annotate 2.00/zipfile27.py @ 31:fe1463e7e24d

Clean up try-except clauses More try-except-else goodness (with the stress on else). Bug fix: compat.py now creates zip_longest in Python 2.5 instead of raising an exception.
author Oleg Oshmyan <chortos@inbox.lv>
date Thu, 25 Nov 2010 00:03:29 +0000
parents a8cc383b787c
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
2 Read and write ZIP files.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
3 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
4 # Improved by Chortos-2 in 2010 (added bzip2 support)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
5 import struct, os, time, sys, shutil
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
6 import binascii, cStringIO, stat
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
7 import io
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
8 import re
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
9
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
10 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
11 import zlib # We may need its compression method
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
12 crc32 = zlib.crc32
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
13 except ImportError:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
14 zlib = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
15 crc32 = binascii.crc32
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
16
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
17 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
18 import bz2 # We may need its compression method
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
19 except ImportError:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
20 bz2 = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
22 __all__ = ["BadZipfile", "error", "ZIP_STORED", "ZIP_DEFLATED", "is_zipfile",
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
23 "ZipInfo", "ZipFile", "PyZipFile", "LargeZipFile", "ZIP_BZIP2" ]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
24
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
25 class BadZipfile(Exception):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
26 pass
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
27
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
28
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
29 class LargeZipFile(Exception):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
30 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
31 Raised when writing a zipfile, the zipfile requires ZIP64 extensions
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
32 and those extensions are disabled.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
33 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
34
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
35 error = BadZipfile # The exception raised by this module
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
36
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
37 ZIP64_LIMIT = (1 << 31) - 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
38 ZIP_FILECOUNT_LIMIT = 1 << 16
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
39 ZIP_MAX_COMMENT = (1 << 16) - 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
40
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
41 # constants for Zip file compression methods
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
42 ZIP_STORED = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
43 ZIP_DEFLATED = 8
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
44 ZIP_BZIP2 = 12
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
45 # Other ZIP compression methods not supported
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
46
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
47 # Below are some formats and associated data for reading/writing headers using
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
48 # the struct module. The names and structures of headers/records are those used
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
49 # in the PKWARE description of the ZIP file format:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
50 # http://www.pkware.com/documents/casestudies/APPNOTE.TXT
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
51 # (URL valid as of January 2008)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
52
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
53 # The "end of central directory" structure, magic number, size, and indices
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
54 # (section V.I in the format document)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
55 structEndArchive = "<4s4H2LH"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
56 stringEndArchive = "PK\005\006"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
57 sizeEndCentDir = struct.calcsize(structEndArchive)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
58
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
59 _ECD_SIGNATURE = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
60 _ECD_DISK_NUMBER = 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
61 _ECD_DISK_START = 2
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
62 _ECD_ENTRIES_THIS_DISK = 3
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
63 _ECD_ENTRIES_TOTAL = 4
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
64 _ECD_SIZE = 5
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
65 _ECD_OFFSET = 6
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
66 _ECD_COMMENT_SIZE = 7
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
67 # These last two indices are not part of the structure as defined in the
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
68 # spec, but they are used internally by this module as a convenience
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
69 _ECD_COMMENT = 8
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
70 _ECD_LOCATION = 9
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
71
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
72 # The "central directory" structure, magic number, size, and indices
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
73 # of entries in the structure (section V.F in the format document)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
74 structCentralDir = "<4s4B4HL2L5H2L"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
75 stringCentralDir = "PK\001\002"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
76 sizeCentralDir = struct.calcsize(structCentralDir)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
77
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
78 # indexes of entries in the central directory structure
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
79 _CD_SIGNATURE = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
80 _CD_CREATE_VERSION = 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
81 _CD_CREATE_SYSTEM = 2
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
82 _CD_EXTRACT_VERSION = 3
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
83 _CD_EXTRACT_SYSTEM = 4
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
84 _CD_FLAG_BITS = 5
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
85 _CD_COMPRESS_TYPE = 6
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
86 _CD_TIME = 7
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
87 _CD_DATE = 8
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
88 _CD_CRC = 9
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
89 _CD_COMPRESSED_SIZE = 10
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
90 _CD_UNCOMPRESSED_SIZE = 11
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
91 _CD_FILENAME_LENGTH = 12
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
92 _CD_EXTRA_FIELD_LENGTH = 13
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
93 _CD_COMMENT_LENGTH = 14
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
94 _CD_DISK_NUMBER_START = 15
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
95 _CD_INTERNAL_FILE_ATTRIBUTES = 16
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
96 _CD_EXTERNAL_FILE_ATTRIBUTES = 17
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
97 _CD_LOCAL_HEADER_OFFSET = 18
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
98
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
99 # The "local file header" structure, magic number, size, and indices
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
100 # (section V.A in the format document)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
101 structFileHeader = "<4s2B4HL2L2H"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
102 stringFileHeader = "PK\003\004"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
103 sizeFileHeader = struct.calcsize(structFileHeader)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
104
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
105 _FH_SIGNATURE = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
106 _FH_EXTRACT_VERSION = 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
107 _FH_EXTRACT_SYSTEM = 2
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
108 _FH_GENERAL_PURPOSE_FLAG_BITS = 3
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
109 _FH_COMPRESSION_METHOD = 4
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
110 _FH_LAST_MOD_TIME = 5
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
111 _FH_LAST_MOD_DATE = 6
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
112 _FH_CRC = 7
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
113 _FH_COMPRESSED_SIZE = 8
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
114 _FH_UNCOMPRESSED_SIZE = 9
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
115 _FH_FILENAME_LENGTH = 10
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
116 _FH_EXTRA_FIELD_LENGTH = 11
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
117
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
118 # The "Zip64 end of central directory locator" structure, magic number, and size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
119 structEndArchive64Locator = "<4sLQL"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
120 stringEndArchive64Locator = "PK\x06\x07"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
121 sizeEndCentDir64Locator = struct.calcsize(structEndArchive64Locator)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
122
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
123 # The "Zip64 end of central directory" record, magic number, size, and indices
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
124 # (section V.G in the format document)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
125 structEndArchive64 = "<4sQ2H2L4Q"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
126 stringEndArchive64 = "PK\x06\x06"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
127 sizeEndCentDir64 = struct.calcsize(structEndArchive64)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
128
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
129 _CD64_SIGNATURE = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
130 _CD64_DIRECTORY_RECSIZE = 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
131 _CD64_CREATE_VERSION = 2
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
132 _CD64_EXTRACT_VERSION = 3
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
133 _CD64_DISK_NUMBER = 4
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
134 _CD64_DISK_NUMBER_START = 5
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
135 _CD64_NUMBER_ENTRIES_THIS_DISK = 6
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
136 _CD64_NUMBER_ENTRIES_TOTAL = 7
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
137 _CD64_DIRECTORY_SIZE = 8
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
138 _CD64_OFFSET_START_CENTDIR = 9
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
139
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
140 def _check_zipfile(fp):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
141 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
142 if _EndRecData(fp):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
143 return True # file has correct magic number
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
144 except IOError:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
145 pass
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
146 return False
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
147
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
148 def is_zipfile(filename):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
149 """Quickly see if a file is a ZIP file by checking the magic number.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
150
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
151 The filename argument may be a file or file-like object too.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
152 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
153 result = False
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
154 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
155 if hasattr(filename, "read"):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
156 result = _check_zipfile(fp=filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
157 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
158 with open(filename, "rb") as fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
159 result = _check_zipfile(fp)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
160 except IOError:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
161 pass
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
162 return result
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
163
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
164 def _EndRecData64(fpin, offset, endrec):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
165 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
166 Read the ZIP64 end-of-archive records and use that to update endrec
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
167 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
168 fpin.seek(offset - sizeEndCentDir64Locator, 2)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
169 data = fpin.read(sizeEndCentDir64Locator)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
170 sig, diskno, reloff, disks = struct.unpack(structEndArchive64Locator, data)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
171 if sig != stringEndArchive64Locator:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
172 return endrec
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
173
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
174 if diskno != 0 or disks != 1:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
175 raise BadZipfile("zipfiles that span multiple disks are not supported")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
176
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
177 # Assume no 'zip64 extensible data'
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
178 fpin.seek(offset - sizeEndCentDir64Locator - sizeEndCentDir64, 2)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
179 data = fpin.read(sizeEndCentDir64)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
180 sig, sz, create_version, read_version, disk_num, disk_dir, \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
181 dircount, dircount2, dirsize, diroffset = \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
182 struct.unpack(structEndArchive64, data)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
183 if sig != stringEndArchive64:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
184 return endrec
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
185
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
186 # Update the original endrec using data from the ZIP64 record
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
187 endrec[_ECD_SIGNATURE] = sig
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
188 endrec[_ECD_DISK_NUMBER] = disk_num
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
189 endrec[_ECD_DISK_START] = disk_dir
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
190 endrec[_ECD_ENTRIES_THIS_DISK] = dircount
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
