annotate 2.00/zipfile31.py @ 29:a8cc383b787c

Clean up zipfiles and diff them to stock ones
author Oleg Oshmyan <chortos@inbox.lv>
date Wed, 24 Nov 2010 23:21:31 +0000
parents 2.00/zipfile3.py@3d535503161f
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 XXX references to utf-8 need further investigation.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
5 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
6 # Improved by Chortos-2 in 2010 (added bzip2 support)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
7 import struct, os, time, sys, shutil
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
8 import binascii, io, stat
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 = b"<4s4H2LH"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
56 stringEndArchive = b"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 = b"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 = b"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 = b"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 = b"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:] == b"\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(b"")
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 = b"" # Comment for each file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
299 self.extra = b"" # 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 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
354 return self.filename.encode('ascii'), self.flag_bits
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
355 except UnicodeEncodeError:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
356 return self.filename.encode('utf-8'), self.flag_bits | 0x800
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
357
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
358 def _decodeExtra(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
359 # Try to decode the extra field.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
360 extra = self.extra
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
361 unpack = struct.unpack
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
362 while extra:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
363 tp, ln = unpack('<HH', extra[:4])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
364 if tp == 1:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
365 if ln >= 24:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
366 counts = unpack('<QQQ', extra[4:28])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
367 elif ln == 16:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
368 counts = unpack('<QQ', extra[4:20])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
369 elif ln == 8:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
370 counts = unpack('<Q', extra[4:12])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
371 elif ln == 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
372 counts = ()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
373 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
374 raise RuntimeError("Corrupt extra field %s"%(ln,))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
375
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
376 idx = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
377
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
378 # ZIP64 extension (large files and/or large archives)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
379 if self.file_size in (0xffffffffffffffff, 0xffffffff):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
380 self.file_size = counts[idx]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
381 idx += 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
382
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
383 if self.compress_size == 0xFFFFFFFF:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
384 self.compress_size = counts[idx]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
385 idx += 1
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 if self.header_offset == 0xffffffff:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
388 old = self.header_offset
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
389 self.header_offset = 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 extra = extra[ln+4:]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
393
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
394
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
395 class _ZipDecrypter:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
396 """Class to handle decryption of files stored within a ZIP archive.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
397
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
398 ZIP supports a password-based form of encryption. Even though known
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
399 plaintext attacks have been found against it, it is still useful
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
400 to be able to get data out of such a file.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
401
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
402 Usage:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
403 zd = _ZipDecrypter(mypwd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
404 plain_char = zd(cypher_char)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
405 plain_text = map(zd, cypher_text)
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
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
408 def _GenerateCRCTable():
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
409 """Generate a CRC-32 table.
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 ZIP encryption uses the CRC32 one-byte primitive for scrambling some
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
412 internal keys. We noticed that a direct implementation is faster than
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
413 relying on binascii.crc32().
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
414 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
415 poly = 0xedb88320
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
416 table = [0] * 256
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
417 for i in range(256):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
418 crc = i
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
419 for j in range(8):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
420 if crc & 1:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
421 crc = ((crc >> 1) & 0x7FFFFFFF) ^ poly
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
422 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
423 crc = ((crc >> 1) & 0x7FFFFFFF)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
424 table[i] = crc
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
425 return table
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
426 crctable = _GenerateCRCTable()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
427
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
428 def _crc32(self, ch, crc):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
429 """Compute the CRC32 primitive on one byte."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
430 return ((crc >> 8) & 0xffffff) ^ self.crctable[(crc ^ ch) & 0xff]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
431
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
432 def __init__(self, pwd):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
433 self.key0 = 305419896
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
434 self.key1 = 591751049
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
435 self.key2 = 878082192
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
436 for p in pwd:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
437 self._UpdateKeys(p)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
438
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
439 def _UpdateKeys(self, c):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
440 self.key0 = self._crc32(c, self.key0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
441 self.key1 = (self.key1 + (self.key0 & 255)) & 4294967295
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
442 self.key1 = (self.key1 * 134775813 + 1) & 4294967295
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
443 self.key2 = self._crc32((self.key1 >> 24) & 255, self.key2)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
444
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
445 def __call__(self, c):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
446 """Decrypt a single character."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
447 assert isinstance(c, int)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
448 k = self.key2 | 2
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
449 c = c ^ (((k * (k^1)) >> 8) & 255)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
450 self._UpdateKeys(c)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
451 return c
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
452
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
453 class ZipExtFile:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
454 """File-like object for reading an archive member.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
