | | assert!(c.is_valid(1)); // "2020-12-15T12:34:56" |
8011 | | assert_eq!(1608035696000, c.value(1)); |
8012 | | assert!(!c.is_valid(2)); // "2020-2-2T12:34:56" |
8013 | | |
8014 | | assert!(!c.is_valid(3)); // "2000-00-00T12:00:00" |
8015 | | assert!(c.is_valid(4)); // "2000-01-01 12:00:00" |
8016 | | assert_eq!(946728000000, c.value(4)); |
8017 | | assert!(c.is_valid(5)); // "2000-01-01" |
8018 | | assert_eq!(946684800000, c.value(5)); |
8019 | | } |
8020 | | |
8021 | | #[test] |
8022 | | fn test_can_cast_fsl_to_fsl() { |
8023 | | let from_array = Arc::new( |
8024 | | FixedSizeListArray::from_iter_primitive::<Float32Type, _, _>( |
8025 | | [Some([Some(1.0), Some(2.0)]), None], |
8026 | | 2, |
8027 | | ), |
8028 | | ) as ArrayRef; |
8029 | | let to_array = Arc::new( |
8030 | | FixedSizeListArray::from_iter_primitive::<Float16Type, _, _>( |
8031 | | [ |
8032 | | Some([Some(f16::from_f32(1.0)), Some(f16::from_f32(2.0))]), |
8033 | | None, |
8034 | | ], |
8035 | | 2, |
8036 | | ), |
8037 | | ) as ArrayRef; |
8038 | | |
8039 | | assert!(can_cast_types(from_array.data_type(), to_array.data_type())); |
8040 | | let actual = cast(&from_array, to_array.data_type()).unwrap(); |
8041 | | assert_eq!(actual.data_type(), to_array.data_type()); |
8042 | | |
8043 | | let invalid_target = |
8044 | | DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Binary, true)), 2); |
8045 | | assert!(!can_cast_types(from_array.data_type(), &invalid_target)); |
8046 | | |
8047 | | let invalid_size = |
8048 | | DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Float16, true)), 5); |
8049 | | assert!(!can_cast_types(from_array.data_type(), &invalid_size)); |
8050 | | } |
8051 | | |
8052 | | #[test] |
8053 | | fn test_can_cast_types_fixed_size_list_to_list() { |
8054 | | // DataType::List |
8055 | | let array1 = Arc::new(make_fixed_size_list_array()) as ArrayRef; |
8056 | | assert!(can_cast_types( |
8057 | | array1.data_type(), |
8058 | | &DataType::List(Arc::new(Field::new("", DataType::Int32, false))) |
8059 | | )); |
8060 | | |
8061 | | // DataType::LargeList |
8062 | | let array2 = Arc::new(make_fixed_size_list_array_for_large_list()) as ArrayRef; |
8063 | | assert!(can_cast_types( |
8064 | | array2.data_type(), |
8065 | | &DataType::LargeList(Arc::new(Field::new("", DataType::Int64, false))) |
8066 | | )); |
8067 | | } |
8068 | | |
8069 | | #[test] |
8070 | | fn test_cast_fixed_size_list_to_list() { |
8071 | | // Important cases: |
8072 | | // 1. With/without nulls |
8073 | | // 2. LargeList and List |
8074 | | // 3. With and without inner casts |
8075 | | |
8076 | | let cases = [ |
8077 | | // fixed_size_list<i32, 2> => list<i32> |
8078 | | ( |
8079 | | Arc::new(FixedSizeListArray::from_iter_primitive::<Int32Type, _, _>( |
8080 | | [[1, 1].map(Some), [2, 2].map(Some)].map(Some), |
8081 | | 2, |
8082 | | )) as ArrayRef, |
8083 | | Arc::new(ListArray::from_iter_primitive::<Int32Type, _, _>([ |
8084 | | Some([Some(1), Some(1)]), |
8085 | | Some([Some(2), Some(2)]), |
8086 | | ])) as ArrayRef, |
8087 | | ), |
8088 | | // fixed_size_list<i32, 2> => list<i32> (nullable) |
8089 | | ( |
8090 | | Arc::new(FixedSizeListArray::from_iter_primitive::<Int32Type, _, _>( |
8091 | | [None, Some([Some(2), Some(2)])], |
8092 | | 2, |
8093 | | )) as ArrayRef, |
8094 | | Arc::new(ListArray::from_iter_primitive::<Int32Type, _, _>([ |
8095 | | None, |
8096 | | Some([Some(2), Some(2)]), |
8097 | | ])) as ArrayRef, |
8098 | | ), |
8099 | | // fixed_size_list<i32, 2> => large_list<i64> |
8100 | | ( |
8101 | | Arc::new(FixedSizeListArray::from_iter_primitive::<Int32Type, _, _>( |
8102 | | [[1, 1].map(Some), [2, 2].map(Some)].map(Some), |
8103 | | 2, |
8104 | | )) as ArrayRef, |
8105 | | Arc::new(LargeListArray::from_iter_primitive::<Int64Type, _, _>([ |
8106 | | Some([Some(1), Some(1)]), |
8107 | | Some([Some(2), Some(2)]), |
8108 | | ])) as ArrayRef, |
8109 | | ), |
8110 | | // fixed_size_list<i32, 2> => large_list<i64> (nullable) |
8111 | | ( |
8112 | | Arc::new(FixedSizeListArray::from_iter_primitive::<Int32Type, _, _>( |
8113 | | [None, Some([Some(2), Some(2)])], |
8114 | | 2, |
8115 | | )) as ArrayRef, |
8116 | | Arc::new(LargeListArray::from_iter_primitive::<Int64Type, _, _>([ |
8117 | | None, |
8118 | | Some([Some(2), Some(2)]), |
8119 | | ])) as ArrayRef, |
8120 | | ), |
8121 | | ]; |
8122 | | |
8123 | | for (array, expected) in cases { |
8124 | | let array = Arc::new(array) as ArrayRef; |
8125 | | |
8126 | | assert!( |
8127 | | can_cast_types(array.data_type(), expected.data_type()), |
8128 | | "can_cast_types claims we cannot cast {:?} to {:?}", |
8129 | | array.data_type(), |
8130 | | expected.data_type() |
8131 | | ); |
8132 | | |
8133 | | let list_array = cast(&array, expected.data_type()) |
8134 | | .unwrap_or_else(|_| panic!("Failed to cast {array:?} to {expected:?}")); |
8135 | | assert_eq!( |
8136 | | list_array.as_ref(), |
8137 | | &expected, |
8138 | | "Incorrect result from casting {array:?} to {expected:?}", |
8139 | | ); |
8140 | | } |
8141 | | } |
8142 | | |
8143 | | #[test] |
8144 | | fn test_cast_utf8_to_list() { |
8145 | | // DataType::List |
8146 | | let array = Arc::new(StringArray::from(vec!["5"])) as ArrayRef; |
8147 | | let field = Arc::new(Field::new("", DataType::Int32, false)); |
8148 | | let list_array = cast(&array, &DataType::List(field.clone())).unwrap(); |
8149 | | let actual = list_array.as_list_opt::<i32>().unwrap(); |
8150 | | let expect = ListArray::from_iter_primitive::<Int32Type, _, _>([Some([Some(5)])]); |
8151 | | assert_eq!(&expect.value(0), &actual.value(0)); |
8152 | | |
8153 | | // DataType::LargeList |
8154 | | let list_array = cast(&array, &DataType::LargeList(field.clone())).unwrap(); |
8155 | | let actual = list_array.as_list_opt::<i64>().unwrap(); |
8156 | | let expect = LargeListArray::from_iter_primitive::<Int32Type, _, _>([Some([Some(5)])]); |
8157 | | assert_eq!(&expect.value(0), &actual.value(0)); |
8158 | | |
8159 | | // DataType::FixedSizeList |
8160 | | let list_array = cast(&array, &DataType::FixedSizeList(field.clone(), 1)).unwrap(); |
8161 | | let actual = list_array.as_fixed_size_list_opt().unwrap(); |
8162 | | let expect = |
8163 | | FixedSizeListArray::from_iter_primitive::<Int32Type, _, _>([Some([Some(5)])], 1); |
8164 | | assert_eq!(&expect.value(0), &actual.value(0)); |
8165 | | } |
8166 | | |
8167 | | #[test] |
8168 | | fn test_cast_single_element_fixed_size_list() { |
8169 | | // FixedSizeList<T>[1] => T |
8170 | | let from_array = Arc::new(FixedSizeListArray::from_iter_primitive::<Int16Type, _, _>( |
8171 | | [(Some([Some(5)]))], |
8172 | | 1, |
8173 | | )) as ArrayRef; |
8174 | | let casted_array = cast(&from_array, &DataType::Int32).unwrap(); |
8175 | | let actual: &Int32Array = casted_array.as_primitive(); |
8176 | | let expected = Int32Array::from(vec![Some(5)]); |
8177 | | assert_eq!(&expected, actual); |
8178 | | |
8179 | | // FixedSizeList<T>[1] => FixedSizeList<U>[1] |
8180 | | let from_array = Arc::new(FixedSizeListArray::from_iter_primitive::<Int16Type, _, _>( |
8181 | | [(Some([Some(5)]))], |
8182 | | 1, |
8183 | | )) as ArrayRef; |
8184 | | let to_field = Arc::new(Field::new("dummy", DataType::Float32, false)); |
8185 | | let actual = cast(&from_array, &DataType::FixedSizeList(to_field.clone(), 1)).unwrap(); |
8186 | | let expected = Arc::new(FixedSizeListArray::new( |
8187 | | to_field.clone(), |
8188 | | 1, |
8189 | | Arc::new(Float32Array::from(vec![Some(5.0)])) as ArrayRef, |
8190 | | None, |
8191 | | )) as ArrayRef; |
8192 | | assert_eq!(*expected, *actual); |
8193 | | |
8194 | | // FixedSizeList<T>[1] => FixedSizeList<FixdSizedList<U>[1]>[1] |
8195 | | let from_array = Arc::new(FixedSizeListArray::from_iter_primitive::<Int16Type, _, _>( |
8196 | | [(Some([Some(5)]))], |
8197 | | 1, |
8198 | | )) as ArrayRef; |
8199 | | let to_field_inner = Arc::new(Field::new_list_field(DataType::Float32, false)); |
8200 | | let to_field = Arc::new(Field::new( |
8201 | | "dummy", |
8202 | | DataType::FixedSizeList(to_field_inner.clone(), 1), |
8203 | | false, |
8204 | | )); |
8205 | | let actual = cast(&from_array, &DataType::FixedSizeList(to_field.clone(), 1)).unwrap(); |
8206 | | let expected = Arc::new(FixedSizeListArray::new( |
8207 | | to_field.clone(), |
8208 | | 1, |
8209 | | Arc::new(FixedSizeListArray::new( |
8210 | | to_field_inner.clone(), |
8211 | | 1, |
8212 | | Arc::new(Float32Array::from(vec![Some(5.0)])) as ArrayRef, |
8213 | | None, |
8214 | | )) as ArrayRef, |
8215 | | None, |
8216 | | )) as ArrayRef; |
8217 | | assert_eq!(*expected, *actual); |
8218 | | |
8219 | | // T => FixedSizeList<T>[1] (non-nullable) |
8220 | | let field = Arc::new(Field::new("dummy", DataType::Float32, false)); |
8221 | | let from_array = Arc::new(Int8Array::from(vec![Some(5)])) as ArrayRef; |
8222 | | let casted_array = cast(&from_array, &DataType::FixedSizeList(field.clone(), 1)).unwrap(); |
8223 | | let actual = casted_array.as_fixed_size_list(); |
8224 | | let expected = Arc::new(FixedSizeListArray::new( |
8225 | | field.clone(), |
8226 | | 1, |
8227 | | Arc::new(Float32Array::from(vec![Some(5.0)])) as ArrayRef, |
8228 | | None, |
8229 | | )) as ArrayRef; |
8230 | | assert_eq!(expected.as_ref(), actual); |
8231 | | |
8232 | | // T => FixedSizeList<T>[1] (nullable) |
8233 | | let field = Arc::new(Field::new("nullable", DataType::Float32, true)); |
8234 | | let from_array = Arc::new(Int8Array::from(vec![None])) as ArrayRef; |
8235 | | let casted_array = cast(&from_array, &DataType::FixedSizeList(field.clone(), 1)).unwrap(); |
8236 | | let actual = casted_array.as_fixed_size_list(); |
8237 | | let expected = Arc::new(FixedSizeListArray::new( |
8238 | | field.clone(), |
8239 | | 1, |
8240 | | Arc::new(Float32Array::from(vec![None])) as ArrayRef, |
8241 | | None, |
8242 | | )) as ArrayRef; |
8243 | | assert_eq!(expected.as_ref(), actual); |
8244 | | } |
8245 | | |
8246 | | #[test] |
8247 | | fn test_cast_list_containers() { |
8248 | | // large-list to list |
8249 | | let array = Arc::new(make_large_list_array()) as ArrayRef; |
8250 | | let list_array = cast( |
8251 | | &array, |
8252 | | &DataType::List(Arc::new(Field::new("", DataType::Int32, false))), |
8253 | | ) |
8254 | | .unwrap(); |
8255 | | let actual = list_array.as_any().downcast_ref::<ListArray>().unwrap(); |
8256 | | let expected = array.as_any().downcast_ref::<LargeListArray>().unwrap(); |
8257 | | |
8258 | | assert_eq!(&expected.value(0), &actual.value(0)); |
8259 | | assert_eq!(&expected.value(1), &actual.value(1)); |
8260 | | assert_eq!(&expected.value(2), &actual.value(2)); |
8261 | | |
8262 | | // list to large-list |
8263 | | let array = Arc::new(make_list_array()) as ArrayRef; |
8264 | | let large_list_array = cast( |
8265 | | &array, |
8266 | | &DataType::LargeList(Arc::new(Field::new("", DataType::Int32, false))), |
8267 | | ) |
8268 | | .unwrap(); |
8269 | | let actual = large_list_array |
8270 | | .as_any() |
8271 | | .downcast_ref::<LargeListArray>() |
8272 | | .unwrap(); |
8273 | | let expected = array.as_any().downcast_ref::<ListArray>().unwrap(); |
8274 | | |
8275 | | assert_eq!(&expected.value(0), &actual.value(0)); |
8276 | | assert_eq!(&expected.value(1), &actual.value(1)); |
8277 | | assert_eq!(&expected.value(2), &actual.value(2)); |
8278 | | } |
8279 | | |
8280 | | #[test] |
8281 | | fn test_cast_list_to_fsl() { |
8282 | | // There four noteworthy cases we should handle: |
8283 | | // 1. No nulls |
8284 | | // 2. Nulls that are always empty |
8285 | | // 3. Nulls that have varying lengths |
8286 | | // 4. Nulls that are correctly sized (same as target list size) |
8287 | | |
8288 | | // Non-null case |
8289 | | let field = Arc::new(Field::new_list_field(DataType::Int32, true)); |
8290 | | let values = vec![ |
8291 | | Some(vec![Some(1), Some(2), Some(3)]), |
8292 | | Some(vec![Some(4), Some(5), Some(6)]), |
8293 | | ]; |
8294 | | let array = Arc::new(ListArray::from_iter_primitive::<Int32Type, _, _>( |
8295 | | values.clone(), |
8296 | | )) as ArrayRef; |
8297 | | let expected = Arc::new(FixedSizeListArray::from_iter_primitive::<Int32Type, _, _>( |
8298 | | values, 3, |
8299 | | )) as ArrayRef; |
8300 | | let actual = cast(array.as_ref(), &DataType::FixedSizeList(field.clone(), 3)).unwrap(); |
8301 | | assert_eq!(expected.as_ref(), actual.as_ref()); |
8302 | | |
8303 | | // Null cases |
8304 | | // Array is [[1, 2, 3], null, [4, 5, 6], null] |
8305 | | let cases = [ |
8306 | | ( |
8307 | | // Zero-length nulls |
8308 | | vec![1, 2, 3, 4, 5, 6], |
8309 | | vec![3, 0, 3, 0], |
8310 | | ), |
8311 | | ( |
8312 | | // Varying-length nulls |
8313 | | vec![1, 2, 3, 0, 0, 4, 5, 6, 0], |
8314 | | vec![3, 2, 3, 1], |
8315 | | ), |
8316 | | ( |
8317 | | // Correctly-sized nulls |
8318 | | vec![1, 2, 3, 0, 0, 0, 4, 5, 6, 0, 0, 0], |
8319 | | vec![3, 3, 3, 3], |
8320 | | ), |
8321 | | ( |
8322 | | // Mixed nulls |
8323 | | vec![1, 2, 3, 4, 5, 6, 0, 0, 0], |
8324 | | vec![3, 0, 3, 3], |
8325 | | ), |
8326 | | ]; |
8327 | | let null_buffer = NullBuffer::from(vec![true, false, true, false]); |
8328 | | |
8329 | | let expected = Arc::new(FixedSizeListArray::from_iter_primitive::<Int32Type, _, _>( |
8330 | | vec![ |
8331 | | Some(vec![Some(1), Some(2), Some(3)]), |
8332 | | None, |
8333 | | Some(vec![Some(4), Some(5), Some(6)]), |
8334 | | None, |
8335 | | ], |
8336 | | 3, |
8337 | | )) as ArrayRef; |
8338 | | |
8339 | | for (values, lengths) in cases.iter() { |
8340 | | let array = Arc::new(ListArray::new( |
8341 | | field.clone(), |
8342 | | OffsetBuffer::from_lengths(lengths.clone()), |
8343 | | Arc::new(Int32Array::from(values.clone())), |
8344 | | Some(null_buffer.clone()), |
8345 | | )) as ArrayRef; |
8346 | | let actual = cast(array.as_ref(), &DataType::FixedSizeList(field.clone(), 3)).unwrap(); |
8347 | | assert_eq!(expected.as_ref(), actual.as_ref()); |
8348 | | } |
8349 | | } |
8350 | | |
8351 | | #[test] |
8352 | | fn test_cast_list_to_fsl_safety() { |
8353 | | let values = vec![ |
8354 | | Some(vec![Some(1), Some(2), Some(3)]), |
8355 | | Some(vec![Some(4), Some(5)]), |
8356 | | Some(vec![Some(6), Some(7), Some(8), Some(9)]), |
8357 | | Some(vec![Some(3), Some(4), Some(5)]), |
8358 | | ]; |
8359 | | let array = Arc::new(ListArray::from_iter_primitive::<Int32Type, _, _>( |
8360 | | values.clone(), |
8361 | | )) as ArrayRef; |
8362 | | |
8363 | | let res = cast_with_options( |
8364 | | array.as_ref(), |
8365 | | &DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int32, true)), 3), |
8366 | | &CastOptions { |
8367 | | safe: false, |
8368 | | ..Default::default() |
8369 | | }, |
8370 | | ); |
8371 | | assert!(res.is_err()); |
8372 | | assert!(format!("{res:?}") |
8373 | | .contains("Cannot cast to FixedSizeList(3): value at index 1 has length 2")); |
8374 | | |
8375 | | // When safe=true (default), the cast will fill nulls for lists that are |
8376 | | // too short and truncate lists that are too long. |
8377 | | let res = cast( |
8378 | | array.as_ref(), |
8379 | | &DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int32, true)), 3), |
8380 | | ) |
8381 | | .unwrap(); |
8382 | | let expected = Arc::new(FixedSizeListArray::from_iter_primitive::<Int32Type, _, _>( |
8383 | | vec![ |
8384 | | Some(vec![Some(1), Some(2), Some(3)]), |
8385 | | None, // Too short -> replaced with null |
8386 | | None, // Too long -> replaced with null |
8387 | | Some(vec![Some(3), Some(4), Some(5)]), |
8388 | | ], |
8389 | | 3, |
8390 | | )) as ArrayRef; |
8391 | | assert_eq!(expected.as_ref(), res.as_ref()); |
8392 | | |
8393 | | // The safe option is false and the source array contains a null list. |
