blob: 6639317537bbf7d95c830553b6124dbf24849bb4 [file] [log] [blame]
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001"""
2Read and write ZIP files.
3"""
Georg Brandl62416bc2008-01-07 18:47:44 +00004import struct, os, time, sys, shutil
Martin v. Löwis0dfcfc82009-01-24 14:00:33 +00005import binascii, cStringIO, stat
Antoine Pitrou94c33eb2010-01-27 20:59:50 +00006import io
7import re
Serhiy Storchaka7c068752013-02-02 12:30:49 +02008import string
Guido van Rossum32abe6f2000-03-31 17:30:02 +00009
10try:
Tim Peterse1190062001-01-15 03:34:38 +000011 import zlib # We may need its compression method
Gregory P. Smithb89a0962008-03-19 01:46:10 +000012 crc32 = zlib.crc32
Guido van Rossum9c673f32001-04-10 15:37:12 +000013except ImportError:
Guido van Rossum32abe6f2000-03-31 17:30:02 +000014 zlib = None
Gregory P. Smithb89a0962008-03-19 01:46:10 +000015 crc32 = binascii.crc32
Guido van Rossum32abe6f2000-03-31 17:30:02 +000016
Skip Montanaro40fc1602001-03-01 04:27:19 +000017__all__ = ["BadZipfile", "error", "ZIP_STORED", "ZIP_DEFLATED", "is_zipfile",
Ronald Oussoren143cefb2006-06-15 08:14:18 +000018 "ZipInfo", "ZipFile", "PyZipFile", "LargeZipFile" ]
Skip Montanaro40fc1602001-03-01 04:27:19 +000019
Fred Drake5db246d2000-09-29 20:44:48 +000020class BadZipfile(Exception):
Guido van Rossum32abe6f2000-03-31 17:30:02 +000021 pass
Ronald Oussoren143cefb2006-06-15 08:14:18 +000022
23
24class LargeZipFile(Exception):
Tim Petersa608bb22006-06-15 18:06:29 +000025 """
Ronald Oussoren143cefb2006-06-15 08:14:18 +000026 Raised when writing a zipfile, the zipfile requires ZIP64 extensions
27 and those extensions are disabled.
28 """
29
Tim Peterse1190062001-01-15 03:34:38 +000030error = BadZipfile # The exception raised by this module
Guido van Rossum32abe6f2000-03-31 17:30:02 +000031
Amaury Forgeot d'Arcd25f87a2009-01-17 16:40:17 +000032ZIP64_LIMIT = (1 << 31) - 1
Martin v. Löwis8c436412008-07-03 12:51:14 +000033ZIP_FILECOUNT_LIMIT = 1 << 16
34ZIP_MAX_COMMENT = (1 << 16) - 1
Ronald Oussoren143cefb2006-06-15 08:14:18 +000035
Guido van Rossum32abe6f2000-03-31 17:30:02 +000036# constants for Zip file compression methods
37ZIP_STORED = 0
38ZIP_DEFLATED = 8
39# Other ZIP compression methods not supported
40
Martin v. Löwis8c436412008-07-03 12:51:14 +000041# Below are some formats and associated data for reading/writing headers using
42# the struct module. The names and structures of headers/records are those used
43# in the PKWARE description of the ZIP file format:
44# http://www.pkware.com/documents/casestudies/APPNOTE.TXT
45# (URL valid as of January 2008)
Ronald Oussoren143cefb2006-06-15 08:14:18 +000046
Martin v. Löwis8c436412008-07-03 12:51:14 +000047# The "end of central directory" structure, magic number, size, and indices
48# (section V.I in the format document)
Amaury Forgeot d'Arcae6d2b92008-07-11 21:28:25 +000049structEndArchive = "<4s4H2LH"
50stringEndArchive = "PK\005\006"
51sizeEndCentDir = struct.calcsize(structEndArchive)
Martin v. Löwis8c436412008-07-03 12:51:14 +000052
53_ECD_SIGNATURE = 0
54_ECD_DISK_NUMBER = 1
55_ECD_DISK_START = 2
56_ECD_ENTRIES_THIS_DISK = 3
57_ECD_ENTRIES_TOTAL = 4
58_ECD_SIZE = 5
59_ECD_OFFSET = 6
60_ECD_COMMENT_SIZE = 7
61# These last two indices are not part of the structure as defined in the
62# spec, but they are used internally by this module as a convenience
63_ECD_COMMENT = 8
64_ECD_LOCATION = 9
65
66# The "central directory" structure, magic number, size, and indices
67# of entries in the structure (section V.F in the format document)
68structCentralDir = "<4s4B4HL2L5H2L"
Amaury Forgeot d'Arcae6d2b92008-07-11 21:28:25 +000069stringCentralDir = "PK\001\002"
Martin v. Löwis8c436412008-07-03 12:51:14 +000070sizeCentralDir = struct.calcsize(structCentralDir)
71
Fred Drake3e038e52001-02-28 17:56:26 +000072# indexes of entries in the central directory structure
73_CD_SIGNATURE = 0
74_CD_CREATE_VERSION = 1
75_CD_CREATE_SYSTEM = 2
76_CD_EXTRACT_VERSION = 3
Martin v. Löwis8c436412008-07-03 12:51:14 +000077_CD_EXTRACT_SYSTEM = 4
Fred Drake3e038e52001-02-28 17:56:26 +000078_CD_FLAG_BITS = 5
79_CD_COMPRESS_TYPE = 6
80_CD_TIME = 7
81_CD_DATE = 8
82_CD_CRC = 9
83_CD_COMPRESSED_SIZE = 10
84_CD_UNCOMPRESSED_SIZE = 11
85_CD_FILENAME_LENGTH = 12
86_CD_EXTRA_FIELD_LENGTH = 13
87_CD_COMMENT_LENGTH = 14
88_CD_DISK_NUMBER_START = 15
89_CD_INTERNAL_FILE_ATTRIBUTES = 16
90_CD_EXTERNAL_FILE_ATTRIBUTES = 17
91_CD_LOCAL_HEADER_OFFSET = 18
92
Martin v. Löwis8c436412008-07-03 12:51:14 +000093# The "local file header" structure, magic number, size, and indices
94# (section V.A in the format document)
95structFileHeader = "<4s2B4HL2L2H"
Amaury Forgeot d'Arcae6d2b92008-07-11 21:28:25 +000096stringFileHeader = "PK\003\004"
Martin v. Löwis8c436412008-07-03 12:51:14 +000097sizeFileHeader = struct.calcsize(structFileHeader)
98
Fred Drake3e038e52001-02-28 17:56:26 +000099_FH_SIGNATURE = 0
100_FH_EXTRACT_VERSION = 1
Martin v. Löwis8c436412008-07-03 12:51:14 +0000101_FH_EXTRACT_SYSTEM = 2
Fred Drake3e038e52001-02-28 17:56:26 +0000102_FH_GENERAL_PURPOSE_FLAG_BITS = 3
103_FH_COMPRESSION_METHOD = 4
104_FH_LAST_MOD_TIME = 5
105_FH_LAST_MOD_DATE = 6
106_FH_CRC = 7
107_FH_COMPRESSED_SIZE = 8
108_FH_UNCOMPRESSED_SIZE = 9
109_FH_FILENAME_LENGTH = 10
110_FH_EXTRA_FIELD_LENGTH = 11
111
Martin v. Löwis8c436412008-07-03 12:51:14 +0000112# The "Zip64 end of central directory locator" structure, magic number, and size
Amaury Forgeot d'Arcae6d2b92008-07-11 21:28:25 +0000113structEndArchive64Locator = "<4sLQL"
114stringEndArchive64Locator = "PK\x06\x07"
115sizeEndCentDir64Locator = struct.calcsize(structEndArchive64Locator)
Martin v. Löwis8c436412008-07-03 12:51:14 +0000116
117# The "Zip64 end of central directory" record, magic number, size, and indices
118# (section V.G in the format document)
Amaury Forgeot d'Arcae6d2b92008-07-11 21:28:25 +0000119structEndArchive64 = "<4sQ2H2L4Q"
120stringEndArchive64 = "PK\x06\x06"
121sizeEndCentDir64 = struct.calcsize(structEndArchive64)
Martin v. Löwis8c436412008-07-03 12:51:14 +0000122
123_CD64_SIGNATURE = 0
124_CD64_DIRECTORY_RECSIZE = 1
125_CD64_CREATE_VERSION = 2
126_CD64_EXTRACT_VERSION = 3
127_CD64_DISK_NUMBER = 4
128_CD64_DISK_NUMBER_START = 5
129_CD64_NUMBER_ENTRIES_THIS_DISK = 6
130_CD64_NUMBER_ENTRIES_TOTAL = 7
131_CD64_DIRECTORY_SIZE = 8
132_CD64_OFFSET_START_CENTDIR = 9
133
Antoine Pitrou6f193e02008-12-27 15:43:12 +0000134def _check_zipfile(fp):
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000135 try:
Antoine Pitrou6f193e02008-12-27 15:43:12 +0000136 if _EndRecData(fp):
137 return True # file has correct magic number
Fred Drake7e473802001-05-11 19:52:57 +0000138 except IOError:
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000139 pass
Guido van Rossum8ca162f2002-04-07 06:36:23 +0000140 return False
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000141
Antoine Pitrou6f193e02008-12-27 15:43:12 +0000142def is_zipfile(filename):
143 """Quickly see if a file is a ZIP file by checking the magic number.
144
145 The filename argument may be a file or file-like object too.
146 """
147 result = False
148 try:
149 if hasattr(filename, "read"):
150 result = _check_zipfile(fp=filename)
151 else:
152 with open(filename, "rb") as fp:
153 result = _check_zipfile(fp)
154 except IOError:
155 pass
156 return result
157
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000158def _EndRecData64(fpin, offset, endrec):
159 """
160 Read the ZIP64 end-of-archive records and use that to update endrec
161 """
Georg Brandl86e0c892010-11-26 07:22:28 +0000162 try:
163 fpin.seek(offset - sizeEndCentDir64Locator, 2)
164 except IOError:
165 # If the seek fails, the file is not large enough to contain a ZIP64
166 # end-of-archive record, so just return the end record we were given.