191 endrec[_ECD_ENTRIES_TOTAL] = dircount2
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
192 endrec[_ECD_SIZE] = dirsize
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
193 endrec[_ECD_OFFSET] = diroffset
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
194 return endrec
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
195
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
196
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
197 def _EndRecData(fpin):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
198 """Return data from the "End of Central Directory" record, or None.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
199
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
200 The data is a list of the nine items in the ZIP "End of central dir"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
201 record followed by a tenth item, the file seek offset of this record."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
202
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
203 # Determine file size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
204 fpin.seek(0, 2)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
205 filesize = fpin.tell()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
206
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
207 # Check to see if this is ZIP file with no archive comment (the
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
208 # "end of central directory" structure should be the last item in the
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
209 # file if this is the case).
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
210 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
211 fpin.seek(-sizeEndCentDir, 2)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
212 except IOError:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
213 return None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
214 data = fpin.read()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
215 if data[0:4] == stringEndArchive and data[-2:] == "\000\000":
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
216 # the signature is correct and there's no comment, unpack structure
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
217 endrec = struct.unpack(structEndArchive, data)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
218 endrec=list(endrec)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
219
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
220 # Append a blank comment and record start offset
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
221 endrec.append("")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
222 endrec.append(filesize - sizeEndCentDir)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
223
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
224 # Try to read the "Zip64 end of central directory" structure
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
225 return _EndRecData64(fpin, -sizeEndCentDir, endrec)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
226
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
227 # Either this is not a ZIP file, or it is a ZIP file with an archive
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
228 # comment. Search the end of the file for the "end of central directory"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
229 # record signature. The comment is the last item in the ZIP file and may be
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
230 # up to 64K long. It is assumed that the "end of central directory" magic
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
231 # number does not appear in the comment.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
232 maxCommentStart = max(filesize - (1 << 16) - sizeEndCentDir, 0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
233 fpin.seek(maxCommentStart, 0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
234 data = fpin.read()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
235 start = data.rfind(stringEndArchive)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
236 if start >= 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
237 # found the magic number; attempt to unpack and interpret
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
238 recData = data[start:start+sizeEndCentDir]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
239 endrec = list(struct.unpack(structEndArchive, recData))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
240 comment = data[start+sizeEndCentDir:]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
241 # check that comment length is correct
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
242 if endrec[_ECD_COMMENT_SIZE] == len(comment):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
243 # Append the archive comment and start offset
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
244 endrec.append(comment)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
245 endrec.append(maxCommentStart + start)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
246
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
247 # Try to read the "Zip64 end of central directory" structure
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
248 return _EndRecData64(fpin, maxCommentStart + start - filesize,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
249 endrec)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
250
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
251 # Unable to find a valid end of central directory structure
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
252 return
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
253
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
254
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
255 class ZipInfo (object):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
256 """Class with attributes describing each file in the ZIP archive."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
257
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
258 __slots__ = (
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
259 'orig_filename',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
260 'filename',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
261 'date_time',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
262 'compress_type',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
263 'comment',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
264 'extra',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
265 'create_system',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
266 'create_version',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
267 'extract_version',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
268 'reserved',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
269 'flag_bits',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
270 'volume',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
271 'internal_attr',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
272 'external_attr',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
273 'header_offset',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
274 'CRC',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
275 'compress_size',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
276 'file_size',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
277 '_raw_time',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
278 )
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
279
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
280 def __init__(self, filename="NoName", date_time=(1980,1,1,0,0,0)):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
281 self.orig_filename = filename # Original file name in archive
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
282
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
283 # Terminate the file name at the first null byte. Null bytes in file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
284 # names are used as tricks by viruses in archives.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
285 null_byte = filename.find(chr(0))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
286 if null_byte >= 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
287 filename = filename[0:null_byte]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
288 # This is used to ensure paths in generated ZIP files always use
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
289 # forward slashes as the directory separator, as required by the
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
290 # ZIP format specification.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
291 if os.sep != "/" and os.sep in filename:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
292 filename = filename.replace(os.sep, "/")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
293
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
294 self.filename = filename # Normalized file name
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
295 self.date_time = date_time # year, month, day, hour, min, sec
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
296 # Standard values:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
297 self.compress_type = ZIP_STORED # Type of compression for the file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
298 self.comment = "" # Comment for each file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
299 self.extra = "" # ZIP extra data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
300 if sys.platform == 'win32':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
301 self.create_system = 0 # System which created ZIP archive
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
302 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
303 # Assume everything else is unix-y
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
304 self.create_system = 3 # System which created ZIP archive
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
305 self.create_version = 20 # Version which created ZIP archive
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
306 self.extract_version = 20 # Version needed to extract archive
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
307 self.reserved = 0 # Must be zero
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
308 self.flag_bits = 0 # ZIP flag bits
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
309 self.volume = 0 # Volume number of file header
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
310 self.internal_attr = 0 # Internal attributes
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
311 self.external_attr = 0 # External file attributes
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
312 # Other attributes are set by class ZipFile:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
313 # header_offset Byte offset to the file header
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
314 # CRC CRC-32 of the uncompressed file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
315 # compress_size Size of the compressed file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
316 # file_size Size of the uncompressed file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
317
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
318 def FileHeader(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
319 """Return the per-file header as a string."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
320 dt = self.date_time
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
321 dosdate = (dt[0] - 1980) << 9 | dt[1] << 5 | dt[2]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
322 dostime = dt[3] << 11 | dt[4] << 5 | (dt[5] // 2)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
323 if self.flag_bits & 0x08:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
324 # Set these to zero because we write them after the file data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
325 CRC = compress_size = file_size = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
326 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
327 CRC = self.CRC
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
328 compress_size = self.compress_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
329 file_size = self.file_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
330
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
331 extra = self.extra
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
332
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
333 if file_size > ZIP64_LIMIT or compress_size > ZIP64_LIMIT:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
334 # File is larger than what fits into a 4 byte integer,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
335 # fall back to the ZIP64 extension
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
336 fmt = '<HHQQ'
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
337 extra = extra + struct.pack(fmt,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
338 1, struct.calcsize(fmt)-4, file_size, compress_size)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
339 file_size = 0xffffffff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
340 compress_size = 0xffffffff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
341 self.extract_version = max(45, self.extract_version)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
342 self.create_version = max(45, self.extract_version)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
343
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
344 filename, flag_bits = self._encodeFilenameFlags()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
345 header = struct.pack(structFileHeader, stringFileHeader,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
346 self.extract_version, self.reserved, flag_bits,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
347 self.compress_type, dostime, dosdate, CRC,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
348 compress_size, file_size,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
349 len(filename), len(extra))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
350 return header + filename + extra
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
351
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
352 def _encodeFilenameFlags(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
353 if isinstance(self.filename, unicode):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
354 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
355 return self.filename.encode('ascii'), self.flag_bits
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
356 except UnicodeEncodeError:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
357 return self.filename.encode('utf-8'), self.flag_bits | 0x800
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
358 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
359 return self.filename, self.flag_bits
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
360
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
361 def _decodeFilename(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
362 if self.flag_bits & 0x800:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
363 return self.filename.decode('utf-8')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
364 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
365 return self.filename
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
366
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
367 def _decodeExtra(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
368 # Try to decode the extra field.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
369 extra = self.extra
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
370 unpack = struct.unpack
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
371 while extra:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
372 tp, ln = unpack('<HH', extra[:4])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
373 if tp == 1:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
374 if ln >= 24:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
375 counts = unpack('<QQQ', extra[4:28])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
