weissman.py

#!/usr/bin/env python
import struct
import sys
import binascii

# def str2bytes(string):
#     return "".join("\\x" + "{0:02x}".format(ord(x)) for x in string)

def compare(bs, ks, tag=None):
    if bs != ks:
        print "ERROR mismatched"
        if tag is not None:
            print "{0:02x}".format(tag)
        for j in range(len(bs)):
            print binascii.hexlify(bs[j]), binascii.hexlify(ks[j]), bs[j], ks[j]
        sys.exit(1)

# 0x13 ^S - read 9 bytes
# even numbers 0x08 through 0x12 -
# take b/2 bytes from block identified by next 2 bytes
def decode(data, known):
    c = 0
    chunks = []
    tagmap = dict()
    mapfile = open('map.csv', 'w')
    while c < len(data):
        b = data[c]; c += 1
        if b == '\x13':
            bs = data[c:c+9]; c += 9
            ks = known.read(9)
            compare(bs, ks)
            chunks.append(bs)
            sys.stdout.write(bs)
        elif b in ['\x08', '\x0a', '\x0c', '\x0e', '\x10', '\x12',]:
            (tag,) = struct.unpack("<H", data[c:c+2]); c += 2
            l = ord(b)/2
            if tag in tagmap:
                bs = chunks[tagmap[tag]][:l]
                ks = known.read(l)
                compare(bs, ks, tag)
                sys.stdout.write(bs)
            else:
                m = known.read(l)
                sys.stdout.write(m)
                # print "$"
                # print "0x{0:x} 0x{1:x}".format(ord(b), tag)
                # print "what are the next {} bytes?".format(l)
                # m = input()
                # print m
                # if len(m) != l:
                #     print "WARNING: that wasn't {} bytes".format(l)
                for (j, chunk) in enumerate(chunks):
                    if chunk.find(m) == 0:
                        tagmap[tag] = j
                        if l == 9:
                            mapfile.write("0x{0:x}, {1:d}, \"{2}\"\n".
                                    format(tag, j, binascii.hexlify(m)))
                            mapfile.flush()
                        break
        else:
            print "oops 0x{0:x}".format(ord(b))
            sys.exit(1)

# Usage: weissman.py weissman.csawlz pg28885.txt
infile = open(sys.argv[1])
(magic, version, num_files) = struct.unpack("<8sii", infile.read(16))
print(magic, version, num_files)
for i in range(num_files):
    (magic, c_size, u_size, filename) = struct.unpack("<iii32s", infile.read(44))
    filename = filename.rstrip('\0')
    print(magic, c_size, u_size, filename)
    data = infile.read(c_size)
    if filename == sys.argv[2]:
        decode(data, open(sys.argv[2]))