Add support for new-style string formatting #38

Activity

1. MattDMo commented on May 26, 2015

   

   MattDMo

   on May 26, 2015

   Owner

   More actions

   Can you please elaborate? What sort of support would you like to see?
2. MattDMo

   self-assigned this

   on May 26, 2015
3. MattDMo

   added

   enhancement

   discussion

   on May 26, 2015
4. sprt commented on May 26, 2015

   

   sprt

   on May 26, 2015

   Author

   More actions

   See https://docs.python.org/2/library/string.html#formatstrings

   For example, {} in '{}'.format(foo) isn't properly highlighted but '%s' % foo is.
5. MattDMo commented on May 26, 2015

   

   MattDMo

   on May 26, 2015

   Owner

   More actions

   OK, I'll see what I can do. Development is kind of at a low ebb at the moment, as I'm pretty busy with my real job, but I'll definitely put this on the list. If you have any contributions to make, feel free to submit a PR.
6. sprt commented on Jun 13, 2015

   

   sprt

   on Jun 13, 2015

   Author

   More actions

   Okay I came up with this regex:

   (?<![^\{]\{)\{(?:(?:(?:[A-Za-z_]\w*|(?:[1-9]\d*|0|0[Oo]?[0-7]+|0[Xx][0-7]+|0[Bb][01]+)))?(?:\.[A-Za-z_]\w*|\[(?:(?:[1-9]\d*|0|0[Oo]?[0-7]+|0[Xx][0-7]+|0[Bb][01]+)|[^\]\}}\{{]+)\])*)?(?:\![rsa])?(?:\:(?:(?:.?[<>=\^])?[\x20+-]?\#?0?(?:[1-9]\d*|0|0[Oo]?[0-7]+|0[Xx][0-7]+|0[Bb][01]+)?,?(?:\.(?:[1-9]\d*|0|0[Oo]?[0-7]+|0[Xx][0-7]+|0[Bb][01]+))?[bcdEeFfGgnosXx%]?))?\}(?!\}[^\}])
   

   This is the script I used to build it:

   import re
   
   def merge(*args):
       return r'(?:{})'.format(r'|'.join(args))
   
   # integer
   bininteger = r'0[Bb][01]+'
   hexinteger = r'0[Xx][0-7]+'
   octinteger = r'0[Oo]?[0-7]+'
   decimalinteger = r'[1-9]\d*|0'
   integer = merge(decimalinteger, octinteger, hexinteger, bininteger)
   
   # format_spec
   type_ = r'[bcdEeFfGgnosXx%]'
   precision = integer
   width = integer
   sign = r'[\x20+-]'
   align = r'[<>=\^]'
   fill = r'.'
   format_spec = r'(?:(?:{fill}?{align})?{sign}?\#?0?{width}?,?(?:\.{precision})?{type_}?)'.format(**locals())
   
   conversion = r'[rsa]'
   index_string = r'[^\]\}}\{{]+'
   identifier = r'[A-Za-z_]\w*'
   element_index = merge(integer, index_string)
   attribute_name = identifier
   arg_name = r'(?:{})?'.format(merge(identifier, integer))
   field_name = r'(?:{arg_name}(?:\.{attribute_name}|\[{element_index}\])*)'.format(**locals())
   replacement_field = r'(?<![^\{{]\{{)\{{{field_name}?(?:\!{conversion})?(?:\:{format_spec})?\}}(?!\}}[^\}}])'.format(**locals())
   
   print(replacement_field)

   You can see here that it works pretty well. I just copy-pasted the examples from the Python docs.

   These are the only strings it can't parse:

   '{:%Y-%m-%d %H:%M:%S}'
   '{0:{fill}{align}16}'
   '{0:{width}{base}}'
   

   but to be honest they don't look valid according to the grammar. There seem to be some discrepancies between the docs and the implementation as I had to edit the index_string regex according to this.
7. MattDMo

   added a commit that references this issue on Nov 1, 2015

   Changed scope for regular_expressions, started working on string_form…

   ...

   8dfdf37
8. sprt

   closed this as completedon Apr 4, 2018