376 elif ln == 16:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
377 counts = unpack('<QQ', extra[4:20])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
378 elif ln == 8:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
379 counts = unpack('<Q', extra[4:12])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
380 elif ln == 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
381 counts = ()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
382 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
383 raise RuntimeError, "Corrupt extra field %s"%(ln,)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
384
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
385 idx = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
386
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
387 # ZIP64 extension (large files and/or large archives)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
388 if self.file_size in (0xffffffffffffffffL, 0xffffffffL):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
389 self.file_size = counts[idx]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
390 idx += 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
391
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
392 if self.compress_size == 0xFFFFFFFFL:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
393 self.compress_size = counts[idx]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
394 idx += 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
395
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
396 if self.header_offset == 0xffffffffL:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
397 old = self.header_offset
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
398 self.header_offset = counts[idx]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
399 idx+=1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
400
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
401 extra = extra[ln+4:]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
402
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
403
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
404 class _ZipDecrypter:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
405 """Class to handle decryption of files stored within a ZIP archive.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
406
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
407 ZIP supports a password-based form of encryption. Even though known
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
408 plaintext attacks have been found against it, it is still useful
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
409 to be able to get data out of such a file.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
410
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
411 Usage:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
412 zd = _ZipDecrypter(mypwd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
413 plain_char = zd(cypher_char)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
414 plain_text = map(zd, cypher_text)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
415 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
416
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
417 def _GenerateCRCTable():
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
418 """Generate a CRC-32 table.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
419
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
420 ZIP encryption uses the CRC32 one-byte primitive for scrambling some
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
421 internal keys. We noticed that a direct implementation is faster than
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
422 relying on binascii.crc32().
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
423 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
424 poly = 0xedb88320
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
425 table = [0] * 256
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
426 for i in range(256):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
427 crc = i
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
428 for j in range(8):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
429 if crc & 1:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
430 crc = ((crc >> 1) & 0x7FFFFFFF) ^ poly
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
431 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
432 crc = ((crc >> 1) & 0x7FFFFFFF)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
433 table[i] = crc
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
434 return table
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
435 crctable = _GenerateCRCTable()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
436
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
437 def _crc32(self, ch, crc):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
438 """Compute the CRC32 primitive on one byte."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
439 return ((crc >> 8) & 0xffffff) ^ self.crctable[(crc ^ ord(ch)) & 0xff]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
440
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
441 def __init__(self, pwd):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
442 self.key0 = 305419896
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
443 self.key1 = 591751049
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
444 self.key2 = 878082192
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
445 for p in pwd:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
446 self._UpdateKeys(p)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
447
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
448 def _UpdateKeys(self, c):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
449 self.key0 = self._crc32(c, self.key0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
450 self.key1 = (self.key1 + (self.key0 & 255)) & 4294967295
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
451 self.key1 = (self.key1 * 134775813 + 1) & 4294967295
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
452 self.key2 = self._crc32(chr((self.key1 >> 24) & 255), self.key2)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
453
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
454 def __call__(self, c):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
455 """Decrypt a single character."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
456 c = ord(c)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
457 k = self.key2 | 2
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
458 c = c ^ (((k * (k^1)) >> 8) & 255)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
459 c = chr(c)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
460 self._UpdateKeys(c)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
461 return c
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
462
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
463 class ZipExtFile(io.BufferedIOBase):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
464 """File-like object for reading an archive member.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
465 Is returned by ZipFile.open().
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
466 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
467
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
468 # Max size supported by decompressor.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
469 MAX_N = 1 << 31 - 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
470
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
471 # Read from compressed files in 4k blocks.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
472 MIN_READ_SIZE = 4096
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
473
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
474 # Search for universal newlines or line chunks.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
475 PATTERN = re.compile(r'^(?P<chunk>[^\r\n]+)|(?P<newline>\n|\r\n?)')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
476
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
477 def __init__(self, fileobj, mode, zipinfo, decrypter=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
478 self._fileobj = fileobj
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
479 self._decrypter = decrypter
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
480
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
481 self._compress_type = zipinfo.compress_type
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
482 self._compress_size = zipinfo.compress_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
483 self._compress_left = zipinfo.compress_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
484
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
485 if self._compress_type == ZIP_DEFLATED:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
486 self._decompressor = zlib.decompressobj(-15)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
487 elif self._compress_type == ZIP_BZIP2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
488 self._decompressor = bz2.BZ2Decompressor()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
489 self.MIN_READ_SIZE = 900000
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
490 self._unconsumed = ''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
491
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
492 self._readbuffer = ''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
493 self._offset = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
494
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
495 self._universal = 'U' in mode
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
496 self.newlines = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
497
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
498 # Adjust read size for encrypted files since the first 12 bytes
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
499 # are for the encryption/password information.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
500 if self._decrypter is not None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
501 self._compress_left -= 12
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
502
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
503 self.mode = mode
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
504 self.name = zipinfo.filename
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
505
29
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
506 if hasattr(zipinfo, 'CRC'):
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
507 self._expected_crc = zipinfo.CRC
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
508 self._running_crc = crc32(b'') & 0xffffffff
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
509 else:
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
510 self._expected_crc = None
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
511
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
512 def readline(self, limit=-1):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
513 """Read and return a line from the stream.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
514
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
515 If limit is specified, at most limit bytes will be read.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
516 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
517
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
518 if not self._universal and limit < 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
519 # Shortcut common case - newline found in buffer.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
520 i = self._readbuffer.find('\n', self._offset) + 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
521 if i > 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
522 line = self._readbuffer[self._offset: i]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
523 self._offset = i
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
524 return line
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
525
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
526 if not self._universal:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
527 return io.BufferedIOBase.readline(self, limit)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
528
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
529 line = ''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
530 while limit < 0 or len(line) < limit:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
531 readahead = self.peek(2)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
532 if readahead == '':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
533 return line
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
534
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
535 #
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
536 # Search for universal newlines or line chunks.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
537 #
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
538 # The pattern returns either a line chunk or a newline, but not
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
539 # both. Combined with peek(2), we are assured that the sequence
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
540 # '\r\n' is always retrieved completely and never split into
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
541 # separate newlines - '\r', '\n' due to coincidental readaheads.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
542 #
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
543 match = self.PATTERN.search(readahead)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
544 newline = match.group('newline')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
545 if newline is not None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
546 if self.newlines is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
547 self.newlines = []
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
548 if newline not in self.newlines:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
549 self.newlines.append(newline)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
550 self._offset += len(newline)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
551 return line + '\n'
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
552
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
553 chunk = match.group('chunk')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
554 if limit >= 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
555 chunk = chunk[: limit - len(line)]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
556
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
557 self._offset += len(chunk)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
558 line += chunk
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
559
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
560 return line
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
561
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
562 def peek(self, n=1):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
563 """Returns buffered bytes without advancing the position."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
564 if n > len(self._readbuffer) - self._offset:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
565 chunk = self.read(n)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
566 self._offset -= len(chunk)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
567
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
568 # Return up to 512 bytes to reduce allocation overhead for tight loops.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
569 return self._readbuffer[self._offset: self._offset + 512]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
570
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
571 def readable(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
572 return True
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
573
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
574 def read(self, n=-1):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
575 """Read and return up to n bytes.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