455 Is returned by ZipFile.open().
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
456 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
457
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
458 def __init__(self, fileobj, zipinfo, decrypt=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
459 self.fileobj = fileobj
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
460 self.decrypter = decrypt
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
461 self.bytes_read = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
462 self.rawbuffer = b''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
463 self.readbuffer = b''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
464 self.linebuffer = b''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
465 self.eof = False
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
466 self.univ_newlines = False
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
467 self.nlSeps = (b"\n", )
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
468 self.lastdiscard = b''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
469
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
470 self.compress_type = zipinfo.compress_type
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
471 self.compress_size = zipinfo.compress_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
472
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
473 self.closed = False
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
474 self.mode = "r"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
475 self.name = zipinfo.filename
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 # read from compressed files in 64k blocks
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
478 self.compreadsize = 64*1024
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
479 if self.compress_type == ZIP_DEFLATED:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
480 self.dc = zlib.decompressobj(-15)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
481 elif self.compress_type == ZIP_BZIP2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
482 self.dc = bz2.BZ2Decompressor()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
483 self.compreadsize = 900000
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
484
29
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
485 if hasattr(zipinfo, 'CRC'):
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
486 self._expected_crc = zipinfo.CRC
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
487 self._running_crc = crc32(b'') & 0xffffffff
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
488 else:
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
489 self._expected_crc = None
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
490
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
491 def set_univ_newlines(self, univ_newlines):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
492 self.univ_newlines = univ_newlines
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
493
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
494 # pick line separator char(s) based on universal newlines flag
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
495 self.nlSeps = (b"\n", )
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
496 if self.univ_newlines:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
497 self.nlSeps = (b"\r\n", b"\r", b"\n")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
498
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
499 def __iter__(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
500 return self
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
501
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
502 def __next__(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
503 nextline = self.readline()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
504 if not nextline:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
505 raise StopIteration()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
506
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
507 return nextline
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
508
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
509 def close(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
510 self.closed = True
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
511
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
512 def _checkfornewline(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
513 nl, nllen = -1, -1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
514 if self.linebuffer:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
515 # ugly check for cases where half of an \r\n pair was
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
516 # read on the last pass, and the \r was discarded. In this
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
517 # case we just throw away the \n at the start of the buffer.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
518 if (self.lastdiscard, self.linebuffer[:1]) == (b'\r', b'\n'):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
519 self.linebuffer = self.linebuffer[1:]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
520
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
521 for sep in self.nlSeps:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
522 nl = self.linebuffer.find(sep)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
523 if nl >= 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
524 nllen = len(sep)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
525 return nl, nllen
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
526
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
527 return nl, nllen
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 def readline(self, size = -1):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
530 """Read a line with approx. size. If size is negative,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
531 read a whole line.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
532 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
533 if size < 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
534 size = sys.maxsize
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
535 elif size == 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
536 return b''
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 # check for a newline already in buffer
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
539 nl, nllen = self._checkfornewline()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
540
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
541 if nl >= 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
542 # the next line was already in the buffer
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
543 nl = min(nl, size)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
544 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
545 # no line break in buffer - try to read more
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
546 size -= len(self.linebuffer)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
547 while nl < 0 and size > 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
548 buf = self.read(min(size, 100))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
549 if not buf:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
550 break
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
551 self.linebuffer += buf
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
552 size -= len(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
553
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
554 # check for a newline in buffer
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
555 nl, nllen = self._checkfornewline()
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 # we either ran out of bytes in the file, or
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
558 # met the specified size limit without finding a newline,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
559 # so return current buffer
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
560 if nl < 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
561 s = self.linebuffer
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
562 self.linebuffer = b''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
563 return s
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
564
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
565 buf = self.linebuffer[:nl]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
566 self.lastdiscard = self.linebuffer[nl:nl + nllen]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
567 self.linebuffer = self.linebuffer[nl + nllen:]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
568
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
569 # line is always returned with \n as newline char (except possibly
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
570 # for a final incomplete line in the file, which is handled above).
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
571 return buf + b"\n"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
572
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
573 def readlines(self, sizehint = -1):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
574 """Return a list with all (following) lines. The sizehint parameter
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
575 is ignored in this implementation.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
576 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
577 result = []
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
578 while True:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
579 line = self.readline()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
580 if not line: break
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
581 result.append(line)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
582 return result
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
583
29
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
584 def _update_crc(self, newdata, eof):
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
585 # Update the CRC using the given data.
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
586 if self._expected_crc is None:
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
587 # 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: 28
diff changeset
588 return
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
589 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: 28
diff changeset
590 # 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: 28
diff changeset
591 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: 28
diff changeset
592 raise BadZipfile("Bad CRC-32 for file %r" % self.name)
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
593
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
594 def read(self, size = None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
595 # act like file obj and return empty string if size is 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
596 if size == 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
597 return b''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
598
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
599 # determine read size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
600 bytesToRead = self.compress_size - self.bytes_read
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 # adjust read size for encrypted files since the first 12 bytes
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
603 # are for the encryption/password information
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
604 if self.decrypter is not None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
605 bytesToRead -= 12
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
606
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
607 if size is not None and size >= 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
608 if self.compress_type == ZIP_STORED:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
609 lr = len(self.readbuffer)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
610 bytesToRead = min(bytesToRead, size - lr)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
611 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
612 if len(self.readbuffer) > size:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
613 # the user has requested fewer bytes than we've already
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
614 # pulled through the decompressor; don't read any more
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
615 bytesToRead = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
616 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
617 # user will use up the buffer, so read some more
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
618 lr = len(self.rawbuffer)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
619 bytesToRead = min(bytesToRead, self.compreadsize - lr)
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 # avoid reading past end of file contents
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
622 if bytesToRead + self.bytes_read > self.compress_size:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
623 bytesToRead = self.compress_size - self.bytes_read
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
624
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
625 # try to read from file (if necessary)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
626 if bytesToRead > 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
627 data = self.fileobj.read(bytesToRead)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
628 self.bytes_read += len(data)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
629 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
630 self.rawbuffer += data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
631 except:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
632 print(repr(self.fileobj), repr(self.rawbuffer),
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
633 repr(data))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
634 raise
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
635
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
636 # handle contents of raw buffer
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
637 if self.rawbuffer:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
638 newdata = self.rawbuffer
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
639 self.rawbuffer = b''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
640
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
641 # decrypt new data if we were given an object to handle that
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
642 if newdata and self.decrypter is not None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
643 newdata = bytes(map(self.decrypter, newdata))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
644
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
645 # decompress newly read data if necessary
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
646 if newdata and self.compress_type != ZIP_STORED:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
647 newdata = self.dc.decompress(newdata)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
648 self.rawbuffer = self.dc.unconsumed_tail if self.compress_type == ZIP_DEFLATED else ''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
649 if self.eof and len(self.rawbuffer) == 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
650 # we're out of raw bytes (both from the file and
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
651 # the local buffer); flush just to make sure the
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
652 # decompressor is done
28
3d535503161f hasattr -> try
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
653 try:
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
654 newdata += self.dc.flush()
28
3d535503161f hasattr -> try
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
655 except AttributeError:
3d535503161f hasattr -> try
Oleg Oshmyan <chortos@inbox.lv>
parents: 21
diff changeset
656 pass
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
657 # prevent decompressor from being used again
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
658 self.dc = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
659
29
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
660 self._update_crc(newdata, eof=(
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
661 self.compress_size == self.bytes_read and
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
662 len(self.rawbuffer) == 0))
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
663 self.readbuffer += newdata
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
664
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
665 # return what the user asked for
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
666 if size is None or len(self.readbuffer) <= size:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
667 data = self.readbuffer
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
668 self.readbuffer = b''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
669 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
670 data = self.readbuffer[:size]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
671 self.readbuffer = self.readbuffer[size:]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
672
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
673 return data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
674
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
675
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
676 class ZipFile:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
677 """ Class with methods to open, read, write, close, list zip files.