8394 | | // issue: https://github.com/apache/arrow-rs/issues/5642 |
8395 | | let array = Arc::new(ListArray::from_iter_primitive::<Int32Type, _, _>(vec![ |
8396 | | Some(vec![Some(1), Some(2), Some(3)]), |
8397 | | None, |
8398 | | ])) as ArrayRef; |
8399 | | let res = cast_with_options( |
8400 | | array.as_ref(), |
8401 | | &DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int32, true)), 3), |
8402 | | &CastOptions { |
8403 | | safe: false, |
8404 | | ..Default::default() |
8405 | | }, |
8406 | | ) |
8407 | | .unwrap(); |
8408 | | let expected = Arc::new(FixedSizeListArray::from_iter_primitive::<Int32Type, _, _>( |
8409 | | vec![Some(vec![Some(1), Some(2), Some(3)]), None], |
8410 | | 3, |
8411 | | )) as ArrayRef; |
8412 | | assert_eq!(expected.as_ref(), res.as_ref()); |
8413 | | } |
8414 | | |
8415 | | #[test] |
8416 | | fn test_cast_large_list_to_fsl() { |
8417 | | let values = vec![Some(vec![Some(1), Some(2)]), Some(vec![Some(3), Some(4)])]; |
8418 | | let array = Arc::new(LargeListArray::from_iter_primitive::<Int32Type, _, _>( |
8419 | | values.clone(), |
8420 | | )) as ArrayRef; |
8421 | | let expected = Arc::new(FixedSizeListArray::from_iter_primitive::<Int32Type, _, _>( |
8422 | | values, 2, |
8423 | | )) as ArrayRef; |
8424 | | let actual = cast( |
8425 | | array.as_ref(), |
8426 | | &DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int32, true)), 2), |
8427 | | ) |
8428 | | .unwrap(); |
8429 | | assert_eq!(expected.as_ref(), actual.as_ref()); |
8430 | | } |
8431 | | |
8432 | | #[test] |
8433 | | fn test_cast_list_to_fsl_subcast() { |
8434 | | let array = Arc::new(LargeListArray::from_iter_primitive::<Int32Type, _, _>( |
8435 | | vec![ |
8436 | | Some(vec![Some(1), Some(2)]), |
8437 | | Some(vec![Some(3), Some(i32::MAX)]), |
8438 | | ], |
8439 | | )) as ArrayRef; |
8440 | | let expected = Arc::new(FixedSizeListArray::from_iter_primitive::<Int64Type, _, _>( |
8441 | | vec![ |
8442 | | Some(vec![Some(1), Some(2)]), |
8443 | | Some(vec![Some(3), Some(i32::MAX as i64)]), |
8444 | | ], |
8445 | | 2, |
8446 | | )) as ArrayRef; |
8447 | | let actual = cast( |
8448 | | array.as_ref(), |
8449 | | &DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int64, true)), 2), |
8450 | | ) |
8451 | | .unwrap(); |
8452 | | assert_eq!(expected.as_ref(), actual.as_ref()); |
8453 | | |
8454 | | let res = cast_with_options( |
8455 | | array.as_ref(), |
8456 | | &DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int16, true)), 2), |
8457 | | &CastOptions { |
8458 | | safe: false, |
8459 | | ..Default::default() |
8460 | | }, |
8461 | | ); |
8462 | | assert!(res.is_err()); |
8463 | | assert!(format!("{res:?}").contains("Can't cast value 2147483647 to type Int16")); |
8464 | | } |
8465 | | |
8466 | | #[test] |
8467 | | fn test_cast_list_to_fsl_empty() { |
8468 | | let field = Arc::new(Field::new_list_field(DataType::Int32, true)); |
8469 | | let array = new_empty_array(&DataType::List(field.clone())); |
8470 | | |
8471 | | let target_type = DataType::FixedSizeList(field.clone(), 3); |
8472 | | let expected = new_empty_array(&target_type); |
8473 | | |
8474 | | let actual = cast(array.as_ref(), &target_type).unwrap(); |
8475 | | assert_eq!(expected.as_ref(), actual.as_ref()); |
8476 | | } |
8477 | | |
8478 | | fn make_list_array() -> ListArray { |
8479 | | // Construct a value array |
8480 | | let value_data = ArrayData::builder(DataType::Int32) |
8481 | | .len(8) |
8482 | | .add_buffer(Buffer::from_slice_ref([0, 1, 2, 3, 4, 5, 6, 7])) |
8483 | | .build() |
8484 | | .unwrap(); |
8485 | | |
8486 | | // Construct a buffer for value offsets, for the nested array: |
8487 | | // [[0, 1, 2], [3, 4, 5], [6, 7]] |
8488 | | let value_offsets = Buffer::from_slice_ref([0, 3, 6, 8]); |
8489 | | |
8490 | | // Construct a list array from the above two |
8491 | | let list_data_type = DataType::List(Arc::new(Field::new_list_field(DataType::Int32, true))); |
8492 | | let list_data = ArrayData::builder(list_data_type) |
8493 | | .len(3) |
8494 | | .add_buffer(value_offsets) |
8495 | | .add_child_data(value_data) |
8496 | | .build() |
8497 | | .unwrap(); |
8498 | | ListArray::from(list_data) |
8499 | | } |
8500 | | |
8501 | | fn make_large_list_array() -> LargeListArray { |
8502 | | // Construct a value array |
8503 | | let value_data = ArrayData::builder(DataType::Int32) |
8504 | | .len(8) |
8505 | | .add_buffer(Buffer::from_slice_ref([0, 1, 2, 3, 4, 5, 6, 7])) |
8506 | | .build() |
8507 | | .unwrap(); |
8508 | | |
8509 | | // Construct a buffer for value offsets, for the nested array: |
8510 | | // [[0, 1, 2], [3, 4, 5], [6, 7]] |
8511 | | let value_offsets = Buffer::from_slice_ref([0i64, 3, 6, 8]); |
8512 | | |
8513 | | // Construct a list array from the above two |
8514 | | let list_data_type = |
8515 | | DataType::LargeList(Arc::new(Field::new_list_field(DataType::Int32, true))); |
8516 | | let list_data = ArrayData::builder(list_data_type) |
8517 | | .len(3) |
8518 | | .add_buffer(value_offsets) |
8519 | | .add_child_data(value_data) |
8520 | | .build() |
8521 | | .unwrap(); |
8522 | | LargeListArray::from(list_data) |
8523 | | } |
8524 | | |
8525 | | fn make_fixed_size_list_array() -> FixedSizeListArray { |
8526 | | // Construct a value array |
8527 | | let value_data = ArrayData::builder(DataType::Int32) |
8528 | | .len(8) |
8529 | | .add_buffer(Buffer::from_slice_ref([0, 1, 2, 3, 4, 5, 6, 7])) |
8530 | | .build() |
8531 | | .unwrap(); |
8532 | | |
8533 | | let list_data_type = |
8534 | | DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int32, true)), 4); |
8535 | | let list_data = ArrayData::builder(list_data_type) |
8536 | | .len(2) |
8537 | | .add_child_data(value_data) |
8538 | | .build() |
8539 | | .unwrap(); |
8540 | | FixedSizeListArray::from(list_data) |
8541 | | } |
8542 | | |
8543 | | fn make_fixed_size_list_array_for_large_list() -> FixedSizeListArray { |
8544 | | // Construct a value array |
8545 | | let value_data = ArrayData::builder(DataType::Int64) |
8546 | | .len(8) |
8547 | | .add_buffer(Buffer::from_slice_ref([0i64, 1, 2, 3, 4, 5, 6, 7])) |
8548 | | .build() |
8549 | | .unwrap(); |
8550 | | |
8551 | | let list_data_type = |
8552 | | DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int64, true)), 4); |
8553 | | let list_data = ArrayData::builder(list_data_type) |
8554 | | .len(2) |
8555 | | .add_child_data(value_data) |
8556 | | .build() |
8557 | | .unwrap(); |
8558 | | FixedSizeListArray::from(list_data) |
8559 | | } |
8560 | | |
8561 | | #[test] |
8562 | | fn test_cast_map_dont_allow_change_of_order() { |
8563 | | let string_builder = StringBuilder::new(); |
8564 | | let value_builder = StringBuilder::new(); |
8565 | | let mut builder = MapBuilder::new( |
8566 | | Some(MapFieldNames { |
8567 | | entry: "entries".to_string(), |
8568 | | key: "key".to_string(), |
8569 | | value: "value".to_string(), |
8570 | | }), |
8571 | | string_builder, |
8572 | | value_builder, |
8573 | | ); |
8574 | | |
8575 | | builder.keys().append_value("0"); |
8576 | | builder.values().append_value("test_val_1"); |
8577 | | builder.append(true).unwrap(); |
8578 | | builder.keys().append_value("1"); |
8579 | | builder.values().append_value("test_val_2"); |
8580 | | builder.append(true).unwrap(); |
8581 | | |
8582 | | // map builder returns unsorted map by default |
8583 | | let array = builder.finish(); |
8584 | | |
8585 | | let new_ordered = true; |
8586 | | let new_type = DataType::Map( |
8587 | | Arc::new(Field::new( |
8588 | | "entries", |
8589 | | DataType::Struct( |
8590 | | vec![ |
8591 | | Field::new("key", DataType::Utf8, false), |
8592 | | Field::new("value", DataType::Utf8, false), |
8593 | | ] |
8594 | | .into(), |
8595 | | ), |
8596 | | false, |
8597 | | )), |
8598 | | new_ordered, |
8599 | | ); |
8600 | | |
8601 | | let new_array_result = cast(&array, &new_type.clone()); |
8602 | | assert!(!can_cast_types(array.data_type(), &new_type)); |
8603 | | assert!( |
8604 | | matches!(new_array_result, Err(ArrowError::CastError(t)) if t == r#"Casting from Map(Field { name: "entries", data_type: Struct([Field { name: "key", data_type: Utf8, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "value", data_type: Utf8, nullable: true, dict_id: 0, dict_is_ordered: false, metadata: {} }]), nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, false) to Map(Field { name: "entries", data_type: Struct([Field { name: "key", data_type: Utf8, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "value", data_type: Utf8, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }]), nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, true) not supported"#) |
8605 | | ); |
8606 | | } |
8607 | | |
8608 | | #[test] |
8609 | | fn test_cast_map_dont_allow_when_container_cant_cast() { |
8610 | | let string_builder = StringBuilder::new(); |
8611 | | let value_builder = IntervalDayTimeArray::builder(2); |
8612 | | let mut builder = MapBuilder::new( |
8613 | | Some(MapFieldNames { |
8614 | | entry: "entries".to_string(), |
8615 | | key: "key".to_string(), |
8616 | | value: "value".to_string(), |
8617 | | }), |
8618 | | string_builder, |
8619 | | value_builder, |
8620 | | ); |
8621 | | |
8622 | | builder.keys().append_value("0"); |
8623 | | builder.values().append_value(IntervalDayTime::new(1, 1)); |
8624 | | builder.append(true).unwrap(); |
8625 | | builder.keys().append_value("1"); |
8626 | | builder.values().append_value(IntervalDayTime::new(2, 2)); |
8627 | | builder.append(true).unwrap(); |
8628 | | |
8629 | | // map builder returns unsorted map by default |
8630 | | let array = builder.finish(); |
8631 | | |
8632 | | let new_ordered = true; |
8633 | | let new_type = DataType::Map( |
8634 | | Arc::new(Field::new( |
8635 | | "entries", |
8636 | | DataType::Struct( |
8637 | | vec![ |
8638 | | Field::new("key", DataType::Utf8, false), |
8639 | | Field::new("value", DataType::Duration(TimeUnit::Second), false), |
8640 | | ] |
8641 | | .into(), |
8642 | | ), |
8643 | | false, |
8644 | | )), |
8645 | | new_ordered, |
8646 | | ); |
8647 | | |
8648 | | let new_array_result = cast(&array, &new_type.clone()); |
8649 | | assert!(!can_cast_types(array.data_type(), &new_type)); |
8650 | | assert!( |
8651 | | matches!(new_array_result, Err(ArrowError::CastError(t)) if t == r#"Casting from Map(Field { name: "entries", data_type: Struct([Field { name: "key", data_type: Utf8, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "value", data_type: Interval(DayTime), nullable: true, dict_id: 0, dict_is_ordered: false, metadata: {} }]), nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, false) to Map(Field { name: "entries", data_type: Struct([Field { name: "key", data_type: Utf8, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, Field { name: "value", data_type: Duration(Second), nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }]), nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }, true) not supported"#) |
8652 | | ); |
8653 | | } |
8654 | | |
8655 | | #[test] |
8656 | | fn test_cast_map_field_names() { |
8657 | | let string_builder = StringBuilder::new(); |
8658 | | let value_builder = StringBuilder::new(); |
8659 | | let mut builder = MapBuilder::new( |
8660 | | Some(MapFieldNames { |
8661 | | entry: "entries".to_string(), |
8662 | | key: "key".to_string(), |
8663 | | value: "value".to_string(), |
8664 | | }), |
8665 | | string_builder, |
8666 | | value_builder, |
8667 | | ); |
8668 | | |
8669 | | builder.keys().append_value("0"); |
8670 | | builder.values().append_value("test_val_1"); |
8671 | | builder.append(true).unwrap(); |
8672 | | builder.keys().append_value("1"); |
8673 | | builder.values().append_value("test_val_2"); |
8674 | | builder.append(true).unwrap(); |
8675 | | builder.append(false).unwrap(); |
8676 | | |
8677 | | let array = builder.finish(); |
8678 | | |
8679 | | let new_type = DataType::Map( |
8680 | | Arc::new(Field::new( |
8681 | | "entries_new", |
8682 | | DataType::Struct( |
8683 | | vec![ |
8684 | | Field::new("key_new", DataType::Utf8, false), |
8685 | | Field::new("value_values", DataType::Utf8, false), |
8686 | | ] |
8687 | | .into(), |
8688 | | ), |
8689 | | false, |
8690 | | )), |
8691 | | false, |
8692 | | ); |
8693 | | |
8694 | | assert_ne!(new_type, array.data_type().clone()); |
8695 | | |
8696 | | let new_array = cast(&array, &new_type.clone()).unwrap(); |
8697 | | assert_eq!(new_type, new_array.data_type().clone()); |
8698 | | let map_array = new_array.as_map(); |
8699 | | |
8700 | | assert_ne!(new_type, array.data_type().clone()); |
8701 | | assert_eq!(new_type, map_array.data_type().clone()); |
8702 | | |
8703 | | let key_string = map_array |
8704 | | .keys() |
8705 | | .as_any() |
8706 | | .downcast_ref::<StringArray>() |
8707 | | .unwrap() |
8708 | | .into_iter() |
8709 | | .flatten() |
8710 | | .collect::<Vec<_>>(); |
8711 | | assert_eq!(&key_string, &vec!["0", "1"]); |
8712 | | |
8713 | | let values_string_array = cast(map_array.values(), &DataType::Utf8).unwrap(); |
8714 | | let values_string = values_string_array |
8715 | | .as_any() |
8716 | | .downcast_ref::<StringArray>() |
8717 | | .unwrap() |
8718 | | .into_iter() |
8719 | | .flatten() |
8720 | | .collect::<Vec<_>>(); |
8721 | | assert_eq!(&values_string, &vec!["test_val_1", "test_val_2"]); |
8722 | | |
8723 | | assert_eq!( |
8724 | | map_array.nulls(), |
8725 | | Some(&NullBuffer::from(vec![true, true, false])) |
8726 | | ); |
8727 | | } |
8728 | | |
8729 | | #[test] |
8730 | | fn test_cast_map_contained_values() { |
8731 | | let string_builder = StringBuilder::new(); |
8732 | | let value_builder = Int8Builder::new(); |
8733 | | let mut builder = MapBuilder::new( |
8734 | | Some(MapFieldNames { |
8735 | | entry: "entries".to_string(), |
8736 | | key: "key".to_string(), |
8737 | | value: "value".to_string(), |
8738 | | }), |
8739 | | string_builder, |
8740 | | value_builder, |
8741 | | ); |
8742 | | |
8743 | | builder.keys().append_value("0"); |
8744 | | builder.values().append_value(44); |
8745 | | builder.append(true).unwrap(); |
8746 | | builder.keys().append_value("1"); |
8747 | | builder.values().append_value(22); |
8748 | | builder.append(true).unwrap(); |
8749 | | |
8750 | | let array = builder.finish(); |
8751 | | |
8752 | | let new_type = DataType::Map( |
8753 | | Arc::new(Field::new( |
8754 | | "entries", |
8755 | | DataType::Struct( |
8756 | | vec![ |
8757 | | Field::new("key", DataType::Utf8, false), |
8758 | | Field::new("value", DataType::Utf8, false), |
8759 | | ] |
8760 | | .into(), |
8761 | | ), |
8762 | | false, |
8763 | | )), |
8764 | | false, |
8765 | | ); |
8766 | | |
8767 | | let new_array = cast(&array, &new_type.clone()).unwrap(); |
8768 | | assert_eq!(new_type, new_array.data_type().clone()); |
8769 | | let map_array = new_array.as_map(); |
8770 | | |
8771 | | assert_ne!(new_type, array.data_type().clone()); |
8772 | | assert_eq!(new_type, map_array.data_type().clone()); |
8773 | | |
8774 | | let key_string = map_array |
8775 | | .keys() |
8776 | | .as_any() |
8777 | | .downcast_ref::<StringArray>() |
8778 | | .unwrap() |
8779 | | .into_iter() |
8780 | | .flatten() |
8781 | | .collect::<Vec<_>>(); |
8782 | | assert_eq!(&key_string, &vec!["0", "1"]); |
8783 | | |
8784 | | let values_string_array = cast(map_array.values(), &DataType::Utf8).unwrap(); |
8785 | | let values_string = values_string_array |
8786 | | .as_any() |
8787 | | .downcast_ref::<StringArray>() |
8788 | | .unwrap() |
8789 | | .into_iter() |
8790 | | .flatten() |
8791 | | .collect::<Vec<_>>(); |
8792 | | assert_eq!(&values_string, &vec!["44", "22"]); |
8793 | | } |
8794 | | |
8795 | | #[test] |
8796 | | fn test_utf8_cast_offsets() { |
