If you only care that the string doesn't end with .jpg
or .png
, you can use this:
^.+$(?<!\.jpg)(?<!\.png)
The ^.+
isn't strictly necessary, but depending on how the JSON parser is coded you might need to force the regex to consume the whole string. If you're using the regex for other validations as well, you might want something more elaborate, like:
^\w+(?:\.\w+)+$(?<!\.jpg)(?<!\.png)
You probably tried to use (?<!\.jpg|\.png)
, which wouldn't work because Python's regex flavor is one of the most restrictive when it comes to lookbehinds. PHP and Ruby 1.9+ would accept it because each of the alternatives has a fixed length. They don't even have to be the same length; (?<!\.jpg|\.jpeg|\.png)
would work, too. Just don't try to factor out the dot, as in (?<!\.(?:jpg|jpeg|png))
; the alternation has to be at the top level of the lookbehind.
Java would accept the factored-out version because it does a little more work at compile time to determine the maximum number of characters the lookbehind might need to match. The lookbehind expression needs to be fairly simple though, and it can't use the +
or *
quantifiers. Finally, the .NET and JGSoft flavors place no restrictions at all on lookbehinds. But Python makes a very simple-minded attempt to figure out the exact number of characters the lookbehind needs to match, generating that cryptic error message when it fails.