At one point it seems we've changed how a regular expression is formed from the user input in the Tokenizer processor, but we didn't update the validation. This means that, in some edge cases, a valid input could lead to a form error or some invalid expression would be accepted.
| Comment | File | Size | Author |
|---|---|---|---|
| #6 | 2871497-6--tokenizer_regex_validation.patch | 1.66 KB | drunken monkey |
Comments
Comment #2
drunken monkeyThis should fix it.
Comment #4
drunken monkeyOh, seems we actually have a test for the wrong behavior!
Comment #6
drunken monkeyHuh, turns out PREG is amazingly fault-tolerant when it comes to character classes. Tried a lot of nonsense, without any errors. But this one should work.
Comment #7
borisson_Comment #9
drunken monkeyThanks for reviewing!
Committed.