Program terminates with uncaught exception if an invalid regex is used with RE2::Scanner#scan

[serch]

If we compile an invalid regex and try to use it the program terminates:

irb(main):001:0> RE2::Regexp.new('???').scan("test").to_a
/tmp/re2-20210214-21445-1a81zn5/re2-2021-02-02/re2/re2.cc:205: Error parsing '???': no argument for repetition operator: ??
libc++abi.dylib: terminating with uncaught exception of type std::length_error: vector

Ruby would instead raise:

irb(main):001:0> Regexp.new('???')
Traceback (most recent call last):
        3: from (irb):1
        2: from (irb):1:in `new'
        1: from (irb):1:in `initialize'
RegexpError (target of repeat operator is not specified: /???/)

I ran into this problem when I tried to compile some ruby-style regexes with RE2 and one had a negative look-behind: RE2::Regexp.new('(?<!foo)test').scan("test").to_a

[mudge]

Hi @serch,

Thanks for reporting this, it looks like it is a problem with RE2::Scanner specifically when using an invalid re2 pattern, e.g. using the RE2::Regexp instance alone does not crash:

> pattern = RE2::Regexp.new('???')
re2/re2.cc:205: Error parsing '???': no argument for repetition operator: ??
=> #<RE2::Regexp /???/>
> pattern.ok?
=> false
> pattern.error
=> "no argument for repetition operator: ??"
> pattern.error_arg
=> "??"
> pattern.match('foo')
re2/re2.cc:624: Invalid RE2: no argument for repetition operator: ??
=> nil
> pattern =~ 'foo'
re2/re2.cc:624: Invalid RE2: no argument for repetition operator: ??
=> false

But as soon as we try to use RE2::Scanner#scan:

> scanner = pattern.scan('foo')
=> #<RE2::Scanner:0x00007fcebf175328>
> scanner.eof?
=> false
> scanner.rewind
=> #<RE2::Scanner:0x00007fcebf175328>
> scanner.scan
libc++abi.dylib: terminating with uncaught exception of type std::length_error: vector
zsh: abort      pry -Ilib

Let me investigate this further and I'll keep you apprised.

[mudge]

Hi @serch,

This should be fixed in https://github.com/mudge/re2/releases/tag/v1.4.0

Could you please try upgrading and let me know if that fixes your issue?

[serch]

It does, thanks for the prompt reply and for the fix!