#!/usr/bin/env python
# Author: Dave Hull <dphull@trustedsignal.com>
# License: Public Domain
# Version: 0.01
# Run this script from the command line as follows:
#
# python yahoo_msg_decoder.py
#
# You will be prompted to provide the archive owner's username and the 
# archive filename. The script will decode (xor) the archive and dump
# the contents to the screen. Currently it barfs at the end because I
# haven't provided a good method to check for the end of file.
#
# To Do:
# Provide a graceful way of finishing up the file parsing.
# Get rid of the interactive mode and use command line args instead.
# Provide command line argument for the "other" conversant's username.
# Prevent output from going bold given certain byte strings.
# Provide output options for HTML.

import re, os, struct
from time import gmtime, strftime

def getUserName():
    username = ''
    m = None
    print '''Yahoo! Messenger archives are xor'd with the archive owner's username. You must 
know the archive owner's username to display the plain text.'''
    while m is None:
        username = raw_input('''Enter the Yahoo! Messenger user's username: ''')
        m = re.match('^[A-Za-z]{1}\w*\.?\w*$', username)
        if m is not None and 3 < len(username) < 33:
            return username
        else:
            m = None    
            print '''That's not a valid Yahoo Messenger ID. Try again.'''

def getFileName():
    filename = raw_input('''Enter the name of the Yahoo! Messenger archive file, use the full path: ''')
    return filename

def openFile(filename):
    filehandle = open(filename, 'rb')
    return filehandle

def getHeader(filehandle):
    packedTime = filehandle.read(4)
    tz = filehandle.read(1)
    return tz

def getMsgTime(filehandle):
    packedTime = filehandle.read(4) # time is always 4 bytes
    utcOffset = filehandle.read(1)  # UTC offset follows time
    unpackedTime = struct.unpack("i", packedTime)[0]
    return strftime("%Y-%m-%d %H:%M:%S UTC", gmtime(unpackedTime))

def getMsg(filehandle, readlength, username):
    msg = ''
    packedMsg = filehandle.read(readlength)
    mLength = str(readlength) + "s"
    unpackedMsg = struct.unpack(mLength, packedMsg)[0]
    msgLen = len(unpackedMsg)
    usernameLen = len(username)
    while msgLen > len(username):
        username += username
    for i in range(msgLen):
        msg += chr(ord(unpackedMsg[i]) ^ ord(username[i]))
    filehandle.read(1) # there's a null byte after the message
    return msg

def getReservedBits(filehandle):
    filehandle.read(3) # reserved bits are always in 3s

def getNBits(filehandle, n):
    filehandle.read(n)

def getSent(filehandle):
    sent = filehandle.read(1)
    return sent

def getMsgLength(filehandle):
    msgLength = filehandle.read(1)
    return ord(msgLength)

username = getUserName()
filename = getFileName()

filehandle = openFile(filename)

# Need to read header to determine how to parse file
if getHeader(filehandle) == 0:
    getReservedBits(filehandle)
    getSent(filehandle)
    getNBits(filehandle, 11)
else:
    filehandle.seek(0)

while 1:
    msgTime = getMsgTime(filehandle)
    getReservedBits(filehandle)
    if ord(getSent(filehandle)):
        print "other (" + msgTime + "):",
    else:
        print username + " (" + msgTime +"):",
    getReservedBits(filehandle)
    msgLen = getMsgLength(filehandle)
    getReservedBits(filehandle)
    print getMsg(filehandle, msgLen, username)
    getReservedBits(filehandle)

filehandle.close()