576 If the argument is omitted, None, or negative, data is read and returned until EOF is reached..
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
577 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
578 buf = ''
29
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
579 if n is None:
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
580 n = -1
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
581 while True:
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
582 if n < 0:
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
583 data = self.read1(n)
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
584 elif n > len(buf):
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
585 data = self.read1(n - len(buf))
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
586 else:
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
587 return buf
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
588 if len(data) == 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
589 return buf
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
590 buf += data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
591
29
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
592 def _update_crc(self, newdata, eof):
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
593 # Update the CRC using the given data.
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
594 if self._expected_crc is None:
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
595 # No need to compute the CRC if we don't have a reference value
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
596 return
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
597 self._running_crc = crc32(newdata, self._running_crc) & 0xffffffff
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
598 # Check the CRC if we're at the end of the file
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
599 if eof and self._running_crc != self._expected_crc:
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
600 raise BadZipfile("Bad CRC-32 for file %r" % self.name)
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
601
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
602 def read1(self, n):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
603 """Read up to n bytes with at most one read() system call."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
604
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
605 # Simplify algorithm (branching) by transforming negative n to large n.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
606 if n < 0 or n is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
607 n = self.MAX_N
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
608
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
609 # Bytes available in read buffer.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
610 len_readbuffer = len(self._readbuffer) - self._offset
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
611
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
612 # Read from file.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
613 if self._compress_left > 0 and n > len_readbuffer + len(self._unconsumed):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
614 nbytes = n - len_readbuffer - len(self._unconsumed)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
615 nbytes = max(nbytes, self.MIN_READ_SIZE)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
616 nbytes = min(nbytes, self._compress_left)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
617
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
618 data = self._fileobj.read(nbytes)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
619 self._compress_left -= len(data)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
620
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
621 if data and self._decrypter is not None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
622 data = ''.join(map(self._decrypter, data))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
623
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
624 if self._compress_type == ZIP_STORED:
29
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
625 self._update_crc(data, eof=(self._compress_left==0))
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
626 self._readbuffer = self._readbuffer[self._offset:] + data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
627 self._offset = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
628 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
629 # Prepare deflated bytes for decompression.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
630 self._unconsumed += data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
631
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
632 # Handle unconsumed data.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
633 if (len(self._unconsumed) > 0 and n > len_readbuffer and
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
634 self._compress_type == ZIP_DEFLATED):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
635 data = self._decompressor.decompress(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
636 self._unconsumed,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
637 max(n - len_readbuffer, self.MIN_READ_SIZE)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
638 )
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
639
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
640 self._unconsumed = self._decompressor.unconsumed_tail
29
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
641 eof = len(self._unconsumed) == 0 and self._compress_left == 0
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
642 if eof:
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
643 data += self._decompressor.flush()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
644
29
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
645 self._update_crc(data, eof=eof)
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
646 self._readbuffer = self._readbuffer[self._offset:] + data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
647 self._offset = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
648 elif (len(self._unconsumed) > 0 and n > len_readbuffer and
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
649 self._compress_type == ZIP_BZIP2):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
650 data = self._decompressor.decompress(self._unconsumed)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
651
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
652 self._unconsumed = ''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
653 self._readbuffer = self._readbuffer[self._offset:] + data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
654 self._offset = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
655
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
656 # Read from buffer.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
657 data = self._readbuffer[self._offset: self._offset + n]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
658 self._offset += len(data)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
659 return data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
660
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
661
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
662
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
663 class ZipFile:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
664 """ Class with methods to open, read, write, close, list zip files.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
665
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
666 z = ZipFile(file, mode="r", compression=ZIP_STORED, allowZip64=False)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
667
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
668 file: Either the path to the file, or a file-like object.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
669 If it is a path, the file will be opened and closed by ZipFile.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
670 mode: The mode can be either read "r", write "w" or append "a".
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
671 compression: ZIP_STORED (no compression), ZIP_DEFLATED (requires zlib),
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
672 or ZIP_BZIP2 (requires bz2).
