Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1 | """ |
| 2 | Read and write ZIP files. |
| 3 | """ |
Georg Brandl | 62416bc | 2008-01-07 18:47:44 +0000 | [diff] [blame] | 4 | import struct, os, time, sys, shutil |
Martin v. Löwis | 0dfcfc8 | 2009-01-24 14:00:33 +0000 | [diff] [blame] | 5 | import binascii, cStringIO, stat |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 6 | import io |
| 7 | import re |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 8 | |
| 9 | try: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 10 | import zlib # We may need its compression method |
Gregory P. Smith | b89a096 | 2008-03-19 01:46:10 +0000 | [diff] [blame] | 11 | crc32 = zlib.crc32 |
Guido van Rossum | 9c673f3 | 2001-04-10 15:37:12 +0000 | [diff] [blame] | 12 | except ImportError: |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 13 | zlib = None |
Gregory P. Smith | b89a096 | 2008-03-19 01:46:10 +0000 | [diff] [blame] | 14 | crc32 = binascii.crc32 |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 15 | |
Skip Montanaro | 40fc160 | 2001-03-01 04:27:19 +0000 | [diff] [blame] | 16 | __all__ = ["BadZipfile", "error", "ZIP_STORED", "ZIP_DEFLATED", "is_zipfile", |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 17 | "ZipInfo", "ZipFile", "PyZipFile", "LargeZipFile" ] |
Skip Montanaro | 40fc160 | 2001-03-01 04:27:19 +0000 | [diff] [blame] | 18 | |
Fred Drake | 5db246d | 2000-09-29 20:44:48 +0000 | [diff] [blame] | 19 | class BadZipfile(Exception): |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 20 | pass |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 21 | |
| 22 | |
| 23 | class LargeZipFile(Exception): |
Tim Peters | a608bb2 | 2006-06-15 18:06:29 +0000 | [diff] [blame] | 24 | """ |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 25 | Raised when writing a zipfile, the zipfile requires ZIP64 extensions |
| 26 | and those extensions are disabled. |
| 27 | """ |
| 28 | |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 29 | error = BadZipfile # The exception raised by this module |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 30 | |
Amaury Forgeot d'Arc | d25f87a | 2009-01-17 16:40:17 +0000 | [diff] [blame] | 31 | ZIP64_LIMIT = (1 << 31) - 1 |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 32 | ZIP_FILECOUNT_LIMIT = 1 << 16 |
| 33 | ZIP_MAX_COMMENT = (1 << 16) - 1 |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 34 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 35 | # constants for Zip file compression methods |
| 36 | ZIP_STORED = 0 |
| 37 | ZIP_DEFLATED = 8 |
| 38 | # Other ZIP compression methods not supported |
| 39 | |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 40 | # Below are some formats and associated data for reading/writing headers using |
| 41 | # the struct module. The names and structures of headers/records are those used |
| 42 | # in the PKWARE description of the ZIP file format: |
| 43 | # http://www.pkware.com/documents/casestudies/APPNOTE.TXT |
| 44 | # (URL valid as of January 2008) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 45 | |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 46 | # The "end of central directory" structure, magic number, size, and indices |
| 47 | # (section V.I in the format document) |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 48 | structEndArchive = "<4s4H2LH" |
| 49 | stringEndArchive = "PK\005\006" |
| 50 | sizeEndCentDir = struct.calcsize(structEndArchive) |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 51 | |
| 52 | _ECD_SIGNATURE = 0 |
| 53 | _ECD_DISK_NUMBER = 1 |
| 54 | _ECD_DISK_START = 2 |
| 55 | _ECD_ENTRIES_THIS_DISK = 3 |
| 56 | _ECD_ENTRIES_TOTAL = 4 |
| 57 | _ECD_SIZE = 5 |
| 58 | _ECD_OFFSET = 6 |
| 59 | _ECD_COMMENT_SIZE = 7 |
| 60 | # These last two indices are not part of the structure as defined in the |
| 61 | # spec, but they are used internally by this module as a convenience |
| 62 | _ECD_COMMENT = 8 |
| 63 | _ECD_LOCATION = 9 |
| 64 | |
| 65 | # The "central directory" structure, magic number, size, and indices |
| 66 | # of entries in the structure (section V.F in the format document) |
| 67 | structCentralDir = "<4s4B4HL2L5H2L" |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 68 | stringCentralDir = "PK\001\002" |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 69 | sizeCentralDir = struct.calcsize(structCentralDir) |
| 70 | |
Fred Drake | 3e038e5 | 2001-02-28 17:56:26 +0000 | [diff] [blame] | 71 | # indexes of entries in the central directory structure |
| 72 | _CD_SIGNATURE = 0 |
| 73 | _CD_CREATE_VERSION = 1 |
| 74 | _CD_CREATE_SYSTEM = 2 |
| 75 | _CD_EXTRACT_VERSION = 3 |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 76 | _CD_EXTRACT_SYSTEM = 4 |
Fred Drake | 3e038e5 | 2001-02-28 17:56:26 +0000 | [diff] [blame] | 77 | _CD_FLAG_BITS = 5 |
| 78 | _CD_COMPRESS_TYPE = 6 |
| 79 | _CD_TIME = 7 |
| 80 | _CD_DATE = 8 |
| 81 | _CD_CRC = 9 |
| 82 | _CD_COMPRESSED_SIZE = 10 |
| 83 | _CD_UNCOMPRESSED_SIZE = 11 |
| 84 | _CD_FILENAME_LENGTH = 12 |
| 85 | _CD_EXTRA_FIELD_LENGTH = 13 |
| 86 | _CD_COMMENT_LENGTH = 14 |
| 87 | _CD_DISK_NUMBER_START = 15 |
| 88 | _CD_INTERNAL_FILE_ATTRIBUTES = 16 |
| 89 | _CD_EXTERNAL_FILE_ATTRIBUTES = 17 |
| 90 | _CD_LOCAL_HEADER_OFFSET = 18 |
| 91 | |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 92 | # The "local file header" structure, magic number, size, and indices |
| 93 | # (section V.A in the format document) |
| 94 | structFileHeader = "<4s2B4HL2L2H" |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 95 | stringFileHeader = "PK\003\004" |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 96 | sizeFileHeader = struct.calcsize(structFileHeader) |
| 97 | |
Fred Drake | 3e038e5 | 2001-02-28 17:56:26 +0000 | [diff] [blame] | 98 | _FH_SIGNATURE = 0 |
| 99 | _FH_EXTRACT_VERSION = 1 |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 100 | _FH_EXTRACT_SYSTEM = 2 |
Fred Drake | 3e038e5 | 2001-02-28 17:56:26 +0000 | [diff] [blame] | 101 | _FH_GENERAL_PURPOSE_FLAG_BITS = 3 |
| 102 | _FH_COMPRESSION_METHOD = 4 |
| 103 | _FH_LAST_MOD_TIME = 5 |
| 104 | _FH_LAST_MOD_DATE = 6 |
| 105 | _FH_CRC = 7 |
| 106 | _FH_COMPRESSED_SIZE = 8 |
| 107 | _FH_UNCOMPRESSED_SIZE = 9 |
| 108 | _FH_FILENAME_LENGTH = 10 |
| 109 | _FH_EXTRA_FIELD_LENGTH = 11 |
| 110 | |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 111 | # The "Zip64 end of central directory locator" structure, magic number, and size |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 112 | structEndArchive64Locator = "<4sLQL" |
| 113 | stringEndArchive64Locator = "PK\x06\x07" |
| 114 | sizeEndCentDir64Locator = struct.calcsize(structEndArchive64Locator) |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 115 | |
| 116 | # The "Zip64 end of central directory" record, magic number, size, and indices |
| 117 | # (section V.G in the format document) |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 118 | structEndArchive64 = "<4sQ2H2L4Q" |
| 119 | stringEndArchive64 = "PK\x06\x06" |
| 120 | sizeEndCentDir64 = struct.calcsize(structEndArchive64) |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 121 | |
| 122 | _CD64_SIGNATURE = 0 |
| 123 | _CD64_DIRECTORY_RECSIZE = 1 |
| 124 | _CD64_CREATE_VERSION = 2 |
| 125 | _CD64_EXTRACT_VERSION = 3 |
| 126 | _CD64_DISK_NUMBER = 4 |
| 127 | _CD64_DISK_NUMBER_START = 5 |
| 128 | _CD64_NUMBER_ENTRIES_THIS_DISK = 6 |
| 129 | _CD64_NUMBER_ENTRIES_TOTAL = 7 |
| 130 | _CD64_DIRECTORY_SIZE = 8 |
| 131 | _CD64_OFFSET_START_CENTDIR = 9 |
| 132 | |
Antoine Pitrou | 6f193e0 | 2008-12-27 15:43:12 +0000 | [diff] [blame] | 133 | def _check_zipfile(fp): |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 134 | try: |
Antoine Pitrou | 6f193e0 | 2008-12-27 15:43:12 +0000 | [diff] [blame] | 135 | if _EndRecData(fp): |
| 136 | return True # file has correct magic number |
Fred Drake | 7e47380 | 2001-05-11 19:52:57 +0000 | [diff] [blame] | 137 | except IOError: |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 138 | pass |
Guido van Rossum | 8ca162f | 2002-04-07 06:36:23 +0000 | [diff] [blame] | 139 | return False |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 140 | |
Antoine Pitrou | 6f193e0 | 2008-12-27 15:43:12 +0000 | [diff] [blame] | 141 | def is_zipfile(filename): |
| 142 | """Quickly see if a file is a ZIP file by checking the magic number. |
| 143 | |
| 144 | The filename argument may be a file or file-like object too. |
| 145 | """ |
| 146 | result = False |
| 147 | try: |
| 148 | if hasattr(filename, "read"): |
| 149 | result = _check_zipfile(fp=filename) |
| 150 | else: |
| 151 | with open(filename, "rb") as fp: |
| 152 | result = _check_zipfile(fp) |
| 153 | except IOError: |
| 154 | pass |
| 155 | return result |
| 156 | |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 157 | def _EndRecData64(fpin, offset, endrec): |
| 158 | """ |
| 159 | Read the ZIP64 end-of-archive records and use that to update endrec |
| 160 | """ |
Georg Brandl | 86e0c89 | 2010-11-26 07:22:28 +0000 | [diff] [blame] | 161 | try: |
| 162 | fpin.seek(offset - sizeEndCentDir64Locator, 2) |
| 163 | except IOError: |
| 164 | # If the seek fails, the file is not large enough to contain a ZIP64 |
| 165 | # end-of-archive record, so just return the end record we were given. |
| 166 | return endrec |
| 167 | |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 168 | data = fpin.read(sizeEndCentDir64Locator) |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 169 | sig, diskno, reloff, disks = struct.unpack(structEndArchive64Locator, data) |
| 170 | if sig != stringEndArchive64Locator: |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 171 | return endrec |
| 172 | |
| 173 | if diskno != 0 or disks != 1: |
| 174 | raise BadZipfile("zipfiles that span multiple disks are not supported") |
| 175 | |
Tim Peters | a608bb2 | 2006-06-15 18:06:29 +0000 | [diff] [blame] | 176 | # Assume no 'zip64 extensible data' |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 177 | fpin.seek(offset - sizeEndCentDir64Locator - sizeEndCentDir64, 2) |
| 178 | data = fpin.read(sizeEndCentDir64) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 179 | sig, sz, create_version, read_version, disk_num, disk_dir, \ |
| 180 | dircount, dircount2, dirsize, diroffset = \ |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 181 | struct.unpack(structEndArchive64, data) |
