tools/msnchat

   1 #!/usr/bin/env python
   2 """
   3 Process packet capture files and produce a nice HTML
   4 report of MSN Chat sessions.
   5
   6 Copyright (c) 2003 by Gilbert Ramirez <gram@alumni.rice.edu>
   7
   8 SPDX-License-Identifier: GPL-2.0-or-later
   9 """
  10
  11 import os
  12 import re
  13 import sys
  14 import array
  15 import string
  16 import WiresharkXML
  17 import getopt
  18
  19 # By default we output the HTML to stdout
  20 out_fh = sys.stdout
  21
  22 class MSNMessage:
  23     pass
  24
  25 class MSN_MSG(MSNMessage):
  26     def __init__(self, timestamp, user, message):
  27         self.timestamp = timestamp
  28         self.user = user
  29         self.message = message
  30
  31
  32 class Conversation:
  33     """Keeps track of a single MSN chat session"""
  34
  35     re_MSG_out = re.compile("MSG (?P<TrID>\d+) (?P<ACKTYPE>[UNA]) (?P<len>\d+)")
  36     re_MSG_in  = re.compile("MSG (?P<user>\S+)@(?P<domain>\S+) (?P<alias>\S+) (?P<len>\d+)")
  37
  38     USER_NOT_FOUND = -1
  39     DEFAULT_USER = None
  40
  41
  42     DEFAULT_USER_COLOR = "#0000ff"
  43     USER_COLORS = [ "#ff0000", "#00ff00",
  44             "#800000", "#008000", "#000080" ]
  45
  46     DEFAULT_USER_TEXT_COLOR = "#000000"
  47     USER_TEXT_COLOR = "#000080"
  48
  49     def __init__(self):
  50         self.packets = []
  51         self.messages = []
  52
  53     def AddPacket(self, packet):
  54         self.packets.append(packet)
  55
  56     def Summarize(self):
  57         for packet in self.packets:
  58             msg = self.CreateMSNMessage(packet)
  59             if msg:
  60                 self.messages.append(msg)
  61             else:
  62                  #XXX
  63                  pass
  64
  65
  66     def CreateMSNMessage(self, packet):
  67         msnms = packet.get_items("msnms")[0]
  68
  69         # Check the first line in the msnms transmission for the user
  70         child = msnms.children[0]
  71         user = self.USER_NOT_FOUND
  72
  73         m = self.re_MSG_out.search(child.show)
  74         if m:
  75             user = self.DEFAULT_USER
  76
  77         else:
  78             m = self.re_MSG_in.search(child.show)
  79             if m:
  80                 user = m.group("alias")
  81
  82         if user == self.USER_NOT_FOUND:
  83             print >> sys.stderr, "No match for", child.show
  84             sys.exit(1)
  85             return None
  86
  87         msg = ""
  88
  89         i = 5
  90         check_trailing = 0
  91         if len(msnms.children) > 5:
  92             check_trailing = 1
  93
  94         while i < len(msnms.children):
  95             msg += msnms.children[i].show
  96             if check_trailing:
  97                 j = msg.find("MSG ")
  98                 if j >= 0:
  99                     msg = msg[:j]
 100                     i += 5
 101                 else:
 102                     i += 6
 103             else:
 104                 i += 6
 105
 106         timestamp = packet.get_items("frame.time")[0].get_show()
 107         i = timestamp.rfind(".")
 108         timestamp = timestamp[:i]
 109
 110         return MSN_MSG(timestamp, user, msg)
 111
 112     def MsgToHTML(self, text):
 113         bytes = array.array("B")
 114
 115         new_string = text
 116         i = new_string.find("\\")
 117
 118         while i > -1:
 119             # At the end?
 120             if i == len(new_string) - 1:
 121                 # Just let the default action
 122                 # copy everything to 'bytes'
 123                 break
 124
 125             if new_string[i+1] in string.digits:
 126                 left = new_string[:i]
 127                 bytes.fromstring(left)
 128
 129                 right = new_string[i+4:]
 130
 131                 oct_string = new_string[i+1:i+4]
 132                 char = int(oct_string, 8)
 133                 bytes.append(char)
 134
 135                 new_string = right
 136
 137             # ignore \r and \n
 138             elif new_string[i+1] in "rn":
 139                 copy_these = new_string[:i]
 140                 bytes.fromstring(copy_these)
 141                 new_string = new_string[i+2:]
 142
 143             else:
 144                 copy_these = new_string[:i+2]
 145                 bytes.fromstring(copy_these)
 146                 new_string = new_string[i+2:]
 147
 148             i = new_string.find("\\")
 149
 150
 151         bytes.fromstring(new_string)
 152
 153         return bytes
 154
 155     def CreateHTML(self, default_user):
 156         if not self.messages:
 157             return
 158
 159         print >> out_fh, """
 160 <HR><BR><H3 Align=Center> ---- New Conversation @ %s ----</H3><BR>""" \
 161             % (self.messages[0].timestamp)
 162
 163         user_color_assignments = {}
 164
 165         for msg in self.messages:
 166             # Calculate 'user' and 'user_color' and 'user_text_color'
 167             if msg.user == self.DEFAULT_USER:
 168                 user = default_user
 169                 user_color = self.DEFAULT_USER_COLOR
 170                 user_text_color = self.DEFAULT_USER_TEXT_COLOR
 171             else:
 172                 user = msg.user
 173                 user_text_color = self.USER_TEXT_COLOR
 174                 if user_color_assignments.has_key(user):
 175                     user_color = user_color_assignments[user]
 176                 else:
 177                     num_assigned = len(user_color_assignments.keys())
 178                     user_color = self.USER_COLORS[num_assigned]
 179                     user_color_assignments[user] = user_color
 180
 181             # "Oct  6, 2003 21:45:25"  --> "21:45:25"
 182             timestamp = msg.timestamp.split()[-1]
 183
 184             htmlmsg = self.MsgToHTML(msg.message)
 185
 186             print >> out_fh, """
 187 <FONT COLOR="%s"><FONT SIZE="2">(%s) </FONT><B>%s:</B></FONT> <FONT COLOR="%s">""" \
 188                 % (user_color, timestamp, user, user_text_color)
 189
 190             htmlmsg.tofile(out_fh)
 191
 192             print >> out_fh, "</FONT><BR>"
 193
 194
 195 class CaptureFile:
 196     """Parses a single a capture file and keeps track of
 197     all chat sessions in the file."""