8797 | | // test if offset of the array is taken into account during cast |
8798 | | let str_array = StringArray::from(vec!["a", "b", "c"]); |
8799 | | let str_array = str_array.slice(1, 2); |
8800 | | |
8801 | | let out = cast(&str_array, &DataType::LargeUtf8).unwrap(); |
8802 | | |
8803 | | let large_str_array = out.as_any().downcast_ref::<LargeStringArray>().unwrap(); |
8804 | | let strs = large_str_array.into_iter().flatten().collect::<Vec<_>>(); |
8805 | | assert_eq!(strs, &["b", "c"]) |
8806 | | } |
8807 | | |
8808 | | #[test] |
8809 | | fn test_list_cast_offsets() { |
8810 | | // test if offset of the array is taken into account during cast |
8811 | | let array1 = make_list_array().slice(1, 2); |
8812 | | let array2 = Arc::new(make_list_array()) as ArrayRef; |
8813 | | |
8814 | | let dt = DataType::LargeList(Arc::new(Field::new_list_field(DataType::Int32, true))); |
8815 | | let out1 = cast(&array1, &dt).unwrap(); |
8816 | | let out2 = cast(&array2, &dt).unwrap(); |
8817 | | |
8818 | | assert_eq!(&out1, &out2.slice(1, 2)) |
8819 | | } |
8820 | | |
8821 | | #[test] |
8822 | | fn test_list_to_string() { |
8823 | | let str_array = StringArray::from(vec!["a", "b", "c", "d", "e", "f", "g", "h"]); |
8824 | | let value_offsets = Buffer::from_slice_ref([0, 3, 6, 8]); |
8825 | | let value_data = str_array.into_data(); |
8826 | | |
8827 | | let list_data_type = DataType::List(Arc::new(Field::new_list_field(DataType::Utf8, true))); |
8828 | | let list_data = ArrayData::builder(list_data_type) |
8829 | | .len(3) |
8830 | | .add_buffer(value_offsets) |
8831 | | .add_child_data(value_data) |
8832 | | .build() |
8833 | | .unwrap(); |
8834 | | let array = Arc::new(ListArray::from(list_data)) as ArrayRef; |
8835 | | |
8836 | | let out = cast(&array, &DataType::Utf8).unwrap(); |
8837 | | let out = out |
8838 | | .as_any() |
8839 | | .downcast_ref::<StringArray>() |
8840 | | .unwrap() |
8841 | | .into_iter() |
8842 | | .flatten() |
8843 | | .collect::<Vec<_>>(); |
8844 | | assert_eq!(&out, &vec!["[a, b, c]", "[d, e, f]", "[g, h]"]); |
8845 | | |
8846 | | let out = cast(&array, &DataType::LargeUtf8).unwrap(); |
8847 | | let out = out |
8848 | | .as_any() |
8849 | | .downcast_ref::<LargeStringArray>() |
8850 | | .unwrap() |
8851 | | .into_iter() |
8852 | | .flatten() |
8853 | | .collect::<Vec<_>>(); |
8854 | | assert_eq!(&out, &vec!["[a, b, c]", "[d, e, f]", "[g, h]"]); |
8855 | | |
8856 | | let array = Arc::new(make_list_array()) as ArrayRef; |
8857 | | let out = cast(&array, &DataType::Utf8).unwrap(); |
8858 | | let out = out |
8859 | | .as_any() |
8860 | | .downcast_ref::<StringArray>() |
8861 | | .unwrap() |
8862 | | .into_iter() |
8863 | | .flatten() |
8864 | | .collect::<Vec<_>>(); |
8865 | | assert_eq!(&out, &vec!["[0, 1, 2]", "[3, 4, 5]", "[6, 7]"]); |
8866 | | |
8867 | | let array = Arc::new(make_large_list_array()) as ArrayRef; |
8868 | | let out = cast(&array, &DataType::LargeUtf8).unwrap(); |
8869 | | let out = out |
8870 | | .as_any() |
8871 | | .downcast_ref::<LargeStringArray>() |
8872 | | .unwrap() |
8873 | | .into_iter() |
8874 | | .flatten() |
8875 | | .collect::<Vec<_>>(); |
8876 | | assert_eq!(&out, &vec!["[0, 1, 2]", "[3, 4, 5]", "[6, 7]"]); |
8877 | | } |
8878 | | |
8879 | | #[test] |
8880 | | fn test_cast_f64_to_decimal128() { |
8881 | | // to reproduce https://github.com/apache/arrow-rs/issues/2997 |
8882 | | |
8883 | | let decimal_type = DataType::Decimal128(18, 2); |
8884 | | let array = Float64Array::from(vec![ |
8885 | | Some(0.0699999999), |
8886 | | Some(0.0659999999), |
8887 | | Some(0.0650000000), |
8888 | | Some(0.0649999999), |
8889 | | ]); |
8890 | | let array = Arc::new(array) as ArrayRef; |
8891 | | generate_cast_test_case!( |
8892 | | &array, |
8893 | | Decimal128Array, |
8894 | | &decimal_type, |
8895 | | vec![ |
8896 | | Some(7_i128), // round up |
8897 | | Some(7_i128), // round up |
8898 | | Some(7_i128), // round up |
8899 | | Some(6_i128), // round down |
8900 | | ] |
8901 | | ); |
8902 | | |
8903 | | let decimal_type = DataType::Decimal128(18, 3); |
8904 | | let array = Float64Array::from(vec![ |
8905 | | Some(0.0699999999), |
8906 | | Some(0.0659999999), |
8907 | | Some(0.0650000000), |
8908 | | Some(0.0649999999), |
8909 | | ]); |
8910 | | let array = Arc::new(array) as ArrayRef; |
8911 | | generate_cast_test_case!( |
8912 | | &array, |
8913 | | Decimal128Array, |
8914 | | &decimal_type, |
8915 | | vec![ |
8916 | | Some(70_i128), // round up |
8917 | | Some(66_i128), // round up |
8918 | | Some(65_i128), // round down |
8919 | | Some(65_i128), // round up |
8920 | | ] |
8921 | | ); |
8922 | | } |
8923 | | |
8924 | | #[test] |
8925 | | fn test_cast_numeric_to_decimal128_overflow() { |
8926 | | let array = Int64Array::from(vec![i64::MAX]); |
8927 | | let array = Arc::new(array) as ArrayRef; |
8928 | | let casted_array = cast_with_options( |
8929 | | &array, |
8930 | | &DataType::Decimal128(38, 30), |
8931 | | &CastOptions { |
8932 | | safe: true, |
8933 | | format_options: FormatOptions::default(), |
8934 | | }, |
8935 | | ); |
8936 | | assert!(casted_array.is_ok()); |
8937 | | assert!(casted_array.unwrap().is_null(0)); |
8938 | | |
8939 | | let casted_array = cast_with_options( |
8940 | | &array, |
8941 | | &DataType::Decimal128(38, 30), |
8942 | | &CastOptions { |
8943 | | safe: false, |
8944 | | format_options: FormatOptions::default(), |
8945 | | }, |
8946 | | ); |
8947 | | assert!(casted_array.is_err()); |
8948 | | } |
8949 | | |
8950 | | #[test] |
8951 | | fn test_cast_numeric_to_decimal256_overflow() { |
8952 | | let array = Int64Array::from(vec![i64::MAX]); |
8953 | | let array = Arc::new(array) as ArrayRef; |
8954 | | let casted_array = cast_with_options( |
8955 | | &array, |
8956 | | &DataType::Decimal256(76, 76), |
8957 | | &CastOptions { |
8958 | | safe: true, |
8959 | | format_options: FormatOptions::default(), |
8960 | | }, |
8961 | | ); |
8962 | | assert!(casted_array.is_ok()); |
8963 | | assert!(casted_array.unwrap().is_null(0)); |
8964 | | |
8965 | | let casted_array = cast_with_options( |
8966 | | &array, |
8967 | | &DataType::Decimal256(76, 76), |
8968 | | &CastOptions { |
8969 | | safe: false, |
8970 | | format_options: FormatOptions::default(), |
8971 | | }, |
8972 | | ); |
8973 | | assert!(casted_array.is_err()); |
8974 | | } |
8975 | | |
8976 | | #[test] |
8977 | | fn test_cast_floating_point_to_decimal128_precision_overflow() { |
8978 | | let array = Float64Array::from(vec![1.1]); |
8979 | | let array = Arc::new(array) as ArrayRef; |
8980 | | let casted_array = cast_with_options( |
8981 | | &array, |
8982 | | &DataType::Decimal128(2, 2), |
8983 | | &CastOptions { |
8984 | | safe: true, |
8985 | | format_options: FormatOptions::default(), |
8986 | | }, |
8987 | | ); |
8988 | | assert!(casted_array.is_ok()); |
8989 | | assert!(casted_array.unwrap().is_null(0)); |
8990 | | |
8991 | | let casted_array = cast_with_options( |
8992 | | &array, |
8993 | | &DataType::Decimal128(2, 2), |
8994 | | &CastOptions { |
8995 | | safe: false, |
8996 | | format_options: FormatOptions::default(), |
8997 | | }, |
8998 | | ); |
8999 | | let err = casted_array.unwrap_err().to_string(); |
9000 | | let expected_error = "Invalid argument error: 110 is too large to store in a Decimal128 of precision 2. Max is 99"; |
9001 | | assert!( |
9002 | | err.contains(expected_error), |
9003 | | "did not find expected error '{expected_error}' in actual error '{err}'" |
9004 | | ); |
9005 | | } |
9006 | | |
9007 | | #[test] |
9008 | | fn test_cast_floating_point_to_decimal256_precision_overflow() { |
9009 | | let array = Float64Array::from(vec![1.1]); |
9010 | | let array = Arc::new(array) as ArrayRef; |
9011 | | let casted_array = cast_with_options( |
9012 | | &array, |
9013 | | &DataType::Decimal256(2, 2), |
9014 | | &CastOptions { |
9015 | | safe: true, |
9016 | | format_options: FormatOptions::default(), |
9017 | | }, |
9018 | | ); |
9019 | | assert!(casted_array.is_ok()); |
9020 | | assert!(casted_array.unwrap().is_null(0)); |
9021 | | |
9022 | | let casted_array = cast_with_options( |
9023 | | &array, |
9024 | | &DataType::Decimal256(2, 2), |
9025 | | &CastOptions { |
9026 | | safe: false, |
9027 | | format_options: FormatOptions::default(), |
9028 | | }, |
9029 | | ); |
9030 | | let err = casted_array.unwrap_err().to_string(); |
9031 | | let expected_error = "Invalid argument error: 110 is too large to store in a Decimal256 of precision 2. Max is 99"; |
9032 | | assert!( |
9033 | | err.contains(expected_error), |
9034 | | "did not find expected error '{expected_error}' in actual error '{err}'" |
9035 | | ); |
9036 | | } |
9037 | | |
9038 | | #[test] |
9039 | | fn test_cast_floating_point_to_decimal128_overflow() { |
9040 | | let array = Float64Array::from(vec![f64::MAX]); |
9041 | | let array = Arc::new(array) as ArrayRef; |
9042 | | let casted_array = cast_with_options( |
9043 | | &array, |
9044 | | &DataType::Decimal128(38, 30), |
9045 | | &CastOptions { |
9046 | | safe: true, |
9047 | | format_options: FormatOptions::default(), |
9048 | | }, |
9049 | | ); |
9050 | | assert!(casted_array.is_ok()); |
9051 | | assert!(casted_array.unwrap().is_null(0)); |
9052 | | |
9053 | | let casted_array = cast_with_options( |
9054 | | &array, |
9055 | | &DataType::Decimal128(38, 30), |
9056 | | &CastOptions { |
9057 | | safe: false, |
9058 | | format_options: FormatOptions::default(), |
9059 | | }, |
9060 | | ); |
9061 | | let err = casted_array.unwrap_err().to_string(); |
9062 | | let expected_error = "Cast error: Cannot cast to Decimal128(38, 30)"; |
9063 | | assert!( |
9064 | | err.contains(expected_error), |
9065 | | "did not find expected error '{expected_error}' in actual error '{err}'" |
9066 | | ); |
9067 | | } |
9068 | | |
9069 | | #[test] |
9070 | | fn test_cast_floating_point_to_decimal256_overflow() { |
9071 | | let array = Float64Array::from(vec![f64::MAX]); |
9072 | | let array = Arc::new(array) as ArrayRef; |
9073 | | let casted_array = cast_with_options( |
9074 | | &array, |
9075 | | &DataType::Decimal256(76, 50), |
9076 | | &CastOptions { |
9077 | | safe: true, |
9078 | | format_options: FormatOptions::default(), |
9079 | | }, |
9080 | | ); |
9081 | | assert!(casted_array.is_ok()); |
9082 | | assert!(casted_array.unwrap().is_null(0)); |
9083 | | |
9084 | | let casted_array = cast_with_options( |
9085 | | &array, |
9086 | | &DataType::Decimal256(76, 50), |
9087 | | &CastOptions { |
9088 | | safe: false, |
9089 | | format_options: FormatOptions::default(), |
9090 | | }, |
9091 | | ); |
9092 | | let err = casted_array.unwrap_err().to_string(); |
9093 | | let expected_error = "Cast error: Cannot cast to Decimal256(76, 50)"; |
9094 | | assert!( |
9095 | | err.contains(expected_error), |
9096 | | "did not find expected error '{expected_error}' in actual error '{err}'" |
9097 | | ); |
9098 | | } |
9099 | | #[test] |
9100 | | fn test_cast_decimal256_to_f64_no_overflow() { |
9101 | | // Test casting i256::MAX: should produce a large finite positive value |
9102 | | let array = vec![Some(i256::MAX)]; |
9103 | | let array = create_decimal256_array(array, 76, 2).unwrap(); |
9104 | | let array = Arc::new(array) as ArrayRef; |
9105 | | |
9106 | | let result = cast(&array, &DataType::Float64).unwrap(); |
9107 | | let result = result.as_primitive::<Float64Type>(); |
9108 | | assert!(result.value(0).is_finite()); |
9109 | | assert!(result.value(0) > 0.0); // Positive result |
9110 | | |
9111 | | // Test casting i256::MIN: should produce a large finite negative value |
9112 | | let array = vec![Some(i256::MIN)]; |
9113 | | let array = create_decimal256_array(array, 76, 2).unwrap(); |
9114 | | let array = Arc::new(array) as ArrayRef; |
9115 | | |
9116 | | let result = cast(&array, &DataType::Float64).unwrap(); |
9117 | | let result = result.as_primitive::<Float64Type>(); |
9118 | | assert!(result.value(0).is_finite()); |
9119 | | assert!(result.value(0) < 0.0); // Negative result |
9120 | | } |
9121 | | |
9122 | | #[test] |
9123 | | fn test_cast_decimal128_to_decimal128_negative_scale() { |
9124 | | let input_type = DataType::Decimal128(20, 0); |
9125 | | let output_type = DataType::Decimal128(20, -1); |
9126 | | assert!(can_cast_types(&input_type, &output_type)); |
9127 | | let array = vec![Some(1123450), Some(2123455), Some(3123456), None]; |
9128 | | let input_decimal_array = create_decimal128_array(array, 20, 0).unwrap(); |
9129 | | let array = Arc::new(input_decimal_array) as ArrayRef; |
9130 | | generate_cast_test_case!( |
9131 | | &array, |
9132 | | Decimal128Array, |
9133 | | &output_type, |
9134 | | vec![ |
9135 | | Some(112345_i128), |
9136 | | Some(212346_i128), |
9137 | | Some(312346_i128), |
9138 | | None |
9139 | | ] |
9140 | | ); |
9141 | | |
9142 | | let casted_array = cast(&array, &output_type).unwrap(); |
9143 | | let decimal_arr = casted_array.as_primitive::<Decimal128Type>(); |
9144 | | |
9145 | | assert_eq!("1123450", decimal_arr.value_as_string(0)); |
9146 | | assert_eq!("2123460", decimal_arr.value_as_string(1)); |
9147 | | assert_eq!("3123460", decimal_arr.value_as_string(2)); |
9148 | | } |
9149 | | |
9150 | | #[test] |
9151 | | fn decimal128_min_max_to_f64() { |
9152 | | // Ensure Decimal128 i128::MIN/MAX round-trip cast |
9153 | | let min128 = i128::MIN; |
9154 | | let max128 = i128::MAX; |
9155 | | assert_eq!(min128 as f64, min128 as f64); |
9156 | | assert_eq!(max128 as f64, max128 as f64); |
9157 | | } |
9158 | | |
9159 | | #[test] |
9160 | | fn test_cast_numeric_to_decimal128_negative() { |
9161 | | let decimal_type = DataType::Decimal128(38, -1); |
9162 | | let array = Arc::new(Int32Array::from(vec![ |
9163 | | Some(1123456), |
9164 | | Some(2123456), |
9165 | | Some(3123456), |
9166 | | ])) as ArrayRef; |
9167 | | |
9168 | | let casted_array = cast(&array, &decimal_type).unwrap(); |
9169 | | let decimal_arr = casted_array.as_primitive::<Decimal128Type>(); |
9170 | | |
9171 | | assert_eq!("1123450", decimal_arr.value_as_string(0)); |
9172 | | assert_eq!("2123450", decimal_arr.value_as_string(1)); |
9173 | | assert_eq!("3123450", decimal_arr.value_as_string(2)); |
9174 | | |
9175 | | let array = Arc::new(Float32Array::from(vec![ |
9176 | | Some(1123.456), |
9177 | | Some(2123.456), |
9178 | | Some(3123.456), |
9179 | | ])) as ArrayRef; |
9180 | | |
9181 | | let casted_array = cast(&array, &decimal_type).unwrap(); |
9182 | | let decimal_arr = casted_array.as_primitive::<Decimal128Type>(); |
9183 | | |
9184 | | assert_eq!("1120", decimal_arr.value_as_string(0)); |
9185 | | assert_eq!("2120", decimal_arr.value_as_string(1)); |
9186 | | assert_eq!("3120", decimal_arr.value_as_string(2)); |
9187 | | } |
9188 | | |
9189 | | #[test] |
9190 | | fn test_cast_decimal128_to_decimal128_negative() { |
9191 | | let input_type = DataType::Decimal128(10, -1); |
9192 | | let output_type = DataType::Decimal128(10, -2); |
9193 | | assert!(can_cast_types(&input_type, &output_type)); |
9194 | | let array = vec![Some(123)]; |
9195 | | let input_decimal_array = create_decimal128_array(array, 10, -1).unwrap(); |
9196 | | let array = Arc::new(input_decimal_array) as ArrayRef; |
9197 | | generate_cast_test_case!(&array, Decimal128Array, &output_type, vec![Some(12_i128),]); |
9198 | | |
9199 | | let casted_array = cast(&array, &output_type).unwrap(); |
9200 | | let decimal_arr = casted_array.as_primitive::<Decimal128Type>(); |
9201 | | |
9202 | | assert_eq!("1200", decimal_arr.value_as_string(0)); |
9203 | | |
9204 | | let array = vec![Some(125)]; |
9205 | | let input_decimal_array = create_decimal128_array(array, 10, -1).unwrap(); |
9206 | | let array = Arc::new(input_decimal_array) as ArrayRef; |
