Bug #20097: Regexp#match? with empty capture group repeat is inconsistent

• Status: Open
• Assignee: -
• Target version: -
• ruby -v: ruby 3.3.0 (2023-12-25 revision 5124f9ac75) +MN [arm64-darwin22]
• Backport: 3.0: UNKNOWN, 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN

[tompng (tomoya ishida)]

It sometimes matches and sometimes does not, depend on repeat quantifier number.

# empty capture group repeat
p /(a(){1,4}){2}/.match?('aa') #=> true
p /(a(){1,5}){2}/.match?('aa') #=> false

# repeat of capture group that match to empty string
p /(a(|){1,2}){2}/.match?('aa') #=> true
p /(a(|){1,3}){2}/.match?('aa') #=> false
p /(a(x?y?|z){1,2}){2}/.match?('aa') #=> false

non-capturing group seems ok.

p /(a(?:){1,4}){2}/.match?('aa') #=> true
p /(a(?:){1,5}){2}/.match?('aa') #=> true
p /(a(?:|){1,2}){2}/.match?('aa') #=> true
p /(a(?:|){1,3}){2}/.match?('aa') #=> true
p /(a(?:x?y?|z){1,2}){2}/.match?('aa') #=> true

Same behavior in ruby 3.0, 3.1, 3.2

[dentarg (Patrik Ragnarsson)]

Same as https://bugs.ruby-lang.org/issues/20083?

[tompng (tomoya ishida)]

Not same. https://bugs.ruby-lang.org/issues/20083 is related to regexp optimization from ruby 3.2.
This issue is not because it's same result in older ruby version 3.0 and 3.1