| 182 | if sig != stringEndArchive64: |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 183 | return endrec |
| 184 | |
| 185 | # Update the original endrec using data from the ZIP64 record |
Antoine Pitrou | ebcd0ce | 2008-09-05 23:30:23 +0000 | [diff] [blame] | 186 | endrec[_ECD_SIGNATURE] = sig |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 187 | endrec[_ECD_DISK_NUMBER] = disk_num |
| 188 | endrec[_ECD_DISK_START] = disk_dir |
| 189 | endrec[_ECD_ENTRIES_THIS_DISK] = dircount |
| 190 | endrec[_ECD_ENTRIES_TOTAL] = dircount2 |
| 191 | endrec[_ECD_SIZE] = dirsize |
| 192 | endrec[_ECD_OFFSET] = diroffset |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 193 | return endrec |
| 194 | |
| 195 | |
Martin v. Löwis | 6f6873b | 2002-10-13 13:54:50 +0000 | [diff] [blame] | 196 | def _EndRecData(fpin): |
| 197 | """Return data from the "End of Central Directory" record, or None. |
| 198 | |
| 199 | The data is a list of the nine items in the ZIP "End of central dir" |
| 200 | record followed by a tenth item, the file seek offset of this record.""" |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 201 | |
| 202 | # Determine file size |
| 203 | fpin.seek(0, 2) |
| 204 | filesize = fpin.tell() |
| 205 | |
| 206 | # Check to see if this is ZIP file with no archive comment (the |
| 207 | # "end of central directory" structure should be the last item in the |
| 208 | # file if this is the case). |
Amaury Forgeot d'Arc | 3e5b027 | 2009-07-28 22:15:30 +0000 | [diff] [blame] | 209 | try: |
| 210 | fpin.seek(-sizeEndCentDir, 2) |
| 211 | except IOError: |
| 212 | return None |
Martin v. Löwis | 6f6873b | 2002-10-13 13:54:50 +0000 | [diff] [blame] | 213 | data = fpin.read() |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 214 | if data[0:4] == stringEndArchive and data[-2:] == "\000\000": |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 215 | # the signature is correct and there's no comment, unpack structure |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 216 | endrec = struct.unpack(structEndArchive, data) |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 217 | endrec=list(endrec) |
| 218 | |
| 219 | # Append a blank comment and record start offset |
| 220 | endrec.append("") |
| 221 | endrec.append(filesize - sizeEndCentDir) |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 222 | |
Amaury Forgeot d'Arc | 2407ac9 | 2009-01-17 22:43:50 +0000 | [diff] [blame] | 223 | # Try to read the "Zip64 end of central directory" structure |
| 224 | return _EndRecData64(fpin, -sizeEndCentDir, endrec) |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 225 | |
| 226 | # Either this is not a ZIP file, or it is a ZIP file with an archive |
| 227 | # comment. Search the end of the file for the "end of central directory" |
| 228 | # record signature. The comment is the last item in the ZIP file and may be |
| 229 | # up to 64K long. It is assumed that the "end of central directory" magic |
| 230 | # number does not appear in the comment. |
| 231 | maxCommentStart = max(filesize - (1 << 16) - sizeEndCentDir, 0) |
| 232 | fpin.seek(maxCommentStart, 0) |
Martin v. Löwis | 6f6873b | 2002-10-13 13:54:50 +0000 | [diff] [blame] | 233 | data = fpin.read() |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 234 | start = data.rfind(stringEndArchive) |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 235 | if start >= 0: |
| 236 | # found the magic number; attempt to unpack and interpret |
| 237 | recData = data[start:start+sizeEndCentDir] |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 238 | endrec = list(struct.unpack(structEndArchive, recData)) |
R David Murray | 873c583 | 2011-06-09 16:01:09 -0400 | [diff] [blame] | 239 | commentSize = endrec[_ECD_COMMENT_SIZE] #as claimed by the zip file |
| 240 | comment = data[start+sizeEndCentDir:start+sizeEndCentDir+commentSize] |
| 241 | endrec.append(comment) |
| 242 | endrec.append(maxCommentStart + start) |
Amaury Forgeot d'Arc | 2407ac9 | 2009-01-17 22:43:50 +0000 | [diff] [blame] | 243 | |
R David Murray | 873c583 | 2011-06-09 16:01:09 -0400 | [diff] [blame] | 244 | # Try to read the "Zip64 end of central directory" structure |
| 245 | return _EndRecData64(fpin, maxCommentStart + start - filesize, |
| 246 | endrec) |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 247 | |
| 248 | # Unable to find a valid end of central directory structure |
| 249 | return |
Martin v. Löwis | 6f6873b | 2002-10-13 13:54:50 +0000 | [diff] [blame] | 250 | |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 251 | |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 252 | class ZipInfo (object): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 253 | """Class with attributes describing each file in the ZIP archive.""" |
| 254 | |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 255 | __slots__ = ( |
| 256 | 'orig_filename', |
| 257 | 'filename', |
| 258 | 'date_time', |
| 259 | 'compress_type', |
| 260 | 'comment', |
| 261 | 'extra', |
| 262 | 'create_system', |
| 263 | 'create_version', |
| 264 | 'extract_version', |
| 265 | 'reserved', |
| 266 | 'flag_bits', |
| 267 | 'volume', |
| 268 | 'internal_attr', |
| 269 | 'external_attr', |
| 270 | 'header_offset', |
| 271 | 'CRC', |
| 272 | 'compress_size', |
| 273 | 'file_size', |
Gregory P. Smith | 0c63fc2 | 2008-01-20 01:21:03 +0000 | [diff] [blame] | 274 | '_raw_time', |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 275 | ) |
| 276 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 277 | def __init__(self, filename="NoName", date_time=(1980,1,1,0,0,0)): |
Greg Ward | 8e36d28 | 2003-06-18 00:53:06 +0000 | [diff] [blame] | 278 | self.orig_filename = filename # Original file name in archive |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 279 | |
| 280 | # Terminate the file name at the first null byte. Null bytes in file |
| 281 | # names are used as tricks by viruses in archives. |
Greg Ward | 8e36d28 | 2003-06-18 00:53:06 +0000 | [diff] [blame] | 282 | null_byte = filename.find(chr(0)) |
| 283 | if null_byte >= 0: |
| 284 | filename = filename[0:null_byte] |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 285 | # This is used to ensure paths in generated ZIP files always use |
| 286 | # forward slashes as the directory separator, as required by the |
| 287 | # ZIP format specification. |
| 288 | if os.sep != "/" and os.sep in filename: |
Greg Ward | 8e36d28 | 2003-06-18 00:53:06 +0000 | [diff] [blame] | 289 | filename = filename.replace(os.sep, "/") |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 290 | |
Greg Ward | 8e36d28 | 2003-06-18 00:53:06 +0000 | [diff] [blame] | 291 | self.filename = filename # Normalized file name |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 292 | self.date_time = date_time # year, month, day, hour, min, sec |
Senthil Kumaran | ddd4031 | 2011-10-20 01:38:35 +0800 | [diff] [blame] | 293 | |
| 294 | if date_time[0] < 1980: |
| 295 | raise ValueError('ZIP does not support timestamps before 1980') |
| 296 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 297 | # Standard values: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 298 | self.compress_type = ZIP_STORED # Type of compression for the file |
| 299 | self.comment = "" # Comment for each file |
| 300 | self.extra = "" # ZIP extra data |
Martin v. Löwis | 0075690 | 2006-02-05 17:09:41 +0000 | [diff] [blame] | 301 | if sys.platform == 'win32': |
| 302 | self.create_system = 0 # System which created ZIP archive |
| 303 | else: |
| 304 | # Assume everything else is unix-y |
| 305 | self.create_system = 3 # System which created ZIP archive |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 306 | self.create_version = 20 # Version which created ZIP archive |
| 307 | self.extract_version = 20 # Version needed to extract archive |
| 308 | self.reserved = 0 # Must be zero |
| 309 | self.flag_bits = 0 # ZIP flag bits |
| 310 | self.volume = 0 # Volume number of file header |
| 311 | self.internal_attr = 0 # Internal attributes |
| 312 | self.external_attr = 0 # External file attributes |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 313 | # Other attributes are set by class ZipFile: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 314 | # header_offset Byte offset to the file header |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 315 | # CRC CRC-32 of the uncompressed file |
| 316 | # compress_size Size of the compressed file |
| 317 | # file_size Size of the uncompressed file |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 318 | |
| 319 | def FileHeader(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 320 | """Return the per-file header as a string.""" |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 321 | dt = self.date_time |
| 322 | dosdate = (dt[0] - 1980) << 9 | dt[1] << 5 | dt[2] |
Tim Peters | 3caca23 | 2001-12-06 06:23:26 +0000 | [diff] [blame] | 323 | dostime = dt[3] << 11 | dt[4] << 5 | (dt[5] // 2) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 324 | if self.flag_bits & 0x08: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 325 | # Set these to zero because we write them after the file data |
| 326 | CRC = compress_size = file_size = 0 |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 327 | else: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 328 | CRC = self.CRC |
| 329 | compress_size = self.compress_size |
| 330 | file_size = self.file_size |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 331 | |
| 332 | extra = self.extra |
| 333 | |
| 334 | if file_size > ZIP64_LIMIT or compress_size > ZIP64_LIMIT: |
| 335 | # File is larger than what fits into a 4 byte integer, |
| 336 | # fall back to the ZIP64 extension |
Gregory P. Smith | bf02e3b | 2008-03-19 03:14:41 +0000 | [diff] [blame] | 337 | fmt = '<HHQQ' |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 338 | extra = extra + struct.pack(fmt, |
| 339 | 1, struct.calcsize(fmt)-4, file_size, compress_size) |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 340 | file_size = 0xffffffff |
| 341 | compress_size = 0xffffffff |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 342 | self.extract_version = max(45, self.extract_version) |
| 343 | self.create_version = max(45, self.extract_version) |
| 344 | |
Martin v. Löwis | 471617d | 2008-05-05 17:16:58 +0000 | [diff] [blame] | 345 | filename, flag_bits = self._encodeFilenameFlags() |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 346 | header = struct.pack(structFileHeader, stringFileHeader, |
