Par2Parser.py

   1 #!/usr/bin/env python
   2 # vim: set ts=4 sts=4 sw=4 textwidth=92:
   3
   4 ################################################################################
   5 # The PAR2 Parser
   6 #
   7 # This was stolen from cfv (see http://cfv.sourceforge.net/ for a copy)
   8 ################################################################################
   9
  10 import struct, errno, os, md5
  11
  12 def chompnulls(line):
  13         p = line.find('\0')
  14         if p < 0: return line
  15         else:     return line[:p]
  16
  17 def get_protected_files (dir, filename):
  18         """Get all of the filenames that are protected by the par2
  19         file given as the filename"""
  20
  21         assert os.path.isdir (dir) # MUST be a valid directory
  22         assert os.path.isfile (os.path.join (dir, filename))
  23
  24         full_filename = os.path.join (dir, filename)
  25
  26         try:
  27                 file = open(full_filename, 'rb')
  28         except:
  29                 print 'Could not open %s' % (full_filename, )
  30                 return []
  31
  32         # We always want to do crc checks
  33         docrcchecks = True
  34
  35         pkt_header_fmt = '< 8s Q 16s 16s 16s'
  36         pkt_header_size = struct.calcsize(pkt_header_fmt)
  37         file_pkt_fmt = '< 16s 16s 16s Q'
  38         file_pkt_size = struct.calcsize(file_pkt_fmt)
  39         main_pkt_fmt = '< Q I'
  40         main_pkt_size = struct.calcsize(main_pkt_fmt)
  41
  42         seen_file_ids = {}
  43         expected_file_ids = None
  44         filenames = []
  45
  46         while 1:
  47                 d = file.read(pkt_header_size)
  48                 if not d:
  49                         break
  50
  51                 magic, pkt_len, pkt_md5, set_id, pkt_type = struct.unpack(pkt_header_fmt, d)
  52
  53                 if docrcchecks:
  54                         control_md5 = md5.new()
  55                         control_md5.update(d[0x20:])
  56                         d = file.read(pkt_len - pkt_header_size)
  57                         control_md5.update(d)
  58
  59                         if control_md5.digest() != pkt_md5:
  60                                 raise EnvironmentError, (errno.EINVAL, \
  61                                         "corrupt par2 file - bad packet hash")
  62
  63                 if pkt_type == 'PAR 2.0\0FileDesc':
  64                         if not docrcchecks:
  65                                 d = file.read(pkt_len - pkt_header_size)
  66
  67                         file_id, file_md5, file_md5_16k, file_size = \
  68                                 struct.unpack(file_pkt_fmt, d[:file_pkt_size])
  69
  70                         if seen_file_ids.get(file_id) is None:
  71                                 seen_file_ids[file_id] = 1
  72                                 filename = chompnulls(d[file_pkt_size:])
  73                                 filenames.append(filename)
  74
  75                 elif pkt_type == "PAR 2.0\0Main\0\0\0\0":
  76                         if not docrcchecks:
  77                                 d = file.read(pkt_len - pkt_header_size)
  78
  79                         if expected_file_ids is None:
  80                                 expected_file_ids = []
  81                                 slice_size, num_files = struct.unpack(main_pkt_fmt, d[:main_pkt_size])
  82                                 num_nonrecovery = (len(d)-main_pkt_size)/16 - num_files
  83
  84                                 for i in range(main_pkt_size,main_pkt_size+(num_files+num_nonrecovery)*16,16):
  85                                         expected_file_ids.append(d[i:i+16])
  86
  87                 else:
  88                         if not docrcchecks:
  89                                 file.seek(pkt_len - pkt_header_size, 1)
  90
  91         if expected_file_ids is None:
  92                 raise EnvironmentError, (errno.EINVAL, \
  93                         "corrupt or unsupported par2 file - no main packet found")
  94
  95         for id in expected_file_ids:
  96                 if not seen_file_ids.has_key(id):
  97                         raise EnvironmentError, (errno.EINVAL, \
  98                                 "corrupt or unsupported par2 file - " \
  99                                 "expected file description packet not found")
 100
 101         return filenames
 102
 103 def main ():
 104         pass
 105
 106 if __name__ == '__main__':
 107         main ()
 108