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 z = ZipFile(file, mode="r", compression=ZIP_STORED, allowZip64=False)
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 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
682 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
683 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
684 compression: ZIP_STORED (no compression), ZIP_DEFLATED (requires zlib),
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
685 or ZIP_BZIP2 (requires bz2).
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
686 allowZip64: if True ZipFile will create files with ZIP64 extensions when
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
687 needed, otherwise it will raise an exception when this would
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
688 be necessary.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
689
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
690 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
691
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
692 fp = None # Set here since __del__ checks it
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
693
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
694 def __init__(self, file, mode="r", compression=ZIP_STORED, allowZip64=False):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
695 """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
696 if mode not in ("r", "w", "a"):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
697 raise RuntimeError('ZipFile() requires mode "r", "w", or "a"')
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 if compression == ZIP_STORED:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
700 pass
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
701 elif compression == ZIP_DEFLATED:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
702 if not zlib:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
703 raise RuntimeError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
704 "Compression requires the (missing) zlib module")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
705 elif compression == ZIP_BZIP2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
706 if not bz2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
707 raise RuntimeError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
708 "Compression requires the (missing) bz2 module")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
709 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
710 raise RuntimeError("That compression method is not supported")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
711
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
712 self._allowZip64 = allowZip64
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
713 self._didModify = False
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
714 self.debug = 0 # Level of printing: 0 through 3
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
715 self.NameToInfo = {} # Find file info given name
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
716 self.filelist = [] # List of ZipInfo instances for archive
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
717 self.compression = compression # Method of compression
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
718 self.mode = key = mode.replace('b', '')[0]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
719 self.pwd = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
720 self.comment = b''
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
721
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
722 # Check if we were passed a file-like object
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
723 if isinstance(file, str):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
724 # No, it's a filename
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
725 self._filePassed = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
726 self.filename = file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
727 modeDict = {'r' : 'rb', 'w': 'wb', 'a' : 'r+b'}
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
728 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
729 self.fp = io.open(file, modeDict[mode])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
730 except IOError:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
731 if mode == 'a':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
732 mode = key = 'w'
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
733 self.fp = io.open(file, modeDict[mode])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
734 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
735 raise
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
736 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
737 self._filePassed = 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
738 self.fp = file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
739 self.filename = getattr(file, 'name', None)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
740
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
741 if key == 'r':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
742 self._GetContents()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
743 elif key == 'w':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
744 pass
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
745 elif key == 'a':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
746 try: # See if file is a zip file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
747 self._RealGetContents()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
748 # seek to start of directory and overwrite
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
749 self.fp.seek(self.start_dir, 0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
750 except BadZipfile: # file is not a zip file, just append
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
751 self.fp.seek(0, 2)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
752 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
753 if not self._filePassed:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
754 self.fp.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
755 self.fp = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
756 raise RuntimeError('Mode must be "r", "w" or "a"')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
757
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
758 def _GetContents(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
759 """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
760 is bad."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
761 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
762 self._RealGetContents()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
763 except BadZipfile:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
764 if not self._filePassed:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
765 self.fp.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
766 self.fp = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
767 raise
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
768
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
769 def _RealGetContents(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
770 """Read in the table of contents for the ZIP file."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
771 fp = self.fp
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
772 endrec = _EndRecData(fp)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
773 if not endrec:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
774 raise BadZipfile("File is not a zip file")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
775 if self.debug > 1:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
776 print(endrec)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
777 size_cd = endrec[_ECD_SIZE] # bytes in central directory
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
778 offset_cd = endrec[_ECD_OFFSET] # offset of central directory
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
779 self.comment = endrec[_ECD_COMMENT] # archive comment
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
780
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
781 # "concat" is zero, unless zip was concatenated to another file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
782 concat = endrec[_ECD_LOCATION] - size_cd - offset_cd
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
783 if endrec[_ECD_SIGNATURE] == stringEndArchive64:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
784 # If Zip64 extension structures are present, account for them
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
785 concat -= (sizeEndCentDir64 + sizeEndCentDir64Locator)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
786
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
787 if self.debug > 2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
788 inferred = concat + offset_cd
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
789 print("given, inferred, offset", offset_cd, inferred, concat)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
790 # self.start_dir: Position of start of central directory
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
791 self.start_dir = offset_cd + concat
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
792 fp.seek(self.start_dir, 0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
793 data = fp.read(size_cd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
794 fp = io.BytesIO(data)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
795 total = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
796 while total < size_cd:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
797 centdir = fp.read(sizeCentralDir)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
798 if centdir[0:4] != stringCentralDir:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
799 raise BadZipfile("Bad magic number for central directory")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
800 centdir = struct.unpack(structCentralDir, centdir)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
801 if self.debug > 2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
802 print(centdir)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
803 filename = fp.read(centdir[_CD_FILENAME_LENGTH])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
