/Users/andrewlamb/Software/arrow-rs/arrow-buffer/src/builder/boolean.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 | | use crate::{bit_mask, bit_util, BooleanBuffer, Buffer, MutableBuffer}; |
19 | | use std::ops::Range; |
20 | | |
21 | | /// Builder for [`BooleanBuffer`] |
22 | | /// |
23 | | /// # See Also |
24 | | /// |
25 | | /// * [`NullBuffer`] for building [`BooleanBuffer`]s for representing nulls |
26 | | /// |
27 | | /// [`NullBuffer`]: crate::NullBuffer |
28 | | #[derive(Debug)] |
29 | | pub struct BooleanBufferBuilder { |
30 | | buffer: MutableBuffer, |
31 | | len: usize, |
32 | | } |
33 | | |
34 | | impl BooleanBufferBuilder { |
35 | | /// Creates a new `BooleanBufferBuilder` with sufficient space for |
36 | | /// `capacity` bits (not bytes). |
37 | | /// |
38 | | /// The capacity is rounded up to the nearest multiple of 8 for the |
39 | | /// allocation. |
40 | | #[inline] |
41 | 207 | pub fn new(capacity: usize) -> Self { |
42 | 207 | let byte_capacity = bit_util::ceil(capacity, 8); |
43 | 207 | let buffer = MutableBuffer::new(byte_capacity); |
44 | 207 | Self { buffer, len: 0 } |
45 | 207 | } |
46 | | |
47 | | /// Creates a new `BooleanBufferBuilder` from [`MutableBuffer`] of `len` |
48 | 0 | pub fn new_from_buffer(buffer: MutableBuffer, len: usize) -> Self { |
49 | 0 | assert!(len <= buffer.len() * 8); |
50 | 0 | let mut s = Self { |
51 | 0 | len: buffer.len() * 8, |
52 | 0 | buffer, |
53 | 0 | }; |
54 | 0 | s.truncate(len); |
55 | 0 | s |
56 | 0 | } |
57 | | |
58 | | /// Returns the length of the buffer |
59 | | #[inline] |
60 | 37 | pub fn len(&self) -> usize { |
61 | 37 | self.len |
62 | 37 | } |
63 | | |
64 | | /// Sets a bit in the buffer at `index` |
65 | | #[inline] |
66 | 11 | pub fn set_bit(&mut self, index: usize, v: bool) { |
67 | 11 | if v { |
68 | 11 | bit_util::set_bit(self.buffer.as_mut(), index); |
69 | 11 | } else { |
70 | 0 | bit_util::unset_bit(self.buffer.as_mut(), index); |
71 | 0 | } |
72 | 11 | } |
73 | | |
74 | | /// Gets a bit in the buffer at `index` |
75 | | #[inline] |
76 | | pub fn get_bit(&self, index: usize) -> bool { |
77 | | bit_util::get_bit(self.buffer.as_slice(), index) |
78 | | } |
79 | | |
80 | | /// Returns true if empty |
81 | | #[inline] |
82 | | pub fn is_empty(&self) -> bool { |
83 | | self.len == 0 |
84 | | } |
85 | | |
86 | | /// Returns the capacity of the buffer, in bits (not bytes) |
87 | | /// |
88 | | /// Note this |
89 | | /// |
90 | | /// # Example |
91 | | /// ``` |
92 | | /// # use arrow_buffer::builder::BooleanBufferBuilder; |
93 | | /// // empty requires 0 bytes |
94 | | /// let b = BooleanBufferBuilder::new(0); |
95 | | /// assert_eq!(0, b.capacity()); |
96 | | /// // Creating space for 1 bit results in 64 bytes (space for 512 bits) |
97 | | /// // (64 is the minimum allocation size for 64 bit architectures) |
98 | | /// let mut b = BooleanBufferBuilder::new(1); |
99 | | /// assert_eq!(512, b.capacity()); |
100 | | /// // 1000 bits requires 128 bytes (space for 1024 bits) |
101 | | /// b.append_n(1000, true); |
