pyfontify.py

"""Module to analyze Python source code; for syntax coloring tools.

Interface:
	tags = fontify(pytext, searchfrom, searchto)

The 'pytext' argument is a string containing Python source code.
The (optional) arguments 'searchfrom' and 'searchto' may contain a slice in pytext. 
The returned value is a list of tuples, formatted like this:
	[('keyword', 0, 6, None), ('keyword', 11, 17, None), ('comment', 23, 53, None), etc. ]
The tuple contents are always like this:
	(tag, startindex, endindex, sublist)
tag is one of ('comment', 'string', 'keyword', 'function', 'class')
sublist is not used, hence always None. 
"""

# Based on FontText.py by Mitchell S. Chapman,
# which was modified by Zachary Roadhouse,
# then un-Tk'd by Just van Rossum.
# Many thanks for regular expression debugging & authoring are due to:
#	Tim (the-incredib-ly y'rs) Peters and Cristian Tismer
# So, who owns the copyright? ;-) How about this:
# Copyright 1996-1997: 
#	Mitchell S. Chapman,
#	Zachary Roadhouse,
#	Tim Peters,
#	Just van Rossum
#
# Version 0.4 - changes copyright (C) 2001 Mark Pilgrim (mark@diveintopython.org)
#   2001/02/05 - MAP - distinguish between class and function identifiers
#   2001/03/21 - MAP - get keywords from keyword module (instead of hard-coded list)
#   2001/03/22 - MAP - use re module instead of deprecated regex module

__version__ = "0.4"

import string, re, keyword

# Build up a regular expression which will match anything
# interesting, including multi-line triple-quoted strings.
commentPat = "#.*"

pat = "q[^\q\n]*(\\\\[\000-\377][^\q\n]*)*q"
quotePat = string.replace(pat, "q", "'") + "|" + string.replace(pat, 'q', '"')

# Way to go, Tim!
pat = """
	qqq
	[^\\q]*
	(
		(	\\\\[\000-\377]
		|	q
			(	\\\\[\000-\377]
			|	[^\\q]
			|	q
				(	\\\\[\000-\377]
				|	[^\\q]
				)
			)
		)
		[^\\q]*
	)*
	qqq
"""
pat = string.join(string.split(pat), '')	# get rid of whitespace
tripleQuotePat = string.replace(pat, "q", "'") + "|" + string.replace(pat, 'q', '"')

# Build up a regular expression which matches all and only
# Python keywords. This will let us skip the uninteresting
# identifier references.
# nonKeyPat identifies characters which may legally precede
# a keyword pattern.
nonKeyPat = "(^|[^a-zA-Z0-9_.\"'])"
keywordsPat = string.join(keyword.kwlist, "|")
keyPat = nonKeyPat + "(" + keywordsPat + ")" + nonKeyPat

matchPat = keyPat + "|" + commentPat + "|" + tripleQuotePat + "|" + quotePat
matchRE = re.compile(matchPat)

idKeyPat = "[ \t]*[A-Za-z_][A-Za-z_0-9.]*"	# Ident w. leading whitespace.
idRE = re.compile(idKeyPat)

def fontify(pytext, searchfrom=0, searchto=None):
	if searchto is None:
		searchto = len(pytext)
	tags = []
	commentTag = 'comment'
	stringTag = 'string'
	keywordTag = 'keyword'
	functionTag = 'function'
	classTag = 'class'

	start = 0
	end = searchfrom
	while 1:
		matchObject = matchRE.search(pytext, end)
		if not matchObject: break
		(start, end) = matchObject.span()
		match = matchObject.group(0)
		c = match[0]
		if c not in "#'\"":
			# Must have matched a keyword.
			if start <> searchfrom:
				# there's still a redundant char before and after it, strip!
				match = match[1:-1]
				start = start + 1
			else:
				# this is the first keyword in the text.
				# Only a space at the end.
				match = match[:-1]
			end = end - 1
			tags.append((keywordTag, start, end, None))
			# If this was a defining keyword, look ahead to the
			# following identifier.
			if match in ["def", "class"]:
				idMatchObject = idRE.search(pytext, end)
				if idMatchObject:
					(start, end) = idMatchObject.span()
					match = idMatchObject.group(0)
					tags.append(((match=='def') and functionTag or classTag, start, end, None))
		elif c == "#":
			tags.append((commentTag, start, end, None))
		else:
			tags.append((stringTag, start, end, None))
	return tags

def test(path):
	f = open(path)
	text = f.read()
	f.close()
	tags = fontify(text)
	for tag, start, end, sublist in tags:
		print tag, `text[start:end]`, start, end