Extractor now has support for mentions and URLs with indices. Conformance test suite updated to the latest version.
There seemed to be a problem with PHP 5.1.x (?) where the generated URL in the hastag was having another link added to it. This could be due to anything (regexs, unicode, ...) as I didn't really diagnose the cause, just found a solution.
Includes many fixes to make the library compliant with the latest test suite including: - No linking of usernames containing latin accented characters. - Adding links to protocol-less URLs - Escaping special characters in URLs Other changes: - Some fixes to unicode handing in regexs.
(I know that this looks weird with 24 and 26, but this makes it pass the conformance tests...)
Although empty array would make more sense, the conformance tests expect an empty string.
Still need to refactor regular expression chunks to reduce duplication.
Will work in the browser and on the command line. Need to convert to PHPUnit tests.