167 return endrec
168
Martin v. Löwis8c436412008-07-03 12:51:14 +0000169 data = fpin.read(sizeEndCentDir64Locator)
Serhiy Storchaka0be506a2013-01-31 15:26:55 +0200170 if len(data) != sizeEndCentDir64Locator:
171 return endrec
Amaury Forgeot d'Arcae6d2b92008-07-11 21:28:25 +0000172 sig, diskno, reloff, disks = struct.unpack(structEndArchive64Locator, data)
173 if sig != stringEndArchive64Locator:
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000174 return endrec
175
176 if diskno != 0 or disks != 1:
177 raise BadZipfile("zipfiles that span multiple disks are not supported")
178
Tim Petersa608bb22006-06-15 18:06:29 +0000179 # Assume no 'zip64 extensible data'
Martin v. Löwis8c436412008-07-03 12:51:14 +0000180 fpin.seek(offset - sizeEndCentDir64Locator - sizeEndCentDir64, 2)
181 data = fpin.read(sizeEndCentDir64)
Serhiy Storchaka0be506a2013-01-31 15:26:55 +0200182 if len(data) != sizeEndCentDir64:
183 return endrec
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000184 sig, sz, create_version, read_version, disk_num, disk_dir, \
185 dircount, dircount2, dirsize, diroffset = \
Amaury Forgeot d'Arcae6d2b92008-07-11 21:28:25 +0000186 struct.unpack(structEndArchive64, data)
187 if sig != stringEndArchive64:
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000188 return endrec
189
190 # Update the original endrec using data from the ZIP64 record
Antoine Pitrouebcd0ce2008-09-05 23:30:23 +0000191 endrec[_ECD_SIGNATURE] = sig
Martin v. Löwis8c436412008-07-03 12:51:14 +0000192 endrec[_ECD_DISK_NUMBER] = disk_num
193 endrec[_ECD_DISK_START] = disk_dir
194 endrec[_ECD_ENTRIES_THIS_DISK] = dircount
195 endrec[_ECD_ENTRIES_TOTAL] = dircount2
196 endrec[_ECD_SIZE] = dirsize
197 endrec[_ECD_OFFSET] = diroffset
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000198 return endrec
199
200
Martin v. Löwis6f6873b2002-10-13 13:54:50 +0000201def _EndRecData(fpin):
202 """Return data from the "End of Central Directory" record, or None.
203
204 The data is a list of the nine items in the ZIP "End of central dir"
205 record followed by a tenth item, the file seek offset of this record."""
Martin v. Löwis8c436412008-07-03 12:51:14 +0000206
207 # Determine file size
208 fpin.seek(0, 2)
209 filesize = fpin.tell()
210
211 # Check to see if this is ZIP file with no archive comment (the
212 # "end of central directory" structure should be the last item in the
213 # file if this is the case).
Amaury Forgeot d'Arc3e5b0272009-07-28 22:15:30 +0000214 try:
215 fpin.seek(-sizeEndCentDir, 2)
216 except IOError:
217 return None
Martin v. Löwis6f6873b2002-10-13 13:54:50 +0000218 data = fpin.read()
Serhiy Storchaka0be506a2013-01-31 15:26:55 +0200219 if (len(data) == sizeEndCentDir and
220 data[0:4] == stringEndArchive and
221 data[-2:] == b"\000\000"):
Martin v. Löwis8c436412008-07-03 12:51:14 +0000222 # the signature is correct and there's no comment, unpack structure
Amaury Forgeot d'Arcae6d2b92008-07-11 21:28:25 +0000223 endrec = struct.unpack(structEndArchive, data)
Martin v. Löwis8c436412008-07-03 12:51:14 +0000224 endrec=list(endrec)
225
226 # Append a blank comment and record start offset
227 endrec.append("")
228 endrec.append(filesize - sizeEndCentDir)
Martin v. Löwis8c436412008-07-03 12:51:14 +0000229
Amaury Forgeot d'Arc2407ac92009-01-17 22:43:50 +0000230 # Try to read the "Zip64 end of central directory" structure
231 return _EndRecData64(fpin, -sizeEndCentDir, endrec)
Martin v. Löwis8c436412008-07-03 12:51:14 +0000232
233 # Either this is not a ZIP file, or it is a ZIP file with an archive
234 # comment. Search the end of the file for the "end of central directory"
235 # record signature. The comment is the last item in the ZIP file and may be
236 # up to 64K long. It is assumed that the "end of central directory" magic
237 # number does not appear in the comment.
238 maxCommentStart = max(filesize - (1 << 16) - sizeEndCentDir, 0)
239 fpin.seek(maxCommentStart, 0)
Martin v. Löwis6f6873b2002-10-13 13:54:50 +0000240 data = fpin.read()
Amaury Forgeot d'Arcae6d2b92008-07-11 21:28:25 +0000241 start = data.rfind(stringEndArchive)
Martin v. Löwis8c436412008-07-03 12:51:14 +0000242 if start >= 0:
243 # found the magic number; attempt to unpack and interpret
244 recData = data[start:start+sizeEndCentDir]
Serhiy Storchaka0be506a2013-01-31 15:26:55 +0200245 if len(recData) != sizeEndCentDir:
246 # Zip file is corrupted.
247 return None
Amaury Forgeot d'Arcae6d2b92008-07-11 21:28:25 +0000248 endrec = list(struct.unpack(structEndArchive, recData))
R David Murray873c5832011-06-09 16:01:09 -0400249 commentSize = endrec[_ECD_COMMENT_SIZE] #as claimed by the zip file
250 comment = data[start+sizeEndCentDir:start+sizeEndCentDir+commentSize]
251 endrec.append(comment)
252 endrec.append(maxCommentStart + start)
Amaury Forgeot d'Arc2407ac92009-01-17 22:43:50 +0000253
R David Murray873c5832011-06-09 16:01:09 -0400254 # Try to read the "Zip64 end of central directory" structure
255 return _EndRecData64(fpin, maxCommentStart + start - filesize,
256 endrec)
Martin v. Löwis8c436412008-07-03 12:51:14 +0000257
258 # Unable to find a valid end of central directory structure
Serhiy Storchaka0be506a2013-01-31 15:26:55 +0200259 return None
Martin v. Löwis6f6873b2002-10-13 13:54:50 +0000260
Fred Drake484d7352000-10-02 21:14:52 +0000261
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000262class ZipInfo (object):
Fred Drake484d7352000-10-02 21:14:52 +0000263 """Class with attributes describing each file in the ZIP archive."""
264
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000265 __slots__ = (
266 'orig_filename',
267 'filename',
268 'date_time',
269 'compress_type',
270 'comment',
271 'extra',
272 'create_system',
273 'create_version',
274 'extract_version',
275 'reserved',
276 'flag_bits',
277 'volume',
278 'internal_attr',
279 'external_attr',
280 'header_offset',
281 'CRC',
282 'compress_size',
283 'file_size',
Gregory P. Smith0c63fc22008-01-20 01:21:03 +0000284 '_raw_time',
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000285 )
286
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000287 def __init__(self, filename="NoName", date_time=(1980,1,1,0,0,0)):
Greg Ward8e36d282003-06-18 00:53:06 +0000288 self.orig_filename = filename # Original file name in archive
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000289
290 # Terminate the file name at the first null byte. Null bytes in file
291 # names are used as tricks by viruses in archives.
Greg Ward8e36d282003-06-18 00:53:06 +0000292 null_byte = filename.find(chr(0))
293 if null_byte >= 0:
294 filename = filename[0:null_byte]
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000295 # This is used to ensure paths in generated ZIP files always use
296 # forward slashes as the directory separator, as required by the
297 # ZIP format specification.
298 if os.sep != "/" and os.sep in filename:
Greg Ward8e36d282003-06-18 00:53:06 +0000299 filename = filename.replace(os.sep, "/")
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000300
Greg Ward8e36d282003-06-18 00:53:06 +0000301 self.filename = filename # Normalized file name
Tim Peterse1190062001-01-15 03:34:38 +0000302 self.date_time = date_time # year, month, day, hour, min, sec
Senthil Kumaranddd40312011-10-20 01:38:35 +0800303
304 if date_time[0] < 1980:
305 raise ValueError('ZIP does not support timestamps before 1980')
306
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000307 # Standard values:
Tim Peterse1190062001-01-15 03:34:38 +0000308 self.compress_type = ZIP_STORED # Type of compression for the file
309 self.comment = "" # Comment for each file
310 self.extra = "" # ZIP extra data
Martin v. Löwis00756902006-02-05 17:09:41 +0000311 if sys.platform == 'win32':
312 self.create_system = 0 # System which created ZIP archive
313 else:
314 # Assume everything else is unix-y
315 self.create_system = 3 # System which created ZIP archive
Tim Peterse1190062001-01-15 03:34:38 +0000316 self.create_version = 20 # Version which created ZIP archive
317 self.extract_version = 20 # Version needed to extract archive
318 self.reserved = 0 # Must be zero
319 self.flag_bits = 0 # ZIP flag bits
320 self.volume = 0 # Volume number of file header
321 self.internal_attr = 0 # Internal attributes
322 self.external_attr = 0 # External file attributes
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000323 # Other attributes are set by class ZipFile:
Tim Peterse1190062001-01-15 03:34:38 +0000324 # header_offset Byte offset to the file header
Tim Peterse1190062001-01-15 03:34:38 +0000325 # CRC CRC-32 of the uncompressed file
326 # compress_size Size of the compressed file
327 # file_size Size of the uncompressed file
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000328
Serhiy Storchaka80a9fd72013-01-15 00:29:51 +0200329 def FileHeader(self, zip64=None):
Fred Drake484d7352000-10-02 21:14:52 +0000330 """Return the per-file header as a string."""
