mirror/ruby
1
0
Fork 0
mirror of https://github.com/ruby/ruby.git synced 2026-01-26 12:14:51 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix regexp performance regression for patterns starting with s/k

Browse Source 
Commit 981ee02c7c ("Fix performance problem with /k/i and /s/i") was
merged for Ruby 4.0 to enable partial Boyer-Moore optimization for
patterns containing 's' or 'k' by using the prefix before those
characters.

However, when 's' or 'k' appears at the start of a pattern (no usable
prefix), set_bm_skip() returns 0 and the code returned early without
setting any optimization mode, leaving reg->optimize at
ONIG_OPTIMIZE_NONE. This caused up to 30x slowdown for patterns like
/slackware/i when matched against strings with non-ASCII characters.

This patch keeps the improvement from 981ee02c7c for patterns with
3+ char prefix, while fixing the regression by falling back to
ONIG_OPTIMIZE_EXACT_IC with the full pattern when the usable prefix
is less than 3 characters.

Before: /\bslackware\b/i with non-ASCII string: 2.24 us/op
After:  /\bslackware\b/i with non-ASCII string: 0.70 us/op (3.2x faster)

[Bug #21824]
This commit is contained in:
Chris Hasiński 2026-01-11 14:12:07 +01:00 committed by Nobuyoshi Nakada
parent 09cd13114a
commit 5de4cc5608
Notes: git 2026-01-13 01:20:10 +00:00 
Merged: https://github.com/ruby/ruby/pull/15848
1 changed files with 10 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
regcomp.c
View File

@ -5264,18 +5264,24 @@ set_optimize_exact_info(regex_t* reg, OptExactInfo* e)
if (e->ignore_case > 0) {
if (e->len >= 3 || (e->len >= 2 && allow_reverse)) {
int orig_len = e->len;
e->len = set_bm_skip(reg->exact, reg->exact_end, reg,
reg->map, 1);
reg->exact_end = reg->exact + e->len;
if (e->len >= 3) {
reg->exact_end = reg->exact + e->len;
reg->optimize = (allow_reverse != 0
? ONIG_OPTIMIZE_EXACT_BM_IC : ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC);
}
else if (e->len > 0) {
else {
/* Even if BM skip table can't be built (e.g., pattern starts with
's' or 'k' which have multi-byte case fold variants), we should
still use EXACT_IC optimization with the original pattern.
Without this fallback, patterns like /slackware/i have no
optimization at all, causing severe performance regression
especially with non-ASCII strings. See [Bug #21824] */
e->len = orig_len; /* Restore original length for EXACT_IC */
reg->optimize = ONIG_OPTIMIZE_EXACT_IC;
}
else
return 0;
}
else {
reg->optimize = ONIG_OPTIMIZE_EXACT_IC;