Martin v. Löwis | 471617d | 2008-05-05 17:16:58 +0000 | [diff] [blame] | 347 | self.extract_version, self.reserved, flag_bits, |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 348 | self.compress_type, dostime, dosdate, CRC, |
| 349 | compress_size, file_size, |
Martin v. Löwis | 471617d | 2008-05-05 17:16:58 +0000 | [diff] [blame] | 350 | len(filename), len(extra)) |
| 351 | return header + filename + extra |
| 352 | |
| 353 | def _encodeFilenameFlags(self): |
| 354 | if isinstance(self.filename, unicode): |
| 355 | try: |
| 356 | return self.filename.encode('ascii'), self.flag_bits |
| 357 | except UnicodeEncodeError: |
| 358 | return self.filename.encode('utf-8'), self.flag_bits | 0x800 |
| 359 | else: |
| 360 | return self.filename, self.flag_bits |
| 361 | |
| 362 | def _decodeFilename(self): |
| 363 | if self.flag_bits & 0x800: |
| 364 | return self.filename.decode('utf-8') |
| 365 | else: |
| 366 | return self.filename |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 367 | |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 368 | def _decodeExtra(self): |
| 369 | # Try to decode the extra field. |
| 370 | extra = self.extra |
| 371 | unpack = struct.unpack |
| 372 | while extra: |
Gregory P. Smith | bf02e3b | 2008-03-19 03:14:41 +0000 | [diff] [blame] | 373 | tp, ln = unpack('<HH', extra[:4]) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 374 | if tp == 1: |
| 375 | if ln >= 24: |
Gregory P. Smith | bf02e3b | 2008-03-19 03:14:41 +0000 | [diff] [blame] | 376 | counts = unpack('<QQQ', extra[4:28]) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 377 | elif ln == 16: |
Gregory P. Smith | bf02e3b | 2008-03-19 03:14:41 +0000 | [diff] [blame] | 378 | counts = unpack('<QQ', extra[4:20]) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 379 | elif ln == 8: |
Gregory P. Smith | bf02e3b | 2008-03-19 03:14:41 +0000 | [diff] [blame] | 380 | counts = unpack('<Q', extra[4:12]) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 381 | elif ln == 0: |
| 382 | counts = () |
| 383 | else: |
| 384 | raise RuntimeError, "Corrupt extra field %s"%(ln,) |
| 385 | |
| 386 | idx = 0 |
| 387 | |
| 388 | # ZIP64 extension (large files and/or large archives) |
Gregory P. Smith | bf02e3b | 2008-03-19 03:14:41 +0000 | [diff] [blame] | 389 | if self.file_size in (0xffffffffffffffffL, 0xffffffffL): |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 390 | self.file_size = counts[idx] |
| 391 | idx += 1 |
| 392 | |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 393 | if self.compress_size == 0xFFFFFFFFL: |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 394 | self.compress_size = counts[idx] |
| 395 | idx += 1 |
| 396 | |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 397 | if self.header_offset == 0xffffffffL: |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 398 | old = self.header_offset |
| 399 | self.header_offset = counts[idx] |
| 400 | idx+=1 |
| 401 | |
| 402 | extra = extra[ln+4:] |
Tim Peters | a608bb2 | 2006-06-15 18:06:29 +0000 | [diff] [blame] | 403 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 404 | |
Martin v. Löwis | c6d626e | 2007-02-13 09:49:38 +0000 | [diff] [blame] | 405 | class _ZipDecrypter: |
| 406 | """Class to handle decryption of files stored within a ZIP archive. |
| 407 | |
| 408 | ZIP supports a password-based form of encryption. Even though known |
| 409 | plaintext attacks have been found against it, it is still useful |
Gregory P. Smith | da40723 | 2008-01-20 01:32:00 +0000 | [diff] [blame] | 410 | to be able to get data out of such a file. |
Martin v. Löwis | c6d626e | 2007-02-13 09:49:38 +0000 | [diff] [blame] | 411 | |
| 412 | Usage: |
| 413 | zd = _ZipDecrypter(mypwd) |
| 414 | plain_char = zd(cypher_char) |
| 415 | plain_text = map(zd, cypher_text) |
| 416 | """ |
| 417 | |
| 418 | def _GenerateCRCTable(): |
| 419 | """Generate a CRC-32 table. |
| 420 | |
| 421 | ZIP encryption uses the CRC32 one-byte primitive for scrambling some |
| 422 | internal keys. We noticed that a direct implementation is faster than |
| 423 | relying on binascii.crc32(). |
| 424 | """ |
| 425 | poly = 0xedb88320 |
| 426 | table = [0] * 256 |
| 427 | for i in range(256): |
| 428 | crc = i |
| 429 | for j in range(8): |
| 430 | if crc & 1: |
| 431 | crc = ((crc >> 1) & 0x7FFFFFFF) ^ poly |
| 432 | else: |
| 433 | crc = ((crc >> 1) & 0x7FFFFFFF) |
| 434 | table[i] = crc |
| 435 | return table |
| 436 | crctable = _GenerateCRCTable() |
| 437 | |
| 438 | def _crc32(self, ch, crc): |
| 439 | """Compute the CRC32 primitive on one byte.""" |
| 440 | return ((crc >> 8) & 0xffffff) ^ self.crctable[(crc ^ ord(ch)) & 0xff] |
| 441 | |
| 442 | def __init__(self, pwd): |
| 443 | self.key0 = 305419896 |
| 444 | self.key1 = 591751049 |
| 445 | self.key2 = 878082192 |
| 446 | for p in pwd: |
| 447 | self._UpdateKeys(p) |
| 448 | |
| 449 | def _UpdateKeys(self, c): |
| 450 | self.key0 = self._crc32(c, self.key0) |
| 451 | self.key1 = (self.key1 + (self.key0 & 255)) & 4294967295 |
| 452 | self.key1 = (self.key1 * 134775813 + 1) & 4294967295 |
| 453 | self.key2 = self._crc32(chr((self.key1 >> 24) & 255), self.key2) |
| 454 | |
| 455 | def __call__(self, c): |
| 456 | """Decrypt a single character.""" |
| 457 | c = ord(c) |
| 458 | k = self.key2 | 2 |
| 459 | c = c ^ (((k * (k^1)) >> 8) & 255) |
| 460 | c = chr(c) |
| 461 | self._UpdateKeys(c) |
| 462 | return c |
| 463 | |
Ezio Melotti | 9e94972 | 2012-11-18 13:18:06 +0200 | [diff] [blame^] | 464 | |
| 465 | compressor_names = { |
| 466 | 0: 'store', |
| 467 | 1: 'shrink', |
| 468 | 2: 'reduce', |
| 469 | 3: 'reduce', |
| 470 | 4: 'reduce', |
| 471 | 5: 'reduce', |
| 472 | 6: 'implode', |
| 473 | 7: 'tokenize', |
| 474 | 8: 'deflate', |
| 475 | 9: 'deflate64', |
| 476 | 10: 'implode', |
| 477 | 12: 'bzip2', |
| 478 | 14: 'lzma', |
| 479 | 18: 'terse', |
| 480 | 19: 'lz77', |
| 481 | 97: 'wavpack', |
| 482 | 98: 'ppmd', |
| 483 | } |
| 484 | |
| 485 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 486 | class ZipExtFile(io.BufferedIOBase): |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 487 | """File-like object for reading an archive member. |
Tim Peters | ea5962f | 2007-03-12 18:07:52 +0000 | [diff] [blame] | 488 | Is returned by ZipFile.open(). |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 489 | """ |
Tim Peters | ea5962f | 2007-03-12 18:07:52 +0000 | [diff] [blame] | 490 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 491 | # Max size supported by decompressor. |
| 492 | MAX_N = 1 << 31 - 1 |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 493 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 494 | # Read from compressed files in 4k blocks. |
| 495 | MIN_READ_SIZE = 4096 |
Tim Peters | ea5962f | 2007-03-12 18:07:52 +0000 | [diff] [blame] | 496 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 497 | # Search for universal newlines or line chunks. |
| 498 | PATTERN = re.compile(r'^(?P<chunk>[^\r\n]+)|(?P<newline>\n|\r\n?)') |
| 499 | |
Jesus Cea | 93d628b | 2012-11-04 02:32:08 +0100 | [diff] [blame] | 500 | def __init__(self, fileobj, mode, zipinfo, decrypter=None, |
| 501 | close_fileobj=False): |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 502 | self._fileobj = fileobj |
| 503 | self._decrypter = decrypter |
Jesus Cea | 93d628b | 2012-11-04 02:32:08 +0100 | [diff] [blame] | 504 | self._close_fileobj = close_fileobj |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 505 | |
Ezio Melotti | 4611b05 | 2010-01-28 01:41:30 +0000 | [diff] [blame] | 506 | self._compress_type = zipinfo.compress_type |
| 507 | self._compress_size = zipinfo.compress_size |
| 508 | self._compress_left = zipinfo.compress_size |
| 509 | |
| 510 | if self._compress_type == ZIP_DEFLATED: |
| 511 | self._decompressor = zlib.decompressobj(-15) |
Ezio Melotti | 9e94972 | 2012-11-18 13:18:06 +0200 | [diff] [blame^] | 512 | elif self._compress_type != ZIP_STORED: |
| 513 | descr = compressor_names.get(self._compress_type) |
| 514 | if descr: |
| 515 | raise NotImplementedError("compression type %d (%s)" % (self._compress_type, descr)) |
| 516 | else: |
| 517 | raise NotImplementedError("compression type %d" % (self._compress_type,)) |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 518 | self._unconsumed = '' |
| 519 | |
| 520 | self._readbuffer = '' |
| 521 | self._offset = 0 |
| 522 | |
| 523 | self._universal = 'U' in mode |
| 524 | self.newlines = None |
| 525 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 526 | # Adjust read size for encrypted files since the first 12 bytes |
| 527 | # are for the encryption/password information. |
| 528 | if self._decrypter is not None: |
| 529 | self._compress_left -= 12 |
| 530 | |
| 531 | self.mode = mode |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 532 | self.name = zipinfo.filename |
| 533 | |
Antoine Pitrou | e1436d1 | 2010-08-12 15:25:51 +0000 | [diff] [blame] | 534 | if hasattr(zipinfo, 'CRC'): |
| 535 | self._expected_crc = zipinfo.CRC |
| 536 | self._running_crc = crc32(b'') & 0xffffffff |
| 537 | else: |
| 538 | self._expected_crc = None |
| 539 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 540 | def readline(self, limit=-1): |
| 541 | """Read and return a line from the stream. |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 542 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 543 | If limit is specified, at most limit bytes will be read. |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 544 | """ |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 545 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 546 | if not self._universal and limit < 0: |
| 547 | # Shortcut common case - newline found in buffer. |
| 548 | i = self._readbuffer.find('\n', self._offset) + 1 |
| 549 | if i > 0: |
| 550 | line = self._readbuffer[self._offset: i] |
| 551 | self._offset = i |
| 552 | return line |
Tim Peters | ea5962f | 2007-03-12 18:07:52 +0000 | [diff] [blame] | 553 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 554 | if not self._universal: |
| 555 | return io.BufferedIOBase.readline(self, limit) |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 556 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 557 | line = '' |
| 558 | while limit < 0 or len(line) < limit: |
| 559 | readahead = self.peek(2) |
| 560 | if readahead == '': |
| 561 | return line |