102 | | /// assert_eq!(1024, b.capacity()); |
103 | | /// ``` |
104 | | #[inline] |
105 | 0 | pub fn capacity(&self) -> usize { |
106 | 0 | self.buffer.capacity() * 8 |
107 | 0 | } |
108 | | |
109 | | /// Advances the buffer by `additional` bits |
110 | | #[inline] |
111 | 760 | pub fn advance(&mut self, additional: usize) { |
112 | 760 | let new_len = self.len + additional; |
113 | 760 | let new_len_bytes = bit_util::ceil(new_len, 8); |
114 | 760 | if new_len_bytes > self.buffer.len() { |
115 | 153 | self.buffer.resize(new_len_bytes, 0); |
116 | 607 | } |
117 | 760 | self.len = new_len; |
118 | 760 | } |
119 | | |
120 | | /// Truncates the builder to the given length |
121 | | /// |
122 | | /// If `len` is greater than the buffer's current length, this has no effect |
123 | | #[inline] |
124 | 0 | pub fn truncate(&mut self, len: usize) { |
125 | 0 | if len > self.len { |
126 | 0 | return; |
127 | 0 | } |
128 | | |
129 | 0 | let new_len_bytes = bit_util::ceil(len, 8); |
130 | 0 | self.buffer.truncate(new_len_bytes); |
131 | 0 | self.len = len; |
132 | | |
133 | 0 | let remainder = self.len % 8; |
134 | 0 | if remainder != 0 { |
135 | 0 | let mask = (1_u8 << remainder).wrapping_sub(1); |
136 | 0 | *self.buffer.as_mut().last_mut().unwrap() &= mask; |
137 | 0 | } |
138 | 0 | } |
139 | | |
140 | | /// Reserve space to at least `additional` new bits. |
141 | | /// Capacity will be `>= self.len() + additional`. |
142 | | /// New bytes are uninitialized and reading them is undefined behavior. |
143 | | #[inline] |
144 | | pub fn reserve(&mut self, additional: usize) { |
145 | | let capacity = self.len + additional; |
146 | | if capacity > self.capacity() { |
147 | | // convert differential to bytes |
148 | | let additional = bit_util::ceil(capacity, 8) - self.buffer.len(); |
149 | | self.buffer.reserve(additional); |
150 | | } |
151 | | } |
152 | | |
153 | | /// Resizes the buffer, either truncating its contents (with no change in capacity), or |
154 | | /// growing it (potentially reallocating it) and writing `false` in the newly available bits. |
155 | | #[inline] |
156 | 0 | pub fn resize(&mut self, len: usize) { |
157 | 0 | match len.checked_sub(self.len) { |
158 | 0 | Some(delta) => self.advance(delta), |
159 | 0 | None => self.truncate(len), |
160 | | } |
161 | 0 | } |
162 | | |
163 | | /// Appends a boolean `v` into the buffer |
164 | | #[inline] |
165 | 708 | pub fn append(&mut self, v: bool) { |
166 | 708 | self.advance(1); |
167 | 708 | if v { |
168 | 347 | unsafe { bit_util::set_bit_raw(self.buffer.as_mut_ptr(), self.len - 1) }; |
169 | 361 | } |
170 | 708 | } |
171 | | |
172 | | /// Appends n `additional` bits of value `v` into the buffer |
173 | | #[inline] |
174 | 148 | pub fn append_n(&mut self, additional: usize, v: bool) { |
175 | 148 | match v { |
176 | | true => { |
177 | 148 | let new_len = self.len + additional; |
178 | 148 | let new_len_bytes = bit_util::ceil(new_len, 8); |
179 | 148 | let cur_remainder = self.len % 8; |
180 | 148 | let new_remainder = new_len % 8; |
181 | | |
182 | 148 | if cur_remainder != 0 { |
