gengliangwang opened a new pull request, #56223:
URL: https://github.com/apache/spark/pull/56223
### What changes were proposed in this pull request?
Add `ExpressionImplUtils.compileRegexPattern(String regex, int flags, String
funcName)`, which wraps `Pattern.compile` and maps a `PatternSyntaxException`
to the user-facing INVALID_PARAMETER_VALUE.PATTERN error. Route both the shared
codegen (`RegExpUtils.initLastMatcherCode`, used by the whole regexp expression
family -- `RLike`, `RegExpReplace`, `RegExpExtract`, `RegExpExtractAll`,
`RegExpInStr`, etc.) and the eval helper (`RegExpUtils.getPatternAndLastRegex`)
through it.
`initLastMatcherCode` previously emitted a 5-line inline `try {
Pattern.compile(...) } catch (PatternSyntaxException)` block; it now emits a
single helper call. The per-stage mutable-state caching (`lastRegex` /
`pattern`) is preserved in the generated code.
### Why are the changes needed?
Part of SPARK-56908 (umbrella). This block is emitted by every regexp
expression in every stage that uses one; collapsing it to a single call shrinks
the generated Java across the whole family, helping with the JVM 64KB method /
constant-pool limits, Janino compile time, and JIT work.
### Does this PR introduce _any_ user-facing change?
No. The compiled behavior is identical; only the emitted Java source text
changes.
### How was this patch tested?
```
build/sbt "catalyst/testOnly *RegexpExpressionsSuite"
```
21/21 pass (exercised both with and without whole-stage codegen).
### Was this patch authored or co-authored using generative AI tooling?
Generated-by: Claude Code (Opus 4.8)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