Tim Peters | ea5962f | 2007-03-12 18:07:52 +0000 | [diff] [blame] | 562 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 563 | # |
| 564 | # Search for universal newlines or line chunks. |
| 565 | # |
| 566 | # The pattern returns either a line chunk or a newline, but not |
| 567 | # both. Combined with peek(2), we are assured that the sequence |
| 568 | # '\r\n' is always retrieved completely and never split into |
| 569 | # separate newlines - '\r', '\n' due to coincidental readaheads. |
| 570 | # |
| 571 | match = self.PATTERN.search(readahead) |
| 572 | newline = match.group('newline') |
| 573 | if newline is not None: |
| 574 | if self.newlines is None: |
| 575 | self.newlines = [] |
| 576 | if newline not in self.newlines: |
| 577 | self.newlines.append(newline) |
| 578 | self._offset += len(newline) |
| 579 | return line + '\n' |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 580 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 581 | chunk = match.group('chunk') |
| 582 | if limit >= 0: |
| 583 | chunk = chunk[: limit - len(line)] |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 584 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 585 | self._offset += len(chunk) |
| 586 | line += chunk |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 587 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 588 | return line |
| 589 | |
| 590 | def peek(self, n=1): |
| 591 | """Returns buffered bytes without advancing the position.""" |
| 592 | if n > len(self._readbuffer) - self._offset: |
| 593 | chunk = self.read(n) |
| 594 | self._offset -= len(chunk) |
| 595 | |
| 596 | # Return up to 512 bytes to reduce allocation overhead for tight loops. |
| 597 | return self._readbuffer[self._offset: self._offset + 512] |
| 598 | |
| 599 | def readable(self): |
| 600 | return True |
| 601 | |
| 602 | def read(self, n=-1): |
| 603 | """Read and return up to n bytes. |
| 604 | If the argument is omitted, None, or negative, data is read and returned until EOF is reached.. |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 605 | """ |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 606 | buf = '' |
Antoine Pitrou | e4195e8 | 2010-09-12 14:56:27 +0000 | [diff] [blame] | 607 | if n is None: |
| 608 | n = -1 |
| 609 | while True: |
| 610 | if n < 0: |
| 611 | data = self.read1(n) |
| 612 | elif n > len(buf): |
| 613 | data = self.read1(n - len(buf)) |
| 614 | else: |
| 615 | return buf |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 616 | if len(data) == 0: |
| 617 | return buf |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 618 | buf += data |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 619 | |
Antoine Pitrou | e1436d1 | 2010-08-12 15:25:51 +0000 | [diff] [blame] | 620 | def _update_crc(self, newdata, eof): |
| 621 | # Update the CRC using the given data. |
| 622 | if self._expected_crc is None: |
| 623 | # No need to compute the CRC if we don't have a reference value |
| 624 | return |
| 625 | self._running_crc = crc32(newdata, self._running_crc) & 0xffffffff |
| 626 | # Check the CRC if we're at the end of the file |
| 627 | if eof and self._running_crc != self._expected_crc: |
| 628 | raise BadZipfile("Bad CRC-32 for file %r" % self.name) |
| 629 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 630 | def read1(self, n): |
| 631 | """Read up to n bytes with at most one read() system call.""" |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 632 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 633 | # Simplify algorithm (branching) by transforming negative n to large n. |
| 634 | if n < 0 or n is None: |
| 635 | n = self.MAX_N |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 636 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 637 | # Bytes available in read buffer. |
| 638 | len_readbuffer = len(self._readbuffer) - self._offset |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 639 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 640 | # Read from file. |
| 641 | if self._compress_left > 0 and n > len_readbuffer + len(self._unconsumed): |
| 642 | nbytes = n - len_readbuffer - len(self._unconsumed) |
| 643 | nbytes = max(nbytes, self.MIN_READ_SIZE) |
| 644 | nbytes = min(nbytes, self._compress_left) |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 645 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 646 | data = self._fileobj.read(nbytes) |
| 647 | self._compress_left -= len(data) |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 648 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 649 | if data and self._decrypter is not None: |
| 650 | data = ''.join(map(self._decrypter, data)) |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 651 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 652 | if self._compress_type == ZIP_STORED: |
Antoine Pitrou | e1436d1 | 2010-08-12 15:25:51 +0000 | [diff] [blame] | 653 | self._update_crc(data, eof=(self._compress_left==0)) |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 654 | self._readbuffer = self._readbuffer[self._offset:] + data |
| 655 | self._offset = 0 |
| 656 | else: |
| 657 | # Prepare deflated bytes for decompression. |
| 658 | self._unconsumed += data |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 659 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 660 | # Handle unconsumed data. |
Ezio Melotti | 4611b05 | 2010-01-28 01:41:30 +0000 | [diff] [blame] | 661 | if (len(self._unconsumed) > 0 and n > len_readbuffer and |
| 662 | self._compress_type == ZIP_DEFLATED): |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 663 | data = self._decompressor.decompress( |
| 664 | self._unconsumed, |
| 665 | max(n - len_readbuffer, self.MIN_READ_SIZE) |
| 666 | ) |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 667 | |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 668 | self._unconsumed = self._decompressor.unconsumed_tail |
Antoine Pitrou | e1436d1 | 2010-08-12 15:25:51 +0000 | [diff] [blame] | 669 | eof = len(self._unconsumed) == 0 and self._compress_left == 0 |
| 670 | if eof: |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 671 | data += self._decompressor.flush() |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 672 | |
Antoine Pitrou | e1436d1 | 2010-08-12 15:25:51 +0000 | [diff] [blame] | 673 | self._update_crc(data, eof=eof) |
Antoine Pitrou | 94c33eb | 2010-01-27 20:59:50 +0000 | [diff] [blame] | 674 | self._readbuffer = self._readbuffer[self._offset:] + data |
| 675 | self._offset = 0 |
| 676 | |
| 677 | # Read from buffer. |
| 678 | data = self._readbuffer[self._offset: self._offset + n] |
| 679 | self._offset += len(data) |
| 680 | return data |
| 681 | |
Jesus Cea | 93d628b | 2012-11-04 02:32:08 +0100 | [diff] [blame] | 682 | def close(self): |
| 683 | try : |
| 684 | if self._close_fileobj: |
| 685 | self._fileobj.close() |
| 686 | finally: |
| 687 | super(ZipExtFile, self).close() |
Tim Peters | ea5962f | 2007-03-12 18:07:52 +0000 | [diff] [blame] | 688 | |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 689 | |
R David Murray | 3f4ccba | 2012-04-12 18:42:47 -0400 | [diff] [blame] | 690 | class ZipFile(object): |
Tim Peters | a19a168 | 2001-03-29 04:36:09 +0000 | [diff] [blame] | 691 | """ Class with methods to open, read, write, close, list zip files. |
| 692 | |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 693 | z = ZipFile(file, mode="r", compression=ZIP_STORED, allowZip64=False) |
Tim Peters | a19a168 | 2001-03-29 04:36:09 +0000 | [diff] [blame] | 694 | |
Fred Drake | 3d9091e | 2001-03-26 15:49:24 +0000 | [diff] [blame] | 695 | file: Either the path to the file, or a file-like object. |
| 696 | If it is a path, the file will be opened and closed by ZipFile. |
| 697 | mode: The mode can be either read "r", write "w" or append "a". |
| 698 | compression: ZIP_STORED (no compression) or ZIP_DEFLATED (requires zlib). |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 699 | allowZip64: if True ZipFile will create files with ZIP64 extensions when |
| 700 | needed, otherwise it will raise an exception when this would |
| 701 | be necessary. |
| 702 | |
Fred Drake | 3d9091e | 2001-03-26 15:49:24 +0000 | [diff] [blame] | 703 | """ |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 704 | |
Fred Drake | 90eac28 | 2001-02-28 05:29:34 +0000 | [diff] [blame] | 705 | fp = None # Set here since __del__ checks it |
| 706 | |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 707 | def __init__(self, file, mode="r", compression=ZIP_STORED, allowZip64=False): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 708 | """Open the ZIP file with mode read "r", write "w" or append "a".""" |
Georg Brandl | 4b3ab6f | 2007-07-12 09:59:22 +0000 | [diff] [blame] | 709 | if mode not in ("r", "w", "a"): |
| 710 | raise RuntimeError('ZipFile() requires mode "r", "w", or "a"') |
| 711 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 712 | if compression == ZIP_STORED: |
| 713 | pass |
| 714 | elif compression == ZIP_DEFLATED: |
| 715 | if not zlib: |
| 716 | raise RuntimeError,\ |
Fred Drake | 5db246d | 2000-09-29 20:44:48 +0000 | [diff] [blame] | 717 | "Compression requires the (missing) zlib module" |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 718 | else: |
| 719 | raise RuntimeError, "That compression method is not supported" |
Georg Brandl | 4b3ab6f | 2007-07-12 09:59:22 +0000 | [diff] [blame] | 720 | |
| 721 | self._allowZip64 = allowZip64 |
| 722 | self._didModify = False |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 723 | self.debug = 0 # Level of printing: 0 through 3 |
| 724 | self.NameToInfo = {} # Find file info given name |
| 725 | self.filelist = [] # List of ZipInfo instances for archive |
| 726 | self.compression = compression # Method of compression |
Raymond Hettinger | 2ca7c19 | 2005-02-16 09:27:49 +0000 | [diff] [blame] | 727 | self.mode = key = mode.replace('b', '')[0] |
Martin v. Löwis | c6d626e | 2007-02-13 09:49:38 +0000 | [diff] [blame] | 728 | self.pwd = None |
R David Murray | 3f4ccba | 2012-04-12 18:42:47 -0400 | [diff] [blame] | 729 | self._comment = '' |
Tim Peters | a19a168 | 2001-03-29 04:36:09 +0000 | [diff] [blame] | 730 | |
Fred Drake | 3d9091e | 2001-03-26 15:49:24 +0000 | [diff] [blame] | 731 | # Check if we were passed a file-like object |
Walter Dörwald | 65230a2 | 2002-06-03 15:58:32 +0000 | [diff] [blame] | 732 | if isinstance(file, basestring): |