183 | | // Pad last byte with 1s |
184 | 17 | *self.buffer.as_slice_mut().last_mut().unwrap() |= !((1 << cur_remainder) - 1) |
185 | 131 | } |
186 | 148 | self.buffer.resize(new_len_bytes, 0xFF); |
187 | 148 | if new_remainder != 0 { |
188 | | // Clear remaining bits |
189 | 99 | *self.buffer.as_slice_mut().last_mut().unwrap() &= (1 << new_remainder) - 1 |
190 | 49 | } |
191 | 148 | self.len = new_len; |
192 | | } |
193 | 0 | false => self.advance(additional), |
194 | | } |
195 | 148 | } |
196 | | |
197 | | /// Appends a slice of booleans into the buffer |
198 | | #[inline] |
199 | 1 | pub fn append_slice(&mut self, slice: &[bool]) { |
200 | 1 | let additional = slice.len(); |
201 | 1 | self.advance(additional); |
202 | | |
203 | 1 | let offset = self.len() - additional; |
204 | 5 | for (i, v) in slice1 .iter1 ().enumerate1 () { |
205 | 5 | if *v { |
206 | 3 | unsafe { bit_util::set_bit_raw(self.buffer.as_mut_ptr(), offset + i) } |
207 | 2 | } |
208 | | } |
209 | 1 | } |
210 | | |
211 | | /// Append `range` bits from `to_set` |
212 | | /// |
213 | | /// `to_set` is a slice of bits packed LSB-first into `[u8]` |
214 | | /// |
215 | | /// # Panics |
216 | | /// |
217 | | /// Panics if `to_set` does not contain `ceil(range.end / 8)` bytes |
218 | 45 | pub fn append_packed_range(&mut self, range: Range<usize>, to_set: &[u8]) { |
219 | 45 | let offset_write = self.len; |
220 | 45 | let len = range.end - range.start; |
221 | 45 | self.advance(len); |
222 | 45 | bit_mask::set_bits( |
223 | 45 | self.buffer.as_slice_mut(), |
224 | 45 | to_set, |
225 | 45 | offset_write, |
226 | 45 | range.start, |
227 | 45 | len, |
228 | | ); |
229 | 45 | } |
230 | | |
231 | | /// Append [`BooleanBuffer`] to this [`BooleanBufferBuilder`] |
232 | 45 | pub fn append_buffer(&mut self, buffer: &BooleanBuffer) { |
233 | 45 | let range = buffer.offset()..buffer.offset() + buffer.len(); |
234 | 45 | self.append_packed_range(range, buffer.values()) |
235 | 45 | } |
236 | | |
237 | | /// Returns the packed bits |
238 | 0 | pub fn as_slice(&self) -> &[u8] { |
239 | 0 | self.buffer.as_slice() |
240 | 0 | } |
241 | | |
242 | | /// Returns the packed bits |
243 | 0 | pub fn as_slice_mut(&mut self) -> &mut [u8] { |
244 | 0 | self.buffer.as_slice_mut() |
245 | 0 | } |
246 | | |
247 | | /// Creates a [`BooleanBuffer`] |
248 | | #[inline] |
249 | 202 | pub fn finish(&mut self) -> BooleanBuffer { |
250 | 202 | let buf = std::mem::replace(&mut self.buffer, MutableBuffer::new(0)); |
251 | 202 | let len = std::mem::replace(&mut self.len, 0); |
252 | 202 | BooleanBuffer::new(buf.into(), 0, len) |
253 | 202 | } |
254 | | |
255 | | /// Builds the [BooleanBuffer] without resetting the builder. |
256 | 0 | pub fn finish_cloned(&self) -> BooleanBuffer { |
257 | 0 | BooleanBuffer::new(Buffer::from_slice_ref(self.as_slice()), 0, self.len) |
258 | 0 | } |
259 | | } |
260 | | |
261 | | impl From<BooleanBufferBuilder> for Buffer { |
262 | | #[inline] |
263 | 15 | fn from(builder: BooleanBufferBuilder) -> Self { |
264 | 15 | builder.buffer.into() |
265 | 15 | } |
266 | | } |
