Mercurial > touhou
annotate pytouhou/formats/pbg3.py @ 536:6b76c9ba3975
Make archives return files by default, instead of bytes.
author | Emmanuel Gil Peyrot <linkmauve@linkmauve.fr> |
---|---|
date | Sat, 05 Apr 2014 18:53:48 +0200 |
parents | 1b532e7dd521 |
children | e15672733c93 |
rev | line source |
---|---|
52
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
1 # -*- encoding: utf-8 -*- |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
2 ## |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
3 ## Copyright (C) 2011 Thibaut Girka <thib@sitedethib.com> |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
4 ## |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
5 ## This program is free software; you can redistribute it and/or modify |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
6 ## it under the terms of the GNU General Public License as published |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
7 ## by the Free Software Foundation; version 3 only. |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
8 ## |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
9 ## This program is distributed in the hope that it will be useful, |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
10 ## but WITHOUT ANY WARRANTY; without even the implied warranty of |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
11 ## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
12 ## GNU General Public License for more details. |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
13 ## |
ab826bc29aa2
Add some documentation, GPLv3 headers, README and COPYING file.
Thibaut Girka <thib@sitedethib.com>
parents:
0
diff
changeset
|
14 |
204
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
15 """PBG3 archive files handling. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
16 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
17 This module provides classes for handling the PBG3 file format. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
18 The PBG3 format is the archive format used by Touhou: EoSD. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
19 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
20 PBG3 files are merely a bitstream composed of a header, |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
21 a file table, and LZSS-compressed files. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
22 """ |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
23 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
24 from collections import namedtuple |
536
6b76c9ba3975
Make archives return files by default, instead of bytes.
Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
parents:
490
diff
changeset
|
25 from io import BytesIO |
204
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
26 |
58 | 27 from pytouhou.utils.bitstream import BitStream |
490
1b532e7dd521
Decrease PBG3 loading time by improving lzss and bitstream integration.
Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
parents:
377
diff
changeset
|
28 from pytouhou.utils import lzss |
58 | 29 |
30 from pytouhou.utils.helpers import get_logger | |
31 | |
377
70e2ed71b09c
Add meaningful exceptions in format parsing.
Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
parents:
252
diff
changeset
|
32 from pytouhou.formats import WrongFormatError |
70e2ed71b09c
Add meaningful exceptions in format parsing.
Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
parents:
252
diff
changeset
|
33 |
58 | 34 logger = get_logger(__name__) |
35 | |
36 | |
0 | 37 class PBG3BitStream(BitStream): |
204
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
38 """Helper class to handle strings and integers in PBG3 bitstreams.""" |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
39 |
0 | 40 def read_int(self): |
204
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
41 """Read an integer from the bitstream. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
42 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
43 Integers have variable sizes. They begin with a two-bit value indicating |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
44 the number of (non-aligned) bytes to read. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
45 """ |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
46 |
0 | 47 size = self.read(2) |
48 return self.read((size + 1) * 8) | |
49 | |
50 | |
51 def read_string(self, maxsize): | |
204
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
52 """Read a string from the bitstream. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
53 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
54 Strings are stored as standard NULL-termianted sequences of bytes. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
55 The only catch is that they are not byte-aligned. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
56 """ |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
57 |
0 | 58 string = [] |
59 for i in range(maxsize): | |
60 byte = self.read(8) | |
61 if byte == 0: | |
62 break | |
63 string.append(byte) | |
64 return ''.join(chr(byte) for byte in string) | |
65 | |
66 | |
67 | |
204
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
68 PBG3Entry = namedtuple('PBG3Entry', 'unknown1 unknown2 checksum offset size') |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
69 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
70 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
71 |
0 | 72 class PBG3(object): |
204
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
73 """Handle PBG3 archive files. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
74 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
75 PBG3 is a file archive format used in Touhou 6: EoSD. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
76 This class provides a representation of such files, as well as functions to |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
77 read and extract files from a PBG3 archive. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
78 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
79 Instance variables: |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
80 entries -- list of PBG3Entry objects describing files present in the archive |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
81 bitstream -- PBG3BitStream object |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
82 """ |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
83 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
84 def __init__(self, entries=None, bitstream=None): |
252
b5c7369abd7c
Improve data reading perfs
Thibaut Girka <thib@sitedethib.com>
parents:
204
diff
changeset
|
85 self.entries = entries or {} |
0 | 86 self.bitstream = bitstream #TODO |
87 | |
88 | |
97 | 89 def __enter__(self): |
90 return self | |
91 | |
92 | |
93 def __exit__(self, type, value, traceback): | |
94 return self.bitstream.__exit__(type, value, traceback) | |
95 | |
96 | |
0 | 97 @classmethod |
98 def read(cls, file): | |
204
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
99 """Read a PBG3 file. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
100 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
101 Raise an exception if the file is invalid. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
102 Return a PBG3 instance otherwise. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
103 """ |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
104 |
0 | 105 magic = file.read(4) |
106 if magic != b'PBG3': | |
377
70e2ed71b09c
Add meaningful exceptions in format parsing.
Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
parents:
252
diff
changeset
|
107 raise WrongFormatError(magic) |
0 | 108 |
109 bitstream = PBG3BitStream(file) | |
110 entries = {} | |
111 | |
112 nb_entries = bitstream.read_int() | |
113 offset = bitstream.read_int() | |
114 bitstream.seek(offset) | |
115 for i in range(nb_entries): | |
116 unknown1 = bitstream.read_int() | |
117 unknown2 = bitstream.read_int() | |
118 checksum = bitstream.read_int() # Checksum of *compressed data* | |
119 offset = bitstream.read_int() | |
120 size = bitstream.read_int() | |
121 name = bitstream.read_string(255).decode('ascii') | |
204
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
122 entries[name] = PBG3Entry(unknown1, unknown2, checksum, offset, size) |
0 | 123 |
124 return PBG3(entries, bitstream) | |
125 | |
126 | |
127 def list_files(self): | |
204
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
128 """List files present in the archive.""" |
0 | 129 return self.entries.keys() |
130 | |
131 | |
536
6b76c9ba3975
Make archives return files by default, instead of bytes.
Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
parents:
490
diff
changeset
|
132 def get_file(self, filename, check=False): |
204
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
133 """Extract a given file. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
134 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
135 If “filename” is in the archive, extract it and return its contents. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
136 Otherwise, raise an exception. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
137 |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
138 By default, the checksum of the file won't be verified, |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
139 you can however force the verification using the “check” argument. |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
140 """ |
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
141 |
0 | 142 unkwn1, unkwn2, checksum, offset, size = self.entries[filename] |
143 self.bitstream.seek(offset) | |
144 data = lzss.decompress(self.bitstream, size) | |
145 if check: | |
204
88361534c77e
Add some documentation (argh, so much left to document!)
Thibaut Girka <thib@sitedethib.com>
parents:
97
diff
changeset
|
146 # Verify the checksum |
0 | 147 compressed_size = self.bitstream.io.tell() - offset |
148 self.bitstream.seek(offset) | |
149 value = 0 | |
150 for c in self.bitstream.io.read(compressed_size): | |
151 value += ord(c) | |
152 value &= 0xFFFFFFFF | |
153 if value != checksum: | |
58 | 154 logger.warn('corrupted data!') |
536
6b76c9ba3975
Make archives return files by default, instead of bytes.
Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
parents:
490
diff
changeset
|
155 return BytesIO(data) |