Can't ENCLOSE_OPTION be allowed in look-behind?

[knu]

Currently Onigmo (or Oniguruma) does not allow option enclosures (like (?i)) in look-behind, but isn't it possible to allow them?

Taking Ruby as example, this restriction prevents a pattern from being interpolated within a look-behind construct of another.

pattern1 = /foo/  # stringified as "(?-mix:foo)"
pattern2 = /(?<=#{pattern1})/
#=> RegexpError: invalid pattern in look-behind: /(?<=(?-mix:foo))/

This pull request is an unconfirmed attempt to fix the problem.
Thanks for your consideration.

[knu]

On second thought, (?i) may lead a subpattern to variable-length in Unicode case fold semantics. Is that right?
Is there any other option that affects a pattern length? We need to case out those if that is the case.

[knu]

On second thought, (?i) may lead a subpattern to variable-length in Unicode case fold semantics. Is that right?
Is there any other option that affects a pattern length? We need to case out those if that is the case.

[knu]

It doesn't seem like it. In Onigmo/Oniguruma, look-behind does not take Unicode case-folding into account in the first place.

Below is taken from an irb session on ruby 2.0.

>> /(?:\u{149})a/ui =~ "[\u{2bc}na]"  # Unicode case-folding match is in effect.
=> 1
>> /(?<=\u{149})a/ui =~ "[\u{2bc}na]"  # But look-behind does not take account of it.
=> nil

This looks like a bug.

[knu]

It doesn't seem like it. In Onigmo/Oniguruma, look-behind does not take Unicode case-folding into account in the first place.

Below is taken from an irb session on ruby 2.0.

>> /(?:\u{149})a/ui =~ "[\u{2bc}na]"  # Unicode case-folding match is in effect.
=> 1
>> /(?<=\u{149})a/ui =~ "[\u{2bc}na]"  # But look-behind does not take account of it.
=> nil

This looks like a bug.

[k-takata]

Thank you for the pull request. I also added tests for this.

[k-takata]

Thank you for the pull request. I also added tests for this.