URL Path is Decoded Twice, leading to errors or worse if the escape character % is present in the decoded string

[dbkang]

$location.path() in AngularJS already returns a decoded string (see how $$path is set), but UrlMatcher.prototype.exec, which takes path from $location.path() calls decodeURIComponent again on decoded string, which results in an error or possibly an incorrect string if the decoded string itself contains the escape character.

Also, I don't know if this is compensating for the other bug, but in UrlMatcher.prototype.format, encodeURIComponent is missing for the path part of the URL - should be result += encodeURIComponent(value).

[timkindberg]

@dkang81, would you be interested in submitting a pull request with passing tests? If you had time to fix the issues, we could def use the help! Otherwise we can look into it when we get a chance.

[nateabele]

@dkang81 Or, if you could just submit a patch with a failing test that demonstrates the issue, I can patch it. Thanks.

[dbkang]

@timkindberg @nateabele wow, thanks for quick responses guys, I do have unit tests and a fix for this now, just need to ask my boss for permission, should be able to submit a pull request sometime tomorrow!

[nateabele]

Awesome, thanks man.

[ksperling]

I think the fix is good as it stands now, however something to think about is if urlMatcher shouldn't really be looking at non-decoded URLs (and we should have some way to get that from $location).

If I have a state 'test' with a URL pattern like '/foo/:whatever/bar', it should round-trip correctly even if I transitionTo('test', { whatever: 'evil/slash' }).

Or am I missing something?