804 flags = centdir[5]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
805 if flags & 0x800:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
806 # UTF-8 file names extension
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
807 filename = filename.decode('utf-8')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
808 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
809 # Historical ZIP filename encoding
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
810 filename = filename.decode('cp437')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
811 # Create ZipInfo instance to store file information
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
812 x = ZipInfo(filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
813 x.extra = fp.read(centdir[_CD_EXTRA_FIELD_LENGTH])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
814 x.comment = fp.read(centdir[_CD_COMMENT_LENGTH])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
815 x.header_offset = centdir[_CD_LOCAL_HEADER_OFFSET]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
816 (x.create_version, x.create_system, x.extract_version, x.reserved,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
817 x.flag_bits, x.compress_type, t, d,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
818 x.CRC, x.compress_size, x.file_size) = centdir[1:12]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
819 x.volume, x.internal_attr, x.external_attr = centdir[15:18]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
820 # Convert date/time code to (year, month, day, hour, min, sec)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
821 x._raw_time = t
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
822 x.date_time = ( (d>>9)+1980, (d>>5)&0xF, d&0x1F,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
823 t>>11, (t>>5)&0x3F, (t&0x1F) * 2 )
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 x._decodeExtra()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
826 x.header_offset = x.header_offset + concat
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
827 self.filelist.append(x)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
828 self.NameToInfo[x.filename] = x
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
829
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
830 # update total bytes read from central directory
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
831 total = (total + sizeCentralDir + centdir[_CD_FILENAME_LENGTH]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
832 + centdir[_CD_EXTRA_FIELD_LENGTH]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
833 + centdir[_CD_COMMENT_LENGTH])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
834
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
835 if self.debug > 2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
836 print("total", total)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
837
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
838
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
839 def namelist(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
840 """Return a list of file names in the archive."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
841 l = []
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
842 for data in self.filelist:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
843 l.append(data.filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
844 return l
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
845
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
846 def infolist(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
847 """Return a list of class ZipInfo instances for files in the
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
848 archive."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
849 return self.filelist
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
850
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
851 def printdir(self, file=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
852 """Print a table of contents for the zip file."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
853 print("%-46s %19s %12s" % ("File Name", "Modified ", "Size"),
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
854 file=file)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
855 for zinfo in self.filelist:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
856 date = "%d-%02d-%02d %02d:%02d:%02d" % zinfo.date_time[:6]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
857 print("%-46s %s %12d" % (zinfo.filename, date, zinfo.file_size),
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
858 file=file)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
859
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
860 def testzip(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
861 """Read all the files and check the CRC."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
862 chunk_size = 2 ** 20
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
863 for zinfo in self.filelist:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
864 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
865 # Read by chunks, to avoid an OverflowError or a
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
866 # MemoryError with very large embedded files.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
867 f = self.open(zinfo.filename, "r")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
868 while f.read(chunk_size): # Check CRC-32
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
869 pass
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
870 except BadZipfile:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
871 return zinfo.filename
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
872
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
873 def getinfo(self, name):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
874 """Return the instance of ZipInfo given 'name'."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
875 info = self.NameToInfo.get(name)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
876 if info is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
877 raise KeyError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
878 'There is no item named %r in the archive' % name)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
879
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
880 return info
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 def setpassword(self, pwd):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
883 """Set default password for encrypted files."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
884 assert isinstance(pwd, bytes)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
885 self.pwd = pwd
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
886
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
887 def read(self, name, pwd=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
888 """Return file bytes (as a string) for name."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
889 return self.open(name, "r", pwd).read()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
890
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
891 def open(self, name, mode="r", pwd=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
892 """Return file-like object for 'name'."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
893 if mode not in ("r", "U", "rU"):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
894 raise RuntimeError('open() requires mode "r", "U", or "rU"')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
895 if not self.fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
896 raise RuntimeError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
897 "Attempt to read ZIP archive that was already closed")
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 # Only open a new file for instances where we were not
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
900 # given a file object in the constructor
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
901 if self._filePassed:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
902 zef_file = self.fp
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
903 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
904 zef_file = io.open(self.filename, 'rb')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
905
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
906 # Make sure we have an info object
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
907 if isinstance(name, ZipInfo):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
908 # 'name' is already an info object
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
909 zinfo = name
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
910 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
911 # Get info object for name
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
912 zinfo = self.getinfo(name)
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 zef_file.seek(zinfo.header_offset, 0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
915
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
916 # Skip the file header:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
917 fheader = zef_file.read(sizeFileHeader)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
918 if fheader[0:4] != stringFileHeader:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
919 raise BadZipfile("Bad magic number for file header")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
920
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
921 fheader = struct.unpack(structFileHeader, fheader)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
922 fname = zef_file.read(fheader[_FH_FILENAME_LENGTH])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
923 if fheader[_FH_EXTRA_FIELD_LENGTH]:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
924 zef_file.read(fheader[_FH_EXTRA_FIELD_LENGTH])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
925
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
926 if fname != zinfo.orig_filename.encode("utf-8"):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
927 raise BadZipfile(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
928 'File name in directory %r and header %r differ.'
