mruby-regexp: add Regexp#names and MatchData#names - #7027

No actionable comments were generated in the recent review. 🎉

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

_{Comment @coderabbitai help to get the list of available commands.}

This needs a rebase: #7026 landed first as cfc496a.

I merged that one first only because it changes behaviour that already existed, where this adds methods that did not. Both are wanted, and I verified them together before merging anything.

The conflict is not in the implementation. Both changes add a line to the same two places:

• mrbgems/mruby-regexp/README.md, the Ruby API block: re.named_captures gained a value shape, and this adds re.names next to it
• mrbgems/mruby-regexp/test/regexp.rb, adjacent assertions

Keeping both sides is the whole resolution. Applied that way on top of #7022 and #7026, the suite passes (2133 OK, 0 KO, 78 bintests) with no sanitizer report, and the eight rows I compared against CRuby all agree:

/(?<a>x)/.named_captures            {"a" => [1]}
/(?<a>x)(?<b>y)/.named_captures     {"a" => [1], "b" => [2]}
re.named_captures["a"] = 99 ; re.named_captures   {"a" => [1]}
/(x)/.named_captures                {}
/(?<a>x)(?<b>y)/.names              ["a", "b"]
/(x)/.names                         []
/(?<x>a)/.match("a").names          ["x"]
/(?<x>a)/.match("a").named_captures {"x" => "a"}

I checked your claim about key order rather than taking it on trust: regexp_init() inserts into @named_captures in the order the compiler registered the groups, which is ascending group number, so names comes out in pattern order without sorting.

One note for whoever picks up duplicate names later, since both of you flagged it and neither change touches it: /(?<a>x)|(?<a>b)/ still answers {"a" => [2]} and match("b")[:a] still answers nil. names reporting the name once is right either way, so nothing here has to change when that is fixed.

Rebased onto cfc496ace. Resolved exactly as you described, keeping both sides in each place:

• mrbgems/mruby-regexp/README.md: re.names now sits right after re.named_captures, matching the md.named_captures / md.names order in the MatchData block below.
• mrbgems/mruby-regexp/test/regexp.rb: assert("Regexp#named_captures") and assert("Regexp#names") are two separate blocks.

mrbgems/mruby-regexp/mrblib/regexp.rb merged cleanly, so the implementation is unchanged from what you reviewed. The tree is the same 3 files and 26 added lines as before.

rake test passes here (1940 OK, 0 KO, 105 bintests), and all eight rows you compared against CRuby reproduce on the rebased branch.

Thanks for checking the key order claim instead of taking it on trust. Agreed on duplicate names: names reporting the name once holds either way, so that fix can land independently of this.