Some Occurance operators seem to error

[BannerBomb]

So if I use a regex like this

new RE2(/((g{2,32}|q){1,32})/iu)

It gives me the error SyntaxError: invalid repetition size: {1,32}

While regular regex allows it to work.

I'm not sure if this is a bug or if it was intended. Just found it weird. And yes, I know a regex like this isn't really good practice as the {2,32} would be irrelevant.

[uhop]

It looks like RE2's code has a hardcoded limit for a maximum repeat count:

• https://github.com/google/re2/blob/2d866a3d0753f4f4fce93cccc6c59c4b052d7db4/re2/parse.cc#L47

Currently, it is set to 1000. It is used for some internal needs. See comments:

• https://github.com/google/re2/blob/2d866a3d0753f4f4fce93cccc6c59c4b052d7db4/re2/regexp.h#L451
• https://github.com/google/re2/blob/2d866a3d0753f4f4fce93cccc6c59c4b052d7db4/re2/fuzzing/re2_fuzzer.cc#L136

While there is a procedure to change the limit, it is unclear how it affects RE2. The available comments suggest that it should be decreased, not increased.

In your example, we have 32 * 32 = 1024, which is greater than 1000.

I don't see how we can handle it on our side. If this restriction is important, don't hesitate to contact google/re2.

[uhop]

I probably should add that it looks like this restriction is used only by a parser and it checks bound repetition operators and ignores unbound ones: ?, +, *, {5,}, and similar ones. It will check both minimal and maximal repetition values if possible.

The parser calculates values recursively so if there are embedded repetitions (like in your example) they will be properly multiplied.