9207 | | generate_cast_test_case!(&array, Decimal128Array, &output_type, vec![Some(13_i128),]); |
9208 | | |
9209 | | let casted_array = cast(&array, &output_type).unwrap(); |
9210 | | let decimal_arr = casted_array.as_primitive::<Decimal128Type>(); |
9211 | | |
9212 | | assert_eq!("1300", decimal_arr.value_as_string(0)); |
9213 | | } |
9214 | | |
9215 | | #[test] |
9216 | | fn test_cast_decimal128_to_decimal256_negative() { |
9217 | | let input_type = DataType::Decimal128(10, 3); |
9218 | | let output_type = DataType::Decimal256(10, 5); |
9219 | | assert!(can_cast_types(&input_type, &output_type)); |
9220 | | let array = vec![Some(123456), Some(-123456)]; |
9221 | | let input_decimal_array = create_decimal128_array(array, 10, 3).unwrap(); |
9222 | | let array = Arc::new(input_decimal_array) as ArrayRef; |
9223 | | |
9224 | | let hundred = i256::from_i128(100); |
9225 | | generate_cast_test_case!( |
9226 | | &array, |
9227 | | Decimal256Array, |
9228 | | &output_type, |
9229 | | vec![ |
9230 | | Some(i256::from_i128(123456).mul_wrapping(hundred)), |
9231 | | Some(i256::from_i128(-123456).mul_wrapping(hundred)) |
9232 | | ] |
9233 | | ); |
9234 | | } |
9235 | | |
9236 | | #[test] |
9237 | | fn test_parse_string_to_decimal() { |
9238 | | assert_eq!( |
9239 | | Decimal128Type::format_decimal( |
9240 | | parse_string_to_decimal_native::<Decimal128Type>("123.45", 2).unwrap(), |
9241 | | 38, |
9242 | | 2, |
9243 | | ), |
9244 | | "123.45" |
9245 | | ); |
9246 | | assert_eq!( |
9247 | | Decimal128Type::format_decimal( |
9248 | | parse_string_to_decimal_native::<Decimal128Type>("12345", 2).unwrap(), |
9249 | | 38, |
9250 | | 2, |
9251 | | ), |
9252 | | "12345.00" |
9253 | | ); |
9254 | | assert_eq!( |
9255 | | Decimal128Type::format_decimal( |
9256 | | parse_string_to_decimal_native::<Decimal128Type>("0.12345", 2).unwrap(), |
9257 | | 38, |
9258 | | 2, |
9259 | | ), |
9260 | | "0.12" |
9261 | | ); |
9262 | | assert_eq!( |
9263 | | Decimal128Type::format_decimal( |
9264 | | parse_string_to_decimal_native::<Decimal128Type>(".12345", 2).unwrap(), |
9265 | | 38, |
9266 | | 2, |
9267 | | ), |
9268 | | "0.12" |
9269 | | ); |
9270 | | assert_eq!( |
9271 | | Decimal128Type::format_decimal( |
9272 | | parse_string_to_decimal_native::<Decimal128Type>(".1265", 2).unwrap(), |
9273 | | 38, |
9274 | | 2, |
9275 | | ), |
9276 | | "0.13" |
9277 | | ); |
9278 | | assert_eq!( |
9279 | | Decimal128Type::format_decimal( |
9280 | | parse_string_to_decimal_native::<Decimal128Type>(".1265", 2).unwrap(), |
9281 | | 38, |
9282 | | 2, |
9283 | | ), |
9284 | | "0.13" |
9285 | | ); |
9286 | | |
9287 | | assert_eq!( |
9288 | | Decimal256Type::format_decimal( |
9289 | | parse_string_to_decimal_native::<Decimal256Type>("123.45", 3).unwrap(), |
9290 | | 38, |
9291 | | 3, |
9292 | | ), |
9293 | | "123.450" |
9294 | | ); |
9295 | | assert_eq!( |
9296 | | Decimal256Type::format_decimal( |
9297 | | parse_string_to_decimal_native::<Decimal256Type>("12345", 3).unwrap(), |
9298 | | 38, |
9299 | | 3, |
9300 | | ), |
9301 | | "12345.000" |
9302 | | ); |
9303 | | assert_eq!( |
9304 | | Decimal256Type::format_decimal( |
9305 | | parse_string_to_decimal_native::<Decimal256Type>("0.12345", 3).unwrap(), |
9306 | | 38, |
9307 | | 3, |
9308 | | ), |
9309 | | "0.123" |
9310 | | ); |
9311 | | assert_eq!( |
9312 | | Decimal256Type::format_decimal( |
9313 | | parse_string_to_decimal_native::<Decimal256Type>(".12345", 3).unwrap(), |
9314 | | 38, |
9315 | | 3, |
9316 | | ), |
9317 | | "0.123" |
9318 | | ); |
9319 | | assert_eq!( |
9320 | | Decimal256Type::format_decimal( |
9321 | | parse_string_to_decimal_native::<Decimal256Type>(".1265", 3).unwrap(), |
9322 | | 38, |
9323 | | 3, |
9324 | | ), |
9325 | | "0.127" |
9326 | | ); |
9327 | | } |
9328 | | |
9329 | | fn test_cast_string_to_decimal(array: ArrayRef) { |
9330 | | // Decimal128 |
9331 | | let output_type = DataType::Decimal128(38, 2); |
9332 | | assert!(can_cast_types(array.data_type(), &output_type)); |
9333 | | |
9334 | | let casted_array = cast(&array, &output_type).unwrap(); |
9335 | | let decimal_arr = casted_array.as_primitive::<Decimal128Type>(); |
9336 | | |
9337 | | assert_eq!("123.45", decimal_arr.value_as_string(0)); |
9338 | | assert_eq!("1.23", decimal_arr.value_as_string(1)); |
9339 | | assert_eq!("0.12", decimal_arr.value_as_string(2)); |
9340 | | assert_eq!("0.13", decimal_arr.value_as_string(3)); |
9341 | | assert_eq!("1.26", decimal_arr.value_as_string(4)); |
9342 | | assert_eq!("12345.00", decimal_arr.value_as_string(5)); |
9343 | | assert_eq!("12345.00", decimal_arr.value_as_string(6)); |
9344 | | assert_eq!("0.12", decimal_arr.value_as_string(7)); |
9345 | | assert_eq!("12.23", decimal_arr.value_as_string(8)); |
9346 | | assert!(decimal_arr.is_null(9)); |
9347 | | assert_eq!("0.00", decimal_arr.value_as_string(10)); |
9348 | | assert_eq!("0.00", decimal_arr.value_as_string(11)); |
9349 | | assert!(decimal_arr.is_null(12)); |
9350 | | assert_eq!("-1.23", decimal_arr.value_as_string(13)); |
9351 | | assert_eq!("-1.24", decimal_arr.value_as_string(14)); |
9352 | | assert_eq!("0.00", decimal_arr.value_as_string(15)); |
9353 | | assert_eq!("-123.00", decimal_arr.value_as_string(16)); |
9354 | | assert_eq!("-123.23", decimal_arr.value_as_string(17)); |
9355 | | assert_eq!("-0.12", decimal_arr.value_as_string(18)); |
9356 | | assert_eq!("1.23", decimal_arr.value_as_string(19)); |
9357 | | assert_eq!("1.24", decimal_arr.value_as_string(20)); |
9358 | | assert_eq!("0.00", decimal_arr.value_as_string(21)); |
9359 | | assert_eq!("123.00", decimal_arr.value_as_string(22)); |
9360 | | assert_eq!("123.23", decimal_arr.value_as_string(23)); |
9361 | | assert_eq!("0.12", decimal_arr.value_as_string(24)); |
9362 | | assert!(decimal_arr.is_null(25)); |
9363 | | assert!(decimal_arr.is_null(26)); |
9364 | | assert!(decimal_arr.is_null(27)); |
9365 | | assert_eq!("0.00", decimal_arr.value_as_string(28)); |
9366 | | assert_eq!("0.00", decimal_arr.value_as_string(29)); |
9367 | | assert_eq!("12345.00", decimal_arr.value_as_string(30)); |
9368 | | assert_eq!(decimal_arr.len(), 31); |
9369 | | |
9370 | | // Decimal256 |
9371 | | let output_type = DataType::Decimal256(76, 3); |
9372 | | assert!(can_cast_types(array.data_type(), &output_type)); |
9373 | | |
9374 | | let casted_array = cast(&array, &output_type).unwrap(); |
9375 | | let decimal_arr = casted_array.as_primitive::<Decimal256Type>(); |
9376 | | |
9377 | | assert_eq!("123.450", decimal_arr.value_as_string(0)); |
9378 | | assert_eq!("1.235", decimal_arr.value_as_string(1)); |
9379 | | assert_eq!("0.123", decimal_arr.value_as_string(2)); |
9380 | | assert_eq!("0.127", decimal_arr.value_as_string(3)); |
9381 | | assert_eq!("1.263", decimal_arr.value_as_string(4)); |
9382 | | assert_eq!("12345.000", decimal_arr.value_as_string(5)); |
9383 | | assert_eq!("12345.000", decimal_arr.value_as_string(6)); |
9384 | | assert_eq!("0.123", decimal_arr.value_as_string(7)); |
9385 | | assert_eq!("12.234", decimal_arr.value_as_string(8)); |
9386 | | assert!(decimal_arr.is_null(9)); |
9387 | | assert_eq!("0.000", decimal_arr.value_as_string(10)); |
9388 | | assert_eq!("0.000", decimal_arr.value_as_string(11)); |
9389 | | assert!(decimal_arr.is_null(12)); |
9390 | | assert_eq!("-1.235", decimal_arr.value_as_string(13)); |
9391 | | assert_eq!("-1.236", decimal_arr.value_as_string(14)); |
9392 | | assert_eq!("0.000", decimal_arr.value_as_string(15)); |
9393 | | assert_eq!("-123.000", decimal_arr.value_as_string(16)); |
9394 | | assert_eq!("-123.234", decimal_arr.value_as_string(17)); |
9395 | | assert_eq!("-0.123", decimal_arr.value_as_string(18)); |
9396 | | assert_eq!("1.235", decimal_arr.value_as_string(19)); |
9397 | | assert_eq!("1.236", decimal_arr.value_as_string(20)); |
9398 | | assert_eq!("0.000", decimal_arr.value_as_string(21)); |
9399 | | assert_eq!("123.000", decimal_arr.value_as_string(22)); |
9400 | | assert_eq!("123.234", decimal_arr.value_as_string(23)); |
9401 | | assert_eq!("0.123", decimal_arr.value_as_string(24)); |
9402 | | assert!(decimal_arr.is_null(25)); |
9403 | | assert!(decimal_arr.is_null(26)); |
9404 | | assert!(decimal_arr.is_null(27)); |
9405 | | assert_eq!("0.000", decimal_arr.value_as_string(28)); |
9406 | | assert_eq!("0.000", decimal_arr.value_as_string(29)); |
9407 | | assert_eq!("12345.000", decimal_arr.value_as_string(30)); |
9408 | | assert_eq!(decimal_arr.len(), 31); |
9409 | | } |
9410 | | |
9411 | | #[test] |
9412 | | fn test_cast_utf8_to_decimal() { |
9413 | | let str_array = StringArray::from(vec![ |
9414 | | Some("123.45"), |
9415 | | Some("1.2345"), |
9416 | | Some("0.12345"), |
9417 | | Some("0.1267"), |
9418 | | Some("1.263"), |
9419 | | Some("12345.0"), |
9420 | | Some("12345"), |
9421 | | Some("000.123"), |
9422 | | Some("12.234000"), |
9423 | | None, |
9424 | | Some(""), |
9425 | | Some(" "), |
9426 | | None, |
9427 | | Some("-1.23499999"), |
9428 | | Some("-1.23599999"), |
9429 | | Some("-0.00001"), |
9430 | | Some("-123"), |
9431 | | Some("-123.234000"), |
9432 | | Some("-000.123"), |
9433 | | Some("+1.23499999"), |
9434 | | Some("+1.23599999"), |
9435 | | Some("+0.00001"), |
9436 | | Some("+123"), |
9437 | | Some("+123.234000"), |
9438 | | Some("+000.123"), |
9439 | | Some("1.-23499999"), |
9440 | | Some("-1.-23499999"), |
9441 | | Some("--1.23499999"), |
9442 | | Some("0"), |
9443 | | Some("000.000"), |
9444 | | Some("0000000000000000012345.000"), |
9445 | | ]); |
9446 | | let array = Arc::new(str_array) as ArrayRef; |
9447 | | |
9448 | | test_cast_string_to_decimal(array); |
9449 | | |
9450 | | let test_cases = [ |
9451 | | (None, None), |
9452 | | // (Some(""), None), |
9453 | | // (Some(" "), None), |
9454 | | (Some("0"), Some("0")), |
9455 | | (Some("000.000"), Some("0")), |
9456 | | (Some("12345"), Some("12345")), |
9457 | | (Some("000000000000000000000000000012345"), Some("12345")), |
9458 | | (Some("-123"), Some("-123")), |
9459 | | (Some("+123"), Some("123")), |
9460 | | ]; |
9461 | | let inputs = test_cases.iter().map(|entry| entry.0).collect::<Vec<_>>(); |
9462 | | let expected = test_cases.iter().map(|entry| entry.1).collect::<Vec<_>>(); |
9463 | | |
9464 | | let array = Arc::new(StringArray::from(inputs)) as ArrayRef; |
9465 | | test_cast_string_to_decimal_scale_zero(array, &expected); |
9466 | | } |
9467 | | |
9468 | | #[test] |
9469 | | fn test_cast_large_utf8_to_decimal() { |
9470 | | let str_array = LargeStringArray::from(vec![ |
9471 | | Some("123.45"), |
9472 | | Some("1.2345"), |
9473 | | Some("0.12345"), |
9474 | | Some("0.1267"), |
9475 | | Some("1.263"), |
9476 | | Some("12345.0"), |
9477 | | Some("12345"), |
9478 | | Some("000.123"), |
9479 | | Some("12.234000"), |
9480 | | None, |
9481 | | Some(""), |
9482 | | Some(" "), |
9483 | | None, |
9484 | | Some("-1.23499999"), |
9485 | | Some("-1.23599999"), |
9486 | | Some("-0.00001"), |
9487 | | Some("-123"), |
9488 | | Some("-123.234000"), |
9489 | | Some("-000.123"), |
9490 | | Some("+1.23499999"), |
9491 | | Some("+1.23599999"), |
9492 | | Some("+0.00001"), |
9493 | | Some("+123"), |
9494 | | Some("+123.234000"), |
9495 | | Some("+000.123"), |
9496 | | Some("1.-23499999"), |
9497 | | Some("-1.-23499999"), |
9498 | | Some("--1.23499999"), |
9499 | | Some("0"), |
9500 | | Some("000.000"), |
9501 | | Some("0000000000000000012345.000"), |
9502 | | ]); |
9503 | | let array = Arc::new(str_array) as ArrayRef; |
9504 | | |
9505 | | test_cast_string_to_decimal(array); |
9506 | | |
9507 | | let test_cases = [ |
9508 | | (None, None), |
9509 | | (Some(""), None), |
9510 | | (Some(" "), None), |
9511 | | (Some("0"), Some("0")), |
9512 | | (Some("000.000"), Some("0")), |
9513 | | (Some("12345"), Some("12345")), |
9514 | | (Some("000000000000000000000000000012345"), Some("12345")), |
9515 | | (Some("-123"), Some("-123")), |
9516 | | (Some("+123"), Some("123")), |
9517 | | ]; |
9518 | | let inputs = test_cases.iter().map(|entry| entry.0).collect::<Vec<_>>(); |
9519 | | let expected = test_cases.iter().map(|entry| entry.1).collect::<Vec<_>>(); |
9520 | | |
9521 | | let array = Arc::new(LargeStringArray::from(inputs)) as ArrayRef; |
9522 | | test_cast_string_to_decimal_scale_zero(array, &expected); |
9523 | | } |
9524 | | |
9525 | | fn test_cast_string_to_decimal_scale_zero( |
9526 | | array: ArrayRef, |
9527 | | expected_as_string: &[Option<&str>], |
9528 | | ) { |
9529 | | // Decimal128 |
9530 | | let output_type = DataType::Decimal128(38, 0); |
9531 | | assert!(can_cast_types(array.data_type(), &output_type)); |
9532 | | let casted_array = cast(&array, &output_type).unwrap(); |
9533 | | let decimal_arr = casted_array.as_primitive::<Decimal128Type>(); |
9534 | | assert_decimal_array_contents(decimal_arr, expected_as_string); |
9535 | | |
9536 | | // Decimal256 |
9537 | | let output_type = DataType::Decimal256(76, 0); |
9538 | | assert!(can_cast_types(array.data_type(), &output_type)); |
9539 | | let casted_array = cast(&array, &output_type).unwrap(); |
9540 | | let decimal_arr = casted_array.as_primitive::<Decimal256Type>(); |
9541 | | assert_decimal_array_contents(decimal_arr, expected_as_string); |
9542 | | } |
9543 | | |
9544 | | fn assert_decimal_array_contents<T>( |
9545 | | array: &PrimitiveArray<T>, |
9546 | | expected_as_string: &[Option<&str>], |
9547 | | ) where |
9548 | | T: DecimalType + ArrowPrimitiveType, |
9549 | | { |
9550 | | assert_eq!(array.len(), expected_as_string.len()); |
9551 | | for (i, expected) in expected_as_string.iter().enumerate() { |
9552 | | let actual = if array.is_null(i) { |
9553 | | None |
9554 | | } else { |
9555 | | Some(array.value_as_string(i)) |
9556 | | }; |
9557 | | let actual = actual.as_ref().map(|s| s.as_ref()); |
9558 | | assert_eq!(*expected, actual, "Expected at position {i}"); |
9559 | | } |
9560 | | } |
9561 | | |
9562 | | #[test] |
9563 | | fn test_cast_invalid_utf8_to_decimal() { |
9564 | | let str_array = StringArray::from(vec!["4.4.5", ". 0.123"]); |
9565 | | let array = Arc::new(str_array) as ArrayRef; |
9566 | | |
9567 | | // Safe cast |
9568 | | let output_type = DataType::Decimal128(38, 2); |
9569 | | let casted_array = cast(&array, &output_type).unwrap(); |
9570 | | assert!(casted_array.is_null(0)); |
9571 | | assert!(casted_array.is_null(1)); |
9572 | | |
9573 | | let output_type = DataType::Decimal256(76, 2); |
9574 | | let casted_array = cast(&array, &output_type).unwrap(); |
9575 | | assert!(casted_array.is_null(0)); |
9576 | | assert!(casted_array.is_null(1)); |
9577 | | |
9578 | | // Non-safe cast |
9579 | | let output_type = DataType::Decimal128(38, 2); |
9580 | | let str_array = StringArray::from(vec!["4.4.5"]); |
9581 | | let array = Arc::new(str_array) as ArrayRef; |
9582 | | let option = CastOptions { |
9583 | | safe: false, |
9584 | | format_options: FormatOptions::default(), |
9585 | | }; |
9586 | | let casted_err = cast_with_options(&array, &output_type, &option).unwrap_err(); |
9587 | | assert!(casted_err |
9588 | | .to_string() |
9589 | | .contains("Cannot cast string '4.4.5' to value of Decimal128(38, 10) type")); |
9590 | | |
9591 | | let str_array = StringArray::from(vec![". 0.123"]); |
9592 | | let array = Arc::new(str_array) as ArrayRef; |
9593 | | let casted_err = cast_with_options(&array, &output_type, &option).unwrap_err(); |
9594 | | assert!(casted_err |
9595 | | .to_string() |
9596 | | .contains("Cannot cast string '. 0.123' to value of Decimal128(38, 10) type")); |
9597 | | } |
9598 | | |
9599 | | fn test_cast_string_to_decimal128_overflow(overflow_array: ArrayRef) { |
9600 | | let output_type = DataType::Decimal128(38, 2); |
9601 | | let casted_array = cast(&overflow_array, &output_type).unwrap(); |