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
929 % (zinfo.orig_filename, fname))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
930
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
931 # check for encrypted flag & handle password
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
932 is_encrypted = zinfo.flag_bits & 0x1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
933 zd = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
934 if is_encrypted:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
935 if not pwd:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
936 pwd = self.pwd
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
937 if not pwd:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
938 raise RuntimeError("File %s is encrypted, "
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
939 "password required for extraction" % 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 zd = _ZipDecrypter(pwd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
942 # 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
943 # used to strengthen the algorithm. The first 11 bytes are
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
944 # completely random, while the 12th contains the MSB of the CRC,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
945 # 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
946 # and is used to check the correctness of the password.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
947 bytes = zef_file.read(12)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
948 h = list(map(zd, bytes[0:12]))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
949 if zinfo.flag_bits & 0x8:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
950 # compare against the file type from extended local headers
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
951 check_byte = (zinfo._raw_time >> 8) & 0xff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
952 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
953 # compare against the CRC otherwise
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
954 check_byte = (zinfo.CRC >> 24) & 0xff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
955 if h[11] != check_byte:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
956 raise RuntimeError("Bad password for file", name)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
957
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
958 # build and return a ZipExtFile
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
959 if zd is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
960 zef = ZipExtFile(zef_file, zinfo)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
961 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
962 zef = ZipExtFile(zef_file, zinfo, zd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
963
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
964 # set universal newlines on ZipExtFile if necessary
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
965 if "U" in mode:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
966 zef.set_univ_newlines(True)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
967 return zef
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(self, member, path=None, pwd=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
970 """Extract a member from the archive to the current working directory,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
971 using its full name. Its file information is extracted as accurately
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
972 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
973 specify a different directory using `path'.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
974 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
975 if not isinstance(member, ZipInfo):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
976 member = self.getinfo(member)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
977
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
978 if path is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
979 path = os.getcwd()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
980
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
981 return self._extract_member(member, path, pwd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
982
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
983 def extractall(self, path=None, members=None, pwd=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
984 """Extract all members from the archive to the current working
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
985 directory. `path' specifies a different directory to extract to.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
986 `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
987 by namelist().
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
988 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
989 if members is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
990 members = self.namelist()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
991
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
992 for zipinfo in members:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
993 self.extract(zipinfo, path, pwd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
994
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
995 def _extract_member(self, member, targetpath, pwd):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
996 """Extract the ZipInfo object 'member' to a physical
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
997 file on the path targetpath.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
998 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
999 # build the destination pathname, replacing
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1000 # forward slashes to platform specific separators.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1001 # Strip trailing path separator, unless it represents the root.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1002 if (targetpath[-1:] in (os.path.sep, os.path.altsep)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1003 and len(os.path.splitdrive(targetpath)[1]) > 1):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1004 targetpath = targetpath[:-1]
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 # don't include leading "/" from file name if present
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1007 if member.filename[0] == '/':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1008 targetpath = os.path.join(targetpath, member.filename[1:])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1009 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1010 targetpath = os.path.join(targetpath, member.filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1011
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1012 targetpath = os.path.normpath(targetpath)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1013
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1014 # Create all upper directories if necessary.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1015 upperdirs = os.path.dirname(targetpath)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1016 if upperdirs and not os.path.exists(upperdirs):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1017 os.makedirs(upperdirs)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1018
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1019 if member.filename[-1] == '/':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1020 if not os.path.isdir(targetpath):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1021 os.mkdir(targetpath)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1022 return targetpath
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1023
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1024 source = self.open(member, pwd=pwd)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1025 target = open(targetpath, "wb")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1026 shutil.copyfileobj(source, target)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1027 source.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1028 target.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1029
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1030 return targetpath
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 _writecheck(self, zinfo):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1033 """Check for errors before writing a file to the archive."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1034 if zinfo.filename in self.NameToInfo:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1035 if self.debug: # Warning for duplicate names
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1036 print("Duplicate name:", zinfo.filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1037 if self.mode not in ("w", "a"):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1038 raise RuntimeError('write() requires mode "w" or "a"')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1039 if not self.fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1040 raise RuntimeError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1041 "Attempt to write ZIP archive that was already closed")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1042 if zinfo.compress_type == ZIP_DEFLATED and not zlib:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1043 raise RuntimeError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1044 "Compression requires the (missing) zlib module")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1045 if zinfo.compress_type == ZIP_BZIP2 and not bz2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1046 raise RuntimeError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1047 "Compression requires the (missing) bz2 module")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1048 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
1049 raise RuntimeError("That compression method is not supported")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1050 if zinfo.file_size > ZIP64_LIMIT:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1051 if not self._allowZip64:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1052 raise LargeZipFile("Filesize would require ZIP64 extensions")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1053 if zinfo.header_offset > ZIP64_LIMIT:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1054 if not self._allowZip64:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1055 raise LargeZipFile(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1056 "Zipfile size would require ZIP64 extensions")
