david-mollitor-db opened a new pull request, #58815:
URL: https://github.com/apache/spark/pull/58815
### What changes were proposed in this pull request?
`RegExpExtractBase.extract`, `RegExpExtractBase.extractAll`, and
`RegExpInStr` call
`Matcher.toMatchResult()` on every match, then read
`groupCount`/`group(idx)` (the extract
helpers) or `start()` (`regexp_instr`) from the returned snapshot:
```scala
if (matcher.find()) {
val mr = matcher.toMatchResult
checkGroupIndex(prettyName, mr.groupCount, idx)
val group = mr.group(idx)
...
}
```
`Matcher.toMatchResult()` allocates an immutable `MatchResult` snapshot of
the current match
state, including a copy of the groups `int[]`. This PR reads the same values
directly from the
`Matcher`:
```scala
if (matcher.find()) {
checkGroupIndex(prettyName, matcher.groupCount, idx)
val group = matcher.group(idx)
...
}
```
and `m.toMatchResult().start()` -> `m.start()` in `RegExpInStr` (both the
interpreted and the
codegen paths).
### Why are the changes needed?
The values are read immediately after `find()`, with nothing else mutating
the matcher in
between, so `Matcher.groupCount()`, `Matcher.group(int)`, and
`Matcher.start()` return exactly
the same values as the corresponding `MatchResult` methods for the current
match. Snapshotting
into a `MatchResult` first only adds one allocation (the snapshot plus a
copy of the groups
`int[]`) per match on the `regexp_extract`, `regexp_extract_all`, and
`regexp_instr` hot paths.
Profiling a `regexp_extract` workload under JFR (`settings=profile`) shows
`java.util.regex.Matcher.toMatchResult` accounting for ~5% of sampled
allocation pressure;
this change removes it. Wall-clock time is unchanged within run-to-run noise
(this is an
allocation / GC-pressure reduction, not a CPU one).
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
Existing tests, which cover `regexp_extract`, `regexp_extract_all`, and
`regexp_instr`:
```
build/sbt 'catalyst/testOnly *RegexpExpressionsSuite'
```
All 24 tests pass.
### Was this patch authored or co-authored using generative AI tooling?
Generated-by: Isaac
This pull request and its description were written by Isaac.
--
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]