9602 | | let decimal_arr = casted_array.as_primitive::<Decimal128Type>(); |
9603 | | |
9604 | | assert!(decimal_arr.is_null(0)); |
9605 | | assert!(decimal_arr.is_null(1)); |
9606 | | assert!(decimal_arr.is_null(2)); |
9607 | | assert_eq!( |
9608 | | "999999999999999999999999999999999999.99", |
9609 | | decimal_arr.value_as_string(3) |
9610 | | ); |
9611 | | assert_eq!( |
9612 | | "100000000000000000000000000000000000.00", |
9613 | | decimal_arr.value_as_string(4) |
9614 | | ); |
9615 | | } |
9616 | | |
9617 | | #[test] |
9618 | | fn test_cast_string_to_decimal128_precision_overflow() { |
9619 | | let array = StringArray::from(vec!["1000".to_string()]); |
9620 | | let array = Arc::new(array) as ArrayRef; |
9621 | | let casted_array = cast_with_options( |
9622 | | &array, |
9623 | | &DataType::Decimal128(10, 8), |
9624 | | &CastOptions { |
9625 | | safe: true, |
9626 | | format_options: FormatOptions::default(), |
9627 | | }, |
9628 | | ); |
9629 | | assert!(casted_array.is_ok()); |
9630 | | assert!(casted_array.unwrap().is_null(0)); |
9631 | | |
9632 | | let err = cast_with_options( |
9633 | | &array, |
9634 | | &DataType::Decimal128(10, 8), |
9635 | | &CastOptions { |
9636 | | safe: false, |
9637 | | format_options: FormatOptions::default(), |
9638 | | }, |
9639 | | ); |
9640 | | assert_eq!("Invalid argument error: 100000000000 is too large to store in a Decimal128 of precision 10. Max is 9999999999", err.unwrap_err().to_string()); |
9641 | | } |
9642 | | |
9643 | | #[test] |
9644 | | fn test_cast_utf8_to_decimal128_overflow() { |
9645 | | let overflow_str_array = StringArray::from(vec![ |
9646 | | i128::MAX.to_string(), |
9647 | | i128::MIN.to_string(), |
9648 | | "99999999999999999999999999999999999999".to_string(), |
9649 | | "999999999999999999999999999999999999.99".to_string(), |
9650 | | "99999999999999999999999999999999999.999".to_string(), |
9651 | | ]); |
9652 | | let overflow_array = Arc::new(overflow_str_array) as ArrayRef; |
9653 | | |
9654 | | test_cast_string_to_decimal128_overflow(overflow_array); |
9655 | | } |
9656 | | |
9657 | | #[test] |
9658 | | fn test_cast_large_utf8_to_decimal128_overflow() { |
9659 | | let overflow_str_array = LargeStringArray::from(vec![ |
9660 | | i128::MAX.to_string(), |
9661 | | i128::MIN.to_string(), |
9662 | | "99999999999999999999999999999999999999".to_string(), |
9663 | | "999999999999999999999999999999999999.99".to_string(), |
9664 | | "99999999999999999999999999999999999.999".to_string(), |
9665 | | ]); |
9666 | | let overflow_array = Arc::new(overflow_str_array) as ArrayRef; |
9667 | | |
9668 | | test_cast_string_to_decimal128_overflow(overflow_array); |
9669 | | } |
9670 | | |
9671 | | fn test_cast_string_to_decimal256_overflow(overflow_array: ArrayRef) { |
9672 | | let output_type = DataType::Decimal256(76, 2); |
9673 | | let casted_array = cast(&overflow_array, &output_type).unwrap(); |
9674 | | let decimal_arr = casted_array.as_primitive::<Decimal256Type>(); |
9675 | | |
9676 | | assert_eq!( |
9677 | | "170141183460469231731687303715884105727.00", |
9678 | | decimal_arr.value_as_string(0) |
9679 | | ); |
9680 | | assert_eq!( |
9681 | | "-170141183460469231731687303715884105728.00", |
9682 | | decimal_arr.value_as_string(1) |
9683 | | ); |
9684 | | assert_eq!( |
9685 | | "99999999999999999999999999999999999999.00", |
9686 | | decimal_arr.value_as_string(2) |
9687 | | ); |
9688 | | assert_eq!( |
9689 | | "999999999999999999999999999999999999.99", |
9690 | | decimal_arr.value_as_string(3) |
9691 | | ); |
9692 | | assert_eq!( |
9693 | | "100000000000000000000000000000000000.00", |
9694 | | decimal_arr.value_as_string(4) |
9695 | | ); |
9696 | | assert!(decimal_arr.is_null(5)); |
9697 | | assert!(decimal_arr.is_null(6)); |
9698 | | } |
9699 | | |
9700 | | #[test] |
9701 | | fn test_cast_string_to_decimal256_precision_overflow() { |
9702 | | let array = StringArray::from(vec!["1000".to_string()]); |
9703 | | let array = Arc::new(array) as ArrayRef; |
9704 | | let casted_array = cast_with_options( |
9705 | | &array, |
9706 | | &DataType::Decimal256(10, 8), |
9707 | | &CastOptions { |
9708 | | safe: true, |
9709 | | format_options: FormatOptions::default(), |
9710 | | }, |
9711 | | ); |
9712 | | assert!(casted_array.is_ok()); |
9713 | | assert!(casted_array.unwrap().is_null(0)); |
9714 | | |
9715 | | let err = cast_with_options( |
9716 | | &array, |
9717 | | &DataType::Decimal256(10, 8), |
9718 | | &CastOptions { |
9719 | | safe: false, |
9720 | | format_options: FormatOptions::default(), |
9721 | | }, |
9722 | | ); |
9723 | | assert_eq!("Invalid argument error: 100000000000 is too large to store in a Decimal256 of precision 10. Max is 9999999999", err.unwrap_err().to_string()); |
9724 | | } |
9725 | | |
9726 | | #[test] |
9727 | | fn test_cast_utf8_to_decimal256_overflow() { |
9728 | | let overflow_str_array = StringArray::from(vec![ |
9729 | | i128::MAX.to_string(), |
9730 | | i128::MIN.to_string(), |
9731 | | "99999999999999999999999999999999999999".to_string(), |
9732 | | "999999999999999999999999999999999999.99".to_string(), |
9733 | | "99999999999999999999999999999999999.999".to_string(), |
9734 | | i256::MAX.to_string(), |
9735 | | i256::MIN.to_string(), |
9736 | | ]); |
9737 | | let overflow_array = Arc::new(overflow_str_array) as ArrayRef; |
9738 | | |
9739 | | test_cast_string_to_decimal256_overflow(overflow_array); |
9740 | | } |
9741 | | |
9742 | | #[test] |
9743 | | fn test_cast_large_utf8_to_decimal256_overflow() { |
9744 | | let overflow_str_array = LargeStringArray::from(vec![ |
9745 | | i128::MAX.to_string(), |
9746 | | i128::MIN.to_string(), |
9747 | | "99999999999999999999999999999999999999".to_string(), |
9748 | | "999999999999999999999999999999999999.99".to_string(), |
9749 | | "99999999999999999999999999999999999.999".to_string(), |
9750 | | i256::MAX.to_string(), |
9751 | | i256::MIN.to_string(), |
9752 | | ]); |
9753 | | let overflow_array = Arc::new(overflow_str_array) as ArrayRef; |
9754 | | |
9755 | | test_cast_string_to_decimal256_overflow(overflow_array); |
9756 | | } |
9757 | | |
9758 | | #[test] |
9759 | | fn test_cast_outside_supported_range_for_nanoseconds() { |
9760 | | const EXPECTED_ERROR_MESSAGE: &str = "The dates that can be represented as nanoseconds have to be between 1677-09-21T00:12:44.0 and 2262-04-11T23:47:16.854775804"; |
9761 | | |
9762 | | let array = StringArray::from(vec![Some("1650-01-01 01:01:01.000001")]); |
9763 | | |
9764 | | let cast_options = CastOptions { |
9765 | | safe: false, |
9766 | | format_options: FormatOptions::default(), |
9767 | | }; |
9768 | | |
9769 | | let result = cast_string_to_timestamp::<i32, TimestampNanosecondType>( |
9770 | | &array, |
9771 | | &None::<Arc<str>>, |
9772 | | &cast_options, |
9773 | | ); |
9774 | | |
9775 | | let err = result.unwrap_err(); |
9776 | | assert_eq!( |
9777 | | err.to_string(), |
9778 | | format!( |
9779 | | "Cast error: Overflow converting {} to Nanosecond. {}", |
9780 | | array.value(0), |
9781 | | EXPECTED_ERROR_MESSAGE |
9782 | | ) |
9783 | | ); |
9784 | | } |
9785 | | |
9786 | | #[test] |
9787 | | fn test_cast_date32_to_timestamp() { |
9788 | | let a = Date32Array::from(vec![Some(18628), Some(18993), None]); // 2021-1-1, 2022-1-1 |
9789 | | let array = Arc::new(a) as ArrayRef; |
9790 | | let b = cast(&array, &DataType::Timestamp(TimeUnit::Second, None)).unwrap(); |
9791 | | let c = b.as_primitive::<TimestampSecondType>(); |
9792 | | assert_eq!(1609459200, c.value(0)); |
9793 | | assert_eq!(1640995200, c.value(1)); |
9794 | | assert!(c.is_null(2)); |
9795 | | } |
9796 | | |
9797 | | #[test] |
9798 | | fn test_cast_date32_to_timestamp_ms() { |
9799 | | let a = Date32Array::from(vec![Some(18628), Some(18993), None]); // 2021-1-1, 2022-1-1 |
9800 | | let array = Arc::new(a) as ArrayRef; |
9801 | | let b = cast(&array, &DataType::Timestamp(TimeUnit::Millisecond, None)).unwrap(); |
9802 | | let c = b |
9803 | | .as_any() |
9804 | | .downcast_ref::<TimestampMillisecondArray>() |
9805 | | .unwrap(); |
9806 | | assert_eq!(1609459200000, c.value(0)); |
9807 | | assert_eq!(1640995200000, c.value(1)); |
9808 | | assert!(c.is_null(2)); |
9809 | | } |
9810 | | |
9811 | | #[test] |
9812 | | fn test_cast_date32_to_timestamp_us() { |
9813 | | let a = Date32Array::from(vec![Some(18628), Some(18993), None]); // 2021-1-1, 2022-1-1 |
9814 | | let array = Arc::new(a) as ArrayRef; |
9815 | | let b = cast(&array, &DataType::Timestamp(TimeUnit::Microsecond, None)).unwrap(); |
9816 | | let c = b |
9817 | | .as_any() |
9818 | | .downcast_ref::<TimestampMicrosecondArray>() |
9819 | | .unwrap(); |
9820 | | assert_eq!(1609459200000000, c.value(0)); |
9821 | | assert_eq!(1640995200000000, c.value(1)); |
9822 | | assert!(c.is_null(2)); |
9823 | | } |
9824 | | |
9825 | | #[test] |
9826 | | fn test_cast_date32_to_timestamp_ns() { |
9827 | | let a = Date32Array::from(vec![Some(18628), Some(18993), None]); // 2021-1-1, 2022-1-1 |
9828 | | let array = Arc::new(a) as ArrayRef; |
9829 | | let b = cast(&array, &DataType::Timestamp(TimeUnit::Nanosecond, None)).unwrap(); |
9830 | | let c = b |
9831 | | .as_any() |
9832 | | .downcast_ref::<TimestampNanosecondArray>() |
9833 | | .unwrap(); |
9834 | | assert_eq!(1609459200000000000, c.value(0)); |
9835 | | assert_eq!(1640995200000000000, c.value(1)); |
9836 | | assert!(c.is_null(2)); |
9837 | | } |
9838 | | |
9839 | | #[test] |
9840 | | fn test_timezone_cast() { |
9841 | | let a = StringArray::from(vec![ |
9842 | | "2000-01-01T12:00:00", // date + time valid |
9843 | | "2020-12-15T12:34:56", // date + time valid |
9844 | | ]); |
9845 | | let array = Arc::new(a) as ArrayRef; |
9846 | | let b = cast(&array, &DataType::Timestamp(TimeUnit::Nanosecond, None)).unwrap(); |
9847 | | let v = b.as_primitive::<TimestampNanosecondType>(); |
9848 | | |
9849 | | assert_eq!(v.value(0), 946728000000000000); |
9850 | | assert_eq!(v.value(1), 1608035696000000000); |
9851 | | |
9852 | | let b = cast( |
9853 | | &b, |
9854 | | &DataType::Timestamp(TimeUnit::Nanosecond, Some("+00:00".into())), |
9855 | | ) |
9856 | | .unwrap(); |
9857 | | let v = b.as_primitive::<TimestampNanosecondType>(); |
9858 | | |
9859 | | assert_eq!(v.value(0), 946728000000000000); |
9860 | | assert_eq!(v.value(1), 1608035696000000000); |
9861 | | |
9862 | | let b = cast( |
9863 | | &b, |
9864 | | &DataType::Timestamp(TimeUnit::Millisecond, Some("+02:00".into())), |
9865 | | ) |
9866 | | .unwrap(); |
9867 | | let v = b.as_primitive::<TimestampMillisecondType>(); |
9868 | | |
9869 | | assert_eq!(v.value(0), 946728000000); |
9870 | | assert_eq!(v.value(1), 1608035696000); |
9871 | | } |
9872 | | |
9873 | | #[test] |
9874 | | fn test_cast_utf8_to_timestamp() { |
9875 | | fn test_tz(tz: Arc<str>) { |
9876 | | let valid = StringArray::from(vec![ |
9877 | | "2023-01-01 04:05:06.789000-08:00", |
9878 | | "2023-01-01 04:05:06.789000-07:00", |
9879 | | "2023-01-01 04:05:06.789 -0800", |
9880 | | "2023-01-01 04:05:06.789 -08:00", |
9881 | | "2023-01-01 040506 +0730", |
9882 | | "2023-01-01 040506 +07:30", |
9883 | | "2023-01-01 04:05:06.789", |
9884 | | "2023-01-01 04:05:06", |
9885 | | "2023-01-01", |
9886 | | ]); |
9887 | | |
9888 | | let array = Arc::new(valid) as ArrayRef; |
9889 | | let b = cast_with_options( |
9890 | | &array, |
9891 | | &DataType::Timestamp(TimeUnit::Nanosecond, Some(tz.clone())), |
9892 | | &CastOptions { |
9893 | | safe: false, |
9894 | | format_options: FormatOptions::default(), |
9895 | | }, |
9896 | | ) |
9897 | | .unwrap(); |
9898 | | |
9899 | | let tz = tz.as_ref().parse().unwrap(); |
9900 | | |
9901 | | let as_tz = |
9902 | | |v: i64| as_datetime_with_timezone::<TimestampNanosecondType>(v, tz).unwrap(); |
9903 | | |
9904 | | let as_utc = |v: &i64| as_tz(*v).naive_utc().to_string(); |
9905 | | let as_local = |v: &i64| as_tz(*v).naive_local().to_string(); |
9906 | | |
9907 | | let values = b.as_primitive::<TimestampNanosecondType>().values(); |
9908 | | let utc_results: Vec<_> = values.iter().map(as_utc).collect(); |
9909 | | let local_results: Vec<_> = values.iter().map(as_local).collect(); |
9910 | | |
9911 | | // Absolute timestamps should be parsed preserving the same UTC instant |
9912 | | assert_eq!( |
9913 | | &utc_results[..6], |
9914 | | &[ |
9915 | | "2023-01-01 12:05:06.789".to_string(), |
9916 | | "2023-01-01 11:05:06.789".to_string(), |
9917 | | "2023-01-01 12:05:06.789".to_string(), |
9918 | | "2023-01-01 12:05:06.789".to_string(), |
9919 | | "2022-12-31 20:35:06".to_string(), |
9920 | | "2022-12-31 20:35:06".to_string(), |
9921 | | ] |
9922 | | ); |
9923 | | // Non-absolute timestamps should be parsed preserving the same local instant |
9924 | | assert_eq!( |
9925 | | &local_results[6..], |
9926 | | &[ |
9927 | | "2023-01-01 04:05:06.789".to_string(), |
9928 | | "2023-01-01 04:05:06".to_string(), |
9929 | | "2023-01-01 00:00:00".to_string() |
9930 | | ] |
9931 | | ) |
9932 | | } |
9933 | | |
9934 | | test_tz("+00:00".into()); |
9935 | | test_tz("+02:00".into()); |
9936 | | } |
9937 | | |
9938 | | #[test] |
9939 | | fn test_cast_invalid_utf8() { |
9940 | | let v1: &[u8] = b"\xFF invalid"; |
9941 | | let v2: &[u8] = b"\x00 Foo"; |
9942 | | let s = BinaryArray::from(vec![v1, v2]); |
9943 | | let options = CastOptions { |
9944 | | safe: true, |
9945 | | format_options: FormatOptions::default(), |
9946 | | }; |
9947 | | let array = cast_with_options(&s, &DataType::Utf8, &options).unwrap(); |
9948 | | let a = array.as_string::<i32>(); |
9949 | | a.to_data().validate_full().unwrap(); |
9950 | | |
9951 | | assert_eq!(a.null_count(), 1); |
9952 | | assert_eq!(a.len(), 2); |
9953 | | assert!(a.is_null(0)); |
9954 | | assert_eq!(a.value(0), ""); |
9955 | | assert_eq!(a.value(1), "\x00 Foo"); |
9956 | | } |
9957 | | |
9958 | | #[test] |
9959 | | fn test_cast_utf8_to_timestamptz() { |
9960 | | let valid = StringArray::from(vec!["2023-01-01"]); |
9961 | | |
9962 | | let array = Arc::new(valid) as ArrayRef; |
9963 | | let b = cast( |
9964 | | &array, |
9965 | | &DataType::Timestamp(TimeUnit::Nanosecond, Some("+00:00".into())), |
9966 | | ) |
9967 | | .unwrap(); |
9968 | | |
9969 | | let expect = DataType::Timestamp(TimeUnit::Nanosecond, Some("+00:00".into())); |
9970 | | |
9971 | | assert_eq!(b.data_type(), &expect); |
9972 | | let c = b |
9973 | | .as_any() |
9974 | | .downcast_ref::<TimestampNanosecondArray>() |
9975 | | .unwrap(); |
9976 | | assert_eq!(1672531200000000000, c.value(0)); |
9977 | | } |
9978 | | |
9979 | | #[test] |
9980 | | fn test_cast_decimal_to_string() { |
9981 | | assert!(can_cast_types( |
9982 | | &DataType::Decimal32(9, 4), |
9983 | | &DataType::Utf8View |
9984 | | )); |
9985 | | assert!(can_cast_types( |
9986 | | &DataType::Decimal64(16, 4), |
9987 | | &DataType::Utf8View |
9988 | | )); |
9989 | | assert!(can_cast_types( |
9990 | | &DataType::Decimal128(10, 4), |
9991 | | &DataType::Utf8View |
9992 | | )); |
9993 | | assert!(can_cast_types( |
9994 | | &DataType::Decimal256(38, 10), |
9995 | | &DataType::Utf8View |
9996 | | )); |
9997 | | |
9998 | | macro_rules! assert_decimal_values { |
9999 | | ($array:expr) => { |
10000 | | let c = $array; |
10001 | | assert_eq!("1123.454", c.value(0)); |
10002 | | assert_eq!("2123.456", c.value(1)); |
10003 | | assert_eq!("-3123.453", c.value(2)); |
10004 | | assert_eq!("-3123.456", c.value(3)); |
10005 | | assert_eq!("0.000", c.value(4)); |
10006 | | assert_eq!("0.123", c.value(5)); |
10007 | | assert_eq!("1234.567", c.value(6)); |
10008 | | assert_eq!("-1234.567", c.value(7)); |
10009 | | assert!(c.is_null(8)); |
10010 | | }; |
10011 | | } |
10012 | | |
10013 | | fn test_decimal_to_string<IN: ArrowPrimitiveType, OffsetSize: OffsetSizeTrait>( |