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 def write(self, filename, arcname=None, compress_type=None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1059 """Put the bytes from filename into the archive under the name
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1060 arcname."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1061 if not self.fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1062 raise RuntimeError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1063 "Attempt to write to ZIP archive that was already closed")
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 st = os.stat(filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1066 isdir = stat.S_ISDIR(st.st_mode)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1067 mtime = time.localtime(st.st_mtime)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1068 date_time = mtime[0:6]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1069 # Create ZipInfo instance to store file information
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1070 if arcname is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1071 arcname = filename
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1072 arcname = os.path.normpath(os.path.splitdrive(arcname)[1])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1073 while arcname[0] in (os.sep, os.altsep):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1074 arcname = arcname[1:]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1075 if isdir:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1076 arcname += '/'
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1077 zinfo = ZipInfo(arcname, date_time)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1078 zinfo.external_attr = (st[0] & 0xFFFF) << 16 # Unix attributes
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1079 if compress_type is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1080 zinfo.compress_type = self.compression
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1081 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1082 zinfo.compress_type = compress_type
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1083
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1084 zinfo.file_size = st.st_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1085 zinfo.flag_bits = 0x00
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1086 zinfo.header_offset = self.fp.tell() # Start of header bytes
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1087
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1088 self._writecheck(zinfo)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1089 self._didModify = True
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1090
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1091 if isdir:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1092 zinfo.file_size = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1093 zinfo.compress_size = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1094 zinfo.CRC = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1095 self.filelist.append(zinfo)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1096 self.NameToInfo[zinfo.filename] = zinfo
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1097 self.fp.write(zinfo.FileHeader())
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1098 return
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1099
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1100 with open(filename, "rb") as fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1101 # Must overwrite CRC and sizes with correct data later
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1102 zinfo.CRC = CRC = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1103 zinfo.compress_size = compress_size = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1104 zinfo.file_size = file_size = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1105 self.fp.write(zinfo.FileHeader())
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1106 if zinfo.compress_type == ZIP_DEFLATED:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1107 cmpr = zlib.compressobj(zlib.Z_DEFAULT_COMPRESSION,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1108 zlib.DEFLATED, -15)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1109 elif zinfo.compress_type == ZIP_BZIP2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1110 cmpr = bz2.BZ2Compressor()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1111 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1112 cmpr = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1113 while 1:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1114 buf = fp.read(1024 * 8)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1115 if not buf:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1116 break
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1117 file_size = file_size + len(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1118 CRC = crc32(buf, CRC) & 0xffffffff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1119 if cmpr:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1120 buf = cmpr.compress(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1121 compress_size = compress_size + len(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1122 self.fp.write(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1123 if cmpr:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1124 buf = cmpr.flush()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1125 compress_size = compress_size + len(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1126 self.fp.write(buf)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1127 zinfo.compress_size = compress_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1128 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1129 zinfo.compress_size = file_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1130 zinfo.CRC = CRC
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1131 zinfo.file_size = file_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1132 # Seek backwards and write CRC and file sizes
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1133 position = self.fp.tell() # Preserve current position in file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1134 self.fp.seek(zinfo.header_offset + 14, 0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1135 self.fp.write(struct.pack("<LLL", zinfo.CRC, zinfo.compress_size,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1136 zinfo.file_size))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1137 self.fp.seek(position, 0)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1138 self.filelist.append(zinfo)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1139 self.NameToInfo[zinfo.filename] = zinfo
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1140
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1141 def writestr(self, zinfo_or_arcname, data):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1142 """Write a file into the archive. The contents is 'data', which
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1143 may be either a 'str' or a 'bytes' instance; if it is a 'str',
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1144 it is encoded as UTF-8 first.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1145 'zinfo_or_arcname' is either a ZipInfo instance or
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1146 the name of the file in the archive."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1147 if isinstance(data, str):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1148 data = data.encode("utf-8")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1149 if not isinstance(zinfo_or_arcname, ZipInfo):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1150 zinfo = ZipInfo(filename=zinfo_or_arcname,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1151 date_time=time.localtime(time.time())[:6])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1152 zinfo.compress_type = self.compression
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1153 zinfo.external_attr = 0o600 << 16
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1154 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1155 zinfo = zinfo_or_arcname
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1156
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1157 if not self.fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1158 raise RuntimeError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1159 "Attempt to write to ZIP archive that was already closed")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1160
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1161 zinfo.file_size = len(data) # Uncompressed size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1162 zinfo.header_offset = self.fp.tell() # Start of header data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1163 self._writecheck(zinfo)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1164 self._didModify = True
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1165 zinfo.CRC = crc32(data) & 0xffffffff # CRC-32 checksum
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1166 if zinfo.compress_type == ZIP_DEFLATED:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1167 co = zlib.compressobj(zlib.Z_DEFAULT_COMPRESSION,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1168 zlib.DEFLATED, -15)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1169 data = co.compress(data) + co.flush()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1170 zinfo.compress_size = len(data) # Compressed size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1171 elif zinfo.compress_type == ZIP_BZIP2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1172 co = bz2.BZ2Compressor()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1173 data = co.compress(data) + co.flush()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1174 zinfo.compress_size = len(data) # Compressed size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1175 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1176 zinfo.compress_size = zinfo.file_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1177 zinfo.header_offset = self.fp.tell() # Start of header data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1178 self.fp.write(zinfo.FileHeader())
