[
https://issues.apache.org/jira/browse/LANG-1830?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Maksym Korshun updated LANG-1830:
---------------------------------
Description:
ClassUtils.toCleanName() currently validates the array suffix using
String.matches(),
which recompiles the regular expression on every call.
Replace the repeated call to String.matches("(?:
[])+")
with a private static final Pattern and reuse it for validation.
This reduces unnecessary regex compilation in a frequently used utility method
without changing behavior.
A simple JMH benchmark comparing the current implementation with a precompiled
{{Pattern}} showed approximately 4x-9x higher throughput, depending on the
input.
Benchmark (tail) Mode Cnt Score
Error Units
ClassUtilsRegexBenchmark.precompiledPattern [] thrpt 2 32800762.230
ops/s
ClassUtilsRegexBenchmark.precompiledPattern [][] thrpt 2 25525841.466
ops/s
ClassUtilsRegexBenchmark.precompiledPattern [][][][] thrpt 2 20276144.578
ops/s
ClassUtilsRegexBenchmark.precompiledPattern [ thrpt 2 59534277.654
ops/s
ClassUtilsRegexBenchmark.precompiledPattern ][ thrpt 2 63698539.468
ops/s
ClassUtilsRegexBenchmark.precompiledPattern [][]x thrpt 2 20136295.686
ops/s
ClassUtilsRegexBenchmark.stringMatches [] thrpt 2 6529300.056
ops/s
ClassUtilsRegexBenchmark.stringMatches [][] thrpt 2 5999979.754
ops/s
ClassUtilsRegexBenchmark.stringMatches [][][][] thrpt 2 5506269.938
ops/s
ClassUtilsRegexBenchmark.stringMatches [ thrpt 2 7142911.032
ops/s
ClassUtilsRegexBenchmark.stringMatches ][ thrpt 2 7220328.509
ops/s
ClassUtilsRegexBenchmark.stringMatches [][]x thrpt 2 5474452.510
ops/s
||Benchmark||Score (ops/s)||
|ClassUtilsRegexBenchmark.precompiledPattern|32800762.230|
|ClassUtilsRegexBenchmark.precompiledPattern|25525841.466|
|ClassUtilsRegexBenchmark.precompiledPattern|59534277.654|
|ClassUtilsRegexBenchmark.stringMatches|6529300.056|
|ClassUtilsRegexBenchmark.stringMatches|5999979.754|
|ClassUtilsRegexBenchmark.stringMatches|7142911.032|
was:
ClassUtils.toCleanName() currently validates the array suffix using
String.matches(),
which recompiles the regular expression on every call.
Replace the repeated call to String.matches("(?:\\[])+")
with a private static final Pattern and reuse it for validation.
This reduces unnecessary regex compilation in a frequently used utility method
without changing behavior.
> Avoid repeated regex compilation in ClassUtils.toCleanName()
> ------------------------------------------------------------
>
> Key: LANG-1830
> URL: https://issues.apache.org/jira/browse/LANG-1830
> Project: Commons Lang
> Issue Type: Improvement
> Components: lang.*
> Reporter: Maksym Korshun
> Priority: Minor
>
> ClassUtils.toCleanName() currently validates the array suffix using
> String.matches(),
> which recompiles the regular expression on every call.
> Replace the repeated call to String.matches("(?:
> [])+")
> with a private static final Pattern and reuse it for validation.
> This reduces unnecessary regex compilation in a frequently used utility
> method without changing behavior.
>
> A simple JMH benchmark comparing the current implementation with a
> precompiled {{Pattern}} showed approximately 4x-9x higher throughput,
> depending on the input.
> Benchmark (tail) Mode Cnt
> Score Error Units
> ClassUtilsRegexBenchmark.precompiledPattern [] thrpt 2
> 32800762.230 ops/s
> ClassUtilsRegexBenchmark.precompiledPattern [][] thrpt 2
> 25525841.466 ops/s
> ClassUtilsRegexBenchmark.precompiledPattern [][][][] thrpt 2
> 20276144.578 ops/s
> ClassUtilsRegexBenchmark.precompiledPattern [ thrpt 2
> 59534277.654 ops/s
> ClassUtilsRegexBenchmark.precompiledPattern ][ thrpt 2
> 63698539.468 ops/s
> ClassUtilsRegexBenchmark.precompiledPattern [][]x thrpt 2
> 20136295.686 ops/s
> ClassUtilsRegexBenchmark.stringMatches [] thrpt 2
> 6529300.056 ops/s
> ClassUtilsRegexBenchmark.stringMatches [][] thrpt 2
> 5999979.754 ops/s
> ClassUtilsRegexBenchmark.stringMatches [][][][] thrpt 2
> 5506269.938 ops/s
> ClassUtilsRegexBenchmark.stringMatches [ thrpt 2
> 7142911.032 ops/s
> ClassUtilsRegexBenchmark.stringMatches ][ thrpt 2
> 7220328.509 ops/s
> ClassUtilsRegexBenchmark.stringMatches [][]x thrpt 2
> 5474452.510 ops/s
> ||Benchmark||Score (ops/s)||
> |ClassUtilsRegexBenchmark.precompiledPattern|32800762.230|
> |ClassUtilsRegexBenchmark.precompiledPattern|25525841.466|
> |ClassUtilsRegexBenchmark.precompiledPattern|59534277.654|
> |ClassUtilsRegexBenchmark.stringMatches|6529300.056|
> |ClassUtilsRegexBenchmark.stringMatches|5999979.754|
> |ClassUtilsRegexBenchmark.stringMatches|7142911.032|
>
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
