<regex>: Use std::search() in skip heuristic

[muellerj2]

Towards #5468. This replaces the weird usage of _Cmp_chrange() by a straightforward call to std::search() in _Matcher2::_Skip().

I fix a copy-paste mistake in the comment describing _Cmp_icase_translateleft as well.

I also made an attempt to replace _Cmp_chrange()'s implementation by a straightforward call to std::mismatch(), but that seems to be a pessimization in practice of about 10 % (probably because the strings tend to be quite short).

There will still be one follow-up to make an obvious improvement to the skip heuristic for regex and wregex in collate mode. But otherwise, I think this is basically it for simple improvements to the skip heuristic. There are still a few opportunities that could lead to some improvement -- handling several branches, avoiding to walk the NFA for each _Skip() call, or avoiding to compare the NFA nodes matched by _Skip() in _Match_pat() again -- but they are not straightforward to implement.

Benchmark

benchmark	before	after	speedup
bm_lorem_search/"^bibe"/2	28.2506	28.8783	0.98
bm_lorem_search/"^bibe"/3	27.6228	29.82	0.93
bm_lorem_search/"^bibe"/4	28.8783	32.4707	0.89
bm_lorem_search/"bibe"/2	43492.7	2622.77	16.58
bm_lorem_search/"bibe"/3	90680.8	5000	18.14
bm_lorem_search/"bibe"/4	172631	9626.07	17.93
bm_lorem_search/"(bibe)"/2	47538.5	4296.88	11.06
bm_lorem_search/"(bibe)"/3	92071.8	8370.5	11.00
bm_lorem_search/"(bibe)"/4	181370	15485.6	11.71
bm_lorem_search/"(bibe)+"/2	64062.5	10253.9	6.25
bm_lorem_search/"(bibe)+"/3	153460	20856.3	7.36
bm_lorem_search/"(bibe)+"/4	249062	40108.8	6.21
bm_lorem_search/"(?:bibe)+"/2	49178	4603.8	10.68
bm_lorem_search/"(?:bibe)+"/3	94164.3	8998.29	10.46
bm_lorem_search/"(?:bibe)+"/4	188354	17578.3	10.72
bm_lorem_search/R"(\bbibe)"/2	96256.9	89979.2	1.07
bm_lorem_search/R"(\bbibe)"/3	194972	188354	1.04
bm_lorem_search/R"(\bbibe)"/4	374930	368369	1.02
bm_lorem_search/R"(\Bibe)"/2	235395	222178	1.06
bm_lorem_search/R"(\Bibe)"/3	404531	461498	0.88
bm_lorem_search/R"(\Bibe)"/4	941265	983099	0.96
bm_lorem_search/R"((?=....)bibe)"/2	48131.7	3138.95	15.33
bm_lorem_search/R"((?=....)bibe)"/3	96256.9	6277.9	15.33
bm_lorem_search/R"((?=....)bibe)"/4	179983	12207	14.74
bm_lorem_search/R"((?=bibe)....)"/2	44327.8	2915.74	15.20
bm_lorem_search/R"((?=bibe)....)"/3	87886.7	5580.36	15.75
bm_lorem_search/R"((?=bibe)....)"/4	179983	10986.3	16.38
bm_lorem_search/R"((?!lorem)bibe)"/2	45515.6	2999.44	15.17
bm_lorem_search/R"((?!lorem)bibe)"/3	92071.8	5859.38	15.71
bm_lorem_search/R"((?!lorem)bibe)"/4	188354	11160.7	16.88

Note that this means that all improvements since #5509 have sped up searching for the regex (bibe)+ by a factor of about 450 and for (?:bibe)+ by a factor of about 1000 in this benchmark.

[StephanTLavavej]

Thanks, looks great! 😻

[StephanTLavavej]

I'm mirroring this to the MSVC-internal repo - please notify me if any further changes are pushed.

[StephanTLavavej]

Thanks again for figuring out how to make this faster! 🚗 🦖 🎉