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000331 dt = self.date_time
332 dosdate = (dt[0] - 1980) << 9 | dt[1] << 5 | dt[2]
Tim Peters3caca232001-12-06 06:23:26 +0000333 dostime = dt[3] << 11 | dt[4] << 5 | (dt[5] // 2)
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000334 if self.flag_bits & 0x08:
Tim Peterse1190062001-01-15 03:34:38 +0000335 # Set these to zero because we write them after the file data
336 CRC = compress_size = file_size = 0
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000337 else:
Tim Peterse1190062001-01-15 03:34:38 +0000338 CRC = self.CRC
339 compress_size = self.compress_size
340 file_size = self.file_size
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000341
342 extra = self.extra
343
Serhiy Storchaka80a9fd72013-01-15 00:29:51 +0200344 if zip64 is None:
345 zip64 = file_size > ZIP64_LIMIT or compress_size > ZIP64_LIMIT
346 if zip64:
Gregory P. Smithbf02e3b2008-03-19 03:14:41 +0000347 fmt = '<HHQQ'
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000348 extra = extra + struct.pack(fmt,
349 1, struct.calcsize(fmt)-4, file_size, compress_size)
Serhiy Storchaka80a9fd72013-01-15 00:29:51 +0200350 if file_size > ZIP64_LIMIT or compress_size > ZIP64_LIMIT:
351 if not zip64:
352 raise LargeZipFile("Filesize would require ZIP64 extensions")
353 # File is larger than what fits into a 4 byte integer,
354 # fall back to the ZIP64 extension
Martin v. Löwis8c436412008-07-03 12:51:14 +0000355 file_size = 0xffffffff
356 compress_size = 0xffffffff
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000357 self.extract_version = max(45, self.extract_version)
358 self.create_version = max(45, self.extract_version)
359
Martin v. Löwis471617d2008-05-05 17:16:58 +0000360 filename, flag_bits = self._encodeFilenameFlags()
Amaury Forgeot d'Arcae6d2b92008-07-11 21:28:25 +0000361 header = struct.pack(structFileHeader, stringFileHeader,
Martin v. Löwis471617d2008-05-05 17:16:58 +0000362 self.extract_version, self.reserved, flag_bits,
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000363 self.compress_type, dostime, dosdate, CRC,
364 compress_size, file_size,
Martin v. Löwis471617d2008-05-05 17:16:58 +0000365 len(filename), len(extra))
366 return header + filename + extra
367
368 def _encodeFilenameFlags(self):
369 if isinstance(self.filename, unicode):
370 try:
371 return self.filename.encode('ascii'), self.flag_bits
372 except UnicodeEncodeError:
373 return self.filename.encode('utf-8'), self.flag_bits | 0x800
374 else:
375 return self.filename, self.flag_bits
376
377 def _decodeFilename(self):
378 if self.flag_bits & 0x800:
379 return self.filename.decode('utf-8')
380 else:
381 return self.filename
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000382
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000383 def _decodeExtra(self):
384 # Try to decode the extra field.
385 extra = self.extra
386 unpack = struct.unpack
387 while extra:
Gregory P. Smithbf02e3b2008-03-19 03:14:41 +0000388 tp, ln = unpack('<HH', extra[:4])
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000389 if tp == 1:
390 if ln >= 24:
Gregory P. Smithbf02e3b2008-03-19 03:14:41 +0000391 counts = unpack('<QQQ', extra[4:28])
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000392 elif ln == 16:
Gregory P. Smithbf02e3b2008-03-19 03:14:41 +0000393 counts = unpack('<QQ', extra[4:20])
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000394 elif ln == 8:
Gregory P. Smithbf02e3b2008-03-19 03:14:41 +0000395 counts = unpack('<Q', extra[4:12])
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000396 elif ln == 0:
397 counts = ()
398 else:
399 raise RuntimeError, "Corrupt extra field %s"%(ln,)
400
401 idx = 0
402
403 # ZIP64 extension (large files and/or large archives)
Gregory P. Smithbf02e3b2008-03-19 03:14:41 +0000404 if self.file_size in (0xffffffffffffffffL, 0xffffffffL):
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000405 self.file_size = counts[idx]
406 idx += 1
407
Martin v. Löwis8c436412008-07-03 12:51:14 +0000408 if self.compress_size == 0xFFFFFFFFL:
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000409 self.compress_size = counts[idx]
410 idx += 1
411
Martin v. Löwis8c436412008-07-03 12:51:14 +0000412 if self.header_offset == 0xffffffffL:
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000413 old = self.header_offset
414 self.header_offset = counts[idx]
415 idx+=1
416
417 extra = extra[ln+4:]
Tim Petersa608bb22006-06-15 18:06:29 +0000418
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000419
Martin v. Löwisc6d626e2007-02-13 09:49:38 +0000420class _ZipDecrypter:
421 """Class to handle decryption of files stored within a ZIP archive.
422
423 ZIP supports a password-based form of encryption. Even though known
424 plaintext attacks have been found against it, it is still useful
Gregory P. Smithda407232008-01-20 01:32:00 +0000425 to be able to get data out of such a file.
Martin v. Löwisc6d626e2007-02-13 09:49:38 +0000426
427 Usage:
428 zd = _ZipDecrypter(mypwd)
429 plain_char = zd(cypher_char)
430 plain_text = map(zd, cypher_text)
431 """
432
433 def _GenerateCRCTable():
434 """Generate a CRC-32 table.
435
436 ZIP encryption uses the CRC32 one-byte primitive for scrambling some
437 internal keys. We noticed that a direct implementation is faster than
438 relying on binascii.crc32().
439 """
440 poly = 0xedb88320
441 table = [0] * 256
442 for i in range(256):
443 crc = i
444 for j in range(8):
445 if crc & 1:
446 crc = ((crc >> 1) & 0x7FFFFFFF) ^ poly
447 else:
448 crc = ((crc >> 1) & 0x7FFFFFFF)
449 table[i] = crc
450 return table
451 crctable = _GenerateCRCTable()
452
453 def _crc32(self, ch, crc):
454 """Compute the CRC32 primitive on one byte."""
455 return ((crc >> 8) & 0xffffff) ^ self.crctable[(crc ^ ord(ch)) & 0xff]
456
457 def __init__(self, pwd):
458 self.key0 = 305419896
459 self.key1 = 591751049
460 self.key2 = 878082192
461 for p in pwd:
462 self._UpdateKeys(p)
463
464 def _UpdateKeys(self, c):
465 self.key0 = self._crc32(c, self.key0)
466 self.key1 = (self.key1 + (self.key0 & 255)) & 4294967295
467 self.key1 = (self.key1 * 134775813 + 1) & 4294967295
468 self.key2 = self._crc32(chr((self.key1 >> 24) & 255), self.key2)
469
470 def __call__(self, c):
471 """Decrypt a single character."""
472 c = ord(c)
473 k = self.key2 | 2
474 c = c ^ (((k * (k^1)) >> 8) & 255)
475 c = chr(c)
476 self._UpdateKeys(c)
477 return c
478
Ezio Melotti9e949722012-11-18 13:18:06 +0200479
480compressor_names = {
481 0: 'store',
482 1: 'shrink',
483 2: 'reduce',
484 3: 'reduce',
485 4: 'reduce',
486 5: 'reduce',
487 6: 'implode',
488 7: 'tokenize',
489 8: 'deflate',
490 9: 'deflate64',
491 10: 'implode',
492 12: 'bzip2',
493 14: 'lzma',
494 18: 'terse',
495 19: 'lz77',
496 97: 'wavpack',
497 98: 'ppmd',
498}
499
500
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000501class ZipExtFile(io.BufferedIOBase):
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000502 """File-like object for reading an archive member.
Tim Petersea5962f2007-03-12 18:07:52 +0000503 Is returned by ZipFile.open().
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000504 """
Tim Petersea5962f2007-03-12 18:07:52 +0000505
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000506 # Max size supported by decompressor.
507 MAX_N = 1 << 31 - 1
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000508
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000509 # Read from compressed files in 4k blocks.
510 MIN_READ_SIZE = 4096
Tim Petersea5962f2007-03-12 18:07:52 +0000511
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000512 # Search for universal newlines or line chunks.
513 PATTERN = re.compile(r'^(?P<chunk>[^\r\n]+)|(?P<newline>\n|\r\n?)')
514
Jesus Cea93d628b2012-11-04 02:32:08 +0100515 def __init__(self, fileobj, mode, zipinfo, decrypter=None,
516 close_fileobj=False):
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000517 self._fileobj = fileobj
518 self._decrypter = decrypter
Jesus Cea93d628b2012-11-04 02:32:08 +0100519 self._close_fileobj = close_fileobj
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000520
Ezio Melotti4611b052010-01-28 01:41:30 +0000521 self._compress_type = zipinfo.compress_type
522 self._compress_size = zipinfo.compress_size
523 self._compress_left = zipinfo.compress_size
524
525 if self._compress_type == ZIP_DEFLATED:
526 self._decompressor = zlib.decompressobj(-15)
Ezio Melotti9e949722012-11-18 13:18:06 +0200527 elif self._compress_type != ZIP_STORED:
528 descr = compressor_names.get(self._compress_type)
529 if descr:
530 raise NotImplementedError("compression type %d (%s)" % (self._compress_type, descr))
531 else:
532 raise NotImplementedError("compression type %d" % (self._compress_type,))
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000533 self._unconsumed = ''
534
535 self._readbuffer = ''
536 self._offset = 0
537
538 self._universal = 'U' in mode
539 self.newlines = None
540
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000541 # Adjust read size for encrypted files since the first 12 bytes
542 # are for the encryption/password information.
543 if self._decrypter is not None:
544 self._compress_left -= 12
545
546 self.mode = mode
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000547 self.name = zipinfo.filename
548
Antoine Pitroue1436d12010-08-12 15:25:51 +0000549 if hasattr(zipinfo, 'CRC'):
550 self._expected_crc = zipinfo.CRC
551 self._running_crc = crc32(b'') & 0xffffffff
552 else:
553 self._expected_crc = None
554
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000555 def readline(self, limit=-1):
556 """Read and return a line from the stream.