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
673 allowZip64: if True ZipFile will create files with ZIP64 extensions when
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
674 needed, otherwise it will raise an exception when this would
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
675 be necessary.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
676
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
677 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
678
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
679 fp = None # Set here since __del__ checks it
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
680
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
681 def __init__(self, file, mode="r", compression=ZIP_STORED, allowZip64=False):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
682 """Open the ZIP file with mode read "r", write "w" or append "a"."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
683 if mode not in ("r", "w", "a"):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
684 raise RuntimeError('ZipFile() requires mode "r", "w", or "a"')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
685
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
686 if compression == ZIP_STORED:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
687 pass
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
688 elif compression == ZIP_DEFLATED:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
689 if not zlib:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
690 raise RuntimeError,\
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
691 "Compression requires the (missing) zlib module"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
692 elif compression == ZIP_BZIP2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
693 if not bz2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
694 raise RuntimeError,\
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
695 "Compression requires the (missing) bz2 module"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
696 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
697 raise RuntimeError, "That compression method is not supported"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
698
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
699 self._allowZip64 = allowZip64
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
700 self._didModify = False
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
701 self.debug = 0 # Level of printing: 0 through 3
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
702 self.NameToInfo = {} # Find file info given name
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
703 self.filelist = [] # List of ZipInfo instances for archive
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
704 self.compression = compression # Method of compression
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
705 self.mode = key = mode.replace('b', '')[0]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
706 self.pwd = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
707 self.comment = ''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
708
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
709 # Check if we were passed a file-like object
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
710 if isinstance(file, basestring):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
711 self._filePassed = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
712 self.filename = file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
713 modeDict = {'r' : 'rb', 'w': 'wb', 'a' : 'r+b'}
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
714 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
715 self.fp = open(file, modeDict[mode])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
716 except IOError:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
717 if mode == 'a':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
718 mode = key = 'w'
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
719 self.fp = open(file, modeDict[mode])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
720 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
721 raise
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
722 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
723 self._filePassed = 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
724 self.fp = file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
725 self.filename = getattr(file, 'name', None)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
726
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
727 if key == 'r':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
728 self._GetContents()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
729 elif key == 'w':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
730 pass
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
731 elif key == 'a':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
732 try: # See if file is a zip file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
733 self._RealGetContents()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
734 # seek to start of directory and overwrite
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
735 self.fp.seek(self.start_dir, 0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
736 except BadZipfile: # file is not a zip file, just append
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
737 self.fp.seek(0, 2)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
738 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
739 if not self._filePassed:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
740 self.fp.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
741 self.fp = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
742 raise RuntimeError, 'Mode must be "r", "w" or "a"'
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
743
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
744 def __enter__(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
745 return self
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
746
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
747 def __exit__(self, type, value, traceback):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
748 self.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
749
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
750 def _GetContents(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
751 """Read the directory, making sure we close the file if the format
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
752 is bad."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
753 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
754 self._RealGetContents()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
755 except BadZipfile:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
756 if not self._filePassed:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
757 self.fp.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
758 self.fp = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
759 raise
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
760
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
761 def _RealGetContents(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
762 """Read in the table of contents for the ZIP file."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
763 fp = self.fp
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
764 endrec = _EndRecData(fp)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
765 if not endrec:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
766 raise BadZipfile, "File is not a zip file"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
767 if self.debug > 1:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
768 print endrec
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
769 size_cd = endrec[_ECD_SIZE] # bytes in central directory
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
770 offset_cd = endrec[_ECD_OFFSET] # offset of central directory
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
771 self.comment = endrec[_ECD_COMMENT] # archive comment
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
772
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
773 # "concat" is zero, unless zip was concatenated to another file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
774 concat = endrec[_ECD_LOCATION] - size_cd - offset_cd
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
775 if endrec[_ECD_SIGNATURE] == stringEndArchive64:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
776 # If Zip64 extension structures are present, account for them
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
777 concat -= (sizeEndCentDir64 + sizeEndCentDir64Locator)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
778
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
779 if self.debug > 2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
780 inferred = concat + offset_cd
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
781 print "given, inferred, offset", offset_cd, inferred, concat
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
782 # self.start_dir: Position of start of central directory
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
783 self.start_dir = offset_cd + concat
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
784 fp.seek(self.start_dir, 0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
785 data = fp.read(size_cd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
786 fp = cStringIO.StringIO(data)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
787 total = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
788 while total < size_cd:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
789 centdir = fp.read(sizeCentralDir)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
790 if centdir[0:4] != stringCentralDir:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
791 raise BadZipfile, "Bad magic number for central directory"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
792 centdir = struct.unpack(structCentralDir, centdir)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
793 if self.debug > 2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
794 print centdir
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
795 filename = fp.read(centdir[_CD_FILENAME_LENGTH])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
796 # Create ZipInfo instance to store file information
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
797 x = ZipInfo(filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
798 x.extra = fp.read(centdir[_CD_EXTRA_FIELD_LENGTH])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
799 x.comment = fp.read(centdir[_CD_COMMENT_LENGTH])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
800 x.header_offset = centdir[_CD_LOCAL_HEADER_OFFSET]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
801 (x.create_version, x.create_system, x.extract_version, x.reserved,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
802 x.flag_bits, x.compress_type, t, d,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
803 x.CRC, x.compress_size, x.file_size) = centdir[1:12]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
804 x.volume, x.internal_attr, x.external_attr = centdir[15:18]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
805 # Convert date/time code to (year, month, day, hour, min, sec)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
806 x._raw_time = t
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
807 x.date_time = ( (d>>9)+1980, (d>>5)&0xF, d&0x1F,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
808 t>>11, (t>>5)&0x3F, (t&0x1F) * 2 )
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
809
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
810 x._decodeExtra()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
811 x.header_offset = x.header_offset + concat
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
812 x.filename = x._decodeFilename()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
813 self.filelist.append(x)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
814 self.NameToInfo[x.filename] = x
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
815
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
816 # update total bytes read from central directory
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
817 total = (total + sizeCentralDir + centdir[_CD_FILENAME_LENGTH]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
818 + centdir[_CD_EXTRA_FIELD_LENGTH]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
819 + centdir[_CD_COMMENT_LENGTH])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
820
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
821 if self.debug > 2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
822 print "total", total
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
823
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
824
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
825 def namelist(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
826 """Return a list of file names in the archive."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
827 l = []
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
828 for data in self.filelist:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
829 l.append(data.filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
830 return l
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
831
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
832 def infolist(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
833 """Return a list of class ZipInfo instances for files in the
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
834 archive."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
835 return self.filelist
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
836
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
837 def printdir(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
838 """Print a table of contents for the zip file."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
839 print "%-46s %19s %12s" % ("File Name", "Modified ", "Size")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
840 for zinfo in self.filelist:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
841 date = "%d-%02d-%02d %02d:%02d:%02d" % zinfo.date_time[:6]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
842 print "%-46s %s %12d" % (zinfo.filename, date, zinfo.file_size)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
843
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
844 def testzip(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
845 """Read all the files and check the CRC."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
846 chunk_size = 2 ** 20
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
847 for zinfo in self.filelist:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
848 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
849 # Read by chunks, to avoid an OverflowError or a
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
850 # MemoryError with very large embedded files.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
851 f = self.open(zinfo.filename, "r")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
852 while f.read(chunk_size): # Check CRC-32
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
853 pass
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
854 except BadZipfile:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
855 return zinfo.filename
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
856
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
857 def getinfo(self, name):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
858 """Return the instance of ZipInfo given 'name'."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
859 info = self.NameToInfo.get(name)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
860 if info is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
861 raise KeyError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
862 'There is no item named %r in the archive' % name)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
863
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
864 return info
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
865
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
866 def setpassword(self, pwd):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
867 """Set default password for encrypted files."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
868 self.pwd = pwd
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
869
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
870 def read(self, name, pwd=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
871 """Return file bytes (as a string) for name."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
872 return self.open(name, "r", pwd).read()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
873
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
874 def open(self, name, mode="r", pwd=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
875 """Return file-like object for 'name'."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
876 if mode not in ("r", "U", "rU"):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
877 raise RuntimeError, 'open() requires mode "r", "U", or "rU"'
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
878 if not self.fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
879 raise RuntimeError, \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
880 "Attempt to read ZIP archive that was already closed"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
881
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
882 # Only open a new file for instances where we were not
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
883 # given a file object in the constructor
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
884 if self._filePassed:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
885 zef_file = self.fp
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
886 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
887 zef_file = open(self.filename, 'rb')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
888
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
889 # Make sure we have an info object
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
890 if isinstance(name, ZipInfo):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
891 # 'name' is already an info object
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
892 zinfo = name
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
893 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
894 # Get info object for name
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
895 zinfo = self.getinfo(name)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
896
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
897 zef_file.seek(zinfo.header_offset, 0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
898
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
899 # Skip the file header:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
900 fheader = zef_file.read(sizeFileHeader)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
901 if fheader[0:4] != stringFileHeader:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
902 raise BadZipfile, "Bad magic number for file header"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
903
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
904 fheader = struct.unpack(structFileHeader, fheader)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
905 fname = zef_file.read(fheader[_FH_FILENAME_LENGTH])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
906 if fheader[_FH_EXTRA_FIELD_LENGTH]:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
907 zef_file.read(fheader[_FH_EXTRA_FIELD_LENGTH])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
908
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
909 if fname != zinfo.orig_filename:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
910 raise BadZipfile, \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
911 'File name in directory "%s" and header "%s" differ.' % (
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
912 zinfo.orig_filename, fname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
913
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
914 # check for encrypted flag & handle password
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
915 is_encrypted = zinfo.flag_bits & 0x1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
916 zd = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
917 if is_encrypted:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
918 if not pwd:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
919 pwd = self.pwd
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
920 if not pwd:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
921 raise RuntimeError, "File %s is encrypted, " \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
922 "password required for extraction" % name
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
923
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
924 zd = _ZipDecrypter(pwd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
925 # The first 12 bytes in the cypher stream is an encryption header
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
926 # used to strengthen the algorithm. The first 11 bytes are
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
927 # completely random, while the 12th contains the MSB of the CRC,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
928 # or the MSB of the file time depending on the header type
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
929 # and is used to check the correctness of the password.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
930 bytes = zef_file.read(12)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
931 h = map(zd, bytes[0:12])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
932 if zinfo.flag_bits & 0x8:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
933 # compare against the file type from extended local headers
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
934 check_byte = (zinfo._raw_time >> 8) & 0xff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
935 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
936 # compare against the CRC otherwise
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
937 check_byte = (zinfo.CRC >> 24) & 0xff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
938 if ord(h[11]) != check_byte:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
939 raise RuntimeError("Bad password for file", name)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
940
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
941 return ZipExtFile(zef_file, mode, zinfo, zd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
942
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
943 def extract(self, member, path=None, pwd=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
944 """Extract a member from the archive to the current working directory,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
945 using its full name. Its file information is extracted as accurately
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
946 as possible. `member' may be a filename or a ZipInfo object. You can
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
947 specify a different directory using `path'.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
948 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
949 if not isinstance(member, ZipInfo):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
950 member = self.getinfo(member)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
951
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
952 if path is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
953 path = os.getcwd()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
954
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
955 return self._extract_member(member, path, pwd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
956
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
957 def extractall(self, path=None, members=None, pwd=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
958 """Extract all members from the archive to the current working
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
959 directory. `path' specifies a different directory to extract to.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
960 `members' is optional and must be a subset of the list returned
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
961 by namelist().