267 | | |
268 | | impl From<BooleanBufferBuilder> for BooleanBuffer { |
269 | | #[inline] |
270 | | fn from(builder: BooleanBufferBuilder) -> Self { |
271 | | BooleanBuffer::new(builder.buffer.into(), 0, builder.len) |
272 | | } |
273 | | } |
274 | | |
275 | | #[cfg(test)] |
276 | | mod tests { |
277 | | use super::*; |
278 | | |
279 | | #[test] |
280 | | fn test_boolean_buffer_builder_write_bytes() { |
281 | | let mut b = BooleanBufferBuilder::new(4); |
282 | | b.append(false); |
283 | | b.append(true); |
284 | | b.append(false); |
285 | | b.append(true); |
286 | | assert_eq!(4, b.len()); |
287 | | assert_eq!(512, b.capacity()); |
288 | | let buffer = b.finish(); |
289 | | assert_eq!(4, buffer.len()); |
290 | | |
291 | | // Overallocate capacity |
292 | | let mut b = BooleanBufferBuilder::new(8); |
293 | | b.append_slice(&[false, true, false, true]); |
294 | | assert_eq!(4, b.len()); |
295 | | assert_eq!(512, b.capacity()); |
296 | | let buffer = b.finish(); |
297 | | assert_eq!(4, buffer.len()); |
298 | | } |
299 | | |
300 | | #[test] |
301 | | fn test_boolean_buffer_builder_unset_first_bit() { |
302 | | let mut buffer = BooleanBufferBuilder::new(4); |
303 | | buffer.append(true); |
304 | | buffer.append(true); |
305 | | buffer.append(false); |
306 | | buffer.append(true); |
307 | | buffer.set_bit(0, false); |
308 | | assert_eq!(buffer.len(), 4); |
309 | | assert_eq!(buffer.finish().values(), &[0b1010_u8]); |
310 | | } |
311 | | |
312 | | #[test] |
313 | | fn test_boolean_buffer_builder_unset_last_bit() { |
314 | | let mut buffer = BooleanBufferBuilder::new(4); |
315 | | buffer.append(true); |
316 | | buffer.append(true); |
317 | | buffer.append(false); |
318 | | buffer.append(true); |
319 | | buffer.set_bit(3, false); |
320 | | assert_eq!(buffer.len(), 4); |
321 | | assert_eq!(buffer.finish().values(), &[0b0011_u8]); |
322 | | } |
323 | | |
324 | | #[test] |
325 | | fn test_boolean_buffer_builder_unset_an_inner_bit() { |
326 | | let mut buffer = BooleanBufferBuilder::new(5); |
327 | | buffer.append(true); |
328 | | buffer.append(true); |
329 | | buffer.append(false); |
330 | | buffer.append(true); |
331 | | buffer.set_bit(1, false); |
332 | | assert_eq!(buffer.len(), 4); |
333 | | assert_eq!(buffer.finish().values(), &[0b1001_u8]); |
334 | | } |
335 | | |
336 | | #[test] |
337 | | fn test_boolean_buffer_builder_unset_several_bits() { |
338 | | let mut buffer = BooleanBufferBuilder::new(5); |
339 | | buffer.append(true); |
340 | | buffer.append(true); |
341 | | buffer.append(true); |
342 | | buffer.append(false); |
343 | | buffer.append(true); |
344 | | buffer.set_bit(1, false); |
345 | | buffer.set_bit(2, false); |
346 | | assert_eq!(buffer.len(), 5); |
347 | | assert_eq!(buffer.finish().values(), &[0b10001_u8]); |
348 | | } |
349 | | |
350 | | #[test] |
351 | | fn test_boolean_buffer_builder_unset_several_bits_bigger_than_one_byte() { |
352 | | let mut buffer = BooleanBufferBuilder::new(16); |
353 | | buffer.append_n(10, true); |
354 | | buffer.set_bit(0, false); |
355 | | buffer.set_bit(3, false); |
