Ruby bindings to RE2, a "fast, safe, thread-friendly alternative to backtracking regular expression engines like those used in PCRE, Perl, and Python".
As re2 will report the number of capturing groups for an invalid regular expression as -1, we need to check whether a pattern is OK or not before using the number to initialize any vectors, etc.
This was causing a crash when using RE2::Scanner#scan with an invalid regular expression as we would attempt to initialize vectors with a length of -1. Instead, set the number of capturing groups to 0 and rely on re2's FindAndConsumeN returning false so we return nil to the user to indicate no match was made.
Following that, ensure we always check a pattern is OK before using its number of capturing groups as a positive integer.
This also fixes an edge case in RE2::Regexp#match when a user passes a negative number for the number of intended matches. Previously, this would throw a misleading NoMemoryError, now it throws a more informative ArgumentError.
Take this opportunity to document the behaviour of the library when given invalid regular expressions in the specs.
As re2 will report the number of capturing groups for an invalid regular expression as -1, we need to check whether a pattern is OK or not before using the number to initialize any vectors, etc.
This was causing a crash when using
RE2::Scanner#scanwith an invalid regular expression as we would attempt to initialize vectors with a length of -1. Instead, set the number of capturing groups to 0 and rely on re2'sFindAndConsumeNreturning false so we return nil to the user to indicate no match was made.Following that, ensure we always check a pattern is OK before using its number of capturing groups as a positive integer.
This also fixes an edge case in
RE2::Regexp#matchwhen a user passes a negative number for the number of intended matches. Previously, this would throw a misleadingNoMemoryError, now it throws a more informativeArgumentError.Take this opportunity to document the behaviour of the library when given invalid regular expressions in the specs.