Fred Drake | 3d9091e | 2001-03-26 15:49:24 +0000 | [diff] [blame] | 733 | self._filePassed = 0 |
| 734 | self.filename = file |
| 735 | modeDict = {'r' : 'rb', 'w': 'wb', 'a' : 'r+b'} |
Martin v. Löwis | 84f6de9 | 2007-02-13 10:10:39 +0000 | [diff] [blame] | 736 | try: |
| 737 | self.fp = open(file, modeDict[mode]) |
| 738 | except IOError: |
| 739 | if mode == 'a': |
| 740 | mode = key = 'w' |
| 741 | self.fp = open(file, modeDict[mode]) |
| 742 | else: |
| 743 | raise |
Fred Drake | 3d9091e | 2001-03-26 15:49:24 +0000 | [diff] [blame] | 744 | else: |
| 745 | self._filePassed = 1 |
| 746 | self.fp = file |
| 747 | self.filename = getattr(file, 'name', None) |
Tim Peters | a19a168 | 2001-03-29 04:36:09 +0000 | [diff] [blame] | 748 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 749 | try: |
| 750 | if key == 'r': |
Martin v. Löwis | 6f6873b | 2002-10-13 13:54:50 +0000 | [diff] [blame] | 751 | self._RealGetContents() |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 752 | elif key == 'w': |
Georg Brandl | 86e0c89 | 2010-11-26 07:22:28 +0000 | [diff] [blame] | 753 | # set the modified flag so central directory gets written |
| 754 | # even if no files are added to the archive |
| 755 | self._didModify = True |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 756 | elif key == 'a': |
| 757 | try: |
| 758 | # See if file is a zip file |
| 759 | self._RealGetContents() |
| 760 | # seek to start of directory and overwrite |
| 761 | self.fp.seek(self.start_dir, 0) |
| 762 | except BadZipfile: |
| 763 | # file is not a zip file, just append |
| 764 | self.fp.seek(0, 2) |
| 765 | |
| 766 | # set the modified flag so central directory gets written |
| 767 | # even if no files are added to the archive |
| 768 | self._didModify = True |
| 769 | else: |
| 770 | raise RuntimeError('Mode must be "r", "w" or "a"') |
| 771 | except: |
| 772 | fp = self.fp |
| 773 | self.fp = None |
Tim Peters | 7d3bad6 | 2001-04-04 18:56:49 +0000 | [diff] [blame] | 774 | if not self._filePassed: |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 775 | fp.close() |
| 776 | raise |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 777 | |
Ezio Melotti | 569e61f | 2009-12-30 06:14:51 +0000 | [diff] [blame] | 778 | def __enter__(self): |
| 779 | return self |
| 780 | |
| 781 | def __exit__(self, type, value, traceback): |
| 782 | self.close() |
| 783 | |
Tim Peters | 7d3bad6 | 2001-04-04 18:56:49 +0000 | [diff] [blame] | 784 | def _RealGetContents(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 785 | """Read in the table of contents for the ZIP file.""" |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 786 | fp = self.fp |
Georg Brandl | 86e0c89 | 2010-11-26 07:22:28 +0000 | [diff] [blame] | 787 | try: |
| 788 | endrec = _EndRecData(fp) |
| 789 | except IOError: |
| 790 | raise BadZipfile("File is not a zip file") |
Martin v. Löwis | 6f6873b | 2002-10-13 13:54:50 +0000 | [diff] [blame] | 791 | if not endrec: |
| 792 | raise BadZipfile, "File is not a zip file" |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 793 | if self.debug > 1: |
| 794 | print endrec |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 795 | size_cd = endrec[_ECD_SIZE] # bytes in central directory |
| 796 | offset_cd = endrec[_ECD_OFFSET] # offset of central directory |
R David Murray | 3f4ccba | 2012-04-12 18:42:47 -0400 | [diff] [blame] | 797 | self._comment = endrec[_ECD_COMMENT] # archive comment |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 798 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 799 | # "concat" is zero, unless zip was concatenated to another file |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 800 | concat = endrec[_ECD_LOCATION] - size_cd - offset_cd |
Antoine Pitrou | ebcd0ce | 2008-09-05 23:30:23 +0000 | [diff] [blame] | 801 | if endrec[_ECD_SIGNATURE] == stringEndArchive64: |
| 802 | # If Zip64 extension structures are present, account for them |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 803 | concat -= (sizeEndCentDir64 + sizeEndCentDir64Locator) |
| 804 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 805 | if self.debug > 2: |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 806 | inferred = concat + offset_cd |
| 807 | print "given, inferred, offset", offset_cd, inferred, concat |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 808 | # self.start_dir: Position of start of central directory |
| 809 | self.start_dir = offset_cd + concat |
| 810 | fp.seek(self.start_dir, 0) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 811 | data = fp.read(size_cd) |
| 812 | fp = cStringIO.StringIO(data) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 813 | total = 0 |
| 814 | while total < size_cd: |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 815 | centdir = fp.read(sizeCentralDir) |
Amaury Forgeot d'Arc | ae6d2b9 | 2008-07-11 21:28:25 +0000 | [diff] [blame] | 816 | if centdir[0:4] != stringCentralDir: |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 817 | raise BadZipfile, "Bad magic number for central directory" |
| 818 | centdir = struct.unpack(structCentralDir, centdir) |
| 819 | if self.debug > 2: |
| 820 | print centdir |
Fred Drake | 3e038e5 | 2001-02-28 17:56:26 +0000 | [diff] [blame] | 821 | filename = fp.read(centdir[_CD_FILENAME_LENGTH]) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 822 | # Create ZipInfo instance to store file information |
| 823 | x = ZipInfo(filename) |
Fred Drake | 3e038e5 | 2001-02-28 17:56:26 +0000 | [diff] [blame] | 824 | x.extra = fp.read(centdir[_CD_EXTRA_FIELD_LENGTH]) |
| 825 | x.comment = fp.read(centdir[_CD_COMMENT_LENGTH]) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 826 | x.header_offset = centdir[_CD_LOCAL_HEADER_OFFSET] |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 827 | (x.create_version, x.create_system, x.extract_version, x.reserved, |
| 828 | x.flag_bits, x.compress_type, t, d, |
| 829 | x.CRC, x.compress_size, x.file_size) = centdir[1:12] |
| 830 | x.volume, x.internal_attr, x.external_attr = centdir[15:18] |
| 831 | # Convert date/time code to (year, month, day, hour, min, sec) |
Gregory P. Smith | 0c63fc2 | 2008-01-20 01:21:03 +0000 | [diff] [blame] | 832 | x._raw_time = t |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 833 | x.date_time = ( (d>>9)+1980, (d>>5)&0xF, d&0x1F, |
Fred Drake | 414ca66 | 2000-06-13 18:49:53 +0000 | [diff] [blame] | 834 | t>>11, (t>>5)&0x3F, (t&0x1F) * 2 ) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 835 | |
| 836 | x._decodeExtra() |
| 837 | x.header_offset = x.header_offset + concat |
Martin v. Löwis | 471617d | 2008-05-05 17:16:58 +0000 | [diff] [blame] | 838 | x.filename = x._decodeFilename() |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 839 | self.filelist.append(x) |
| 840 | self.NameToInfo[x.filename] = x |
Martin v. Löwis | 8c43641 | 2008-07-03 12:51:14 +0000 | [diff] [blame] | 841 | |
| 842 | # update total bytes read from central directory |
| 843 | total = (total + sizeCentralDir + centdir[_CD_FILENAME_LENGTH] |
| 844 | + centdir[_CD_EXTRA_FIELD_LENGTH] |
| 845 | + centdir[_CD_COMMENT_LENGTH]) |
| 846 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 847 | if self.debug > 2: |
| 848 | print "total", total |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 849 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 850 | |
| 851 | def namelist(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 852 | """Return a list of file names in the archive.""" |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 853 | l = [] |
| 854 | for data in self.filelist: |
| 855 | l.append(data.filename) |
| 856 | return l |
| 857 | |
| 858 | def infolist(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 859 | """Return a list of class ZipInfo instances for files in the |
| 860 | archive.""" |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 861 | return self.filelist |
| 862 | |
| 863 | def printdir(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 864 | """Print a table of contents for the zip file.""" |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 865 | print "%-46s %19s %12s" % ("File Name", "Modified ", "Size") |
| 866 | for zinfo in self.filelist: |
Raymond Hettinger | 351e1a3 | 2008-01-14 22:58:05 +0000 | [diff] [blame] | 867 | date = "%d-%02d-%02d %02d:%02d:%02d" % zinfo.date_time[:6] |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 868 | print "%-46s %s %12d" % (zinfo.filename, date, zinfo.file_size) |
| 869 | |
| 870 | def testzip(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 871 | """Read all the files and check the CRC.""" |
Antoine Pitrou | c534270 | 2008-08-17 13:06:29 +0000 | [diff] [blame] | 872 | chunk_size = 2 ** 20 |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 873 | for zinfo in self.filelist: |
| 874 | try: |
Antoine Pitrou | c534270 | 2008-08-17 13:06:29 +0000 | [diff] [blame] | 875 | # Read by chunks, to avoid an OverflowError or a |
| 876 | # MemoryError with very large embedded files. |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 877 | with self.open(zinfo.filename, "r") as f: |
| 878 | while f.read(chunk_size): # Check CRC-32 |
| 879 | pass |
Raymond Hettinger | c0fac96 | 2003-06-27 22:25:03 +0000 | [diff] [blame] | 880 | except BadZipfile: |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 881 | return zinfo.filename |
| 882 | |
| 883 | def getinfo(self, name): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 884 | """Return the instance of ZipInfo given 'name'.""" |
Georg Brandl | 4b3ab6f | 2007-07-12 09:59:22 +0000 | [diff] [blame] | 885 | info = self.NameToInfo.get(name) |
| 886 | if info is None: |
| 887 | raise KeyError( |
| 888 | 'There is no item named %r in the archive' % name) |
| 889 | |
| 890 | return info |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 891 | |
Martin v. Löwis | c6d626e | 2007-02-13 09:49:38 +0000 | [diff] [blame] | 892 | def setpassword(self, pwd): |
| 893 | """Set default password for encrypted files.""" |
| 894 | self.pwd = pwd |
| 895 | |
R David Murray | 3f4ccba | 2012-04-12 18:42:47 -0400 | [diff] [blame] | 896 | @property |
| 897 | def comment(self): |
| 898 | """The comment text associated with the ZIP file.""" |
| 899 | return self._comment |
| 900 | |
| 901 | @comment.setter |
| 902 | def comment(self, comment): |
| 903 | # check for valid comment length |
| 904 | if len(comment) >= ZIP_MAX_COMMENT: |
