TestON/bin/unpep8

#!/usr/bin/python

"""
Translate from PEP8 Python style to Mininet (i.e. Arista-like)
Python style

usage: unpep8 < old.py > new.py

- Reinstates CapWords for methods and instance variables
- Gets rid of triple single quotes
- Eliminates triple quotes on single lines
- Inserts extra spaces to improve readability
- Fixes Doxygen (or doxypy) ugliness

Does the following translations:

ClassName.method_name(foo = bar) -> ClassName.methodName( foo=bar )

Triple-single-quotes -> triple-double-quotes

@param foo description -> foo: description
@return description    -> returns: description
@author me             -> author: me
@todo(me)              -> TODO(me)

Bugs/Limitations:

- Hack to restore strings is ugly
- Multiline strings get mangled
- Comments are mangled (which is arguably the "right thing" to do, except
  that, for example, the left hand sides of the above would get translated!)
- Doesn't eliminate unnecessary backslashes
- Has no opinion on tab size
- complicated indented docstrings get flattened
- We don't (yet) have a filter to generate Doxygen/Doxypy
- Currently leaves indents on blank comment lines
- May lead to namespace collisions (e.g. some_thing and someThing)

Bob Lantz, rlantz@cs.stanford.edu
1/24/2010
"""

import re, sys

def fixUnderscoreTriplet( match ):
   "Translate a matched triplet of the form a_b to aB."
   triplet = match.group()
   return triplet[ :-2 ] + triplet[ -1 ].capitalize()

def reinstateCapWords( text ):
   underscoreTriplet = re.compile( r'[A-Za-z0-9]_[A-Za-z0-9]' )
   return underscoreTriplet.sub( fixUnderscoreTriplet, text )
 
def replaceTripleApostrophes( text ):
   "Replace triple apostrophes with triple quotes."
   return text.replace( "'''", '"""')

def simplifyTripleQuotes( text ):
   "Fix single-line doc strings."
   r = re.compile( r'"""([^\"\n]+)"""' )
   return r.sub( r'"\1"', text )
   
def insertExtraSpaces( text ):
   "Insert extra spaces inside of parentheses and brackets/curly braces."
   lparen = re.compile( r'\((?![\s\)])' )
   text = lparen.sub( r'( ', text )
   rparen = re.compile( r'([^\s\(])(?=\))' )
   text = rparen.sub( r'\1 ', text)
   # brackets
   lbrack = re.compile( r'\[(?![\s\]])' )
   text = lbrack.sub( r'[ ', text )
   rbrack = re.compile( r'([^\s\[])(?=\])' )
   text = rbrack.sub( r'\1 ', text)
   # curly braces
   lcurly = re.compile( r'\{(?![\s\}])' )
   text = lcurly.sub( r'{ ', text )
   rcurly = re.compile( r'([^\s\{])(?=\})' )
   text = rcurly.sub( r'\1 ', text)   
   return text
   
def fixDoxygen( text ):
   """Translate @param foo to foo:, @return bar to returns: bar, and
      @author me to author: me"""
   param = re.compile( r'@param (\w+)' )
   text = param.sub( r'\1:', text )
   returns = re.compile( r'@return' )
   text = returns.sub( r'returns:', text )
   author = re.compile( r'@author' )
   text = author.sub( r'author:', text)
   # @todo -> TODO
   text = text.replace( '@todo', 'TODO' )
   return text

def removeCommentFirstBlankLine( text ):
   "Remove annoying blank lines after first line in comments."
   line = re.compile( r'("""[^\n]*\n)\s*\n', re.MULTILINE )
   return line.sub( r'\1', text )
   
def fixArgs( match, kwarg = re.compile( r'(\w+) = ' ) ):
   "Replace foo = bar with foo=bar."
   return kwarg.sub( r'\1=', match.group() )
   
def fixKeywords( text ):
   "Change keyword argumentsfrom foo = bar to foo=bar."
   args = re.compile( r'\(([^\)]+)\)', re.MULTILINE )
   return args.sub( fixArgs, text )

# Unfortunately, Python doesn't natively support balanced or recursive
# regular expressions. We could use PyParsing, but that opens another can
# of worms. For now, we just have a cheap hack to restore strings,
# so we don't end up accidentally mangling things like messages, search strings,
# and regular expressions.

def lineIter( text ):
   "Simple iterator over lines in text."
   for line in text.splitlines(): yield line
   
def stringIter( strList ):
   "Yield strings in strList."
   for s in strList: yield s

def restoreRegex( regex, old, new ):
   "Find regexes in old and restore them into new."
   oldStrs = regex.findall( old )
   # Sanity check - count should be the same!
   newStrs = regex.findall( new )
   assert len( oldStrs ) == len( newStrs )
   # Replace newStrs with oldStrs
   siter = stringIter( oldStrs )
   reps = lambda dummy: siter.next()
   return regex.sub( reps, new )

# This is a cheap hack, and it may not work 100%, since
# it doesn't handle multiline strings.
# However, it should be mostly harmless...
   
def restoreStrings( oldText, newText ):
   "Restore strings from oldText into newText, returning result."
   oldLines, newLines = lineIter( oldText ), lineIter( newText )
   quoteStrings = re.compile( r'("[^"]*")' )
   tickStrings = re.compile( r"('[^']*')" )
   result = ''
   # It would be nice if we could blast the whole file, but for
   # now it seems to work line-by-line
   for newLine in newLines:
      oldLine = oldLines.next()
      newLine = restoreRegex( quoteStrings, oldLine, newLine )
      newLine = restoreRegex( tickStrings, oldLine, newLine )
      result += newLine + '\n'
   return result
   
# This might be slightly controversial, since it uses
# three spaces to line up multiline comments. However,
# I much prefer it. Limitations: if you have deeper
# indents in comments, they will be eliminated. ;-(

def fixComment( match, 
   indentExp=re.compile( r'\n([ ]*)(?=[^/s])', re.MULTILINE ),
   trailingQuotes=re.compile( r'\s+"""' ) ):
   "Re-indent comment, and join trailing quotes."
   originalIndent = match.group( 1 )
   comment = match.group( 2 )
   indent = '\n' + originalIndent
   # Exception: leave unindented things unindented!
   if len( originalIndent ) is not 0: indent += '   '
   comment = indentExp.sub( indent, comment )
   return originalIndent + trailingQuotes.sub( '"""', comment )
   
def fixCommentIndents( text ):
   "Fix multiline comment indentation."
   comments = re.compile( r'^([ ]*)("""[^"]*""")$', re.MULTILINE )
   return comments.sub( fixComment, text )
 
def removeBogusLinefeeds( text ):
   "Remove extra linefeeds at the end of single-line comments."
   bogusLfs = re.compile( r'"([^"\n]*)\n"', re.MULTILINE )
   return bogusLfs.sub( '"\1"', text)
   
def convertFromPep8( program ):
   oldProgram = program
   # Program text transforms
   program = reinstateCapWords( program )  # Turning off for now
   program = reinstateCapWords( program )
   program = fixKeywords( program )
   program = insertExtraSpaces( program )
   # Undo string damage
   program = restoreStrings( oldProgram, program )
   # Docstring transforms
   program = replaceTripleApostrophes( program )
   program = simplifyTripleQuotes( program )
   program = fixDoxygen( program )
   # program = fixCommentIndents( program ) # Turning off as it breaks tests in TestON
   program = removeBogusLinefeeds( program )
   # Destructive transforms (these can delete lines)
   program = removeCommentFirstBlankLine( program )
   return program

if __name__ == '__main__':
   print convertFromPep8( sys.stdin.read() )