 198
 199     def __init__(self, capture_filename, tshark):
 200         """Run tshark on the capture file and parse
 201         the data."""
 202         self.conversations = []
 203         self.conversations_map = {}
 204
 205         pipe = os.popen(tshark + " -Tpdml -n -R "
 206             "'msnms contains \"X-MMS-IM-Format\"' "
 207             "-r " + capture_filename, "r")
 208
 209         WiresharkXML.parse_fh(pipe, self.collect_packets)
 210
 211         for conv in self.conversations:
 212             conv.Summarize()
 213
 214     def collect_packets(self, packet):
 215         """Collect the packets passed back from WiresharkXML.
 216         Sort them by TCP/IP conversation, as there could be multiple
 217         clients per machine."""
 218         # Just in case we're looking at tunnelling protocols where
 219         # more than one IP or TCP header exists, look at the last one,
 220         # which would be the one inside the tunnel.
 221         src_ip = packet.get_items("ip.src")[-1].get_show()
 222         dst_ip = packet.get_items("ip.dst")[-1].get_show()
 223         src_tcp = packet.get_items("tcp.srcport")[-1].get_show()
 224         dst_tcp = packet.get_items("tcp.dstport")[-1].get_show()
 225
 226         key_params = [src_ip, dst_ip, src_tcp, dst_tcp]
 227         key_params.sort()
 228         key = '|'.join(key_params)
 229
 230         if not self.conversations_map.has_key(key):
 231             conv = self.conversations_map[key] = Conversation()
 232             self.conversations.append(conv)
 233         else:
 234             conv = self.conversations_map[key]
 235
 236         conv.AddPacket(packet)
 237
 238
 239     def CreateHTML(self, default_user):
 240         if not self.conversations:
 241             return
 242
 243         for conv in self.conversations:
 244             conv.CreateHTML(default_user)
 245
 246
 247 def run_filename(filename, default_user, tshark):
 248     """Process one capture file."""
 249
 250     capture = CaptureFile(filename, tshark)
 251     capture.CreateHTML(default_user)
 252
 253
 254 def run(filenames, default_user, tshark):
 255     # HTML Header
 256     print >> out_fh, """
 257 <HTML><TITLE>MSN Conversation</TITLE>
 258 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
 259 <BODY>
 260 """
 261     for filename in filenames:
 262         run_filename(filename, default_user, tshark)
 263
 264     # HTML Footer
 265     print >> out_fh, """
 266 <HR>
 267 </BODY>
 268 </HTML>
 269 """
 270
 271
 272 def usage():
 273     print >> sys.stderr, "msnchat [OPTIONS] CAPTURE_FILE [...]"
 274     print >> sys.stderr, "  -o FILE       name of output file"
 275     print >> sys.stderr, "  -t TSHARK  location of tshark binary"
 276     print >> sys.stderr, "  -u USER       name for unknown user"
 277     sys.exit(1)
 278
 279 def main():
 280     default_user = "Unknown"
 281     tshark = "tshark"
 282
 283     optstring = "ho:t:u:"
 284     longopts = ["help"]
 285
 286     try:
 287         opts, args = getopt.getopt(sys.argv[1:], optstring, longopts)
 288     except getopt.GetoptError:
 289         usage()
 290
 291     for opt, arg in opts:
 292         if opt == "-h" or opt == "--help":
 293             usage()
 294
 295         elif opt == "-o":
 296             filename = arg
 297             global out_fh
 298             try:
 299                 out_fh = open(filename, "w")
 300             except IOError:
 301                 sys.exit("Could not open %s for writing." % (filename,))
 302
 303         elif opt == "-u":
 304             default_user = arg
 305
 306         elif opt == "-t":
 307             tshark = arg
 308
 309         else:
 310             sys.exit("Unhandled command-line option: " + opt)
 311
 312     run(args, default_user, tshark)
 313
 314 if __name__ == '__main__':
 315     main()