| 905 | if self.debug: |
| 906 | print('Archive comment is too long; truncating to %d bytes' |
| 907 | % ZIP_MAX_COMMENT) |
| 908 | comment = comment[:ZIP_MAX_COMMENT] |
| 909 | self._comment = comment |
| 910 | self._didModify = True |
| 911 | |
Martin v. Löwis | c6d626e | 2007-02-13 09:49:38 +0000 | [diff] [blame] | 912 | def read(self, name, pwd=None): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 913 | """Return file bytes (as a string) for name.""" |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 914 | return self.open(name, "r", pwd).read() |
| 915 | |
| 916 | def open(self, name, mode="r", pwd=None): |
| 917 | """Return file-like object for 'name'.""" |
| 918 | if mode not in ("r", "U", "rU"): |
| 919 | raise RuntimeError, 'open() requires mode "r", "U", or "rU"' |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 920 | if not self.fp: |
| 921 | raise RuntimeError, \ |
Fred Drake | 5db246d | 2000-09-29 20:44:48 +0000 | [diff] [blame] | 922 | "Attempt to read ZIP archive that was already closed" |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 923 | |
Tim Peters | ea5962f | 2007-03-12 18:07:52 +0000 | [diff] [blame] | 924 | # Only open a new file for instances where we were not |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 925 | # given a file object in the constructor |
| 926 | if self._filePassed: |
| 927 | zef_file = self.fp |
Jesus Cea | 93d628b | 2012-11-04 02:32:08 +0100 | [diff] [blame] | 928 | should_close = False |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 929 | else: |
| 930 | zef_file = open(self.filename, 'rb') |
Jesus Cea | 93d628b | 2012-11-04 02:32:08 +0100 | [diff] [blame] | 931 | should_close = True |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 932 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 933 | try: |
| 934 | # Make sure we have an info object |
| 935 | if isinstance(name, ZipInfo): |
| 936 | # 'name' is already an info object |
| 937 | zinfo = name |
Gregory P. Smith | 0c63fc2 | 2008-01-20 01:21:03 +0000 | [diff] [blame] | 938 | else: |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 939 | # Get info object for name |
| 940 | zinfo = self.getinfo(name) |
Martin v. Löwis | 3eb7648 | 2007-03-06 10:41:24 +0000 | [diff] [blame] | 941 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 942 | zef_file.seek(zinfo.header_offset, 0) |
| 943 | |
| 944 | # Skip the file header: |
| 945 | fheader = zef_file.read(sizeFileHeader) |
| 946 | if fheader[0:4] != stringFileHeader: |
| 947 | raise BadZipfile, "Bad magic number for file header" |
| 948 | |
| 949 | fheader = struct.unpack(structFileHeader, fheader) |
| 950 | fname = zef_file.read(fheader[_FH_FILENAME_LENGTH]) |
| 951 | if fheader[_FH_EXTRA_FIELD_LENGTH]: |
| 952 | zef_file.read(fheader[_FH_EXTRA_FIELD_LENGTH]) |
| 953 | |
| 954 | if fname != zinfo.orig_filename: |
| 955 | raise BadZipfile, \ |
| 956 | 'File name in directory "%s" and header "%s" differ.' % ( |
| 957 | zinfo.orig_filename, fname) |
| 958 | |
| 959 | # check for encrypted flag & handle password |
| 960 | is_encrypted = zinfo.flag_bits & 0x1 |
| 961 | zd = None |
| 962 | if is_encrypted: |
| 963 | if not pwd: |
| 964 | pwd = self.pwd |
| 965 | if not pwd: |
| 966 | raise RuntimeError, "File %s is encrypted, " \ |
| 967 | "password required for extraction" % name |
| 968 | |
| 969 | zd = _ZipDecrypter(pwd) |
| 970 | # The first 12 bytes in the cypher stream is an encryption header |
| 971 | # used to strengthen the algorithm. The first 11 bytes are |
| 972 | # completely random, while the 12th contains the MSB of the CRC, |
| 973 | # or the MSB of the file time depending on the header type |
| 974 | # and is used to check the correctness of the password. |
| 975 | bytes = zef_file.read(12) |
| 976 | h = map(zd, bytes[0:12]) |
| 977 | if zinfo.flag_bits & 0x8: |
| 978 | # compare against the file type from extended local headers |
| 979 | check_byte = (zinfo._raw_time >> 8) & 0xff |
| 980 | else: |
| 981 | # compare against the CRC otherwise |
| 982 | check_byte = (zinfo.CRC >> 24) & 0xff |
| 983 | if ord(h[11]) != check_byte: |
| 984 | raise RuntimeError("Bad password for file", name) |
| 985 | |
| 986 | return ZipExtFile(zef_file, mode, zinfo, zd, |
| 987 | close_fileobj=should_close) |
| 988 | except: |
| 989 | if should_close: |
| 990 | zef_file.close() |
| 991 | raise |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 992 | |
Georg Brandl | 62416bc | 2008-01-07 18:47:44 +0000 | [diff] [blame] | 993 | def extract(self, member, path=None, pwd=None): |
| 994 | """Extract a member from the archive to the current working directory, |
| 995 | using its full name. Its file information is extracted as accurately |
| 996 | as possible. `member' may be a filename or a ZipInfo object. You can |
| 997 | specify a different directory using `path'. |
| 998 | """ |
| 999 | if not isinstance(member, ZipInfo): |
| 1000 | member = self.getinfo(member) |
| 1001 | |
| 1002 | if path is None: |
| 1003 | path = os.getcwd() |
| 1004 | |
| 1005 | return self._extract_member(member, path, pwd) |
| 1006 | |
| 1007 | def extractall(self, path=None, members=None, pwd=None): |
| 1008 | """Extract all members from the archive to the current working |
| 1009 | directory. `path' specifies a different directory to extract to. |
| 1010 | `members' is optional and must be a subset of the list returned |
| 1011 | by namelist(). |
| 1012 | """ |
| 1013 | if members is None: |
| 1014 | members = self.namelist() |
| 1015 | |
| 1016 | for zipinfo in members: |
| 1017 | self.extract(zipinfo, path, pwd) |
| 1018 | |
| 1019 | def _extract_member(self, member, targetpath, pwd): |
| 1020 | """Extract the ZipInfo object 'member' to a physical |
| 1021 | file on the path targetpath. |
| 1022 | """ |
| 1023 | # build the destination pathname, replacing |
| 1024 | # forward slashes to platform specific separators. |
Antoine Pitrou | 97377bf | 2009-05-04 21:17:17 +0000 | [diff] [blame] | 1025 | # Strip trailing path separator, unless it represents the root. |
| 1026 | if (targetpath[-1:] in (os.path.sep, os.path.altsep) |
| 1027 | and len(os.path.splitdrive(targetpath)[1]) > 1): |
Georg Brandl | 62416bc | 2008-01-07 18:47:44 +0000 | [diff] [blame] | 1028 | targetpath = targetpath[:-1] |
| 1029 | |
| 1030 | # don't include leading "/" from file name if present |
Martin v. Löwis | 0dfcfc8 | 2009-01-24 14:00:33 +0000 | [diff] [blame] | 1031 | if member.filename[0] == '/': |
Georg Brandl | 62416bc | 2008-01-07 18:47:44 +0000 | [diff] [blame] | 1032 | targetpath = os.path.join(targetpath, member.filename[1:]) |
| 1033 | else: |
| 1034 | targetpath = os.path.join(targetpath, member.filename) |
| 1035 | |
| 1036 | targetpath = os.path.normpath(targetpath) |
| 1037 | |
| 1038 | # Create all upper directories if necessary. |
| 1039 | upperdirs = os.path.dirname(targetpath) |
| 1040 | if upperdirs and not os.path.exists(upperdirs): |
| 1041 | os.makedirs(upperdirs) |
| 1042 | |
Martin v. Löwis | 0dfcfc8 | 2009-01-24 14:00:33 +0000 | [diff] [blame] | 1043 | if member.filename[-1] == '/': |
Martin v. Löwis | 0b09c42 | 2009-05-24 19:30:52 +0000 | [diff] [blame] | 1044 | if not os.path.isdir(targetpath): |
| 1045 | os.mkdir(targetpath) |
Martin v. Löwis | 0dfcfc8 | 2009-01-24 14:00:33 +0000 | [diff] [blame] | 1046 | return targetpath |
| 1047 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1048 | with self.open(member, pwd=pwd) as source, \ |
| 1049 | file(targetpath, "wb") as target: |
| 1050 | shutil.copyfileobj(source, target) |
Georg Brandl | 62416bc | 2008-01-07 18:47:44 +0000 | [diff] [blame] | 1051 | |
| 1052 | return targetpath |
| 1053 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1054 | def _writecheck(self, zinfo): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1055 | """Check for errors before writing a file to the archive.""" |
Raymond Hettinger | 54f0222 | 2002-06-01 14:18:47 +0000 | [diff] [blame] | 1056 | if zinfo.filename in self.NameToInfo: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 1057 | if self.debug: # Warning for duplicate names |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1058 | print "Duplicate name:", zinfo.filename |
| 1059 | if self.mode not in ("w", "a"): |
| 1060 | raise RuntimeError, 'write() requires mode "w" or "a"' |
| 1061 | if not self.fp: |
| 1062 | raise RuntimeError, \ |
Fred Drake | 5db246d | 2000-09-29 20:44:48 +0000 | [diff] [blame] | 1063 | "Attempt to write ZIP archive that was already closed" |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1064 | if zinfo.compress_type == ZIP_DEFLATED and not zlib: |
| 1065 | raise RuntimeError, \ |
Fred Drake | 5db246d | 2000-09-29 20:44:48 +0000 | [diff] [blame] | 1066 | "Compression requires the (missing) zlib module" |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1067 | if zinfo.compress_type not in (ZIP_STORED, ZIP_DEFLATED): |
| 1068 | raise RuntimeError, \ |
Fred Drake | 5db246d | 2000-09-29 20:44:48 +0000 | [diff] [blame] | 1069 | "That compression method is not supported" |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1070 | if zinfo.file_size > ZIP64_LIMIT: |
| 1071 | if not self._allowZip64: |
| 1072 | raise LargeZipFile("Filesize would require ZIP64 extensions") |
| 1073 | if zinfo.header_offset > ZIP64_LIMIT: |
| 1074 | if not self._allowZip64: |
| 1075 | raise LargeZipFile("Zipfile size would require ZIP64 extensions") |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1076 | |
| 1077 | def write(self, filename, arcname=None, compress_type=None): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1078 | """Put the bytes from filename into the archive under the name |
| 1079 | arcname.""" |
Georg Brandl | 4b3ab6f | 2007-07-12 09:59:22 +0000 | [diff] [blame] | 1080 | if not self.fp: |
| 1081 | raise RuntimeError( |
| 1082 | "Attempt to write to ZIP archive that was already closed") |
| 1083 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1084 | st = os.stat(filename) |
Martin v. Löwis | 0dfcfc8 | 2009-01-24 14:00:33 +0000 | [diff] [blame] | 1085 | isdir = stat.S_ISDIR(st.st_mode) |
Raymond Hettinger | 32200ae | 2002-06-01 19:51:15 +0000 | [diff] [blame] | 1086 | mtime = time.localtime(st.st_mtime) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1087 | date_time = mtime[0:6] |
| 1088 | # Create ZipInfo instance to store file information |
| 1089 | if arcname is None: |