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1179 self.fp.write(data)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1180 self.fp.flush()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1181 if zinfo.flag_bits & 0x08:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1182 # Write CRC and file sizes after the file data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1183 self.fp.write(struct.pack("<LLL", zinfo.CRC, zinfo.compress_size,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1184 zinfo.file_size))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1185 self.filelist.append(zinfo)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1186 self.NameToInfo[zinfo.filename] = zinfo
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1187
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1188 def __del__(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1189 """Call the "close()" method in case the user forgot."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1190 self.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1191
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1192 def close(self):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1193 """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
1194 records."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1195 if self.fp is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1196 return
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1197
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1198 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
1199 count = 0
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1200 pos1 = self.fp.tell()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1201 for zinfo in self.filelist: # write central directory
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1202 count = count + 1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1203 dt = zinfo.date_time
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1204 dosdate = (dt[0] - 1980) << 9 | dt[1] << 5 | dt[2]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1205 dostime = dt[3] << 11 | dt[4] << 5 | (dt[5] // 2)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1206 extra = []
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1207 if zinfo.file_size > ZIP64_LIMIT \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1208 or zinfo.compress_size > ZIP64_LIMIT:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1209 extra.append(zinfo.file_size)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1210 extra.append(zinfo.compress_size)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1211 file_size = 0xffffffff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1212 compress_size = 0xffffffff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1213 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1214 file_size = zinfo.file_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1215 compress_size = zinfo.compress_size
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1216
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1217 if zinfo.header_offset > ZIP64_LIMIT:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1218 extra.append(zinfo.header_offset)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1219 header_offset = 0xffffffff
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1220 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1221 header_offset = zinfo.header_offset
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1222
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1223 extra_data = zinfo.extra
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1224 if extra:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1225 # Append a ZIP64 field to the extra's
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1226 extra_data = struct.pack(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1227 '<HH' + 'Q'*len(extra),
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1228 1, 8*len(extra), *extra) + extra_data
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1229
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1230 extract_version = max(45, zinfo.extract_version)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1231 create_version = max(45, zinfo.create_version)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1232 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1233 extract_version = zinfo.extract_version
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1234 create_version = zinfo.create_version
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1235
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1236 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1237 filename, flag_bits = zinfo._encodeFilenameFlags()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1238 centdir = struct.pack(structCentralDir,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1239 stringCentralDir, create_version,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1240 zinfo.create_system, extract_version, zinfo.reserved,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1241 flag_bits, zinfo.compress_type, dostime, dosdate,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1242 zinfo.CRC, compress_size, file_size,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1243 len(filename), len(extra_data), len(zinfo.comment),
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1244 0, zinfo.internal_attr, zinfo.external_attr,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1245 header_offset)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1246 except DeprecationWarning:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1247 print((structCentralDir, stringCentralDir, create_version,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1248 zinfo.create_system, extract_version, zinfo.reserved,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1249 zinfo.flag_bits, zinfo.compress_type, dostime, dosdate,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1250 zinfo.CRC, compress_size, file_size,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1251 len(zinfo.filename), len(extra_data), len(zinfo.comment),
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1252 0, zinfo.internal_attr, zinfo.external_attr,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1253 header_offset), file=sys.stderr)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1254 raise
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1255 self.fp.write(centdir)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1256 self.fp.write(filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1257 self.fp.write(extra_data)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1258 self.fp.write(zinfo.comment)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1259
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1260 pos2 = self.fp.tell()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1261 # Write end-of-zip-archive record
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1262 centDirCount = count
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1263 centDirSize = pos2 - pos1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1264 centDirOffset = pos1
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1265 if (centDirCount >= ZIP_FILECOUNT_LIMIT or
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1266 centDirOffset > ZIP64_LIMIT or
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1267 centDirSize > ZIP64_LIMIT):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1268 # Need to write the ZIP64 end-of-archive records
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1269 zip64endrec = struct.pack(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1270 structEndArchive64, stringEndArchive64,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1271 44, 45, 45, 0, 0, centDirCount, centDirCount,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1272 centDirSize, centDirOffset)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1273 self.fp.write(zip64endrec)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1274
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1275 zip64locrec = struct.pack(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1276 structEndArchive64Locator,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1277 stringEndArchive64Locator, 0, pos2, 1)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1278 self.fp.write(zip64locrec)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1279 centDirCount = min(centDirCount, 0xFFFF)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1280 centDirSize = min(centDirSize, 0xFFFFFFFF)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1281 centDirOffset = min(centDirOffset, 0xFFFFFFFF)
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 # check for valid comment length
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1284 if len(self.comment) >= ZIP_MAX_COMMENT:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1285 if self.debug > 0:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1286 msg = 'Archive comment is too long; truncating to %d bytes' \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1287 % ZIP_MAX_COMMENT
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1288 self.comment = self.comment[:ZIP_MAX_COMMENT]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1289
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1290 endrec = struct.pack(structEndArchive, stringEndArchive,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1291 0, 0, centDirCount, centDirCount,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1292 centDirSize, centDirOffset, len(self.comment))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1293 self.fp.write(endrec)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1294 self.fp.write(self.comment)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1295 self.fp.flush()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1296
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1297 if not self._filePassed:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1298 self.fp.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1299 self.fp = None
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1300
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1301
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1302 class PyZipFile(ZipFile):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1303 """Class to create ZIP archives with Python library files and packages."""