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000557
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000558 If limit is specified, at most limit bytes will be read.
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000559 """
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000560
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000561 if not self._universal and limit < 0:
562 # Shortcut common case - newline found in buffer.
563 i = self._readbuffer.find('\n', self._offset) + 1
564 if i > 0:
565 line = self._readbuffer[self._offset: i]
566 self._offset = i
567 return line
Tim Petersea5962f2007-03-12 18:07:52 +0000568
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000569 if not self._universal:
570 return io.BufferedIOBase.readline(self, limit)
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000571
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000572 line = ''
573 while limit < 0 or len(line) < limit:
574 readahead = self.peek(2)
575 if readahead == '':
576 return line
Tim Petersea5962f2007-03-12 18:07:52 +0000577
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000578 #
579 # Search for universal newlines or line chunks.
580 #
581 # The pattern returns either a line chunk or a newline, but not
582 # both. Combined with peek(2), we are assured that the sequence
583 # '\r\n' is always retrieved completely and never split into
584 # separate newlines - '\r', '\n' due to coincidental readaheads.
585 #
586 match = self.PATTERN.search(readahead)
587 newline = match.group('newline')
588 if newline is not None:
589 if self.newlines is None:
590 self.newlines = []
591 if newline not in self.newlines:
592 self.newlines.append(newline)
593 self._offset += len(newline)
594 return line + '\n'
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000595
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000596 chunk = match.group('chunk')
597 if limit >= 0:
598 chunk = chunk[: limit - len(line)]
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000599
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000600 self._offset += len(chunk)
601 line += chunk
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000602
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000603 return line
604
605 def peek(self, n=1):
606 """Returns buffered bytes without advancing the position."""
607 if n > len(self._readbuffer) - self._offset:
608 chunk = self.read(n)
609 self._offset -= len(chunk)
610
611 # Return up to 512 bytes to reduce allocation overhead for tight loops.
612 return self._readbuffer[self._offset: self._offset + 512]
613
614 def readable(self):
615 return True
616
617 def read(self, n=-1):
618 """Read and return up to n bytes.
619 If the argument is omitted, None, or negative, data is read and returned until EOF is reached..
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000620 """
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000621 buf = ''
Antoine Pitroue4195e82010-09-12 14:56:27 +0000622 if n is None:
623 n = -1
624 while True:
625 if n < 0:
626 data = self.read1(n)
627 elif n > len(buf):
628 data = self.read1(n - len(buf))
629 else:
630 return buf
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000631 if len(data) == 0:
632 return buf
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000633 buf += data
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000634
Antoine Pitroue1436d12010-08-12 15:25:51 +0000635 def _update_crc(self, newdata, eof):
636 # Update the CRC using the given data.
637 if self._expected_crc is None:
638 # No need to compute the CRC if we don't have a reference value
639 return
640 self._running_crc = crc32(newdata, self._running_crc) & 0xffffffff
641 # Check the CRC if we're at the end of the file
642 if eof and self._running_crc != self._expected_crc:
643 raise BadZipfile("Bad CRC-32 for file %r" % self.name)
644
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000645 def read1(self, n):
646 """Read up to n bytes with at most one read() system call."""
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000647
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000648 # Simplify algorithm (branching) by transforming negative n to large n.
649 if n < 0 or n is None:
650 n = self.MAX_N
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000651
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000652 # Bytes available in read buffer.
653 len_readbuffer = len(self._readbuffer) - self._offset
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000654
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000655 # Read from file.
656 if self._compress_left > 0 and n > len_readbuffer + len(self._unconsumed):
657 nbytes = n - len_readbuffer - len(self._unconsumed)
658 nbytes = max(nbytes, self.MIN_READ_SIZE)
659 nbytes = min(nbytes, self._compress_left)
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000660
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000661 data = self._fileobj.read(nbytes)
662 self._compress_left -= len(data)
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000663
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000664 if data and self._decrypter is not None:
665 data = ''.join(map(self._decrypter, data))
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000666
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000667 if self._compress_type == ZIP_STORED:
Antoine Pitroue1436d12010-08-12 15:25:51 +0000668 self._update_crc(data, eof=(self._compress_left==0))
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000669 self._readbuffer = self._readbuffer[self._offset:] + data
670 self._offset = 0
671 else:
672 # Prepare deflated bytes for decompression.
673 self._unconsumed += data
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000674
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000675 # Handle unconsumed data.
Ezio Melotti4611b052010-01-28 01:41:30 +0000676 if (len(self._unconsumed) > 0 and n > len_readbuffer and
677 self._compress_type == ZIP_DEFLATED):
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000678 data = self._decompressor.decompress(
679 self._unconsumed,
680 max(n - len_readbuffer, self.MIN_READ_SIZE)
681 )
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000682
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000683 self._unconsumed = self._decompressor.unconsumed_tail
Antoine Pitroue1436d12010-08-12 15:25:51 +0000684 eof = len(self._unconsumed) == 0 and self._compress_left == 0
685 if eof:
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000686 data += self._decompressor.flush()
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000687
Antoine Pitroue1436d12010-08-12 15:25:51 +0000688 self._update_crc(data, eof=eof)
Antoine Pitrou94c33eb2010-01-27 20:59:50 +0000689 self._readbuffer = self._readbuffer[self._offset:] + data
690 self._offset = 0
691
692 # Read from buffer.
693 data = self._readbuffer[self._offset: self._offset + n]
694 self._offset += len(data)
695 return data
696
Jesus Cea93d628b2012-11-04 02:32:08 +0100697 def close(self):
698 try :
699 if self._close_fileobj:
700 self._fileobj.close()
701 finally:
702 super(ZipExtFile, self).close()
Tim Petersea5962f2007-03-12 18:07:52 +0000703
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000704
R David Murray3f4ccba2012-04-12 18:42:47 -0400705class ZipFile(object):
Tim Petersa19a1682001-03-29 04:36:09 +0000706 """ Class with methods to open, read, write, close, list zip files.
707
Martin v. Löwis8c436412008-07-03 12:51:14 +0000708 z = ZipFile(file, mode="r", compression=ZIP_STORED, allowZip64=False)
Tim Petersa19a1682001-03-29 04:36:09 +0000709
Fred Drake3d9091e2001-03-26 15:49:24 +0000710 file: Either the path to the file, or a file-like object.
711 If it is a path, the file will be opened and closed by ZipFile.
712 mode: The mode can be either read "r", write "w" or append "a".
713 compression: ZIP_STORED (no compression) or ZIP_DEFLATED (requires zlib).
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000714 allowZip64: if True ZipFile will create files with ZIP64 extensions when
715 needed, otherwise it will raise an exception when this would
716 be necessary.
717
Fred Drake3d9091e2001-03-26 15:49:24 +0000718 """
Fred Drake484d7352000-10-02 21:14:52 +0000719
Fred Drake90eac282001-02-28 05:29:34 +0000720 fp = None # Set here since __del__ checks it
721
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000722 def __init__(self, file, mode="r", compression=ZIP_STORED, allowZip64=False):
Fred Drake484d7352000-10-02 21:14:52 +0000723 """Open the ZIP file with mode read "r", write "w" or append "a"."""
Georg Brandl4b3ab6f2007-07-12 09:59:22 +0000724 if mode not in ("r", "w", "a"):
725 raise RuntimeError('ZipFile() requires mode "r", "w", or "a"')
726
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000727 if compression == ZIP_STORED:
728 pass
729 elif compression == ZIP_DEFLATED:
730 if not zlib:
731 raise RuntimeError,\
Fred Drake5db246d2000-09-29 20:44:48 +0000732 "Compression requires the (missing) zlib module"
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000733 else:
734 raise RuntimeError, "That compression method is not supported"
Georg Brandl4b3ab6f2007-07-12 09:59:22 +0000735
736 self._allowZip64 = allowZip64
737 self._didModify = False
Tim Peterse1190062001-01-15 03:34:38 +0000738 self.debug = 0 # Level of printing: 0 through 3
739 self.NameToInfo = {} # Find file info given name
740 self.filelist = [] # List of ZipInfo instances for archive
741 self.compression = compression # Method of compression
Raymond Hettinger2ca7c192005-02-16 09:27:49 +0000742 self.mode = key = mode.replace('b', '')[0]
Martin v. Löwisc6d626e2007-02-13 09:49:38 +0000743 self.pwd = None
R David Murray3f4ccba2012-04-12 18:42:47 -0400744 self._comment = ''
Tim Petersa19a1682001-03-29 04:36:09 +0000745
Fred Drake3d9091e2001-03-26 15:49:24 +0000746 # Check if we were passed a file-like object
Walter Dörwald65230a22002-06-03 15:58:32 +0000747 if isinstance(file, basestring):
Fred Drake3d9091e2001-03-26 15:49:24 +0000748 self._filePassed = 0
749 self.filename = file
750 modeDict = {'r' : 'rb', 'w': 'wb', 'a' : 'r+b'}
Martin v. Löwis84f6de92007-02-13 10:10:39 +0000751 try:
752 self.fp = open(file, modeDict[mode])
753 except IOError:
754 if mode == 'a':
755 mode = key = 'w'
756 self.fp = open(file, modeDict[mode])
757 else:
758 raise
Fred Drake3d9091e2001-03-26 15:49:24 +0000759 else:
760 self._filePassed = 1
761 self.fp = file
762 self.filename = getattr(file, 'name', None)
Tim Petersa19a1682001-03-29 04:36:09 +0000763
Antoine Pitrou02512fb2012-11-17 23:56:53 +0100764 try:
765 if key == 'r':
Martin v. Löwis6f6873b2002-10-13 13:54:50 +0000766 self._RealGetContents()
Antoine Pitrou02512fb2012-11-17 23:56:53 +0100767 elif key == 'w':
Georg Brandl86e0c892010-11-26 07:22:28 +0000768 # set the modified flag so central directory gets written
769 # even if no files are added to the archive
770 self._didModify = True
Antoine Pitrou02512fb2012-11-17 23:56:53 +0100771 elif key == 'a':
772 try:
773 # See if file is a zip file
774 self._RealGetContents()
775 # seek to start of directory and overwrite
776 self.fp.seek(self.start_dir, 0)
777 except BadZipfile:
778 # file is not a zip file, just append
779 self.fp.seek(0, 2)
780
781 # set the modified flag so central directory gets written
782 # even if no files are added to the archive
783 self._didModify = True
784 else:
785 raise RuntimeError('Mode must be "r", "w" or "a"')
786 except:
787 fp = self.fp
788 self.fp = None
Tim Peters7d3bad62001-04-04 18:56:49 +0000789 if not self._filePassed:
Antoine Pitrou02512fb2012-11-17 23:56:53 +0100790 fp.close()
791 raise
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000792
Ezio Melotti569e61f2009-12-30 06:14:51 +0000793 def __enter__(self):
794 return self
795
796 def __exit__(self, type, value, traceback):
797 self.close()
798
Tim Peters7d3bad62001-04-04 18:56:49 +0000799 def _RealGetContents(self):
Fred Drake484d7352000-10-02 21:14:52 +0000800 """Read in the table of contents for the ZIP file."""
