Mechanize parses <script> contents with bad results

[stoduk]

I've got a page with a bunch of javascript (which I believe Mechanize will make no effort to handle) - and helpfully there is something in that page that looks like the start of a tag. This ends badly! I get the ParseError("nested SELECTs") exception raised when a genuine select is found. So, two questions long term, is it right that Mechanize is parsing the contents of the <script> tag at all? Not handling javascript is fine, but trying to parse that javascript as if it was HTML won't work well. can anyone think of a short term fix to get me moving? Can I manipulate pages within Mechanize (in this case so I can delete the <script> elements)?

[stoduk]

I found an answer for (2) - see below. Epic hack, but works :)

No idea how to fix (1), won't waste time looking until someone confirms it is at least something that should be fixed. [For instance, I don't know if a script tag can ever have children - if so that would complicate things, as we'd only want to exclude the script contents and not its children]

_sgmllib_copy.py:parse_starttag()
         self.__starttag_text = rawdata[start_pos:j]
+        if not self.rawdata[i-1] == "'": # HACK - quick way to exclude fake tags hiding in javascript
+           self.finish_starttag(tag, attrs)
         return j