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1304
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1305 def writepy(self, pathname, basename=""):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1306 """Add all files from "pathname" to the ZIP archive.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1307
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1308 If pathname is a package directory, search the directory and
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1309 all package subdirectories recursively for all *.py and enter
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1310 the modules into the archive. If pathname is a plain
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1311 directory, listdir *.py and enter all modules. Else, pathname
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1312 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
1313 archive. Added modules are always module.pyo or module.pyc.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1314 This method will compile the module.py into module.pyc if
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1315 necessary.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1316 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1317 dir, name = os.path.split(pathname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1318 if os.path.isdir(pathname):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1319 initname = os.path.join(pathname, "__init__.py")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1320 if os.path.isfile(initname):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1321 # This is a package directory, add it
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1322 if basename:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1323 basename = "%s/%s" % (basename, name)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1324 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1325 basename = name
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1326 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1327 print("Adding package in", pathname, "as", basename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1328 fname, arcname = self._get_codename(initname[0:-3], basename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1329 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1330 print("Adding", arcname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1331 self.write(fname, arcname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1332 dirlist = os.listdir(pathname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1333 dirlist.remove("__init__.py")
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1334 # Add all *.py files and package subdirectories
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1335 for filename in dirlist:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1336 path = os.path.join(pathname, filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1337 root, ext = os.path.splitext(filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1338 if os.path.isdir(path):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1339 if os.path.isfile(os.path.join(path, "__init__.py")):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1340 # This is a package directory, add it
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1341 self.writepy(path, basename) # Recursive call
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1342 elif ext == ".py":
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1343 fname, arcname = self._get_codename(path[0:-3],
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1344 basename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1345 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1346 print("Adding", arcname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1347 self.write(fname, arcname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1348 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1349 # 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
1350 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1351 print("Adding files from directory", pathname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1352 for filename in os.listdir(pathname):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1353 path = os.path.join(pathname, filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1354 root, ext = os.path.splitext(filename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1355 if ext == ".py":
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1356 fname, arcname = self._get_codename(path[0:-3],
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1357 basename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1358 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1359 print("Adding", arcname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1360 self.write(fname, arcname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1361 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1362 if pathname[-3:] != ".py":
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1363 raise RuntimeError(
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1364 'Files added with writepy() must end with ".py"')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1365 fname, arcname = self._get_codename(pathname[0:-3], basename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1366 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1367 print("Adding file", arcname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1368 self.write(fname, arcname)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1369
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1370 def _get_codename(self, pathname, basename):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1371 """Return (filename, archivename) for the path.
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1372
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1373 Given a module name path, return the correct file path and
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1374 archive name, compiling if necessary. For example, given
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1375 /python/lib/string, return (/python/lib/string.pyc, string).
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1376 """
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1377 file_py = pathname + ".py"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1378 file_pyc = pathname + ".pyc"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1379 file_pyo = pathname + ".pyo"
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1380 if os.path.isfile(file_pyo) and \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1381 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
1382 fname = file_pyo # Use .pyo file
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1383 elif not os.path.isfile(file_pyc) or \
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1384 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
1385 import py_compile
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1386 if self.debug:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1387 print("Compiling", file_py)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1388 try:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1389 py_compile.compile(file_py, file_pyc, None, True)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1390 except py_compile.PyCompileError as err:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1391 print(err.msg)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1392 fname = file_pyc
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1393 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1394 fname = file_pyc
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1395 archivename = os.path.split(fname)[1]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1396 if basename:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1397 archivename = "%s/%s" % (basename, archivename)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1398 return (fname, archivename)
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
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1401 def main(args = None):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1402 import textwrap
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1403 USAGE=textwrap.dedent("""\
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1404 Usage:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1405 zipfile.py -l zipfile.zip # Show listing of a zipfile
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1406 zipfile.py -t zipfile.zip # Test if a zipfile is valid
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1407 zipfile.py -e zipfile.zip target # Extract zipfile into target dir
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1408 zipfile.py -c zipfile.zip src ... # Create zipfile from sources
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 if args is None:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1411 args = sys.argv[1:]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1412
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1413 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
1414 print(USAGE)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1415 sys.exit(1)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1416
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1417 if args[0] == '-l':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1418 if len(args) != 2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1419 print(USAGE)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1420 sys.exit(1)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1421 zf = ZipFile(args[1], 'r')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1422 zf.printdir()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1423 zf.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1424
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1425 elif args[0] == '-t':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1426 if len(args) != 2:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1427 print(USAGE)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1428 sys.exit(1)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1429 zf = ZipFile(args[1], 'r')
29
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
1430 badfile = zf.testzip()
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
1431 if badfile:
a8cc383b787c Clean up zipfiles and diff them to stock ones
Oleg Oshmyan <chortos@inbox.lv>
parents: 28
diff changeset
1432 print("The following enclosed file is corrupted: {!r}".format(badfile))
21
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1433 print("Done testing")
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 elif args[0] == '-e':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1436 if len(args) != 3:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1437 print(USAGE)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1438 sys.exit(1)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1439
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1440 zf = ZipFile(args[1], 'r')
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1441 out = args[2]
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1442 for path in zf.namelist():
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1443 if path.startswith('./'):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1444 tgt = os.path.join(out, path[2:])
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1445 else:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1446 tgt = os.path.join(out, path)
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 tgtdir = os.path.dirname(tgt)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1449 if not os.path.exists(tgtdir):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1450 os.makedirs(tgtdir)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1451 with open(tgt, 'wb') as fp:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1452 fp.write(zf.read(path))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1453 zf.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1454
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1455 elif args[0] == '-c':
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1456 if len(args) < 3:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1457 print(USAGE)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1458 sys.exit(1)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1459
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1460 def addToZip(zf, path, zippath):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1461 if os.path.isfile(path):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1462 zf.write(path, zippath, ZIP_DEFLATED)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1463 elif os.path.isdir(path):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1464 for nm in os.listdir(path):
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1465 addToZip(zf,
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1466 os.path.join(path, nm), os.path.join(zippath, nm))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1467 # else: ignore
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1468
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1469 zf = ZipFile(args[1], 'w', allowZip64=True)
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1470 for src in args[2:]:
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1471 addToZip(zf, src, os.path.basename(src))
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1472
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1473 zf.close()
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1474
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1475 if __name__ == "__main__":
ec6f1a132109 A pretty usable version
Oleg Oshmyan <chortos@inbox.lv>
parents:
diff changeset
1476 main()