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000801 fp = self.fp
Georg Brandl86e0c892010-11-26 07:22:28 +0000802 try:
803 endrec = _EndRecData(fp)
804 except IOError:
805 raise BadZipfile("File is not a zip file")
Martin v. Löwis6f6873b2002-10-13 13:54:50 +0000806 if not endrec:
807 raise BadZipfile, "File is not a zip file"
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000808 if self.debug > 1:
809 print endrec
Martin v. Löwis8c436412008-07-03 12:51:14 +0000810 size_cd = endrec[_ECD_SIZE] # bytes in central directory
811 offset_cd = endrec[_ECD_OFFSET] # offset of central directory
R David Murray3f4ccba2012-04-12 18:42:47 -0400812 self._comment = endrec[_ECD_COMMENT] # archive comment
Martin v. Löwis8c436412008-07-03 12:51:14 +0000813
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000814 # "concat" is zero, unless zip was concatenated to another file
Martin v. Löwis8c436412008-07-03 12:51:14 +0000815 concat = endrec[_ECD_LOCATION] - size_cd - offset_cd
Antoine Pitrouebcd0ce2008-09-05 23:30:23 +0000816 if endrec[_ECD_SIGNATURE] == stringEndArchive64:
817 # If Zip64 extension structures are present, account for them
Martin v. Löwis8c436412008-07-03 12:51:14 +0000818 concat -= (sizeEndCentDir64 + sizeEndCentDir64Locator)
819
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000820 if self.debug > 2:
Martin v. Löwis8c436412008-07-03 12:51:14 +0000821 inferred = concat + offset_cd
822 print "given, inferred, offset", offset_cd, inferred, concat
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000823 # self.start_dir: Position of start of central directory
824 self.start_dir = offset_cd + concat
825 fp.seek(self.start_dir, 0)
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000826 data = fp.read(size_cd)
827 fp = cStringIO.StringIO(data)
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000828 total = 0
829 while total < size_cd:
Martin v. Löwis8c436412008-07-03 12:51:14 +0000830 centdir = fp.read(sizeCentralDir)
Serhiy Storchaka0be506a2013-01-31 15:26:55 +0200831 if len(centdir) != sizeCentralDir:
832 raise BadZipfile("Truncated central directory")
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000833 centdir = struct.unpack(structCentralDir, centdir)
Serhiy Storchaka0be506a2013-01-31 15:26:55 +0200834 if centdir[_CD_SIGNATURE] != stringCentralDir:
835 raise BadZipfile("Bad magic number for central directory")
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000836 if self.debug > 2:
837 print centdir
Fred Drake3e038e52001-02-28 17:56:26 +0000838 filename = fp.read(centdir[_CD_FILENAME_LENGTH])
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000839 # Create ZipInfo instance to store file information
840 x = ZipInfo(filename)
Fred Drake3e038e52001-02-28 17:56:26 +0000841 x.extra = fp.read(centdir[_CD_EXTRA_FIELD_LENGTH])
842 x.comment = fp.read(centdir[_CD_COMMENT_LENGTH])
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000843 x.header_offset = centdir[_CD_LOCAL_HEADER_OFFSET]
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000844 (x.create_version, x.create_system, x.extract_version, x.reserved,
845 x.flag_bits, x.compress_type, t, d,
846 x.CRC, x.compress_size, x.file_size) = centdir[1:12]
847 x.volume, x.internal_attr, x.external_attr = centdir[15:18]
848 # Convert date/time code to (year, month, day, hour, min, sec)
Gregory P. Smith0c63fc22008-01-20 01:21:03 +0000849 x._raw_time = t
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000850 x.date_time = ( (d>>9)+1980, (d>>5)&0xF, d&0x1F,
Fred Drake414ca662000-06-13 18:49:53 +0000851 t>>11, (t>>5)&0x3F, (t&0x1F) * 2 )
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000852
853 x._decodeExtra()
854 x.header_offset = x.header_offset + concat
Martin v. Löwis471617d2008-05-05 17:16:58 +0000855 x.filename = x._decodeFilename()
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000856 self.filelist.append(x)
857 self.NameToInfo[x.filename] = x
Martin v. Löwis8c436412008-07-03 12:51:14 +0000858
859 # update total bytes read from central directory
860 total = (total + sizeCentralDir + centdir[_CD_FILENAME_LENGTH]
861 + centdir[_CD_EXTRA_FIELD_LENGTH]
862 + centdir[_CD_COMMENT_LENGTH])
863
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000864 if self.debug > 2:
865 print "total", total
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000866
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000867
868 def namelist(self):
Fred Drake484d7352000-10-02 21:14:52 +0000869 """Return a list of file names in the archive."""
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000870 l = []
871 for data in self.filelist:
872 l.append(data.filename)
873 return l
874
875 def infolist(self):
Fred Drake484d7352000-10-02 21:14:52 +0000876 """Return a list of class ZipInfo instances for files in the
877 archive."""
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000878 return self.filelist
879
880 def printdir(self):
Fred Drake484d7352000-10-02 21:14:52 +0000881 """Print a table of contents for the zip file."""
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000882 print "%-46s %19s %12s" % ("File Name", "Modified ", "Size")
883 for zinfo in self.filelist:
Raymond Hettinger351e1a32008-01-14 22:58:05 +0000884 date = "%d-%02d-%02d %02d:%02d:%02d" % zinfo.date_time[:6]
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000885 print "%-46s %s %12d" % (zinfo.filename, date, zinfo.file_size)
886
887 def testzip(self):
Fred Drake484d7352000-10-02 21:14:52 +0000888 """Read all the files and check the CRC."""
Antoine Pitrouc5342702008-08-17 13:06:29 +0000889 chunk_size = 2 ** 20
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000890 for zinfo in self.filelist:
891 try:
Antoine Pitrouc5342702008-08-17 13:06:29 +0000892 # Read by chunks, to avoid an OverflowError or a
893 # MemoryError with very large embedded files.
Antoine Pitrou02512fb2012-11-17 23:56:53 +0100894 with self.open(zinfo.filename, "r") as f:
895 while f.read(chunk_size): # Check CRC-32
896 pass
Raymond Hettingerc0fac962003-06-27 22:25:03 +0000897 except BadZipfile:
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000898 return zinfo.filename
899
900 def getinfo(self, name):
Fred Drake484d7352000-10-02 21:14:52 +0000901 """Return the instance of ZipInfo given 'name'."""
Georg Brandl4b3ab6f2007-07-12 09:59:22 +0000902 info = self.NameToInfo.get(name)
903 if info is None:
904 raise KeyError(
905 'There is no item named %r in the archive' % name)
906
907 return info
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000908
Martin v. Löwisc6d626e2007-02-13 09:49:38 +0000909 def setpassword(self, pwd):
910 """Set default password for encrypted files."""
911 self.pwd = pwd
912
R David Murray3f4ccba2012-04-12 18:42:47 -0400913 @property
914 def comment(self):
915 """The comment text associated with the ZIP file."""
916 return self._comment
917
918 @comment.setter
919 def comment(self, comment):
920 # check for valid comment length
921 if len(comment) >= ZIP_MAX_COMMENT:
922 if self.debug:
923 print('Archive comment is too long; truncating to %d bytes'
924 % ZIP_MAX_COMMENT)
925 comment = comment[:ZIP_MAX_COMMENT]
926 self._comment = comment
927 self._didModify = True
928
Martin v. Löwisc6d626e2007-02-13 09:49:38 +0000929 def read(self, name, pwd=None):
Fred Drake484d7352000-10-02 21:14:52 +0000930 """Return file bytes (as a string) for name."""
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000931 return self.open(name, "r", pwd).read()
932
933 def open(self, name, mode="r", pwd=None):
934 """Return file-like object for 'name'."""
