Improve SQL validation logic and add related test

[zaleslaw]

Enhanced containsForbiddenPatterns to use regex for stricter checks on SQL keywords and patterns, disallowing forbidden elements like comments and multiple statements. Updated isValidSqlQuery documentation for clarity. Added a test to validate reading from a table with column names containing reserved SQL keywords.

Fixes #1307

[Copilot]

Pull Request Overview

The PR strengthens SQL validation by switching to regex-based forbidden-pattern checks, clarifies the isValidSqlQuery documentation, and adds a test for reading from tables with columns named using reserved keywords.

• Use regex patterns in containsForbiddenPatterns for stricter detection.
• Minor documentation updates in isValidSqlQuery.
• New H2 test for tables whose column names include reserved SQL keywords.

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
dataframe-jdbc/src/test/kotlin/org/jetbrains/kotlinx/dataframe/io/h2/h2Test.kt	Added a test verifying SELECT works when column names include reserved keywords
dataframe-jdbc/src/main/kotlin/org/jetbrains/kotlinx/dataframe/io/readJdbc.kt	Switched forbiddenPatterns to regex, improved error logging, and updated Javadoc examples

Comments suppressed due to low confidence (2)

dataframe-jdbc/src/test/kotlin/org/jetbrains/kotlinx/dataframe/io/h2/h2Test.kt:796

• [nitpick] Consider simplifying the test function name (e.g., read from table with column name containing reserved SQL keyword) for clarity and conciseness.

    fun `read from table with column name containing the reserved SQL keywords`() {

dataframe-jdbc/src/test/kotlin/org/jetbrains/kotlinx/dataframe/io/h2/h2Test.kt:814

• Add tests for the newly documented forbidden patterns (e.g., queries containing --, /* */, or multiple statements separated by ;) to verify that containsForbiddenPatterns correctly rejects them.

    }

[Copilot]

[nitpick] For the multi-line SQL string, consider appending trimIndent() (or trimMargin()) to remove leading whitespace and ensure the query is formatted as intended.

[Copilot]

Consider precompiling the forbiddenPatterns into Regex instances once (e.g., at initialization) instead of compiling a new Regex on every call to improve performance.