356 | | buffer.set_bit(9, false); |
357 | | assert_eq!(buffer.len(), 10); |
358 | | assert_eq!(buffer.finish().values(), &[0b11110110_u8, 0b01_u8]); |
359 | | } |
360 | | |
361 | | #[test] |
362 | | fn test_boolean_buffer_builder_flip_several_bits_bigger_than_one_byte() { |
363 | | let mut buffer = BooleanBufferBuilder::new(16); |
364 | | buffer.append_n(5, true); |
365 | | buffer.append_n(5, false); |
366 | | buffer.append_n(5, true); |
367 | | buffer.set_bit(0, false); |
368 | | buffer.set_bit(3, false); |
369 | | buffer.set_bit(9, false); |
370 | | buffer.set_bit(6, true); |
371 | | buffer.set_bit(14, true); |
372 | | buffer.set_bit(13, false); |
373 | | assert_eq!(buffer.len(), 15); |
374 | | assert_eq!(buffer.finish().values(), &[0b01010110_u8, 0b1011100_u8]); |
375 | | } |
376 | | |
377 | | #[test] |
378 | | fn test_bool_buffer_builder_get_first_bit() { |
379 | | let mut buffer = BooleanBufferBuilder::new(16); |
380 | | buffer.append_n(8, true); |
381 | | buffer.append_n(8, false); |
382 | | assert!(buffer.get_bit(0)); |
383 | | } |
384 | | |
385 | | #[test] |
386 | | fn test_bool_buffer_builder_get_first_bit_not_requires_mutability() { |
387 | | let buffer = { |
388 | | let mut buffer = BooleanBufferBuilder::new(16); |
389 | | buffer.append_n(8, true); |
390 | | buffer |
391 | | }; |
392 | | |
393 | | assert!(buffer.get_bit(0)); |
394 | | } |
395 | | |
396 | | #[test] |
397 | | fn test_bool_buffer_builder_get_last_bit() { |
398 | | let mut buffer = BooleanBufferBuilder::new(16); |
399 | | buffer.append_n(8, true); |
400 | | buffer.append_n(8, false); |
401 | | assert!(!buffer.get_bit(15)); |
402 | | } |
403 | | |
404 | | #[test] |
405 | | fn test_bool_buffer_builder_get_an_inner_bit() { |
406 | | let mut buffer = BooleanBufferBuilder::new(16); |
407 | | buffer.append_n(4, false); |
408 | | buffer.append_n(8, true); |
409 | | buffer.append_n(4, false); |
410 | | assert!(buffer.get_bit(11)); |
411 | | } |
412 | | |
413 | | #[test] |
414 | | fn test_bool_buffer_fuzz() { |
415 | | use rand::prelude::*; |
416 | | |
417 | | let mut buffer = BooleanBufferBuilder::new(12); |
418 | | let mut all_bools = vec![]; |
419 | | let mut rng = rand::rng(); |
420 | | |
421 | | let src_len = 32; |
422 | | let (src, compacted_src) = { |
423 | | let src: Vec<_> = std::iter::from_fn(|| Some(rng.next_u32() & 1 == 0)) |
424 | | .take(src_len) |
425 | | .collect(); |
426 | | |
427 | | let mut compacted_src = BooleanBufferBuilder::new(src_len); |
428 | | compacted_src.append_slice(&src); |
429 | | (src, compacted_src.finish()) |
430 | | }; |
431 | | |
432 | | for _ in 0..100 { |
433 | | let a = rng.next_u32() as usize % src_len; |
434 | | let b = rng.next_u32() as usize % src_len; |
435 | | |
436 | | let start = a.min(b); |
437 | | let end = a.max(b); |
438 | | |
439 | | buffer.append_packed_range(start..end, compacted_src.values()); |
440 | | all_bools.extend_from_slice(&src[start..end]); |
441 | | } |
442 | | |
443 | | let mut compacted = BooleanBufferBuilder::new(all_bools.len()); |
444 | | compacted.append_slice(&all_bools); |
445 | | |