935 if mode not in ("r", "U", "rU"):
936 raise RuntimeError, 'open() requires mode "r", "U", or "rU"'
Guido van Rossum32abe6f2000-03-31 17:30:02 +0000937 if not self.fp:
938 raise RuntimeError, \
Fred Drake5db246d2000-09-29 20:44:48 +0000939 "Attempt to read ZIP archive that was already closed"
Ronald Oussoren143cefb2006-06-15 08:14:18 +0000940
Tim Petersea5962f2007-03-12 18:07:52 +0000941 # Only open a new file for instances where we were not
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000942 # given a file object in the constructor
943 if self._filePassed:
944 zef_file = self.fp
Jesus Cea93d628b2012-11-04 02:32:08 +0100945 should_close = False
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000946 else:
947 zef_file = open(self.filename, 'rb')
Jesus Cea93d628b2012-11-04 02:32:08 +0100948 should_close = True
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000949
Antoine Pitrou02512fb2012-11-17 23:56:53 +0100950 try:
951 # Make sure we have an info object
952 if isinstance(name, ZipInfo):
953 # 'name' is already an info object
954 zinfo = name
Gregory P. Smith0c63fc22008-01-20 01:21:03 +0000955 else:
Antoine Pitrou02512fb2012-11-17 23:56:53 +0100956 # Get info object for name
957 zinfo = self.getinfo(name)
Martin v. Löwis3eb76482007-03-06 10:41:24 +0000958
Antoine Pitrou02512fb2012-11-17 23:56:53 +0100959 zef_file.seek(zinfo.header_offset, 0)
960
961 # Skip the file header:
962 fheader = zef_file.read(sizeFileHeader)
Serhiy Storchaka0be506a2013-01-31 15:26:55 +0200963 if len(fheader) != sizeFileHeader:
964 raise BadZipfile("Truncated file header")
Antoine Pitrou02512fb2012-11-17 23:56:53 +0100965 fheader = struct.unpack(structFileHeader, fheader)
Serhiy Storchaka0be506a2013-01-31 15:26:55 +0200966 if fheader[_FH_SIGNATURE] != stringFileHeader:
967 raise BadZipfile("Bad magic number for file header")
968
Antoine Pitrou02512fb2012-11-17 23:56:53 +0100969 fname = zef_file.read(fheader[_FH_FILENAME_LENGTH])
970 if fheader[_FH_EXTRA_FIELD_LENGTH]:
971 zef_file.read(fheader[_FH_EXTRA_FIELD_LENGTH])
972
973 if fname != zinfo.orig_filename:
974 raise BadZipfile, \
975 'File name in directory "%s" and header "%s" differ.' % (
976 zinfo.orig_filename, fname)
977
978 # check for encrypted flag & handle password
979 is_encrypted = zinfo.flag_bits & 0x1
980 zd = None
981 if is_encrypted:
982 if not pwd:
983 pwd = self.pwd
984 if not pwd:
985 raise RuntimeError, "File %s is encrypted, " \
986 "password required for extraction" % name
987
988 zd = _ZipDecrypter(pwd)
989 # The first 12 bytes in the cypher stream is an encryption header
990 # used to strengthen the algorithm. The first 11 bytes are
991 # completely random, while the 12th contains the MSB of the CRC,
992 # or the MSB of the file time depending on the header type
993 # and is used to check the correctness of the password.
994 bytes = zef_file.read(12)
995 h = map(zd, bytes[0:12])
996 if zinfo.flag_bits & 0x8:
997 # compare against the file type from extended local headers
998 check_byte = (zinfo._raw_time >> 8) & 0xff
999 else:
1000 # compare against the CRC otherwise
1001 check_byte = (zinfo.CRC >> 24) & 0xff
1002 if ord(h[11]) != check_byte:
1003 raise RuntimeError("Bad password for file", name)
1004
1005 return ZipExtFile(zef_file, mode, zinfo, zd,
1006 close_fileobj=should_close)
1007 except:
1008 if should_close:
1009 zef_file.close()
1010 raise
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001011
Georg Brandl62416bc2008-01-07 18:47:44 +00001012 def extract(self, member, path=None, pwd=None):
1013 """Extract a member from the archive to the current working directory,
1014 using its full name. Its file information is extracted as accurately
1015 as possible. `member' may be a filename or a ZipInfo object. You can
1016 specify a different directory using `path'.
1017 """
1018 if not isinstance(member, ZipInfo):
1019 member = self.getinfo(member)
1020
1021 if path is None:
1022 path = os.getcwd()
1023
1024 return self._extract_member(member, path, pwd)
1025
1026 def extractall(self, path=None, members=None, pwd=None):
1027 """Extract all members from the archive to the current working
1028 directory. `path' specifies a different directory to extract to.
1029 `members' is optional and must be a subset of the list returned
1030 by namelist().
1031 """
1032 if members is None:
1033 members = self.namelist()
1034
1035 for zipinfo in members:
1036 self.extract(zipinfo, path, pwd)
1037
1038 def _extract_member(self, member, targetpath, pwd):
1039 """Extract the ZipInfo object 'member' to a physical
1040 file on the path targetpath.
1041 """
1042 # build the destination pathname, replacing
1043 # forward slashes to platform specific separators.
Gregory P. Smith608cc452013-02-01 11:40:18 -08001044 arcname = member.filename.replace('/', os.path.sep)
Georg Brandl62416bc2008-01-07 18:47:44 +00001045
Gregory P. Smith608cc452013-02-01 11:40:18 -08001046 if os.path.altsep:
1047 arcname = arcname.replace(os.path.altsep, os.path.sep)
1048 # interpret absolute pathname as relative, remove drive letter or
1049 # UNC path, redundant separators, "." and ".." components.
1050 arcname = os.path.splitdrive(arcname)[1]
1051 arcname = os.path.sep.join(x for x in arcname.split(os.path.sep)
1052 if x not in ('', os.path.curdir, os.path.pardir))
Gregory P. Smith608cc452013-02-01 11:40:18 -08001053 if os.path.sep == '\\':
Serhiy Storchaka13e56c72013-02-02 17:46:33 +02001054 # filter illegal characters on Windows
Gregory P. Smith608cc452013-02-01 11:40:18 -08001055 illegal = ':<>|"?*'
Serhiy Storchaka6fa83f92013-04-13 12:28:17 +03001056 if isinstance(arcname, unicode):
1057 table = {ord(c): ord('_') for c in illegal}
1058 else:
1059 table = string.maketrans(illegal, '_' * len(illegal))
Gregory P. Smith608cc452013-02-01 11:40:18 -08001060 arcname = arcname.translate(table)
Serhiy Storchaka13e56c72013-02-02 17:46:33 +02001061 # remove trailing dots
1062 arcname = (x.rstrip('.') for x in arcname.split(os.path.sep))
1063 arcname = os.path.sep.join(x for x in arcname if x)
Georg Brandl62416bc2008-01-07 18:47:44 +00001064
Gregory P. Smith608cc452013-02-01 11:40:18 -08001065 targetpath = os.path.join(targetpath, arcname)
Georg Brandl62416bc2008-01-07 18:47:44 +00001066 targetpath = os.path.normpath(targetpath)
1067
1068 # Create all upper directories if necessary.
1069 upperdirs = os.path.dirname(targetpath)
1070 if upperdirs and not os.path.exists(upperdirs):
1071 os.makedirs(upperdirs)
1072
Martin v. Löwis0dfcfc82009-01-24 14:00:33 +00001073 if member.filename[-1] == '/':
Martin v. Löwis0b09c422009-05-24 19:30:52 +00001074 if not os.path.isdir(targetpath):
1075 os.mkdir(targetpath)
Martin v. Löwis0dfcfc82009-01-24 14:00:33 +00001076 return targetpath
1077
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001078 with self.open(member, pwd=pwd) as source, \
1079 file(targetpath, "wb") as target:
1080 shutil.copyfileobj(source, target)
Georg Brandl62416bc2008-01-07 18:47:44 +00001081
1082 return targetpath
1083
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001084 def _writecheck(self, zinfo):
Fred Drake484d7352000-10-02 21:14:52 +00001085 """Check for errors before writing a file to the archive."""
Raymond Hettinger54f02222002-06-01 14:18:47 +00001086 if zinfo.filename in self.NameToInfo:
Tim Peterse1190062001-01-15 03:34:38 +00001087 if self.debug: # Warning for duplicate names
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001088 print "Duplicate name:", zinfo.filename
1089 if self.mode not in ("w", "a"):
1090 raise RuntimeError, 'write() requires mode "w" or "a"'
1091 if not self.fp:
1092 raise RuntimeError, \
Fred Drake5db246d2000-09-29 20:44:48 +00001093 "Attempt to write ZIP archive that was already closed"
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001094 if zinfo.compress_type == ZIP_DEFLATED and not zlib:
1095 raise RuntimeError, \
Fred Drake5db246d2000-09-29 20:44:48 +00001096 "Compression requires the (missing) zlib module"
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001097 if zinfo.compress_type not in (ZIP_STORED, ZIP_DEFLATED):
1098 raise RuntimeError, \
Fred Drake5db246d2000-09-29 20:44:48 +00001099 "That compression method is not supported"
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001100 if zinfo.file_size > ZIP64_LIMIT:
1101 if not self._allowZip64:
1102 raise LargeZipFile("Filesize would require ZIP64 extensions")
1103 if zinfo.header_offset > ZIP64_LIMIT:
1104 if not self._allowZip64:
1105 raise LargeZipFile("Zipfile size would require ZIP64 extensions")
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001106
1107 def write(self, filename, arcname=None, compress_type=None):
Fred Drake484d7352000-10-02 21:14:52 +00001108 """Put the bytes from filename into the archive under the name
1109 arcname."""