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
962 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
963 if members is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
964 members = self.namelist()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
965
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
966 for zipinfo in members:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
967 self.extract(zipinfo, path, pwd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
968
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
969 def _extract_member(self, member, targetpath, pwd):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
970 """Extract the ZipInfo object 'member' to a physical
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
971 file on the path targetpath.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
972 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
973 # build the destination pathname, replacing
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
974 # forward slashes to platform specific separators.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
975 # Strip trailing path separator, unless it represents the root.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
976 if (targetpath[-1:] in (os.path.sep, os.path.altsep)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
977 and len(os.path.splitdrive(targetpath)[1]) > 1):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
978 targetpath = targetpath[:-1]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
979
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
980 # don't include leading "/" from file name if present
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
981 if member.filename[0] == '/':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
982 targetpath = os.path.join(targetpath, member.filename[1:])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
983 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
984 targetpath = os.path.join(targetpath, member.filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
985
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
986 targetpath = os.path.normpath(targetpath)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
987
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
988 # Create all upper directories if necessary.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
989 upperdirs = os.path.dirname(targetpath)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
990 if upperdirs and not os.path.exists(upperdirs):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
991 os.makedirs(upperdirs)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
992
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
993 if member.filename[-1] == '/':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
994 if not os.path.isdir(targetpath):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
995 os.mkdir(targetpath)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
996 return targetpath
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
997
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
998 source = self.open(member, pwd=pwd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
999 target = file(targetpath, "wb")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1000 shutil.copyfileobj(source, target)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1001 source.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1002 target.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1003
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1004 return targetpath
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1005
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1006 def _writecheck(self, zinfo):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1007 """Check for errors before writing a file to the archive."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1008 if zinfo.filename in self.NameToInfo:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1009 if self.debug: # Warning for duplicate names
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1010 print "Duplicate name:", zinfo.filename
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1011 if self.mode not in ("w", "a"):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1012 raise RuntimeError, 'write() requires mode "w" or "a"'
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1013 if not self.fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1014 raise RuntimeError, \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1015 "Attempt to write ZIP archive that was already closed"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1016 if zinfo.compress_type == ZIP_DEFLATED and not zlib:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1017 raise RuntimeError, \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1018 "Compression requires the (missing) zlib module"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1019 if zinfo.compress_type == ZIP_BZIP2 and not bz2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1020 raise RuntimeError, \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1021 "Compression requires the (missing) bz2 module"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1022 if zinfo.compress_type not in (ZIP_STORED, ZIP_DEFLATED, ZIP_BZIP2):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1023 raise RuntimeError, \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1024 "That compression method is not supported"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1025 if zinfo.file_size > ZIP64_LIMIT:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1026 if not self._allowZip64:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1027 raise LargeZipFile("Filesize would require ZIP64 extensions")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1028 if zinfo.header_offset > ZIP64_LIMIT:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1029 if not self._allowZip64:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1030 raise LargeZipFile("Zipfile size would require ZIP64 extensions")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1031
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1032 def write(self, filename, arcname=None, compress_type=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1033 """Put the bytes from filename into the archive under the name
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1034 arcname."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1035 if not self.fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1036 raise RuntimeError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1037 "Attempt to write to ZIP archive that was already closed")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1038
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1039 st = os.stat(filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1040 isdir = stat.S_ISDIR(st.st_mode)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1041 mtime = time.localtime(st.st_mtime)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1042 date_time = mtime[0:6]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1043 # Create ZipInfo instance to store file information
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1044 if arcname is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1045 arcname = filename
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1046 arcname = os.path.normpath(os.path.splitdrive(arcname)[1])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1047 while arcname[0] in (os.sep, os.altsep):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1048 arcname = arcname[1:]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1049 if isdir:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1050 arcname += '/'
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1051 zinfo = ZipInfo(arcname, date_time)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1052 zinfo.external_attr = (st[0] & 0xFFFF) << 16L # Unix attributes
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1053 if compress_type is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1054 zinfo.compress_type = self.compression
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1055 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1056 zinfo.compress_type = compress_type
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1057
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1058 zinfo.file_size = st.st_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1059 zinfo.flag_bits = 0x00
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1060 zinfo.header_offset = self.fp.tell() # Start of header bytes
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1061
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1062 self._writecheck(zinfo)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1063 self._didModify = True
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1064
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1065 if isdir:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1066 zinfo.file_size = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1067 zinfo.compress_size = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1068 zinfo.CRC = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1069 self.filelist.append(zinfo)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1070 self.NameToInfo[zinfo.filename] = zinfo
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1071 self.fp.write(zinfo.FileHeader())
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1072 return
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1073
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1074 with open(filename, "rb") as fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1075 # Must overwrite CRC and sizes with correct data later
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1076 zinfo.CRC = CRC = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1077 zinfo.compress_size = compress_size = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1078 zinfo.file_size = file_size = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1079 self.fp.write(zinfo.FileHeader())
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1080 if zinfo.compress_type == ZIP_DEFLATED:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1081 cmpr = zlib.compressobj(zlib.Z_DEFAULT_COMPRESSION,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1082 zlib.DEFLATED, -15)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1083 elif zinfo.compress_type == ZIP_BZIP2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1084 cmpr = bz2.BZ2Compressor()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1085 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1086 cmpr = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1087 while 1:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1088 buf = fp.read(1024 * 8)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1089 if not buf:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1090 break
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1091 file_size = file_size + len(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1092 CRC = crc32(buf, CRC) & 0xffffffff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1093 if cmpr:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1094 buf = cmpr.compress(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1095 compress_size = compress_size + len(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1096 self.fp.write(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1097 if cmpr:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1098 buf = cmpr.flush()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1099 compress_size = compress_size + len(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1100 self.fp.write(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1101 zinfo.compress_size = compress_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1102 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1103 zinfo.compress_size = file_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1104 zinfo.CRC = CRC
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1105 zinfo.file_size = file_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1106 # Seek backwards and write CRC and file sizes
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1107 position = self.fp.tell() # Preserve current position in file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1108 self.fp.seek(zinfo.header_offset + 14, 0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1109 self.fp.write(struct.pack("<LLL", zinfo.CRC, zinfo.compress_size,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1110 zinfo.file_size))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1111 self.fp.seek(position, 0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1112 self.filelist.append(zinfo)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1113 self.NameToInfo[zinfo.filename] = zinfo
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1114
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1115 def writestr(self, zinfo_or_arcname, bytes, compress_type=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1116 """Write a file into the archive. The contents is the string
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1117 'bytes'. 'zinfo_or_arcname' is either a ZipInfo instance or
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1118 the name of the file in the archive."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1119 if not isinstance(zinfo_or_arcname, ZipInfo):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1120 zinfo = ZipInfo(filename=zinfo_or_arcname,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1121 date_time=time.localtime(time.time())[:6])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1122