446 | | assert_eq!(buffer.finish(), compacted.finish()) |
447 | | } |
448 | | |
449 | | #[test] |
450 | | fn test_boolean_array_builder_resize() { |
451 | | let mut builder = BooleanBufferBuilder::new(20); |
452 | | builder.append_n(4, true); |
453 | | builder.append_n(7, false); |
454 | | builder.append_n(2, true); |
455 | | builder.resize(20); |
456 | | |
457 | | assert_eq!(builder.len(), 20); |
458 | | assert_eq!(builder.as_slice(), &[0b00001111, 0b00011000, 0b00000000]); |
459 | | |
460 | | builder.resize(5); |
461 | | assert_eq!(builder.len(), 5); |
462 | | assert_eq!(builder.as_slice(), &[0b00001111]); |
463 | | |
464 | | builder.append_n(4, true); |
465 | | assert_eq!(builder.len(), 9); |
466 | | assert_eq!(builder.as_slice(), &[0b11101111, 0b00000001]); |
467 | | } |
468 | | |
469 | | #[test] |
470 | | fn test_truncate() { |
471 | | let b = MutableBuffer::from_iter([true, true, true, true]); |
472 | | let mut builder = BooleanBufferBuilder::new_from_buffer(b, 2); |
473 | | builder.advance(2); |
474 | | let finished = builder.finish(); |
475 | | assert_eq!(finished.values(), &[0b00000011]); |
476 | | |
477 | | let mut builder = BooleanBufferBuilder::new(10); |
478 | | builder.append_n(5, true); |
479 | | builder.resize(3); |
480 | | builder.advance(2); |
481 | | let finished = builder.finish(); |
482 | | assert_eq!(finished.values(), &[0b00000111]); |
483 | | |
484 | | let mut builder = BooleanBufferBuilder::new(10); |
485 | | builder.append_n(16, true); |
486 | | assert_eq!(builder.as_slice(), &[0xFF, 0xFF]); |
487 | | builder.truncate(20); |
488 | | assert_eq!(builder.as_slice(), &[0xFF, 0xFF]); |
489 | | builder.truncate(14); |
490 | | assert_eq!(builder.as_slice(), &[0xFF, 0b00111111]); |
491 | | builder.append(false); |
492 | | builder.append(true); |
493 | | assert_eq!(builder.as_slice(), &[0xFF, 0b10111111]); |
494 | | builder.append_packed_range(0..3, &[0xFF]); |
495 | | assert_eq!(builder.as_slice(), &[0xFF, 0b10111111, 0b00000111]); |
496 | | builder.truncate(17); |
497 | | assert_eq!(builder.as_slice(), &[0xFF, 0b10111111, 0b00000001]); |
498 | | builder.append_packed_range(0..2, &[2]); |
499 | | assert_eq!(builder.as_slice(), &[0xFF, 0b10111111, 0b0000101]); |
500 | | builder.truncate(8); |
501 | | assert_eq!(builder.as_slice(), &[0xFF]); |
502 | | builder.resize(14); |
503 | | assert_eq!(builder.as_slice(), &[0xFF, 0x00]); |
504 | | builder.truncate(0); |
505 | | assert_eq!(builder.as_slice(), &[]); |
506 | | } |
507 | | |
508 | | #[test] |
509 | | fn test_boolean_builder_increases_buffer_len() { |
510 | | // 00000010 01001000 |
511 | | let buf = Buffer::from([72_u8, 2_u8]); |
512 | | let mut builder = BooleanBufferBuilder::new(8); |
513 | | |
514 | | for i in 0..16 { |
515 | | if i == 3 || i == 6 || i == 9 { |
516 | | builder.append(true); |
517 | | } else { |
518 | | builder.append(false); |
519 | | } |
520 | | } |
521 | | let buf2 = builder.finish(); |
522 | | |
523 | | assert_eq!(buf.len(), buf2.inner().len()); |
524 | | assert_eq!(buf.as_slice(), buf2.values()); |
525 | | } |
526 | | } |