Georg Brandl4b3ab6f2007-07-12 09:59:22 +00001110 if not self.fp:
1111 raise RuntimeError(
1112 "Attempt to write to ZIP archive that was already closed")
1113
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001114 st = os.stat(filename)
Martin v. Löwis0dfcfc82009-01-24 14:00:33 +00001115 isdir = stat.S_ISDIR(st.st_mode)
Raymond Hettinger32200ae2002-06-01 19:51:15 +00001116 mtime = time.localtime(st.st_mtime)
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001117 date_time = mtime[0:6]
1118 # Create ZipInfo instance to store file information
1119 if arcname is None:
Georg Brandl8f7c54e2006-02-20 08:40:38 +00001120 arcname = filename
1121 arcname = os.path.normpath(os.path.splitdrive(arcname)[1])
1122 while arcname[0] in (os.sep, os.altsep):
1123 arcname = arcname[1:]
Martin v. Löwis0dfcfc82009-01-24 14:00:33 +00001124 if isdir:
1125 arcname += '/'
Georg Brandl8f7c54e2006-02-20 08:40:38 +00001126 zinfo = ZipInfo(arcname, date_time)
Andrew M. Kuchling55430212004-07-10 15:40:29 +00001127 zinfo.external_attr = (st[0] & 0xFFFF) << 16L # Unix attributes
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001128 if compress_type is None:
Tim Peterse1190062001-01-15 03:34:38 +00001129 zinfo.compress_type = self.compression
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001130 else:
Tim Peterse1190062001-01-15 03:34:38 +00001131 zinfo.compress_type = compress_type
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001132
1133 zinfo.file_size = st.st_size
Finn Bock03a3bb82001-09-05 18:40:33 +00001134 zinfo.flag_bits = 0x00
Tim Peterse1190062001-01-15 03:34:38 +00001135 zinfo.header_offset = self.fp.tell() # Start of header bytes
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001136
1137 self._writecheck(zinfo)
1138 self._didModify = True
Martin v. Löwis0dfcfc82009-01-24 14:00:33 +00001139
1140 if isdir:
1141 zinfo.file_size = 0
1142 zinfo.compress_size = 0
1143 zinfo.CRC = 0
1144 self.filelist.append(zinfo)
1145 self.NameToInfo[zinfo.filename] = zinfo
Serhiy Storchaka80a9fd72013-01-15 00:29:51 +02001146 self.fp.write(zinfo.FileHeader(False))
Martin v. Löwis0dfcfc82009-01-24 14:00:33 +00001147 return
1148
Benjamin Petersonb91e8ed2009-05-10 02:29:00 +00001149 with open(filename, "rb") as fp:
1150 # Must overwrite CRC and sizes with correct data later
1151 zinfo.CRC = CRC = 0
1152 zinfo.compress_size = compress_size = 0
Serhiy Storchaka80a9fd72013-01-15 00:29:51 +02001153 # Compressed size can be larger than uncompressed size
1154 zip64 = self._allowZip64 and \
1155 zinfo.file_size * 1.05 > ZIP64_LIMIT
1156 self.fp.write(zinfo.FileHeader(zip64))
Benjamin Petersonb91e8ed2009-05-10 02:29:00 +00001157 if zinfo.compress_type == ZIP_DEFLATED:
1158 cmpr = zlib.compressobj(zlib.Z_DEFAULT_COMPRESSION,
1159 zlib.DEFLATED, -15)
1160 else:
1161 cmpr = None
Serhiy Storchaka80a9fd72013-01-15 00:29:51 +02001162 file_size = 0
Benjamin Petersonb91e8ed2009-05-10 02:29:00 +00001163 while 1:
1164 buf = fp.read(1024 * 8)
1165 if not buf:
1166 break
1167 file_size = file_size + len(buf)
1168 CRC = crc32(buf, CRC) & 0xffffffff
1169 if cmpr:
1170 buf = cmpr.compress(buf)
1171 compress_size = compress_size + len(buf)
1172 self.fp.write(buf)
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001173 if cmpr:
1174 buf = cmpr.flush()
1175 compress_size = compress_size + len(buf)
1176 self.fp.write(buf)
1177 zinfo.compress_size = compress_size
1178 else:
1179 zinfo.compress_size = file_size
1180 zinfo.CRC = CRC
1181 zinfo.file_size = file_size
Serhiy Storchaka80a9fd72013-01-15 00:29:51 +02001182 if not zip64 and self._allowZip64:
1183 if file_size > ZIP64_LIMIT:
1184 raise RuntimeError('File size has increased during compressing')
1185 if compress_size > ZIP64_LIMIT:
1186 raise RuntimeError('Compressed size larger than uncompressed size')
1187 # Seek backwards and write file header (which will now include
1188 # correct CRC and file sizes)
Tim Petersb64bec32001-09-18 02:26:39 +00001189 position = self.fp.tell() # Preserve current position in file
Serhiy Storchaka80a9fd72013-01-15 00:29:51 +02001190 self.fp.seek(zinfo.header_offset, 0)
1191 self.fp.write(zinfo.FileHeader(zip64))
Finn Bock03a3bb82001-09-05 18:40:33 +00001192 self.fp.seek(position, 0)
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001193 self.filelist.append(zinfo)
1194 self.NameToInfo[zinfo.filename] = zinfo
1195
Ronald Oussorendd25e862010-02-07 20:18:02 +00001196 def writestr(self, zinfo_or_arcname, bytes, compress_type=None):
Fred Drake484d7352000-10-02 21:14:52 +00001197 """Write a file into the archive. The contents is the string
Just van Rossumb083cb32002-12-12 12:23:32 +00001198 'bytes'. 'zinfo_or_arcname' is either a ZipInfo instance or
1199 the name of the file in the archive."""
1200 if not isinstance(zinfo_or_arcname, ZipInfo):
1201 zinfo = ZipInfo(filename=zinfo_or_arcname,
Raymond Hettinger351e1a32008-01-14 22:58:05 +00001202 date_time=time.localtime(time.time())[:6])
Ronald Oussorendd25e862010-02-07 20:18:02 +00001203
Just van Rossumb083cb32002-12-12 12:23:32 +00001204 zinfo.compress_type = self.compression
Antoine Pitrou5fdfa3e2008-07-25 19:42:26 +00001205 zinfo.external_attr = 0600 << 16
Just van Rossumb083cb32002-12-12 12:23:32 +00001206 else:
1207 zinfo = zinfo_or_arcname
Georg Brandl4b3ab6f2007-07-12 09:59:22 +00001208
1209 if not self.fp:
1210 raise RuntimeError(
1211 "Attempt to write to ZIP archive that was already closed")
1212
Ronald Oussorendd25e862010-02-07 20:18:02 +00001213 if compress_type is not None:
1214 zinfo.compress_type = compress_type
1215
Tim Peterse1190062001-01-15 03:34:38 +00001216 zinfo.file_size = len(bytes) # Uncompressed size
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001217 zinfo.header_offset = self.fp.tell() # Start of header bytes
1218 self._writecheck(zinfo)
1219 self._didModify = True
Gregory P. Smithbf02e3b2008-03-19 03:14:41 +00001220 zinfo.CRC = crc32(bytes) & 0xffffffff # CRC-32 checksum
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001221 if zinfo.compress_type == ZIP_DEFLATED:
1222 co = zlib.compressobj(zlib.Z_DEFAULT_COMPRESSION,
1223 zlib.DEFLATED, -15)
1224 bytes = co.compress(bytes) + co.flush()
Tim Peterse1190062001-01-15 03:34:38 +00001225 zinfo.compress_size = len(bytes) # Compressed size
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001226 else:
1227 zinfo.compress_size = zinfo.file_size
Serhiy Storchaka80a9fd72013-01-15 00:29:51 +02001228 zip64 = zinfo.file_size > ZIP64_LIMIT or \
1229 zinfo.compress_size > ZIP64_LIMIT
1230 if zip64 and not self._allowZip64:
1231 raise LargeZipFile("Filesize would require ZIP64 extensions")
1232 self.fp.write(zinfo.FileHeader(zip64))
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001233 self.fp.write(bytes)
1234 if zinfo.flag_bits & 0x08:
Tim Peterse1190062001-01-15 03:34:38 +00001235 # Write CRC and file sizes after the file data
Serhiy Storchaka80a9fd72013-01-15 00:29:51 +02001236 fmt = '<LQQ' if zip64 else '<LLL'
1237 self.fp.write(struct.pack(fmt, zinfo.CRC, zinfo.compress_size,
Tim Peterse1190062001-01-15 03:34:38 +00001238 zinfo.file_size))
Serhiy Storchaka80a9fd72013-01-15 00:29:51 +02001239 self.fp.flush()
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001240 self.filelist.append(zinfo)
1241 self.NameToInfo[zinfo.filename] = zinfo
1242
1243 def __del__(self):
Fred Drake484d7352000-10-02 21:14:52 +00001244 """Call the "close()" method in case the user forgot."""
Tim Petersd15f8bb2001-11-28 23:16:40 +00001245 self.close()
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001246
1247 def close(self):
Fred Drake484d7352000-10-02 21:14:52 +00001248 """Close the file, and for mode "w" and "a" write the ending
1249 records."""
