changelog shortlog tags changeset files revisions annotate raw

parse_sessionsstore.py

changeset 20: b6b0e9bc226d
author: Greg Darke <greg+laptop@tsukasa.net.au>
date: Thu Apr 16 00:45:45 2009 +1000 (15 months ago)
permissions: -rwxr-xr-x
description: Ignore some stuff 
     1#!/usr/bin/env python
     2# Copyright 2008 Greg Darke <greg+laptop@tsukasa.net.au>
     3# Licensed for distribution under the GPL version 2, check COPYING for details
     4# Program to parse Firefox 3's sessionstore.js file (the list of webpages to restore)
     5
     6import collections
     7
     8class ParseException(Exception):
     9	pass
    10
    11class JSBooleanTrue(object):
    12	@staticmethod
    13	def isValidStart(ch):
    14		return ch =='t'
    15
    16	@staticmethod
    17	def parse(data):
    18		assert data.popleft() == 't'
    19		assert data.popleft() == 'r'
    20		assert data.popleft() == 'u'
    21		assert data.popleft() == 'e'
    22		return True
    23
    24class JSBooleanFalse(object):
    25	@staticmethod
    26	def isValidStart(ch):
    27		return ch =='f'
    28
    29	@staticmethod
    30	def parse(data):
    31		assert data.popleft() == 'f'
    32		assert data.popleft() == 'a'
    33		assert data.popleft() == 'l'
    34		assert data.popleft() == 's'
    35		assert data.popleft() == 'e'
    36		return False
    37
    38class JSDigits(object):
    39	@staticmethod
    40	def isValidStart(ch):
    41		return ch.isdigit()
    42
    43	@staticmethod
    44	def parse(data):
    45		assert data[0].isdigit()
    46		ret = []
    47
    48		while data[0].isdigit():
    49			ret.append(data.popleft())
    50
    51		return int(''.join(ret))
    52
    53class JSDoubleString(object):
    54	@staticmethod
    55	def isValidStart(ch):
    56		return ch == '"'
    57
    58	@staticmethod
    59	def parse(data):
    60		assert data.popleft() == '"'
    61		ret = []
    62
    63		while data[0] != '"':
    64			ret.append(data.popleft())
    65		assert data.popleft() == '"'
    66
    67		return ''.join(ret)
    68
    69class JSList(object):
    70	@staticmethod
    71	def isValidStart(ch):
    72		return ch == '['
    73
    74	@staticmethod
    75	def parse(data):
    76		VALID_VALUES = (JSDict, )
    77		assert data.popleft() == '['
    78		ret = []
    79
    80		while True:
    81			for valueType in VALID_VALUES:
    82				if valueType.isValidStart(data[0]):
    83					value = valueType.parse(data)
    84					break
    85			else:
    86				raise ParseException, 'Unknown valueType start "%s"' % (data[0], )
    87
    88			ret.append(value)
    89
    90			if data[0] == ',':
    91				data.popleft() # There are more entries
    92				while data[0] == ' ':
    93					data.popleft()
    94			else:
    95				break
    96
    97		assert data.popleft() == ']'
    98		return ret
    99
   100class JSDictIdentifier(object):
   101	_validChars = "_abcdefghijklmnopqrstuvwxyABCDEFGHIJKLMNOPQRSTUVWXYZ'"
   102	@classmethod
   103	def isValidStart(cls, ch):
   104		return ch in cls._validChars
   105
   106	@classmethod
   107	def parse(cls, data):
   108		if data[0] not in cls._validChars:
   109			raise ParseException, 'Expected character, but got "%s"' % (data[0], )
   110		ret = []
   111
   112		if data[0] == "'": # Data is quoted
   113			assert data.popleft() == "'"
   114			while data[0] != "'":
   115				ret.append(data.popleft())
   116			assert data.popleft() == "'"
   117		else:
   118			while data[0] != ':':
   119				ret.append(data.popleft())
   120
   121		return ''.join(ret)
   122
   123class JSDict(object):
   124	@staticmethod
   125	def isValidStart(ch):
   126		return ch == '{'
   127
   128	@staticmethod
   129	def parse(data):
   130		VALID_VALUES = (
   131			JSList,
   132			JSDoubleString,
   133			JSDigits,
   134			JSDict,
   135			JSBooleanTrue, JSBooleanFalse,
   136		)
   137		assert data.popleft() == '{'
   138		ret = {}
   139
   140		if data[0] == '}':
   141			# Empty dictionary
   142			return ret
   143
   144		while True:
   145			identifier = JSDictIdentifier.parse(data)
   146			assert data.popleft() == ':'
   147
   148			for valueType in VALID_VALUES:
   149				if valueType.isValidStart(data[0]):
   150					value = valueType.parse(data)
   151					break
   152			else:
   153				raise ParseException, 'Unknown valueType start %s' % (data[0], )
   154
   155			if identifier in ret:
   156				raise ParseException, 'The identifier "%s" is already in the dict' % (identifier, )
   157
   158			ret[identifier] = value
   159			if data[0] == ',':
   160				data.popleft() # There are more entires
   161				while data[0] == ' ':
   162					data.popleft()
   163			else:
   164				break
   165
   166		assert data.popleft() == '}'
   167		return ret
   168
   169
   170def parseSessionStore(data):
   171	data = collections.deque(data)
   172	assert data.popleft() == '('
   173	ret = JSDict.parse(data)
   174	assert data.popleft() == ')'
   175	assert len(data) == 0
   176	return ret
   177
   178def main(filename):
   179	import cgitb
   180	import pprint
   181	cgitb.enable(format='text')
   182
   183	pprint.pprint(parseSessionStore(file(filename).read()))
   184
   185if __name__ == "__main__":
   186	import sys
   187	try:
   188		filename = sys.argv[1]
   189	except IndexError:
   190		print >>sys.stderr, 'Usage: %s path_to_sessionstore.js' % (sys.argv[0], )
   191		sys.exit(1)
   192	main(filename)
   193
   194__all__ = ('parseSessionStore', )