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1123 zinfo.compress_type = self.compression
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1124 zinfo.external_attr = 0600 << 16
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1125 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1126 zinfo = zinfo_or_arcname
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1127
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1128 if not self.fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1129 raise RuntimeError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1130 "Attempt to write to ZIP archive that was already closed")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1131
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1132 if compress_type is not None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1133 zinfo.compress_type = compress_type
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1134
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1135 zinfo.file_size = len(bytes) # Uncompressed size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1136 zinfo.header_offset = self.fp.tell() # Start of header bytes
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1137 self._writecheck(zinfo)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1138 self._didModify = True
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1139 zinfo.CRC = crc32(bytes) & 0xffffffff # CRC-32 checksum
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1140 if zinfo.compress_type == ZIP_DEFLATED:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1141 co = zlib.compressobj(zlib.Z_DEFAULT_COMPRESSION,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1142 zlib.DEFLATED, -15)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1143 bytes = co.compress(bytes) + co.flush()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1144 zinfo.compress_size = len(bytes) # Compressed size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1145 elif zinfo.compress_type == ZIP_BZIP2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1146 co = bz2.BZ2Compressor()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1147 bytes = co.compress(bytes) + co.flush()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1148 zinfo.compress_size = len(bytes) # Compressed size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1149 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1150 zinfo.compress_size = zinfo.file_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1151 zinfo.header_offset = self.fp.tell() # Start of header bytes
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1152 self.fp.write(zinfo.FileHeader())
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1153 self.fp.write(bytes)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1154 self.fp.flush()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1155 if zinfo.flag_bits & 0x08:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1156 # Write CRC and file sizes after the file data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1157 self.fp.write(struct.pack("<LLL", zinfo.CRC, zinfo.compress_size,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1158 zinfo.file_size))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1159 self.filelist.append(zinfo)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1160 self.NameToInfo[zinfo.filename] = zinfo
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1161
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1162 def __del__(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1163 """Call the "close()" method in case the user forgot."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1164 self.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1165
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1166 def close(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1167 """Close the file, and for mode "w" and "a" write the ending
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1168 records."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1169 if self.fp is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1170 return
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1171
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1172 if self.mode in ("w", "a") and self._didModify: # write ending records
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1173 count = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1174 pos1 = self.fp.tell()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1175 for zinfo in self.filelist: # write central directory
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1176 count = count + 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1177 dt = zinfo.date_time
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1178 dosdate = (dt[0] - 1980) << 9 | dt[1] << 5 | dt[2]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1179 dostime = dt[3] << 11 | dt[4] << 5 | (dt[5] // 2)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1180 extra = []
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1181 if zinfo.file_size > ZIP64_LIMIT \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1182 or zinfo.compress_size > ZIP64_LIMIT:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1183 extra.append(zinfo.file_size)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1184 extra.append(zinfo.compress_size)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1185 file_size = 0xffffffff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1186 compress_size = 0xffffffff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1187 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1188 file_size = zinfo.file_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1189 compress_size = zinfo.compress_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1190
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1191 if zinfo.header_offset > ZIP64_LIMIT:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1192 extra.append(zinfo.header_offset)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1193 header_offset = 0xffffffffL
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1194 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1195 header_offset = zinfo.header_offset
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1196
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1197 extra_data = zinfo.extra
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1198 if extra:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1199 # Append a ZIP64 field to the extra's
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1200 extra_data = struct.pack(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1201 '<HH' + 'Q'*len(extra),
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1202 1, 8*len(extra), *extra) + extra_data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1203
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1204 extract_version = max(45, zinfo.extract_version)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1205 create_version = max(45, zinfo.create_version)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1206 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1207 extract_version = zinfo.extract_version
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1208 create_version = zinfo.create_version
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1209
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1210 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1211 filename, flag_bits = zinfo._encodeFilenameFlags()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1212 centdir = struct.pack(structCentralDir,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1213 stringCentralDir, create_version,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1214 zinfo.create_system, extract_version, zinfo.reserved,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1215 flag_bits, zinfo.compress_type, dostime, dosdate,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1216 zinfo.CRC, compress_size, file_size,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1217 len(filename), len(extra_data), len(zinfo.comment),
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1218 0, zinfo.internal_attr, zinfo.external_attr,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1219 header_offset)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1220 except DeprecationWarning:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1221 print >>sys.stderr, (structCentralDir,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1222 stringCentralDir, create_version,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1223 zinfo.create_system, extract_version, zinfo.reserved,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1224 zinfo.flag_bits, zinfo.compress_type, dostime, dosdate,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1225 zinfo.CRC, compress_size, file_size,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1226 len(zinfo.filename), len(extra_data), len(zinfo.comment),
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1227 0, zinfo.internal_attr, zinfo.external_attr,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1228 header_offset)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1229 raise
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1230 self.fp.write(centdir)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1231 self.fp.write(filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1232 self.fp.write(extra_data)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1233 self.fp.write(zinfo.comment)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1234
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1235 pos2 = self.fp.tell()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1236 # Write end-of-zip-archive record
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1237 centDirCount = count
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1238 centDirSize = pos2 - pos1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1239 centDirOffset = pos1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1240 if (centDirCount >= ZIP_FILECOUNT_LIMIT or
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1241 centDirOffset > ZIP64_LIMIT or
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1242 centDirSize > ZIP64_LIMIT):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1243 # Need to write the ZIP64 end-of-archive records
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1244 zip64endrec = struct.pack(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1245 structEndArchive64, stringEndArchive64,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1246 44, 45, 45, 0, 0, centDirCount, centDirCount,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1247 centDirSize, centDirOffset)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1248 self.fp.write(zip64endrec)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1249
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1250 zip64locrec = struct.pack(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1251 structEndArchive64Locator,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1252 stringEndArchive64Locator, 0, pos2, 1)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1253 self.fp.write(zip64locrec)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1254 centDirCount = min(centDirCount, 0xFFFF)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1255 centDirSize = min(centDirSize, 0xFFFFFFFF)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1256 centDirOffset = min(centDirOffset, 0xFFFFFFFF)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1257
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1258 # check for valid comment length
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1259 if len(self.comment) >= ZIP_MAX_COMMENT:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1260 if self.debug > 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1261 msg = 'Archive comment is too long; truncating to %d bytes' \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1262 % ZIP_MAX_COMMENT
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1263 self.comment = self.comment[:ZIP_MAX_COMMENT]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1264
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1265 endrec = struct.pack(structEndArchive, stringEndArchive,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1266 0, 0, centDirCount, centDirCount,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1267 centDirSize, centDirOffset, len(self.comment))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1268 self.fp.write(endrec)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1269 self.fp.write(self.comment)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1270 self.fp.flush()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1271
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1272 if not self._filePassed:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1273 self.fp.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1274 self.fp = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1275
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1276
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1277 class PyZipFile(ZipFile):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1278 """Class to create ZIP archives with Python library files and packages."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1279
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1280 def writepy(self, pathname, basename = ""):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1281 """Add all files from "pathname" to the ZIP archive.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1282
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1283 If pathname is a package directory, search the directory and