Tim Petersd15f8bb2001-11-28 23:16:40 +00001250 if self.fp is None:
1251 return
Tim Petersa608bb22006-06-15 18:06:29 +00001252
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001253 try:
1254 if self.mode in ("w", "a") and self._didModify: # write ending records
1255 count = 0
1256 pos1 = self.fp.tell()
1257 for zinfo in self.filelist: # write central directory
1258 count = count + 1
1259 dt = zinfo.date_time
1260 dosdate = (dt[0] - 1980) << 9 | dt[1] << 5 | dt[2]
1261 dostime = dt[3] << 11 | dt[4] << 5 | (dt[5] // 2)
1262 extra = []
1263 if zinfo.file_size > ZIP64_LIMIT \
1264 or zinfo.compress_size > ZIP64_LIMIT:
1265 extra.append(zinfo.file_size)
1266 extra.append(zinfo.compress_size)
1267 file_size = 0xffffffff
1268 compress_size = 0xffffffff
1269 else:
1270 file_size = zinfo.file_size
1271 compress_size = zinfo.compress_size
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001272
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001273 if zinfo.header_offset > ZIP64_LIMIT:
1274 extra.append(zinfo.header_offset)
1275 header_offset = 0xffffffffL
1276 else:
1277 header_offset = zinfo.header_offset
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001278
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001279 extra_data = zinfo.extra
1280 if extra:
1281 # Append a ZIP64 field to the extra's
1282 extra_data = struct.pack(
1283 '<HH' + 'Q'*len(extra),
1284 1, 8*len(extra), *extra) + extra_data
Tim Petersa608bb22006-06-15 18:06:29 +00001285
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001286 extract_version = max(45, zinfo.extract_version)
1287 create_version = max(45, zinfo.create_version)
1288 else:
1289 extract_version = zinfo.extract_version
1290 create_version = zinfo.create_version
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001291
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001292 try:
1293 filename, flag_bits = zinfo._encodeFilenameFlags()
1294 centdir = struct.pack(structCentralDir,
1295 stringCentralDir, create_version,
1296 zinfo.create_system, extract_version, zinfo.reserved,
1297 flag_bits, zinfo.compress_type, dostime, dosdate,
1298 zinfo.CRC, compress_size, file_size,
1299 len(filename), len(extra_data), len(zinfo.comment),
1300 0, zinfo.internal_attr, zinfo.external_attr,
1301 header_offset)
1302 except DeprecationWarning:
1303 print >>sys.stderr, (structCentralDir,
1304 stringCentralDir, create_version,
1305 zinfo.create_system, extract_version, zinfo.reserved,
1306 zinfo.flag_bits, zinfo.compress_type, dostime, dosdate,
1307 zinfo.CRC, compress_size, file_size,
1308 len(zinfo.filename), len(extra_data), len(zinfo.comment),
1309 0, zinfo.internal_attr, zinfo.external_attr,
1310 header_offset)
1311 raise
1312 self.fp.write(centdir)
1313 self.fp.write(filename)
1314 self.fp.write(extra_data)
1315 self.fp.write(zinfo.comment)
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001316
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001317 pos2 = self.fp.tell()
1318 # Write end-of-zip-archive record
1319 centDirCount = count
1320 centDirSize = pos2 - pos1
1321 centDirOffset = pos1
1322 if (centDirCount >= ZIP_FILECOUNT_LIMIT or
1323 centDirOffset > ZIP64_LIMIT or
1324 centDirSize > ZIP64_LIMIT):
1325 # Need to write the ZIP64 end-of-archive records
1326 zip64endrec = struct.pack(
1327 structEndArchive64, stringEndArchive64,
1328 44, 45, 45, 0, 0, centDirCount, centDirCount,
1329 centDirSize, centDirOffset)
1330 self.fp.write(zip64endrec)
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001331
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001332 zip64locrec = struct.pack(
1333 structEndArchive64Locator,
1334 stringEndArchive64Locator, 0, pos2, 1)
1335 self.fp.write(zip64locrec)
1336 centDirCount = min(centDirCount, 0xFFFF)
1337 centDirSize = min(centDirSize, 0xFFFFFFFF)
1338 centDirOffset = min(centDirOffset, 0xFFFFFFFF)
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001339
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001340 endrec = struct.pack(structEndArchive, stringEndArchive,
1341 0, 0, centDirCount, centDirCount,
1342 centDirSize, centDirOffset, len(self._comment))
1343 self.fp.write(endrec)
1344 self.fp.write(self._comment)
1345 self.fp.flush()
1346 finally:
1347 fp = self.fp
1348 self.fp = None
1349 if not self._filePassed:
1350 fp.close()
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001351
1352
1353class PyZipFile(ZipFile):
Fred Drake484d7352000-10-02 21:14:52 +00001354 """Class to create ZIP archives with Python library files and packages."""
1355
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001356 def writepy(self, pathname, basename = ""):
1357 """Add all files from "pathname" to the ZIP archive.
1358
Fred Drake484d7352000-10-02 21:14:52 +00001359 If pathname is a package directory, search the directory and
1360 all package subdirectories recursively for all *.py and enter
1361 the modules into the archive. If pathname is a plain
1362 directory, listdir *.py and enter all modules. Else, pathname
1363 must be a Python *.py file and the module will be put into the
1364 archive. Added modules are always module.pyo or module.pyc.
1365 This method will compile the module.py into module.pyc if
1366 necessary.
1367 """
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001368 dir, name = os.path.split(pathname)
1369 if os.path.isdir(pathname):
1370 initname = os.path.join(pathname, "__init__.py")
1371 if os.path.isfile(initname):
1372 # This is a package directory, add it
1373 if basename:
1374 basename = "%s/%s" % (basename, name)
1375 else:
1376 basename = name
1377 if self.debug:
1378 print "Adding package in", pathname, "as", basename
1379 fname, arcname = self._get_codename(initname[0:-3], basename)
1380 if self.debug:
1381 print "Adding", arcname
1382 self.write(fname, arcname)
1383 dirlist = os.listdir(pathname)
1384 dirlist.remove("__init__.py")
1385 # Add all *.py files and package subdirectories
1386 for filename in dirlist:
1387 path = os.path.join(pathname, filename)
1388 root, ext = os.path.splitext(filename)
1389 if os.path.isdir(path):
1390 if os.path.isfile(os.path.join(path, "__init__.py")):
1391 # This is a package directory, add it
1392 self.writepy(path, basename) # Recursive call
1393 elif ext == ".py":
1394 fname, arcname = self._get_codename(path[0:-3],
1395 basename)
1396 if self.debug:
1397 print "Adding", arcname
1398 self.write(fname, arcname)
1399 else:
1400 # This is NOT a package directory, add its files at top level
1401 if self.debug:
1402 print "Adding files from directory", pathname
1403 for filename in os.listdir(pathname):
1404 path = os.path.join(pathname, filename)
1405 root, ext = os.path.splitext(filename)
1406 if ext == ".py":
1407 fname, arcname = self._get_codename(path[0:-3],
1408 basename)
1409 if self.debug:
1410 print "Adding", arcname
1411 self.write(fname, arcname)
1412 else:
1413 if pathname[-3:] != ".py":
1414 raise RuntimeError, \
Fred Drake5db246d2000-09-29 20:44:48 +00001415 'Files added with writepy() must end with ".py"'
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001416 fname, arcname = self._get_codename(pathname[0:-3], basename)
1417 if self.debug:
1418 print "Adding file", arcname
1419 self.write(fname, arcname)
1420
1421 def _get_codename(self, pathname, basename):
1422 """Return (filename, archivename) for the path.
1423
Fred Drake484d7352000-10-02 21:14:52 +00001424 Given a module name path, return the correct file path and
1425 archive name, compiling if necessary. For example, given
1426 /python/lib/string, return (/python/lib/string.pyc, string).
1427 """
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001428 file_py = pathname + ".py"
1429 file_pyc = pathname + ".pyc"
1430 file_pyo = pathname + ".pyo"
1431 if os.path.isfile(file_pyo) and \
Raymond Hettinger32200ae2002-06-01 19:51:15 +00001432 os.stat(file_pyo).st_mtime >= os.stat(file_py).st_mtime:
Tim Peterse1190062001-01-15 03:34:38 +00001433 fname = file_pyo # Use .pyo file
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001434 elif not os.path.isfile(file_pyc) or \
Raymond Hettinger32200ae2002-06-01 19:51:15 +00001435 os.stat(file_pyc).st_mtime < os.stat(file_py).st_mtime:
Fred Drake484d7352000-10-02 21:14:52 +00001436 import py_compile
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001437 if self.debug:
1438 print "Compiling", file_py
Martin v. Löwis0c6774d2003-01-15 11:51:06 +00001439 try:
1440 py_compile.compile(file_py, file_pyc, None, True)
1441 except py_compile.PyCompileError,err:
1442 print err.msg
Guido van Rossum32abe6f2000-03-31 17:30:02 +00001443 fname = file_pyc
1444 else:
1445 fname = file_pyc
1446 archivename = os.path.split(fname)[1]
1447 if basename:
1448 archivename = "%s/%s" % (basename, archivename)
1449 return (fname, archivename)
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001450
1451
1452def main(args = None):
1453 import textwrap
1454 USAGE=textwrap.dedent("""\
1455 Usage:
1456 zipfile.py -l zipfile.zip # Show listing of a zipfile
1457 zipfile.py -t zipfile.zip # Test if a zipfile is valid
1458 zipfile.py -e zipfile.zip target # Extract zipfile into target dir
1459 zipfile.py -c zipfile.zip src ... # Create zipfile from sources
1460 """)
1461 if args is None:
1462 args = sys.argv[1:]
1463
1464 if not args or args[0] not in ('-l', '-c', '-e', '-t'):
1465 print USAGE
1466 sys.exit(1)
1467
1468 if args[0] == '-l':
1469 if len(args) != 2:
1470 print USAGE
1471 sys.exit(1)
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001472 with ZipFile(args[1], 'r') as zf:
1473 zf.printdir()
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001474
1475 elif args[0] == '-t':
1476 if len(args) != 2:
1477 print USAGE
1478 sys.exit(1)
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001479 with ZipFile(args[1], 'r') as zf:
1480 badfile = zf.testzip()
Antoine Pitroue1436d12010-08-12 15:25:51 +00001481 if badfile:
1482 print("The following enclosed file is corrupted: {!r}".format(badfile))
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001483 print "Done testing"
1484
1485 elif args[0] == '-e':
1486 if len(args) != 3:
1487 print USAGE
1488 sys.exit(1)
1489
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001490 with ZipFile(args[1], 'r') as zf:
1491 out = args[2]
1492 for path in zf.namelist():
1493 if path.startswith('./'):
1494 tgt = os.path.join(out, path[2:])
1495 else:
1496 tgt = os.path.join(out, path)
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001497
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001498 tgtdir = os.path.dirname(tgt)
1499 if not os.path.exists(tgtdir):
1500 os.makedirs(tgtdir)
1501 with open(tgt, 'wb') as fp:
1502 fp.write(zf.read(path))
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001503
1504 elif args[0] == '-c':
1505 if len(args) < 3:
1506 print USAGE
1507 sys.exit(1)
1508
1509 def addToZip(zf, path, zippath):
1510 if os.path.isfile(path):
1511 zf.write(path, zippath, ZIP_DEFLATED)
1512 elif os.path.isdir(path):
1513 for nm in os.listdir(path):
Tim Petersa608bb22006-06-15 18:06:29 +00001514 addToZip(zf,
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001515 os.path.join(path, nm), os.path.join(zippath, nm))
Tim Petersa608bb22006-06-15 18:06:29 +00001516 # else: ignore
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001517
Antoine Pitrou02512fb2012-11-17 23:56:53 +01001518 with ZipFile(args[1], 'w', allowZip64=True) as zf:
1519 for src in args[2:]:
1520 addToZip(zf, src, os.path.basename(src))
Ronald Oussoren143cefb2006-06-15 08:14:18 +00001521
1522if __name__ == "__main__":
1523 main()