10014 | | output_type: DataType, |
10015 | | array: PrimitiveArray<IN>, |
10016 | | ) { |
10017 | | let b = cast(&array, &output_type).unwrap(); |
10018 | | |
10019 | | assert_eq!(b.data_type(), &output_type); |
10020 | | match b.data_type() { |
10021 | | DataType::Utf8View => { |
10022 | | let c = b.as_string_view(); |
10023 | | assert_decimal_values!(c); |
10024 | | } |
10025 | | DataType::Utf8 | DataType::LargeUtf8 => { |
10026 | | let c = b.as_string::<OffsetSize>(); |
10027 | | assert_decimal_values!(c); |
10028 | | } |
10029 | | _ => (), |
10030 | | } |
10031 | | } |
10032 | | |
10033 | | let array32: Vec<Option<i32>> = vec![ |
10034 | | Some(1123454), |
10035 | | Some(2123456), |
10036 | | Some(-3123453), |
10037 | | Some(-3123456), |
10038 | | Some(0), |
10039 | | Some(123), |
10040 | | Some(123456789), |
10041 | | Some(-123456789), |
10042 | | None, |
10043 | | ]; |
10044 | | let array64: Vec<Option<i64>> = array32.iter().map(|num| num.map(|x| x as i64)).collect(); |
10045 | | let array128: Vec<Option<i128>> = |
10046 | | array64.iter().map(|num| num.map(|x| x as i128)).collect(); |
10047 | | let array256: Vec<Option<i256>> = array128 |
10048 | | .iter() |
10049 | | .map(|num| num.map(i256::from_i128)) |
10050 | | .collect(); |
10051 | | |
10052 | | test_decimal_to_string::<Decimal32Type, i32>( |
10053 | | DataType::Utf8View, |
10054 | | create_decimal32_array(array32.clone(), 7, 3).unwrap(), |
10055 | | ); |
10056 | | test_decimal_to_string::<Decimal32Type, i32>( |
10057 | | DataType::Utf8, |
10058 | | create_decimal32_array(array32.clone(), 7, 3).unwrap(), |
10059 | | ); |
10060 | | test_decimal_to_string::<Decimal32Type, i64>( |
10061 | | DataType::LargeUtf8, |
10062 | | create_decimal32_array(array32, 7, 3).unwrap(), |
10063 | | ); |
10064 | | |
10065 | | test_decimal_to_string::<Decimal64Type, i32>( |
10066 | | DataType::Utf8View, |
10067 | | create_decimal64_array(array64.clone(), 7, 3).unwrap(), |
10068 | | ); |
10069 | | test_decimal_to_string::<Decimal64Type, i32>( |
10070 | | DataType::Utf8, |
10071 | | create_decimal64_array(array64.clone(), 7, 3).unwrap(), |
10072 | | ); |
10073 | | test_decimal_to_string::<Decimal64Type, i64>( |
10074 | | DataType::LargeUtf8, |
10075 | | create_decimal64_array(array64, 7, 3).unwrap(), |
10076 | | ); |
10077 | | |
10078 | | test_decimal_to_string::<Decimal128Type, i32>( |
10079 | | DataType::Utf8View, |
10080 | | create_decimal128_array(array128.clone(), 7, 3).unwrap(), |
10081 | | ); |
10082 | | test_decimal_to_string::<Decimal128Type, i32>( |
10083 | | DataType::Utf8, |
10084 | | create_decimal128_array(array128.clone(), 7, 3).unwrap(), |
10085 | | ); |
10086 | | test_decimal_to_string::<Decimal128Type, i64>( |
10087 | | DataType::LargeUtf8, |
10088 | | create_decimal128_array(array128, 7, 3).unwrap(), |
10089 | | ); |
10090 | | |
10091 | | test_decimal_to_string::<Decimal256Type, i32>( |
10092 | | DataType::Utf8View, |
10093 | | create_decimal256_array(array256.clone(), 7, 3).unwrap(), |
10094 | | ); |
10095 | | test_decimal_to_string::<Decimal256Type, i32>( |
10096 | | DataType::Utf8, |
10097 | | create_decimal256_array(array256.clone(), 7, 3).unwrap(), |
10098 | | ); |
10099 | | test_decimal_to_string::<Decimal256Type, i64>( |
10100 | | DataType::LargeUtf8, |
10101 | | create_decimal256_array(array256, 7, 3).unwrap(), |
10102 | | ); |
10103 | | } |
10104 | | |
10105 | | #[test] |
10106 | | fn test_cast_numeric_to_decimal128_precision_overflow() { |
10107 | | let array = Int64Array::from(vec![1234567]); |
10108 | | let array = Arc::new(array) as ArrayRef; |
10109 | | let casted_array = cast_with_options( |
10110 | | &array, |
10111 | | &DataType::Decimal128(7, 3), |
10112 | | &CastOptions { |
10113 | | safe: true, |
10114 | | format_options: FormatOptions::default(), |
10115 | | }, |
10116 | | ); |
10117 | | assert!(casted_array.is_ok()); |
10118 | | assert!(casted_array.unwrap().is_null(0)); |
10119 | | |
10120 | | let err = cast_with_options( |
10121 | | &array, |
10122 | | &DataType::Decimal128(7, 3), |
10123 | | &CastOptions { |
10124 | | safe: false, |
10125 | | format_options: FormatOptions::default(), |
10126 | | }, |
10127 | | ); |
10128 | | assert_eq!("Invalid argument error: 1234567000 is too large to store in a Decimal128 of precision 7. Max is 9999999", err.unwrap_err().to_string()); |
10129 | | } |
10130 | | |
10131 | | #[test] |
10132 | | fn test_cast_numeric_to_decimal256_precision_overflow() { |
10133 | | let array = Int64Array::from(vec![1234567]); |
10134 | | let array = Arc::new(array) as ArrayRef; |
10135 | | let casted_array = cast_with_options( |
10136 | | &array, |
10137 | | &DataType::Decimal256(7, 3), |
10138 | | &CastOptions { |
10139 | | safe: true, |
10140 | | format_options: FormatOptions::default(), |
10141 | | }, |
10142 | | ); |
10143 | | assert!(casted_array.is_ok()); |
10144 | | assert!(casted_array.unwrap().is_null(0)); |
10145 | | |
10146 | | let err = cast_with_options( |
10147 | | &array, |
10148 | | &DataType::Decimal256(7, 3), |
10149 | | &CastOptions { |
10150 | | safe: false, |
10151 | | format_options: FormatOptions::default(), |
10152 | | }, |
10153 | | ); |
10154 | | assert_eq!("Invalid argument error: 1234567000 is too large to store in a Decimal256 of precision 7. Max is 9999999", err.unwrap_err().to_string()); |
10155 | | } |
10156 | | |
10157 | | /// helper function to test casting from duration to interval |
10158 | | fn cast_from_duration_to_interval<T: ArrowTemporalType<Native = i64>>( |
10159 | | array: Vec<i64>, |
10160 | | cast_options: &CastOptions, |
10161 | | ) -> Result<PrimitiveArray<IntervalMonthDayNanoType>, ArrowError> { |
10162 | | let array = PrimitiveArray::<T>::new(array.into(), None); |
10163 | | let array = Arc::new(array) as ArrayRef; |
10164 | | let interval = DataType::Interval(IntervalUnit::MonthDayNano); |
10165 | | let out = cast_with_options(&array, &interval, cast_options)?; |
10166 | | let out = out.as_primitive::<IntervalMonthDayNanoType>().clone(); |
10167 | | Ok(out) |
10168 | | } |
10169 | | |
10170 | | #[test] |
10171 | | fn test_cast_from_duration_to_interval() { |
10172 | | // from duration second to interval month day nano |
10173 | | let array = vec![1234567]; |
10174 | | let casted_array = |
10175 | | cast_from_duration_to_interval::<DurationSecondType>(array, &CastOptions::default()) |
10176 | | .unwrap(); |
10177 | | assert_eq!( |
10178 | | casted_array.data_type(), |
10179 | | &DataType::Interval(IntervalUnit::MonthDayNano) |
10180 | | ); |
10181 | | assert_eq!( |
10182 | | casted_array.value(0), |
10183 | | IntervalMonthDayNano::new(0, 0, 1234567000000000) |
10184 | | ); |
10185 | | |
10186 | | let array = vec![i64::MAX]; |
10187 | | let casted_array = cast_from_duration_to_interval::<DurationSecondType>( |
10188 | | array.clone(), |
10189 | | &CastOptions::default(), |
10190 | | ) |
10191 | | .unwrap(); |
10192 | | assert!(!casted_array.is_valid(0)); |
10193 | | |
10194 | | let casted_array = cast_from_duration_to_interval::<DurationSecondType>( |
10195 | | array, |
10196 | | &CastOptions { |
10197 | | safe: false, |
10198 | | format_options: FormatOptions::default(), |
10199 | | }, |
10200 | | ); |
10201 | | assert!(casted_array.is_err()); |
10202 | | |
10203 | | // from duration millisecond to interval month day nano |
10204 | | let array = vec![1234567]; |
10205 | | let casted_array = cast_from_duration_to_interval::<DurationMillisecondType>( |
10206 | | array, |
10207 | | &CastOptions::default(), |
10208 | | ) |
10209 | | .unwrap(); |
10210 | | assert_eq!( |
10211 | | casted_array.data_type(), |
10212 | | &DataType::Interval(IntervalUnit::MonthDayNano) |
10213 | | ); |
10214 | | assert_eq!( |
10215 | | casted_array.value(0), |
10216 | | IntervalMonthDayNano::new(0, 0, 1234567000000) |
10217 | | ); |
10218 | | |
10219 | | let array = vec![i64::MAX]; |
10220 | | let casted_array = cast_from_duration_to_interval::<DurationMillisecondType>( |
10221 | | array.clone(), |
10222 | | &CastOptions::default(), |
10223 | | ) |
10224 | | .unwrap(); |
10225 | | assert!(!casted_array.is_valid(0)); |
10226 | | |
10227 | | let casted_array = cast_from_duration_to_interval::<DurationMillisecondType>( |
10228 | | array, |
10229 | | &CastOptions { |
10230 | | safe: false, |
10231 | | format_options: FormatOptions::default(), |
10232 | | }, |
10233 | | ); |
10234 | | assert!(casted_array.is_err()); |
10235 | | |
10236 | | // from duration microsecond to interval month day nano |
10237 | | let array = vec![1234567]; |
10238 | | let casted_array = cast_from_duration_to_interval::<DurationMicrosecondType>( |
10239 | | array, |
10240 | | &CastOptions::default(), |
10241 | | ) |
10242 | | .unwrap(); |
10243 | | assert_eq!( |
10244 | | casted_array.data_type(), |
10245 | | &DataType::Interval(IntervalUnit::MonthDayNano) |
10246 | | ); |
10247 | | assert_eq!( |
10248 | | casted_array.value(0), |
10249 | | IntervalMonthDayNano::new(0, 0, 1234567000) |
10250 | | ); |
10251 | | |
10252 | | let array = vec![i64::MAX]; |
10253 | | let casted_array = cast_from_duration_to_interval::<DurationMicrosecondType>( |
10254 | | array.clone(), |
10255 | | &CastOptions::default(), |
10256 | | ) |
10257 | | .unwrap(); |
10258 | | assert!(!casted_array.is_valid(0)); |
10259 | | |
10260 | | let casted_array = cast_from_duration_to_interval::<DurationMicrosecondType>( |
10261 | | array, |
10262 | | &CastOptions { |
10263 | | safe: false, |
10264 | | format_options: FormatOptions::default(), |
10265 | | }, |
10266 | | ); |
10267 | | assert!(casted_array.is_err()); |
10268 | | |
10269 | | // from duration nanosecond to interval month day nano |
10270 | | let array = vec![1234567]; |
10271 | | let casted_array = cast_from_duration_to_interval::<DurationNanosecondType>( |
10272 | | array, |
10273 | | &CastOptions::default(), |
10274 | | ) |
10275 | | .unwrap(); |
10276 | | assert_eq!( |
10277 | | casted_array.data_type(), |
10278 | | &DataType::Interval(IntervalUnit::MonthDayNano) |
10279 | | ); |
10280 | | assert_eq!( |
10281 | | casted_array.value(0), |
10282 | | IntervalMonthDayNano::new(0, 0, 1234567) |
10283 | | ); |
10284 | | |
10285 | | let array = vec![i64::MAX]; |
10286 | | let casted_array = cast_from_duration_to_interval::<DurationNanosecondType>( |
10287 | | array, |
10288 | | &CastOptions { |
10289 | | safe: false, |
10290 | | format_options: FormatOptions::default(), |
10291 | | }, |
10292 | | ) |
10293 | | .unwrap(); |
10294 | | assert_eq!( |
10295 | | casted_array.value(0), |
10296 | | IntervalMonthDayNano::new(0, 0, i64::MAX) |
10297 | | ); |
10298 | | } |
10299 | | |
10300 | | /// helper function to test casting from interval to duration |
10301 | | fn cast_from_interval_to_duration<T: ArrowTemporalType>( |
10302 | | array: &IntervalMonthDayNanoArray, |
10303 | | cast_options: &CastOptions, |
10304 | | ) -> Result<PrimitiveArray<T>, ArrowError> { |
10305 | | let casted_array = cast_with_options(&array, &T::DATA_TYPE, cast_options)?; |
10306 | | casted_array |
10307 | | .as_any() |
10308 | | .downcast_ref::<PrimitiveArray<T>>() |
10309 | | .ok_or_else(|| { |
10310 | | ArrowError::ComputeError(format!("Failed to downcast to {}", T::DATA_TYPE)) |
10311 | | }) |
10312 | | .cloned() |
10313 | | } |
10314 | | |
10315 | | #[test] |
10316 | | fn test_cast_from_interval_to_duration() { |
10317 | | let nullable = CastOptions::default(); |
10318 | | let fallible = CastOptions { |
10319 | | safe: false, |
10320 | | format_options: FormatOptions::default(), |
10321 | | }; |
10322 | | let v = IntervalMonthDayNano::new(0, 0, 1234567); |
10323 | | |
10324 | | // from interval month day nano to duration second |
10325 | | let array = vec![v].into(); |
10326 | | let casted_array: DurationSecondArray = |
10327 | | cast_from_interval_to_duration(&array, &nullable).unwrap(); |
10328 | | assert_eq!(casted_array.value(0), 0); |
10329 | | |
10330 | | let array = vec![IntervalMonthDayNano::MAX].into(); |
10331 | | let casted_array: DurationSecondArray = |
10332 | | cast_from_interval_to_duration(&array, &nullable).unwrap(); |
10333 | | assert!(!casted_array.is_valid(0)); |
10334 | | |
10335 | | let res = cast_from_interval_to_duration::<DurationSecondType>(&array, &fallible); |
10336 | | assert!(res.is_err()); |
10337 | | |
10338 | | // from interval month day nano to duration millisecond |
10339 | | let array = vec![v].into(); |
10340 | | let casted_array: DurationMillisecondArray = |
10341 | | cast_from_interval_to_duration(&array, &nullable).unwrap(); |
10342 | | assert_eq!(casted_array.value(0), 1); |
10343 | | |
10344 | | let array = vec![IntervalMonthDayNano::MAX].into(); |
10345 | | let casted_array: DurationMillisecondArray = |
10346 | | cast_from_interval_to_duration(&array, &nullable).unwrap(); |
10347 | | assert!(!casted_array.is_valid(0)); |
10348 | | |
10349 | | let res = cast_from_interval_to_duration::<DurationMillisecondType>(&array, &fallible); |
10350 | | assert!(res.is_err()); |
10351 | | |
10352 | | // from interval month day nano to duration microsecond |
10353 | | let array = vec![v].into(); |
10354 | | let casted_array: DurationMicrosecondArray = |
10355 | | cast_from_interval_to_duration(&array, &nullable).unwrap(); |
10356 | | assert_eq!(casted_array.value(0), 1234); |
10357 | | |
10358 | | let array = vec![IntervalMonthDayNano::MAX].into(); |
10359 | | let casted_array = |
10360 | | cast_from_interval_to_duration::<DurationMicrosecondType>(&array, &nullable).unwrap(); |
10361 | | assert!(!casted_array.is_valid(0)); |
10362 | | |
10363 | | let casted_array = |
10364 | | cast_from_interval_to_duration::<DurationMicrosecondType>(&array, &fallible); |
10365 | | assert!(casted_array.is_err()); |
10366 | | |
10367 | | // from interval month day nano to duration nanosecond |
10368 | | let array = vec![v].into(); |
10369 | | let casted_array: DurationNanosecondArray = |
10370 | | cast_from_interval_to_duration(&array, &nullable).unwrap(); |
10371 | | assert_eq!(casted_array.value(0), 1234567); |
10372 | | |
10373 | | let array = vec![IntervalMonthDayNano::MAX].into(); |
10374 | | let casted_array: DurationNanosecondArray = |
10375 | | cast_from_interval_to_duration(&array, &nullable).unwrap(); |
10376 | | assert!(!casted_array.is_valid(0)); |
10377 | | |
10378 | | let casted_array = |
10379 | | cast_from_interval_to_duration::<DurationNanosecondType>(&array, &fallible); |
10380 | | assert!(casted_array.is_err()); |
10381 | | |
10382 | | let array = vec![ |
10383 | | IntervalMonthDayNanoType::make_value(0, 1, 0), |
10384 | | IntervalMonthDayNanoType::make_value(-1, 0, 0), |
10385 | | IntervalMonthDayNanoType::make_value(1, 1, 0), |
10386 | | IntervalMonthDayNanoType::make_value(1, 0, 1), |
10387 | | IntervalMonthDayNanoType::make_value(0, 0, -1), |
10388 | | ] |
10389 | | .into(); |
10390 | | let casted_array = |
10391 | | cast_from_interval_to_duration::<DurationNanosecondType>(&array, &nullable).unwrap(); |
10392 | | assert!(!casted_array.is_valid(0)); |
10393 | | assert!(!casted_array.is_valid(1)); |
10394 | | assert!(!casted_array.is_valid(2)); |
10395 | | assert!(!casted_array.is_valid(3)); |
10396 | | assert!(casted_array.is_valid(4)); |
10397 | | assert_eq!(casted_array.value(4), -1); |
10398 | | } |
10399 | | |
10400 | | /// helper function to test casting from interval year month to interval month day nano |
10401 | | fn cast_from_interval_year_month_to_interval_month_day_nano( |
10402 | | array: Vec<i32>, |
10403 | | cast_options: &CastOptions, |
10404 | | ) -> Result<PrimitiveArray<IntervalMonthDayNanoType>, ArrowError> { |
