Incorrect docstring handling #55

Activity

1. erbridge commented on Mar 18, 2015

   

   erbridge

   on Mar 18, 2015

   More actions

   See also e.g. https://github.com/cobrateam/splinter/blob/master/splinter/browser.py#L46. Everything between the """s should be quoted.
2. winstliu

   added

   bug

   on Mar 18, 2015
3. alexjurkiewicz commented on Apr 13, 2015

   

   alexjurkiewicz

   on Apr 13, 2015

   More actions

   Seems to be that the issue relates to the following corner case:

   • raw string prefix
   • triple-quoted string
   • containing backslashes
   • related to brackets in the string

   Here are three instances of the bug that result in slightly different syntax:

   Break 1:

   before()
   r'''(\)'''
   after()

   Break 2:

   before()
   r'''(\'''
   after()

   Break 3:

   before()
   r'''[\'''
   after()

   The last example shows different highlighting for after() -- I think in the original case it's not that the docstring fails to end, but that the active "highlighting mode" is affected by characters in the string somehow.

   Some working examples:

   Works 1:

   before()
   r'(\)'
   after()

   Works 2:

   before()
   r'''\''''
   after()

   Works 3:

   before()
   r'''\]'''
   after()
4. auscompgeek commented on Sep 13, 2015

   

   auscompgeek

   on Sep 13, 2015

   More actions

   Looks like the grammar is possibly trying to greedily highlight regex and hence not bailing when the regex is invalid but the string is closed.

   This doesn't quite explain what's going on in the MySQLdb1 file linked in the OP though, though that might just be the regex matching gone wrong as well.
5. auscompgeek commented on Oct 4, 2015

   

   auscompgeek

   on Oct 4, 2015

   More actions

   @alexjurkiewicz actually, after testing your snippets in a Python REPL, your second snippet is semi-intended behaviour. Python considers r'''(\''' to be an unclosed string, although it seems to be stuck in regex mode after the open parenthesis:

   r'''(\'''
   '''
   hello()

   Similar issue here, except we're dealing with regex classes:

   r'''[\'''
   '''
   hello()

   This works as expected/intended:

   r'''(\'''
   )'''
   test()
   
   r'''[\'''
   ]'''
   test()
6. auscompgeek commented on Oct 4, 2015

   

   auscompgeek

   on Oct 4, 2015

   More actions

   Guess this is just #39 then?
7. alexjurkiewicz commented on Oct 6, 2015

   

   alexjurkiewicz

   on Oct 6, 2015

   More actions

   yes, confusing!
8. winstliu

   self-assigned this

   on Oct 6, 2015
9. winstliu

   removed their assignment

   on Aug 4, 2016