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1284 all package subdirectories recursively for all *.py and enter
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1285 the modules into the archive. If pathname is a plain
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1286 directory, listdir *.py and enter all modules. Else, pathname
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1287 must be a Python *.py file and the module will be put into the
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1288 archive. Added modules are always module.pyo or module.pyc.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1289 This method will compile the module.py into module.pyc if
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1290 necessary.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1291 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1292 dir, name = os.path.split(pathname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1293 if os.path.isdir(pathname):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1294 initname = os.path.join(pathname, "__init__.py")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1295 if os.path.isfile(initname):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1296 # This is a package directory, add it
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1297 if basename:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1298 basename = "%s/%s" % (basename, name)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1299 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1300 basename = name
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1301 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1302 print "Adding package in", pathname, "as", basename
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1303 fname, arcname = self._get_codename(initname[0:-3], basename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1304 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1305 print "Adding", arcname
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1306 self.write(fname, arcname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1307 dirlist = os.listdir(pathname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1308 dirlist.remove("__init__.py")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1309 # Add all *.py files and package subdirectories
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1310 for filename in dirlist:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1311 path = os.path.join(pathname, filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1312 root, ext = os.path.splitext(filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1313 if os.path.isdir(path):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1314 if os.path.isfile(os.path.join(path, "__init__.py")):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1315 # This is a package directory, add it
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1316 self.writepy(path, basename) # Recursive call
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1317 elif ext == ".py":
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1318 fname, arcname = self._get_codename(path[0:-3],
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1319 basename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1320 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1321 print "Adding", arcname
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1322 self.write(fname, arcname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1323 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1324 # This is NOT a package directory, add its files at top level
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1325 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1326 print "Adding files from directory", pathname
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1327 for filename in os.listdir(pathname):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1328 path = os.path.join(pathname, filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1329 root, ext = os.path.splitext(filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1330 if ext == ".py":
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1331 fname, arcname = self._get_codename(path[0:-3],
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1332 basename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1333 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1334 print "Adding", arcname
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1335 self.write(fname, arcname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1336 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1337 if pathname[-3:] != ".py":
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1338 raise RuntimeError, \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1339 'Files added with writepy() must end with ".py"'
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1340 fname, arcname = self._get_codename(pathname[0:-3], basename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1341 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1342 print "Adding file", arcname
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1343 self.write(fname, arcname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1344
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1345 def _get_codename(self, pathname, basename):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1346 """Return (filename, archivename) for the path.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1347
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1348 Given a module name path, return the correct file path and
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1349 archive name, compiling if necessary. For example, given
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1350 /python/lib/string, return (/python/lib/string.pyc, string).
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1351 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1352 file_py = pathname + ".py"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1353 file_pyc = pathname + ".pyc"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1354 file_pyo = pathname + ".pyo"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1355 if os.path.isfile(file_pyo) and \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1356 os.stat(file_pyo).st_mtime >= os.stat(file_py).st_mtime:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1357 fname = file_pyo # Use .pyo file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1358 elif not os.path.isfile(file_pyc) or \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1359 os.stat(file_pyc).st_mtime < os.stat(file_py).st_mtime:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1360 import py_compile
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1361 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1362 print "Compiling", file_py
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1363 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1364 py_compile.compile(file_py, file_pyc, None, True)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1365 except py_compile.PyCompileError,err:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1366 print err.msg
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1367 fname = file_pyc
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1368 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1369 fname = file_pyc
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1370 archivename = os.path.split(fname)[1]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1371 if basename:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1372 archivename = "%s/%s" % (basename, archivename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1373 return (fname, archivename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1374
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1375
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1376 def main(args = None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1377 import textwrap
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1378 USAGE=textwrap.dedent("""\
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1379 Usage:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1380 zipfile.py -l zipfile.zip # Show listing of a zipfile
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1381 zipfile.py -t zipfile.zip # Test if a zipfile is valid
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1382 zipfile.py -e zipfile.zip target # Extract zipfile into target dir
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1383 zipfile.py -c zipfile.zip src ... # Create zipfile from sources
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1384 """)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1385 if args is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1386 args = sys.argv[1:]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1387
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1388 if not args or args[0] not in ('-l', '-c', '-e', '-t'):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1389 print USAGE
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1390 sys.exit(1)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1391
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1392 if args[0] == '-l':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1393 if len(args) != 2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1394 print USAGE
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1395 sys.exit(1)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1396 zf = ZipFile(args[1], 'r')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1397 zf.printdir()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1398 zf.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1399
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1400 elif args[0] == '-t':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1401 if len(args) != 2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1402 print USAGE
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1403 sys.exit(1)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1404 zf = ZipFile(args[1], 'r')
29
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
1405 badfile = zf.testzip()
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
1406 if badfile:
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
1407 print("The following enclosed file is corrupted: {!r}".format(badfile))
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1408 print "Done testing"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1409
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1410 elif args[0] == '-e':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1411 if len(args) != 3:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1412 print USAGE
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1413 sys.exit(1)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1414
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1415 zf = ZipFile(args[1], 'r')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1416 out = args[2]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1417 for path in zf.namelist():
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1418 if path.startswith('./'):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1419 tgt = os.path.join(out, path[2:])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1420 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1421 tgt = os.path.join(out, path)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1422
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1423 tgtdir = os.path.dirname(tgt)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1424 if not os.path.exists(tgtdir):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1425 os.makedirs(tgtdir)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1426 with open(tgt, 'wb') as fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1427 fp.write(zf.read(path))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1428 zf.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1429
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1430 elif args[0] == '-c':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1431 if len(args) < 3:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1432 print USAGE
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1433 sys.exit(1)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1434
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1435 def addToZip(zf, path, zippath):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1436 if os.path.isfile(path):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1437 zf.write(path, zippath, ZIP_DEFLATED)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1438 elif os.path.isdir(path):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1439 for nm in os.listdir(path):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1440 addToZip(zf,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1441 os.path.join(path, nm), os.path.join(zippath, nm))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1442 # else: ignore
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1443
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1444 zf = ZipFile(args[1], 'w', allowZip64=True)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1445 for src in args[2:]:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1446 addToZip(zf, src, os.path.basename(src))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1447
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1448 zf.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1449
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1450 if __name__ == "__main__":
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1451 main()