10405 | | let array = PrimitiveArray::<IntervalYearMonthType>::from(array); |
10406 | | let array = Arc::new(array) as ArrayRef; |
10407 | | let casted_array = cast_with_options( |
10408 | | &array, |
10409 | | &DataType::Interval(IntervalUnit::MonthDayNano), |
10410 | | cast_options, |
10411 | | )?; |
10412 | | casted_array |
10413 | | .as_any() |
10414 | | .downcast_ref::<IntervalMonthDayNanoArray>() |
10415 | | .ok_or_else(|| { |
10416 | | ArrowError::ComputeError( |
10417 | | "Failed to downcast to IntervalMonthDayNanoArray".to_string(), |
10418 | | ) |
10419 | | }) |
10420 | | .cloned() |
10421 | | } |
10422 | | |
10423 | | #[test] |
10424 | | fn test_cast_from_interval_year_month_to_interval_month_day_nano() { |
10425 | | // from interval year month to interval month day nano |
10426 | | let array = vec![1234567]; |
10427 | | let casted_array = cast_from_interval_year_month_to_interval_month_day_nano( |
10428 | | array, |
10429 | | &CastOptions::default(), |
10430 | | ) |
10431 | | .unwrap(); |
10432 | | assert_eq!( |
10433 | | casted_array.data_type(), |
10434 | | &DataType::Interval(IntervalUnit::MonthDayNano) |
10435 | | ); |
10436 | | assert_eq!( |
10437 | | casted_array.value(0), |
10438 | | IntervalMonthDayNano::new(1234567, 0, 0) |
10439 | | ); |
10440 | | } |
10441 | | |
10442 | | /// helper function to test casting from interval day time to interval month day nano |
10443 | | fn cast_from_interval_day_time_to_interval_month_day_nano( |
10444 | | array: Vec<IntervalDayTime>, |
10445 | | cast_options: &CastOptions, |
10446 | | ) -> Result<PrimitiveArray<IntervalMonthDayNanoType>, ArrowError> { |
10447 | | let array = PrimitiveArray::<IntervalDayTimeType>::from(array); |
10448 | | let array = Arc::new(array) as ArrayRef; |
10449 | | let casted_array = cast_with_options( |
10450 | | &array, |
10451 | | &DataType::Interval(IntervalUnit::MonthDayNano), |
10452 | | cast_options, |
10453 | | )?; |
10454 | | Ok(casted_array |
10455 | | .as_primitive::<IntervalMonthDayNanoType>() |
10456 | | .clone()) |
10457 | | } |
10458 | | |
10459 | | #[test] |
10460 | | fn test_cast_from_interval_day_time_to_interval_month_day_nano() { |
10461 | | // from interval day time to interval month day nano |
10462 | | let array = vec![IntervalDayTime::new(123, 0)]; |
10463 | | let casted_array = |
10464 | | cast_from_interval_day_time_to_interval_month_day_nano(array, &CastOptions::default()) |
10465 | | .unwrap(); |
10466 | | assert_eq!( |
10467 | | casted_array.data_type(), |
10468 | | &DataType::Interval(IntervalUnit::MonthDayNano) |
10469 | | ); |
10470 | | assert_eq!(casted_array.value(0), IntervalMonthDayNano::new(0, 123, 0)); |
10471 | | } |
10472 | | |
10473 | | #[test] |
10474 | | fn test_cast_below_unixtimestamp() { |
10475 | | let valid = StringArray::from(vec![ |
10476 | | "1900-01-03 23:59:59", |
10477 | | "1969-12-31 00:00:01", |
10478 | | "1989-12-31 00:00:01", |
10479 | | ]); |
10480 | | |
10481 | | let array = Arc::new(valid) as ArrayRef; |
10482 | | let casted_array = cast_with_options( |
10483 | | &array, |
10484 | | &DataType::Timestamp(TimeUnit::Nanosecond, Some("+00:00".into())), |
10485 | | &CastOptions { |
10486 | | safe: false, |
10487 | | format_options: FormatOptions::default(), |
10488 | | }, |
10489 | | ) |
10490 | | .unwrap(); |
10491 | | |
10492 | | let ts_array = casted_array |
10493 | | .as_primitive::<TimestampNanosecondType>() |
10494 | | .values() |
10495 | | .iter() |
10496 | | .map(|ts| ts / 1_000_000) |
10497 | | .collect::<Vec<_>>(); |
10498 | | |
10499 | | let array = TimestampMillisecondArray::from(ts_array).with_timezone("+00:00".to_string()); |
10500 | | let casted_array = cast(&array, &DataType::Date32).unwrap(); |
10501 | | let date_array = casted_array.as_primitive::<Date32Type>(); |
10502 | | let casted_array = cast(&date_array, &DataType::Utf8).unwrap(); |
10503 | | let string_array = casted_array.as_string::<i32>(); |
10504 | | assert_eq!("1900-01-03", string_array.value(0)); |
10505 | | assert_eq!("1969-12-31", string_array.value(1)); |
10506 | | assert_eq!("1989-12-31", string_array.value(2)); |
10507 | | } |
10508 | | |
10509 | | #[test] |
10510 | | fn test_nested_list() { |
10511 | | let mut list = ListBuilder::new(Int32Builder::new()); |
10512 | | list.append_value([Some(1), Some(2), Some(3)]); |
10513 | | list.append_value([Some(4), None, Some(6)]); |
10514 | | let list = list.finish(); |
10515 | | |
10516 | | let to_field = Field::new("nested", list.data_type().clone(), false); |
10517 | | let to = DataType::List(Arc::new(to_field)); |
10518 | | let out = cast(&list, &to).unwrap(); |
10519 | | let opts = FormatOptions::default().with_null("null"); |
10520 | | let formatted = ArrayFormatter::try_new(out.as_ref(), &opts).unwrap(); |
10521 | | |
10522 | | assert_eq!(formatted.value(0).to_string(), "[[1], [2], [3]]"); |
10523 | | assert_eq!(formatted.value(1).to_string(), "[[4], [null], [6]]"); |
10524 | | } |
10525 | | |
10526 | | #[test] |
10527 | | fn test_nested_list_cast() { |
10528 | | let mut builder = ListBuilder::new(ListBuilder::new(Int32Builder::new())); |
10529 | | builder.append_value([Some([Some(1), Some(2), None]), None]); |
10530 | | builder.append_value([None, Some([]), None]); |
10531 | | builder.append_null(); |
10532 | | builder.append_value([Some([Some(2), Some(3)])]); |
10533 | | let start = builder.finish(); |
10534 | | |
10535 | | let mut builder = LargeListBuilder::new(LargeListBuilder::new(Int8Builder::new())); |
10536 | | builder.append_value([Some([Some(1), Some(2), None]), None]); |
10537 | | builder.append_value([None, Some([]), None]); |
10538 | | builder.append_null(); |
10539 | | builder.append_value([Some([Some(2), Some(3)])]); |
10540 | | let expected = builder.finish(); |
10541 | | |
10542 | | let actual = cast(&start, expected.data_type()).unwrap(); |
10543 | | assert_eq!(actual.as_ref(), &expected); |
10544 | | } |
10545 | | |
10546 | | const CAST_OPTIONS: CastOptions<'static> = CastOptions { |
10547 | | safe: true, |
10548 | | format_options: FormatOptions::new(), |
10549 | | }; |
10550 | | |
10551 | | #[test] |
10552 | | #[allow(clippy::assertions_on_constants)] |
10553 | | fn test_const_options() { |
10554 | | assert!(CAST_OPTIONS.safe) |
10555 | | } |
10556 | | |
10557 | | #[test] |
10558 | | fn test_list_format_options() { |
10559 | | let options = CastOptions { |
10560 | | safe: false, |
10561 | | format_options: FormatOptions::default().with_null("null"), |
10562 | | }; |
10563 | | let array = ListArray::from_iter_primitive::<Int32Type, _, _>(vec![ |
10564 | | Some(vec![Some(0), Some(1), Some(2)]), |
10565 | | Some(vec![Some(0), None, Some(2)]), |
10566 | | ]); |
10567 | | let a = cast_with_options(&array, &DataType::Utf8, &options).unwrap(); |
10568 | | let r: Vec<_> = a.as_string::<i32>().iter().flatten().collect(); |
10569 | | assert_eq!(r, &["[0, 1, 2]", "[0, null, 2]"]); |
10570 | | } |
10571 | | #[test] |
10572 | | fn test_cast_string_to_timestamp_invalid_tz() { |
10573 | | // content after Z should be ignored |
10574 | | let bad_timestamp = "2023-12-05T21:58:10.45ZZTOP"; |
10575 | | let array = StringArray::from(vec![Some(bad_timestamp)]); |
10576 | | |
10577 | | let data_types = [ |
10578 | | DataType::Timestamp(TimeUnit::Second, None), |
10579 | | DataType::Timestamp(TimeUnit::Millisecond, None), |
10580 | | DataType::Timestamp(TimeUnit::Microsecond, None), |
10581 | | DataType::Timestamp(TimeUnit::Nanosecond, None), |
10582 | | ]; |
10583 | | |
10584 | | let cast_options = CastOptions { |
10585 | | safe: false, |
10586 | | ..Default::default() |
10587 | | }; |
10588 | | |
10589 | | for dt in data_types { |
10590 | | assert_eq!( |
10591 | | cast_with_options(&array, &dt, &cast_options) |
10592 | | .unwrap_err() |
10593 | | .to_string(), |
10594 | | "Parser error: Invalid timezone \"ZZTOP\": only offset based timezones supported without chrono-tz feature" |
10595 | | ); |
10596 | | } |
10597 | | } |
10598 | | #[test] |
10599 | | fn test_cast_struct_to_struct() { |
10600 | | let struct_type = DataType::Struct( |
10601 | | vec![ |
10602 | | Field::new("a", DataType::Boolean, false), |
10603 | | Field::new("b", DataType::Int32, false), |
10604 | | ] |
10605 | | .into(), |
10606 | | ); |
10607 | | let to_type = DataType::Struct( |
10608 | | vec![ |
10609 | | Field::new("a", DataType::Utf8, false), |
10610 | | Field::new("b", DataType::Utf8, false), |
10611 | | ] |
10612 | | .into(), |
10613 | | ); |
10614 | | let boolean = Arc::new(BooleanArray::from(vec![false, false, true, true])); |
10615 | | let int = Arc::new(Int32Array::from(vec![42, 28, 19, 31])); |
10616 | | let struct_array = StructArray::from(vec![ |
10617 | | ( |
10618 | | Arc::new(Field::new("b", DataType::Boolean, false)), |
10619 | | boolean.clone() as ArrayRef, |
10620 | | ), |
10621 | | ( |
10622 | | Arc::new(Field::new("c", DataType::Int32, false)), |
10623 | | int.clone() as ArrayRef, |
10624 | | ), |
10625 | | ]); |
10626 | | let casted_array = cast(&struct_array, &to_type).unwrap(); |
10627 | | let casted_array = casted_array.as_struct(); |
10628 | | assert_eq!(casted_array.data_type(), &to_type); |
10629 | | let casted_boolean_array = casted_array |
10630 | | .column(0) |
10631 | | .as_string::<i32>() |
10632 | | .into_iter() |
10633 | | .flatten() |
10634 | | .collect::<Vec<_>>(); |
10635 | | let casted_int_array = casted_array |
10636 | | .column(1) |
10637 | | .as_string::<i32>() |
10638 | | .into_iter() |
10639 | | .flatten() |
10640 | | .collect::<Vec<_>>(); |
10641 | | assert_eq!(casted_boolean_array, vec!["false", "false", "true", "true"]); |
10642 | | assert_eq!(casted_int_array, vec!["42", "28", "19", "31"]); |
10643 | | |
10644 | | // test for can't cast |
10645 | | let to_type = DataType::Struct( |
10646 | | vec![ |
10647 | | Field::new("a", DataType::Date32, false), |
10648 | | Field::new("b", DataType::Utf8, false), |
10649 | | ] |
10650 | | .into(), |
10651 | | ); |
10652 | | assert!(!can_cast_types(&struct_type, &to_type)); |
10653 | | let result = cast(&struct_array, &to_type); |
10654 | | assert_eq!( |
10655 | | "Cast error: Casting from Boolean to Date32 not supported", |
10656 | | result.unwrap_err().to_string() |
10657 | | ); |
10658 | | } |
10659 | | |
10660 | | #[test] |
10661 | | fn test_cast_struct_to_struct_nullability() { |
10662 | | let boolean = Arc::new(BooleanArray::from(vec![false, false, true, true])); |
10663 | | let int = Arc::new(Int32Array::from(vec![Some(42), None, Some(19), None])); |
10664 | | let struct_array = StructArray::from(vec![ |
10665 | | ( |
10666 | | Arc::new(Field::new("b", DataType::Boolean, false)), |
10667 | | boolean.clone() as ArrayRef, |
10668 | | ), |
10669 | | ( |
10670 | | Arc::new(Field::new("c", DataType::Int32, true)), |
10671 | | int.clone() as ArrayRef, |
10672 | | ), |
10673 | | ]); |
10674 | | |
10675 | | // okay: nullable to nullable |
10676 | | let to_type = DataType::Struct( |
10677 | | vec![ |
10678 | | Field::new("a", DataType::Utf8, false), |
10679 | | Field::new("b", DataType::Utf8, true), |
10680 | | ] |
10681 | | .into(), |
10682 | | ); |
10683 | | cast(&struct_array, &to_type).expect("Cast nullable to nullable struct field should work"); |
10684 | | |
10685 | | // error: nullable to non-nullable |
10686 | | let to_type = DataType::Struct( |
10687 | | vec![ |
10688 | | Field::new("a", DataType::Utf8, false), |
10689 | | Field::new("b", DataType::Utf8, false), |
10690 | | ] |
10691 | | .into(), |
10692 | | ); |
10693 | | cast(&struct_array, &to_type) |
10694 | | .expect_err("Cast nullable to non-nullable struct field should fail"); |
10695 | | |
10696 | | let boolean = Arc::new(BooleanArray::from(vec![false, false, true, true])); |
10697 | | let int = Arc::new(Int32Array::from(vec![i32::MAX, 25, 1, 100])); |
10698 | | let struct_array = StructArray::from(vec![ |
10699 | | ( |
10700 | | Arc::new(Field::new("b", DataType::Boolean, false)), |
10701 | | boolean.clone() as ArrayRef, |
10702 | | ), |
10703 | | ( |
10704 | | Arc::new(Field::new("c", DataType::Int32, false)), |
10705 | | int.clone() as ArrayRef, |
10706 | | ), |
10707 | | ]); |
10708 | | |
10709 | | // okay: non-nullable to non-nullable |
10710 | | let to_type = DataType::Struct( |
10711 | | vec![ |
10712 | | Field::new("a", DataType::Utf8, false), |
10713 | | Field::new("b", DataType::Utf8, false), |
10714 | | ] |
10715 | | .into(), |
10716 | | ); |
10717 | | cast(&struct_array, &to_type) |
10718 | | .expect("Cast non-nullable to non-nullable struct field should work"); |
10719 | | |
10720 | | // err: non-nullable to non-nullable but overflowing return null during casting |
10721 | | let to_type = DataType::Struct( |
10722 | | vec![ |
10723 | | Field::new("a", DataType::Utf8, false), |
10724 | | Field::new("b", DataType::Int8, false), |
10725 | | ] |
10726 | | .into(), |
10727 | | ); |
10728 | | cast(&struct_array, &to_type).expect_err( |
10729 | | "Cast non-nullable to non-nullable struct field returning null should fail", |
10730 | | ); |
10731 | | } |
10732 | | |
10733 | | #[test] |
10734 | | fn test_cast_struct_to_non_struct() { |
10735 | | let boolean = Arc::new(BooleanArray::from(vec![true, false])); |
10736 | | let struct_array = StructArray::from(vec![( |
10737 | | Arc::new(Field::new("a", DataType::Boolean, false)), |
10738 | | boolean.clone() as ArrayRef, |
10739 | | )]); |
10740 | | let to_type = DataType::Utf8; |
10741 | | let result = cast(&struct_array, &to_type); |
10742 | | assert_eq!( |
10743 | | r#"Cast error: Casting from Struct([Field { name: "a", data_type: Boolean, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }]) to Utf8 not supported"#, |
10744 | | result.unwrap_err().to_string() |
10745 | | ); |
10746 | | } |
10747 | | |
10748 | | #[test] |
10749 | | fn test_cast_non_struct_to_struct() { |
10750 | | let array = StringArray::from(vec!["a", "b"]); |
10751 | | let to_type = DataType::Struct(vec![Field::new("a", DataType::Boolean, false)].into()); |
10752 | | let result = cast(&array, &to_type); |
10753 | | assert_eq!( |
10754 | | r#"Cast error: Casting from Utf8 to Struct([Field { name: "a", data_type: Boolean, nullable: false, dict_id: 0, dict_is_ordered: false, metadata: {} }]) not supported"#, |
10755 | | result.unwrap_err().to_string() |
10756 | | ); |
10757 | | } |
10758 | | |
10759 | | fn run_decimal_cast_test_case_between_multiple_types(t: DecimalCastTestConfig) { |
10760 | | run_decimal_cast_test_case::<Decimal128Type, Decimal128Type>(t.clone()); |
10761 | | run_decimal_cast_test_case::<Decimal128Type, Decimal256Type>(t.clone()); |
10762 | | run_decimal_cast_test_case::<Decimal256Type, Decimal128Type>(t.clone()); |
10763 | | run_decimal_cast_test_case::<Decimal256Type, Decimal256Type>(t.clone()); |
10764 | | } |
10765 | | |
10766 | | #[test] |
10767 | | fn test_decimal_to_decimal_coverage() { |
10768 | | let test_cases = [ |
10769 | | // increase precision, increase scale, infallible |
10770 | | DecimalCastTestConfig { |
10771 | | input_prec: 5, |
10772 | | input_scale: 1, |
10773 | | input_repr: 99999, // 9999.9 |
10774 | | output_prec: 10, |
10775 | | output_scale: 6, |
10776 | | expected_output_repr: Ok(9999900000), // 9999.900000 |
10777 | | }, |
10778 | | // increase precision, increase scale, fallible, safe |
10779 | | DecimalCastTestConfig { |
10780 | | input_prec: 5, |
10781 | | input_scale: 1, |
10782 | | input_repr: 99, // 9999.9 |
10783 | | output_prec: 7, |
10784 | | output_scale: 6, |
10785 | | expected_output_repr: Ok(9900000), // 9.900000 |
10786 | | }, |
10787 | | // increase precision, increase scale, fallible, unsafe |
10788 | | DecimalCastTestConfig { |
10789 | | input_prec: 5, |
10790 | | input_scale: 1, |