Georg Brandl | 8f7c54e | 2006-02-20 08:40:38 +0000 | [diff] [blame] | 1090 | arcname = filename |
| 1091 | arcname = os.path.normpath(os.path.splitdrive(arcname)[1]) |
| 1092 | while arcname[0] in (os.sep, os.altsep): |
| 1093 | arcname = arcname[1:] |
Martin v. Löwis | 0dfcfc8 | 2009-01-24 14:00:33 +0000 | [diff] [blame] | 1094 | if isdir: |
| 1095 | arcname += '/' |
Georg Brandl | 8f7c54e | 2006-02-20 08:40:38 +0000 | [diff] [blame] | 1096 | zinfo = ZipInfo(arcname, date_time) |
Andrew M. Kuchling | 5543021 | 2004-07-10 15:40:29 +0000 | [diff] [blame] | 1097 | zinfo.external_attr = (st[0] & 0xFFFF) << 16L # Unix attributes |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1098 | if compress_type is None: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 1099 | zinfo.compress_type = self.compression |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1100 | else: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 1101 | zinfo.compress_type = compress_type |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1102 | |
| 1103 | zinfo.file_size = st.st_size |
Finn Bock | 03a3bb8 | 2001-09-05 18:40:33 +0000 | [diff] [blame] | 1104 | zinfo.flag_bits = 0x00 |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 1105 | zinfo.header_offset = self.fp.tell() # Start of header bytes |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1106 | |
| 1107 | self._writecheck(zinfo) |
| 1108 | self._didModify = True |
Martin v. Löwis | 0dfcfc8 | 2009-01-24 14:00:33 +0000 | [diff] [blame] | 1109 | |
| 1110 | if isdir: |
| 1111 | zinfo.file_size = 0 |
| 1112 | zinfo.compress_size = 0 |
| 1113 | zinfo.CRC = 0 |
| 1114 | self.filelist.append(zinfo) |
| 1115 | self.NameToInfo[zinfo.filename] = zinfo |
| 1116 | self.fp.write(zinfo.FileHeader()) |
| 1117 | return |
| 1118 | |
Benjamin Peterson | b91e8ed | 2009-05-10 02:29:00 +0000 | [diff] [blame] | 1119 | with open(filename, "rb") as fp: |
| 1120 | # Must overwrite CRC and sizes with correct data later |
| 1121 | zinfo.CRC = CRC = 0 |
| 1122 | zinfo.compress_size = compress_size = 0 |
| 1123 | zinfo.file_size = file_size = 0 |
| 1124 | self.fp.write(zinfo.FileHeader()) |
| 1125 | if zinfo.compress_type == ZIP_DEFLATED: |
| 1126 | cmpr = zlib.compressobj(zlib.Z_DEFAULT_COMPRESSION, |
| 1127 | zlib.DEFLATED, -15) |
| 1128 | else: |
| 1129 | cmpr = None |
| 1130 | while 1: |
| 1131 | buf = fp.read(1024 * 8) |
| 1132 | if not buf: |
| 1133 | break |
| 1134 | file_size = file_size + len(buf) |
| 1135 | CRC = crc32(buf, CRC) & 0xffffffff |
| 1136 | if cmpr: |
| 1137 | buf = cmpr.compress(buf) |
| 1138 | compress_size = compress_size + len(buf) |
| 1139 | self.fp.write(buf) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1140 | if cmpr: |
| 1141 | buf = cmpr.flush() |
| 1142 | compress_size = compress_size + len(buf) |
| 1143 | self.fp.write(buf) |
| 1144 | zinfo.compress_size = compress_size |
| 1145 | else: |
| 1146 | zinfo.compress_size = file_size |
| 1147 | zinfo.CRC = CRC |
| 1148 | zinfo.file_size = file_size |
Finn Bock | 03a3bb8 | 2001-09-05 18:40:33 +0000 | [diff] [blame] | 1149 | # Seek backwards and write CRC and file sizes |
Tim Peters | b64bec3 | 2001-09-18 02:26:39 +0000 | [diff] [blame] | 1150 | position = self.fp.tell() # Preserve current position in file |
Finn Bock | 03a3bb8 | 2001-09-05 18:40:33 +0000 | [diff] [blame] | 1151 | self.fp.seek(zinfo.header_offset + 14, 0) |
Gregory P. Smith | bf02e3b | 2008-03-19 03:14:41 +0000 | [diff] [blame] | 1152 | self.fp.write(struct.pack("<LLL", zinfo.CRC, zinfo.compress_size, |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1153 | zinfo.file_size)) |
Finn Bock | 03a3bb8 | 2001-09-05 18:40:33 +0000 | [diff] [blame] | 1154 | self.fp.seek(position, 0) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1155 | self.filelist.append(zinfo) |
| 1156 | self.NameToInfo[zinfo.filename] = zinfo |
| 1157 | |
Ronald Oussoren | dd25e86 | 2010-02-07 20:18:02 +0000 | [diff] [blame] | 1158 | def writestr(self, zinfo_or_arcname, bytes, compress_type=None): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1159 | """Write a file into the archive. The contents is the string |
Just van Rossum | b083cb3 | 2002-12-12 12:23:32 +0000 | [diff] [blame] | 1160 | 'bytes'. 'zinfo_or_arcname' is either a ZipInfo instance or |
| 1161 | the name of the file in the archive.""" |
| 1162 | if not isinstance(zinfo_or_arcname, ZipInfo): |
| 1163 | zinfo = ZipInfo(filename=zinfo_or_arcname, |
Raymond Hettinger | 351e1a3 | 2008-01-14 22:58:05 +0000 | [diff] [blame] | 1164 | date_time=time.localtime(time.time())[:6]) |
Ronald Oussoren | dd25e86 | 2010-02-07 20:18:02 +0000 | [diff] [blame] | 1165 | |
Just van Rossum | b083cb3 | 2002-12-12 12:23:32 +0000 | [diff] [blame] | 1166 | zinfo.compress_type = self.compression |
Antoine Pitrou | 5fdfa3e | 2008-07-25 19:42:26 +0000 | [diff] [blame] | 1167 | zinfo.external_attr = 0600 << 16 |
Just van Rossum | b083cb3 | 2002-12-12 12:23:32 +0000 | [diff] [blame] | 1168 | else: |
| 1169 | zinfo = zinfo_or_arcname |
Georg Brandl | 4b3ab6f | 2007-07-12 09:59:22 +0000 | [diff] [blame] | 1170 | |
| 1171 | if not self.fp: |
| 1172 | raise RuntimeError( |
| 1173 | "Attempt to write to ZIP archive that was already closed") |
| 1174 | |
Ronald Oussoren | dd25e86 | 2010-02-07 20:18:02 +0000 | [diff] [blame] | 1175 | if compress_type is not None: |
| 1176 | zinfo.compress_type = compress_type |
| 1177 | |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 1178 | zinfo.file_size = len(bytes) # Uncompressed size |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1179 | zinfo.header_offset = self.fp.tell() # Start of header bytes |
| 1180 | self._writecheck(zinfo) |
| 1181 | self._didModify = True |
Gregory P. Smith | bf02e3b | 2008-03-19 03:14:41 +0000 | [diff] [blame] | 1182 | zinfo.CRC = crc32(bytes) & 0xffffffff # CRC-32 checksum |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1183 | if zinfo.compress_type == ZIP_DEFLATED: |
| 1184 | co = zlib.compressobj(zlib.Z_DEFAULT_COMPRESSION, |
| 1185 | zlib.DEFLATED, -15) |
| 1186 | bytes = co.compress(bytes) + co.flush() |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 1187 | zinfo.compress_size = len(bytes) # Compressed size |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1188 | else: |
| 1189 | zinfo.compress_size = zinfo.file_size |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 1190 | zinfo.header_offset = self.fp.tell() # Start of header bytes |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1191 | self.fp.write(zinfo.FileHeader()) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1192 | self.fp.write(bytes) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1193 | self.fp.flush() |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1194 | if zinfo.flag_bits & 0x08: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 1195 | # Write CRC and file sizes after the file data |
Gregory P. Smith | 2662733 | 2009-06-26 07:50:21 +0000 | [diff] [blame] | 1196 | self.fp.write(struct.pack("<LLL", zinfo.CRC, zinfo.compress_size, |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 1197 | zinfo.file_size)) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1198 | self.filelist.append(zinfo) |
| 1199 | self.NameToInfo[zinfo.filename] = zinfo |
| 1200 | |
| 1201 | def __del__(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1202 | """Call the "close()" method in case the user forgot.""" |
Tim Peters | d15f8bb | 2001-11-28 23:16:40 +0000 | [diff] [blame] | 1203 | self.close() |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1204 | |
| 1205 | def close(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1206 | """Close the file, and for mode "w" and "a" write the ending |
| 1207 | records.""" |
Tim Peters | d15f8bb | 2001-11-28 23:16:40 +0000 | [diff] [blame] | 1208 | if self.fp is None: |
| 1209 | return |
Tim Peters | a608bb2 | 2006-06-15 18:06:29 +0000 | [diff] [blame] | 1210 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1211 | try: |
| 1212 | if self.mode in ("w", "a") and self._didModify: # write ending records |
| 1213 | count = 0 |
| 1214 | pos1 = self.fp.tell() |
| 1215 | for zinfo in self.filelist: # write central directory |
| 1216 | count = count + 1 |
| 1217 | dt = zinfo.date_time |
| 1218 | dosdate = (dt[0] - 1980) << 9 | dt[1] << 5 | dt[2] |
| 1219 | dostime = dt[3] << 11 | dt[4] << 5 | (dt[5] // 2) |
| 1220 | extra = [] |
| 1221 | if zinfo.file_size > ZIP64_LIMIT \ |
| 1222 | or zinfo.compress_size > ZIP64_LIMIT: |
| 1223 | extra.append(zinfo.file_size) |
| 1224 | extra.append(zinfo.compress_size) |
| 1225 | file_size = 0xffffffff |
| 1226 | compress_size = 0xffffffff |
| 1227 | else: |
| 1228 | file_size = zinfo.file_size |
| 1229 | compress_size = zinfo.compress_size |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1230 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1231 | if zinfo.header_offset > ZIP64_LIMIT: |
| 1232 | extra.append(zinfo.header_offset) |
| 1233 | header_offset = 0xffffffffL |
| 1234 | else: |
| 1235 | header_offset = zinfo.header_offset |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1236 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1237 | extra_data = zinfo.extra |
| 1238 | if extra: |
| 1239 | # Append a ZIP64 field to the extra's |
| 1240 | extra_data = struct.pack( |
| 1241 | '<HH' + 'Q'*len(extra), |
| 1242 | 1, 8*len(extra), *extra) + extra_data |
Tim Peters | a608bb2 | 2006-06-15 18:06:29 +0000 | [diff] [blame] | 1243 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1244 | extract_version = max(45, zinfo.extract_version) |
| 1245 | create_version = max(45, zinfo.create_version) |
| 1246 | else: |
| 1247 | extract_version = zinfo.extract_version |
| 1248 | create_version = zinfo.create_version |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1249 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1250 | try: |
| 1251 | filename, flag_bits = zinfo._encodeFilenameFlags() |
| 1252 | centdir = struct.pack(structCentralDir, |
| 1253 | stringCentralDir, create_version, |
| 1254 | zinfo.create_system, extract_version, zinfo.reserved, |
| 1255 | flag_bits, zinfo.compress_type, dostime, dosdate, |
| 1256 | zinfo.CRC, compress_size, file_size, |
| 1257 | len(filename), len(extra_data), len(zinfo.comment), |
| 1258 | 0, zinfo.internal_attr, zinfo.external_attr, |
| 1259 | header_offset) |
| 1260 | except DeprecationWarning: |
| 1261 | print >>sys.stderr, (structCentralDir, |
