Error on save in buffer_post_write - missing endif #452

Activity

1. krader1961 commented on Jul 11, 2014

   

   krader1961

   on Jul 11, 2014

   Contributor

   More actions

   I can't reproduce that error. However, putting foo() = 1 in one of my modules does cause this failure which makes more sense:

   Run pyflakes {'builtins': '_'}
   <traceback object at 0x101bd5fc8>
   SyntaxError can't assign to function call (main.py, line 442)
   

   What is slightly perplexing is that other types of syntax errors I introduced in my test module are detected and reported without causing an uncaught exception. Based on several other recently reported bugs that resulted in the linters throwing exceptions it might be a good idea to wrap the check_path() call in lint.py in a try block to catch and report unexpected exceptions.
2. aseyboldt commented on Jul 25, 2014

   

   aseyboldt

   on Jul 25, 2014

   More actions

   I get the same error message on python code like this with version g2df674c

   foo(self._a=5)
   
3. kirkboyer-wf commented on Aug 25, 2014

   

   kirkboyer-wf

   on Aug 25, 2014

   More actions

   I got nearly exact same Traceback as bgr:

   E121: Undefined variable: null
   E116: Invalid arguments for function 12
   Traceback (most recent call last):
   File "", line 1, in
   File "/Users/.../.vim/bundle/python-mode/pymode/lint.py", line 71, in code_check
   env.run('g:PymodeLocList.current().extend', [e._info for e in errors])
   File "/Users/.../.vim/bundle/python-mode/pymode/environment.py", line 186, in run
   self.prepare_value(a) for a in args
   vim.error

   This happens when I make a particular kind of error in using google's app engine entity queries:
   If I have a model called MyModel with a property called value, in models.py, then writing a file containing the following causes this error:

   import models
   something = models.MyModel.query(models.MyModel.value=3)

   (This code, correctly written, would have the line
   something = models.MyModel.query(models.MyModel.value == 3)
   )
4. bsandrow commented on Oct 24, 2014

   

   bsandrow

   on Oct 24, 2014

   More actions

   I will toss my +1 in with @kirkboyer-wf . I'm using SQLAlchemy, and I wrote a query as:

   import models
   
   result = models.MyModel.query.filter(models.MyModel.id = 1)

   Rather than:

   import models
   
   result = models.MyModel.query.filter(models.MyModel.id == 1)

   It caused this issue. I pulled the latest version of the develop branch (I use Vundle), but the issue is still present.
5. bsandrow commented on Oct 24, 2014

   

   bsandrow

   on Oct 24, 2014

   More actions

   More info:

   It's blowing up on this line:

   env.run('g:PymodeLocList.current().extend', [e._info for e in errors])

   env.run() does some preparation on the e._infos that are passed through. The e._info that's blowing up in this case looks like:

   ERROR[0]._info = {'bufnr': 1,
    'col': None,
    'filename': 'sowingo/api/resources/office_inventory.py',
    'linter': 'pyflakes',
    'lnum': 290,
    'number': 'keyword',
    'rel': 'sowingo/api/resources/office_inventory.py',
    'text': "keyword can't be an expression [pyflakes]",
    'type': 'E'}

   The output of pyflakes looks like:

   sowingo/api/resources/office_inventory.py:290: keyword can't be an expression
   office_inventory_lists_to_items.office_inventory_list_id = list_id
   

   So the None column is there because PyFlakes gives no column. So...

   env.run() calls env.prepare_args() on each of the arguments (in this case each e._info instance). This basically just runs json.dumps() on the value, and does some unicode cleanup if we're in on Python 2.x (but this part isn't blowing up). The actual error is vim.error: Vim(call):E121: Undefined variable: null. Since we're basically turning the arguments of env.run() into a :call command:

       def run(self, name, *args):
           """ Run vim function. """
           vim.command('call %s(%s)' % (name, ", ".join([
               self.prepare_value(a) for a in args
           ])))

   I don't think that it likes the null column (since that's the only value that could end up as null that's being passed through).

   Patching it from:

       for e in errors:
           e._info['bufnr'] = env.curbuf.number

   to:

       for e in errors:
           e._info['bufnr'] = env.curbuf.number
           if e._info['col'] is None:
               e._info['col'] = 1

   seems to do the trick (at least for now).
6. bsandrow commented on Oct 24, 2014

   

   bsandrow

   on Oct 24, 2014

   More actions

   There is probably a better way to deal with this (as the error report now says that the column is "1" which isn't true), but for those that want a quick patch this should work. Here's the patch:

   diff --git a/pymode/lint.py b/pymode/lint.py
   index 5053f7f..57fda6f 100644
   --- a/pymode/lint.py
   +++ b/pymode/lint.py
   @@ -67,6 +67,16 @@ def code_check():
   
        for e in errors:
            e._info['bufnr'] = env.curbuf.number
   +        if e._info['col'] is None:
   +            e._info['col'] = 1
   +
   +    with open('/tmp/dump.txt', 'wb') as fh:
   +        import pprint
   +        for i, e in enumerate(errors):
   +            fh.write('ERROR[%d]._info = %s' % (
   +                i,
   +                pprint.pformat(e._info)
   +            ))
   
        env.run('g:PymodeLocList.current().extend', [e._info for e in errors])

   You can remove the error dump if you like, but it was useful for debugging.
7. stale commented on Feb 13, 2018

   

   staleBot

   on Feb 13, 2018

   More actions

   This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.
8. stale

   added

   inactive

   on Feb 13, 2018
9. stale commented on Mar 6, 2018

   

   staleBot

   on Mar 6, 2018

   More actions

   Closed due to inactivity.
10. stale

    closed this as completedon Mar 6, 2018