10791 | | input_repr: 99999, // 9999.9 |
10792 | | output_prec: 7, |
10793 | | output_scale: 6, |
10794 | | expected_output_repr: Err("Invalid argument error: 9999900000 is too large to store in a {} of precision 7. Max is 9999999".to_string()) // max is 9.999999 |
10795 | | }, |
10796 | | // increase precision, decrease scale, always infallible |
10797 | | DecimalCastTestConfig { |
10798 | | input_prec: 5, |
10799 | | input_scale: 3, |
10800 | | input_repr: 99999, // 99.999 |
10801 | | output_prec: 10, |
10802 | | output_scale: 2, |
10803 | | expected_output_repr: Ok(10000), // 100.00 |
10804 | | }, |
10805 | | // increase precision, decrease scale, no rouding |
10806 | | DecimalCastTestConfig { |
10807 | | input_prec: 5, |
10808 | | input_scale: 3, |
10809 | | input_repr: 99994, // 99.994 |
10810 | | output_prec: 10, |
10811 | | output_scale: 2, |
10812 | | expected_output_repr: Ok(9999), // 99.99 |
10813 | | }, |
10814 | | // increase precision, don't change scale, always infallible |
10815 | | DecimalCastTestConfig { |
10816 | | input_prec: 5, |
10817 | | input_scale: 3, |
10818 | | input_repr: 99999, // 99.999 |
10819 | | output_prec: 10, |
10820 | | output_scale: 3, |
10821 | | expected_output_repr: Ok(99999), // 99.999 |
10822 | | }, |
10823 | | // decrease precision, increase scale, safe |
10824 | | DecimalCastTestConfig { |
10825 | | input_prec: 10, |
10826 | | input_scale: 5, |
10827 | | input_repr: 999999, // 9.99999 |
10828 | | output_prec: 8, |
10829 | | output_scale: 7, |
10830 | | expected_output_repr: Ok(99999900), // 9.9999900 |
10831 | | }, |
10832 | | // decrease precision, increase scale, unsafe |
10833 | | DecimalCastTestConfig { |
10834 | | input_prec: 10, |
10835 | | input_scale: 5, |
10836 | | input_repr: 9999999, // 99.99999 |
10837 | | output_prec: 8, |
10838 | | output_scale: 7, |
10839 | | expected_output_repr: Err("Invalid argument error: 999999900 is too large to store in a {} of precision 8. Max is 99999999".to_string()) // max is 9.9999999 |
10840 | | }, |
10841 | | // decrease precision, decrease scale, safe, infallible |
10842 | | DecimalCastTestConfig { |
10843 | | input_prec: 7, |
10844 | | input_scale: 4, |
10845 | | input_repr: 9999999, // 999.9999 |
10846 | | output_prec: 6, |
10847 | | output_scale: 2, |
10848 | | expected_output_repr: Ok(100000), |
10849 | | }, |
10850 | | // decrease precision, decrease scale, safe, fallible |
10851 | | DecimalCastTestConfig { |
10852 | | input_prec: 10, |
10853 | | input_scale: 5, |
10854 | | input_repr: 12345678, // 123.45678 |
10855 | | output_prec: 8, |
10856 | | output_scale: 3, |
10857 | | expected_output_repr: Ok(123457), // 123.457 |
10858 | | }, |
10859 | | // decrease precision, decrease scale, unsafe |
10860 | | DecimalCastTestConfig { |
10861 | | input_prec: 10, |
10862 | | input_scale: 5, |
10863 | | input_repr: 9999999, // 99.99999 |
10864 | | output_prec: 4, |
10865 | | output_scale: 3, |
10866 | | expected_output_repr: Err("Invalid argument error: 100000 is too large to store in a {} of precision 4. Max is 9999".to_string()) // max is 9.999 |
10867 | | }, |
10868 | | // decrease precision, same scale, safe |
10869 | | DecimalCastTestConfig { |
10870 | | input_prec: 10, |
10871 | | input_scale: 5, |
10872 | | input_repr: 999999, // 9.99999 |
10873 | | output_prec: 6, |
10874 | | output_scale: 5, |
10875 | | expected_output_repr: Ok(999999), // 9.99999 |
10876 | | }, |
10877 | | // decrease precision, same scale, unsafe |
10878 | | DecimalCastTestConfig { |
10879 | | input_prec: 10, |
10880 | | input_scale: 5, |
10881 | | input_repr: 9999999, // 99.99999 |
10882 | | output_prec: 6, |
10883 | | output_scale: 5, |
10884 | | expected_output_repr: Err("Invalid argument error: 9999999 is too large to store in a {} of precision 6. Max is 999999".to_string()) // max is 9.99999 |
10885 | | }, |
10886 | | // same precision, increase scale, safe |
10887 | | DecimalCastTestConfig { |
10888 | | input_prec: 7, |
10889 | | input_scale: 4, |
10890 | | input_repr: 12345, // 1.2345 |
10891 | | output_prec: 7, |
10892 | | output_scale: 6, |
10893 | | expected_output_repr: Ok(1234500), // 1.234500 |
10894 | | }, |
10895 | | // same precision, increase scale, unsafe |
10896 | | DecimalCastTestConfig { |
10897 | | input_prec: 7, |
10898 | | input_scale: 4, |
10899 | | input_repr: 123456, // 12.3456 |
10900 | | output_prec: 7, |
10901 | | output_scale: 6, |
10902 | | expected_output_repr: Err("Invalid argument error: 12345600 is too large to store in a {} of precision 7. Max is 9999999".to_string()) // max is 9.99999 |
10903 | | }, |
10904 | | // same precision, decrease scale, infallible |
10905 | | DecimalCastTestConfig { |
10906 | | input_prec: 7, |
10907 | | input_scale: 5, |
10908 | | input_repr: 1234567, // 12.34567 |
10909 | | output_prec: 7, |
10910 | | output_scale: 4, |
10911 | | expected_output_repr: Ok(123457), // 12.3457 |
10912 | | }, |
10913 | | // same precision, same scale, infallible |
10914 | | DecimalCastTestConfig { |
10915 | | input_prec: 7, |
10916 | | input_scale: 5, |
10917 | | input_repr: 9999999, // 99.99999 |
10918 | | output_prec: 7, |
10919 | | output_scale: 5, |
10920 | | expected_output_repr: Ok(9999999), // 99.99999 |
10921 | | }, |
10922 | | // precision increase, input scale & output scale = 0, infallible |
10923 | | DecimalCastTestConfig { |
10924 | | input_prec: 7, |
10925 | | input_scale: 0, |
10926 | | input_repr: 1234567, // 1234567 |
10927 | | output_prec: 8, |
10928 | | output_scale: 0, |
10929 | | expected_output_repr: Ok(1234567), // 1234567 |
10930 | | }, |
10931 | | // precision decrease, input scale & output scale = 0, failure |
10932 | | DecimalCastTestConfig { |
10933 | | input_prec: 7, |
10934 | | input_scale: 0, |
10935 | | input_repr: 1234567, // 1234567 |
10936 | | output_prec: 6, |
10937 | | output_scale: 0, |
10938 | | expected_output_repr: Err("Invalid argument error: 1234567 is too large to store in a {} of precision 6. Max is 999999".to_string()) |
10939 | | }, |
10940 | | // precision decrease, input scale & output scale = 0, success |
10941 | | DecimalCastTestConfig { |
10942 | | input_prec: 7, |
10943 | | input_scale: 0, |
10944 | | input_repr: 123456, // 123456 |
10945 | | output_prec: 6, |
10946 | | output_scale: 0, |
10947 | | expected_output_repr: Ok(123456), // 123456 |
10948 | | }, |
10949 | | ]; |
10950 | | |
10951 | | for t in test_cases { |
10952 | | run_decimal_cast_test_case_between_multiple_types(t); |
10953 | | } |
10954 | | } |
10955 | | |
10956 | | #[test] |
10957 | | fn test_decimal_to_decimal_increase_scale_and_precision_unchecked() { |
10958 | | let test_cases = [ |
10959 | | DecimalCastTestConfig { |
10960 | | input_prec: 5, |
10961 | | input_scale: 0, |
10962 | | input_repr: 99999, |
10963 | | output_prec: 10, |
10964 | | output_scale: 5, |
10965 | | expected_output_repr: Ok(9999900000), |
10966 | | }, |
10967 | | DecimalCastTestConfig { |
10968 | | input_prec: 5, |
10969 | | input_scale: 0, |
10970 | | input_repr: -99999, |
10971 | | output_prec: 10, |
10972 | | output_scale: 5, |
10973 | | expected_output_repr: Ok(-9999900000), |
10974 | | }, |
10975 | | DecimalCastTestConfig { |
10976 | | input_prec: 5, |
10977 | | input_scale: 2, |
10978 | | input_repr: 99999, |
10979 | | output_prec: 10, |
10980 | | output_scale: 5, |
10981 | | expected_output_repr: Ok(99999000), |
10982 | | }, |
10983 | | DecimalCastTestConfig { |
10984 | | input_prec: 5, |
10985 | | input_scale: -2, |
10986 | | input_repr: -99999, |
10987 | | output_prec: 10, |
10988 | | output_scale: 3, |
10989 | | expected_output_repr: Ok(-9999900000), |
10990 | | }, |
10991 | | DecimalCastTestConfig { |
10992 | | input_prec: 5, |
10993 | | input_scale: 3, |
10994 | | input_repr: -12345, |
10995 | | output_prec: 6, |
10996 | | output_scale: 5, |
10997 | | expected_output_repr: Err("Invalid argument error: -1234500 is too small to store in a {} of precision 6. Min is -999999".to_string()) |
10998 | | }, |
10999 | | ]; |
11000 | | |
11001 | | for t in test_cases { |
11002 | | run_decimal_cast_test_case_between_multiple_types(t); |
11003 | | } |
11004 | | } |
11005 | | |
11006 | | #[test] |
11007 | | fn test_decimal_to_decimal_decrease_scale_and_precision_unchecked() { |
11008 | | let test_cases = [ |
11009 | | DecimalCastTestConfig { |
11010 | | input_prec: 5, |
11011 | | input_scale: 0, |
11012 | | input_repr: 99999, |
11013 | | output_scale: -3, |
11014 | | output_prec: 3, |
11015 | | expected_output_repr: Ok(100), |
11016 | | }, |
11017 | | DecimalCastTestConfig { |
11018 | | input_prec: 5, |
11019 | | input_scale: 0, |
11020 | | input_repr: -99999, |
11021 | | output_prec: 1, |
11022 | | output_scale: -5, |
11023 | | expected_output_repr: Ok(-1), |
11024 | | }, |
11025 | | DecimalCastTestConfig { |
11026 | | input_prec: 10, |
11027 | | input_scale: 2, |
11028 | | input_repr: 123456789, |
11029 | | output_prec: 5, |
11030 | | output_scale: -2, |
11031 | | expected_output_repr: Ok(12346), |
11032 | | }, |
11033 | | DecimalCastTestConfig { |
11034 | | input_prec: 10, |
11035 | | input_scale: 4, |
11036 | | input_repr: -9876543210, |
11037 | | output_prec: 7, |
11038 | | output_scale: 0, |
11039 | | expected_output_repr: Ok(-987654), |
11040 | | }, |
11041 | | DecimalCastTestConfig { |
11042 | | input_prec: 7, |
11043 | | input_scale: 4, |
11044 | | input_repr: 9999999, |
11045 | | output_prec: 6, |
11046 | | output_scale: 3, |
11047 | | expected_output_repr: |
11048 | | Err("Invalid argument error: 1000000 is too large to store in a {} of precision 6. Max is 999999".to_string()), |
11049 | | }, |
11050 | | ]; |
11051 | | for t in test_cases { |
11052 | | run_decimal_cast_test_case_between_multiple_types(t); |
11053 | | } |
11054 | | } |
11055 | | |
11056 | | #[test] |
11057 | | fn test_decimal_to_decimal_throw_error_on_precision_overflow_same_scale() { |
11058 | | let array = vec![Some(123456789)]; |
11059 | | let array = create_decimal128_array(array, 24, 2).unwrap(); |
11060 | | let input_type = DataType::Decimal128(24, 2); |
11061 | | let output_type = DataType::Decimal128(6, 2); |
11062 | | assert!(can_cast_types(&input_type, &output_type)); |
11063 | | |
11064 | | let options = CastOptions { |
11065 | | safe: false, |
11066 | | ..Default::default() |
11067 | | }; |
11068 | | let result = cast_with_options(&array, &output_type, &options); |
11069 | | assert_eq!(result.unwrap_err().to_string(), |
11070 | | "Invalid argument error: 123456789 is too large to store in a Decimal128 of precision 6. Max is 999999"); |
11071 | | } |
11072 | | |
11073 | | #[test] |
11074 | | fn test_decimal_to_decimal_same_scale() { |
11075 | | let array = vec![Some(520)]; |
11076 | | let array = create_decimal128_array(array, 4, 2).unwrap(); |
11077 | | let input_type = DataType::Decimal128(4, 2); |
11078 | | let output_type = DataType::Decimal128(3, 2); |
11079 | | assert!(can_cast_types(&input_type, &output_type)); |
11080 | | |
11081 | | let options = CastOptions { |
11082 | | safe: false, |
11083 | | ..Default::default() |
11084 | | }; |
11085 | | let result = cast_with_options(&array, &output_type, &options); |
11086 | | assert_eq!( |
11087 | | result.unwrap().as_primitive::<Decimal128Type>().value(0), |
11088 | | 520 |
11089 | | ); |
11090 | | |
11091 | | // Cast 0 of decimal(3, 0) type to decimal(2, 0) |
11092 | | assert_eq!( |
11093 | | &cast( |
11094 | | &create_decimal128_array(vec![Some(0)], 3, 0).unwrap(), |
11095 | | &DataType::Decimal128(2, 0) |
11096 | | ) |
11097 | | .unwrap(), |
11098 | | &(Arc::new(create_decimal128_array(vec![Some(0)], 2, 0).unwrap()) as ArrayRef) |
11099 | | ); |
11100 | | } |
11101 | | |
11102 | | #[test] |
11103 | | fn test_decimal_to_decimal_throw_error_on_precision_overflow_lower_scale() { |
11104 | | let array = vec![Some(123456789)]; |
11105 | | let array = create_decimal128_array(array, 24, 4).unwrap(); |
11106 | | let input_type = DataType::Decimal128(24, 4); |
11107 | | let output_type = DataType::Decimal128(6, 2); |
11108 | | assert!(can_cast_types(&input_type, &output_type)); |
11109 | | |
11110 | | let options = CastOptions { |
11111 | | safe: false, |
11112 | | ..Default::default() |
11113 | | }; |
11114 | | let result = cast_with_options(&array, &output_type, &options); |
11115 | | assert_eq!(result.unwrap_err().to_string(), |
11116 | | "Invalid argument error: 1234568 is too large to store in a Decimal128 of precision 6. Max is 999999"); |
11117 | | } |
11118 | | |
11119 | | #[test] |
11120 | | fn test_decimal_to_decimal_throw_error_on_precision_overflow_greater_scale() { |
11121 | | let array = vec![Some(123456789)]; |
11122 | | let array = create_decimal128_array(array, 24, 2).unwrap(); |
11123 | | let input_type = DataType::Decimal128(24, 2); |
11124 | | let output_type = DataType::Decimal128(6, 3); |
11125 | | assert!(can_cast_types(&input_type, &output_type)); |
11126 | | |
11127 | | let options = CastOptions { |
11128 | | safe: false, |
11129 | | ..Default::default() |
11130 | | }; |
11131 | | let result = cast_with_options(&array, &output_type, &options); |
11132 | | assert_eq!(result.unwrap_err().to_string(), |
11133 | | "Invalid argument error: 1234567890 is too large to store in a Decimal128 of precision 6. Max is 999999"); |
11134 | | } |
11135 | | |
11136 | | #[test] |
11137 | | fn test_decimal_to_decimal_throw_error_on_precision_overflow_diff_type() { |
11138 | | let array = vec![Some(123456789)]; |
11139 | | let array = create_decimal128_array(array, 24, 2).unwrap(); |
11140 | | let input_type = DataType::Decimal128(24, 2); |
11141 | | let output_type = DataType::Decimal256(6, 2); |
11142 | | assert!(can_cast_types(&input_type, &output_type)); |
11143 | | |
11144 | | let options = CastOptions { |
11145 | | safe: false, |
11146 | | ..Default::default() |
11147 | | }; |
11148 | | let result = cast_with_options(&array, &output_type, &options); |
11149 | | assert_eq!(result.unwrap_err().to_string(), |
11150 | | "Invalid argument error: 123456789 is too large to store in a Decimal256 of precision 6. Max is 999999"); |
11151 | | } |
11152 | | |
11153 | | #[test] |
11154 | | fn test_first_none() { |
11155 | | let array = Arc::new(ListArray::from_iter_primitive::<Int64Type, _, _>(vec![ |
11156 | | None, |
11157 | | Some(vec![Some(1), Some(2)]), |
11158 | | ])) as ArrayRef; |
11159 | | let data_type = |
11160 | | DataType::FixedSizeList(FieldRef::new(Field::new("item", DataType::Int64, true)), 2); |
11161 | | let opt = CastOptions::default(); |
11162 | | let r = cast_with_options(&array, &data_type, &opt).unwrap(); |
11163 | | |
11164 | | let fixed_array = Arc::new(FixedSizeListArray::from_iter_primitive::<Int64Type, _, _>( |
11165 | | vec![None, Some(vec![Some(1), Some(2)])], |
11166 | | 2, |
11167 | | )) as ArrayRef; |
11168 | | assert_eq!(*fixed_array, *r); |
11169 | | } |
11170 | | |
11171 | | #[test] |
11172 | | fn test_first_last_none() { |
11173 | | let array = Arc::new(ListArray::from_iter_primitive::<Int64Type, _, _>(vec![ |
11174 | | None, |
11175 | | Some(vec![Some(1), Some(2)]), |
11176 | | None, |
11177 | | ])) as ArrayRef; |
11178 | | let data_type = |
11179 | | DataType::FixedSizeList(FieldRef::new(Field::new("item", DataType::Int64, true)), 2); |
11180 | | let opt = CastOptions::default(); |
11181 | | let r = cast_with_options(&array, &data_type, &opt).unwrap(); |
11182 | | |
11183 | | let fixed_array = Arc::new(FixedSizeListArray::from_iter_primitive::<Int64Type, _, _>( |
11184 | | vec![None, Some(vec![Some(1), Some(2)]), None], |
11185 | | 2, |
11186 | | )) as ArrayRef; |
11187 | | assert_eq!(*fixed_array, *r); |
11188 | | } |
11189 | | } |