SourceJavaScriptTokenizer has two issues with keyword parsing: 1. Keywords are tested as "token in this._keywords", thus "__proto__" and all methods found on a generic objects (e.g. "toString" are considered to be keywords) 2. "true}" is parsed as a single "javascript-ident", since the yych = "}"; if (yych <= 0x7F)... check always fails (JavaScript does not define comparisons of strings vs. integers the way we need).
Created attachment 93467 [details] [PATCH] Suggested fix
Comment on attachment 93467 [details] [PATCH] Suggested fix Could you a test for these cases into the existing tokenizer tests? Otherwise looks good.
Created attachment 93628 [details] [PATCH] Test added
Comment on attachment 93628 [details] [PATCH] Test added Clearing flags on attachment: 93628 Committed r86554: <http://trac.webkit.org/changeset/86554>
All reviewed patches have been landed. Closing bug.