| 1262 | stringCentralDir, create_version, |
| 1263 | zinfo.create_system, extract_version, zinfo.reserved, |
| 1264 | zinfo.flag_bits, zinfo.compress_type, dostime, dosdate, |
| 1265 | zinfo.CRC, compress_size, file_size, |
| 1266 | len(zinfo.filename), len(extra_data), len(zinfo.comment), |
| 1267 | 0, zinfo.internal_attr, zinfo.external_attr, |
| 1268 | header_offset) |
| 1269 | raise |
| 1270 | self.fp.write(centdir) |
| 1271 | self.fp.write(filename) |
| 1272 | self.fp.write(extra_data) |
| 1273 | self.fp.write(zinfo.comment) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1274 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1275 | pos2 = self.fp.tell() |
| 1276 | # Write end-of-zip-archive record |
| 1277 | centDirCount = count |
| 1278 | centDirSize = pos2 - pos1 |
| 1279 | centDirOffset = pos1 |
| 1280 | if (centDirCount >= ZIP_FILECOUNT_LIMIT or |
| 1281 | centDirOffset > ZIP64_LIMIT or |
| 1282 | centDirSize > ZIP64_LIMIT): |
| 1283 | # Need to write the ZIP64 end-of-archive records |
| 1284 | zip64endrec = struct.pack( |
| 1285 | structEndArchive64, stringEndArchive64, |
| 1286 | 44, 45, 45, 0, 0, centDirCount, centDirCount, |
| 1287 | centDirSize, centDirOffset) |
| 1288 | self.fp.write(zip64endrec) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1289 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1290 | zip64locrec = struct.pack( |
| 1291 | structEndArchive64Locator, |
| 1292 | stringEndArchive64Locator, 0, pos2, 1) |
| 1293 | self.fp.write(zip64locrec) |
| 1294 | centDirCount = min(centDirCount, 0xFFFF) |
| 1295 | centDirSize = min(centDirSize, 0xFFFFFFFF) |
| 1296 | centDirOffset = min(centDirOffset, 0xFFFFFFFF) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1297 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1298 | endrec = struct.pack(structEndArchive, stringEndArchive, |
| 1299 | 0, 0, centDirCount, centDirCount, |
| 1300 | centDirSize, centDirOffset, len(self._comment)) |
| 1301 | self.fp.write(endrec) |
| 1302 | self.fp.write(self._comment) |
| 1303 | self.fp.flush() |
| 1304 | finally: |
| 1305 | fp = self.fp |
| 1306 | self.fp = None |
| 1307 | if not self._filePassed: |
| 1308 | fp.close() |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1309 | |
| 1310 | |
| 1311 | class PyZipFile(ZipFile): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1312 | """Class to create ZIP archives with Python library files and packages.""" |
| 1313 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1314 | def writepy(self, pathname, basename = ""): |
| 1315 | """Add all files from "pathname" to the ZIP archive. |
| 1316 | |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1317 | If pathname is a package directory, search the directory and |
| 1318 | all package subdirectories recursively for all *.py and enter |
| 1319 | the modules into the archive. If pathname is a plain |
| 1320 | directory, listdir *.py and enter all modules. Else, pathname |
| 1321 | must be a Python *.py file and the module will be put into the |
| 1322 | archive. Added modules are always module.pyo or module.pyc. |
| 1323 | This method will compile the module.py into module.pyc if |
| 1324 | necessary. |
| 1325 | """ |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1326 | dir, name = os.path.split(pathname) |
| 1327 | if os.path.isdir(pathname): |
| 1328 | initname = os.path.join(pathname, "__init__.py") |
| 1329 | if os.path.isfile(initname): |
| 1330 | # This is a package directory, add it |
| 1331 | if basename: |
| 1332 | basename = "%s/%s" % (basename, name) |
| 1333 | else: |
| 1334 | basename = name |
| 1335 | if self.debug: |
| 1336 | print "Adding package in", pathname, "as", basename |
| 1337 | fname, arcname = self._get_codename(initname[0:-3], basename) |
| 1338 | if self.debug: |
| 1339 | print "Adding", arcname |
| 1340 | self.write(fname, arcname) |
| 1341 | dirlist = os.listdir(pathname) |
| 1342 | dirlist.remove("__init__.py") |
| 1343 | # Add all *.py files and package subdirectories |
| 1344 | for filename in dirlist: |
| 1345 | path = os.path.join(pathname, filename) |
| 1346 | root, ext = os.path.splitext(filename) |
| 1347 | if os.path.isdir(path): |
| 1348 | if os.path.isfile(os.path.join(path, "__init__.py")): |
| 1349 | # This is a package directory, add it |
| 1350 | self.writepy(path, basename) # Recursive call |
| 1351 | elif ext == ".py": |
| 1352 | fname, arcname = self._get_codename(path[0:-3], |
| 1353 | basename) |
| 1354 | if self.debug: |
| 1355 | print "Adding", arcname |
| 1356 | self.write(fname, arcname) |
| 1357 | else: |
| 1358 | # This is NOT a package directory, add its files at top level |
| 1359 | if self.debug: |
| 1360 | print "Adding files from directory", pathname |
| 1361 | for filename in os.listdir(pathname): |
| 1362 | path = os.path.join(pathname, filename) |
| 1363 | root, ext = os.path.splitext(filename) |
| 1364 | if ext == ".py": |
| 1365 | fname, arcname = self._get_codename(path[0:-3], |
| 1366 | basename) |
| 1367 | if self.debug: |
| 1368 | print "Adding", arcname |
| 1369 | self.write(fname, arcname) |
| 1370 | else: |
| 1371 | if pathname[-3:] != ".py": |
| 1372 | raise RuntimeError, \ |
Fred Drake | 5db246d | 2000-09-29 20:44:48 +0000 | [diff] [blame] | 1373 | 'Files added with writepy() must end with ".py"' |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1374 | fname, arcname = self._get_codename(pathname[0:-3], basename) |
| 1375 | if self.debug: |
| 1376 | print "Adding file", arcname |
| 1377 | self.write(fname, arcname) |
| 1378 | |
| 1379 | def _get_codename(self, pathname, basename): |
| 1380 | """Return (filename, archivename) for the path. |
| 1381 | |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1382 | Given a module name path, return the correct file path and |
| 1383 | archive name, compiling if necessary. For example, given |
| 1384 | /python/lib/string, return (/python/lib/string.pyc, string). |
| 1385 | """ |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1386 | file_py = pathname + ".py" |
| 1387 | file_pyc = pathname + ".pyc" |
| 1388 | file_pyo = pathname + ".pyo" |
| 1389 | if os.path.isfile(file_pyo) and \ |
Raymond Hettinger | 32200ae | 2002-06-01 19:51:15 +0000 | [diff] [blame] | 1390 | os.stat(file_pyo).st_mtime >= os.stat(file_py).st_mtime: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 1391 | fname = file_pyo # Use .pyo file |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1392 | elif not os.path.isfile(file_pyc) or \ |
Raymond Hettinger | 32200ae | 2002-06-01 19:51:15 +0000 | [diff] [blame] | 1393 | os.stat(file_pyc).st_mtime < os.stat(file_py).st_mtime: |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1394 | import py_compile |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1395 | if self.debug: |
| 1396 | print "Compiling", file_py |
Martin v. Löwis | 0c6774d | 2003-01-15 11:51:06 +0000 | [diff] [blame] | 1397 | try: |
| 1398 | py_compile.compile(file_py, file_pyc, None, True) |
| 1399 | except py_compile.PyCompileError,err: |
| 1400 | print err.msg |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1401 | fname = file_pyc |
| 1402 | else: |
| 1403 | fname = file_pyc |
| 1404 | archivename = os.path.split(fname)[1] |
| 1405 | if basename: |
| 1406 | archivename = "%s/%s" % (basename, archivename) |
| 1407 | return (fname, archivename) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1408 | |
| 1409 | |
| 1410 | def main(args = None): |
| 1411 | import textwrap |
| 1412 | USAGE=textwrap.dedent("""\ |
| 1413 | Usage: |
| 1414 | zipfile.py -l zipfile.zip # Show listing of a zipfile |
| 1415 | zipfile.py -t zipfile.zip # Test if a zipfile is valid |
| 1416 | zipfile.py -e zipfile.zip target # Extract zipfile into target dir |
| 1417 | zipfile.py -c zipfile.zip src ... # Create zipfile from sources |
| 1418 | """) |
| 1419 | if args is None: |
| 1420 | args = sys.argv[1:] |
| 1421 | |
| 1422 | if not args or args[0] not in ('-l', '-c', '-e', '-t'): |
| 1423 | print USAGE |
| 1424 | sys.exit(1) |
| 1425 | |
| 1426 | if args[0] == '-l': |
| 1427 | if len(args) != 2: |
| 1428 | print USAGE |
| 1429 | sys.exit(1) |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1430 | with ZipFile(args[1], 'r') as zf: |
| 1431 | zf.printdir() |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1432 | |
| 1433 | elif args[0] == '-t': |
| 1434 | if len(args) != 2: |
| 1435 | print USAGE |
| 1436 | sys.exit(1) |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1437 | with ZipFile(args[1], 'r') as zf: |
| 1438 | badfile = zf.testzip() |
Antoine Pitrou | e1436d1 | 2010-08-12 15:25:51 +0000 | [diff] [blame] | 1439 | if badfile: |
| 1440 | print("The following enclosed file is corrupted: {!r}".format(badfile)) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1441 | print "Done testing" |
| 1442 | |
| 1443 | elif args[0] == '-e': |
| 1444 | if len(args) != 3: |
| 1445 | print USAGE |
| 1446 | sys.exit(1) |
| 1447 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1448 | with ZipFile(args[1], 'r') as zf: |
| 1449 | out = args[2] |
| 1450 | for path in zf.namelist(): |
| 1451 | if path.startswith('./'): |
| 1452 | tgt = os.path.join(out, path[2:]) |
| 1453 | else: |
| 1454 | tgt = os.path.join(out, path) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1455 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1456 | tgtdir = os.path.dirname(tgt) |
| 1457 | if not os.path.exists(tgtdir): |
| 1458 | os.makedirs(tgtdir) |
| 1459 | with open(tgt, 'wb') as fp: |
| 1460 | fp.write(zf.read(path)) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1461 | |
| 1462 | elif args[0] == '-c': |
| 1463 | if len(args) < 3: |
| 1464 | print USAGE |
| 1465 | sys.exit(1) |
| 1466 | |
| 1467 | def addToZip(zf, path, zippath): |
| 1468 | if os.path.isfile(path): |
| 1469 | zf.write(path, zippath, ZIP_DEFLATED) |
| 1470 | elif os.path.isdir(path): |
| 1471 | for nm in os.listdir(path): |
Tim Peters | a608bb2 | 2006-06-15 18:06:29 +0000 | [diff] [blame] | 1472 | addToZip(zf, |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1473 | os.path.join(path, nm), os.path.join(zippath, nm)) |
Tim Peters | a608bb2 | 2006-06-15 18:06:29 +0000 | [diff] [blame] | 1474 | # else: ignore |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1475 | |
Antoine Pitrou | 02512fb | 2012-11-17 23:56:53 +0100 | [diff] [blame] | 1476 | with ZipFile(args[1], 'w', allowZip64=True) as zf: |
| 1477 | for src in args[2:]: |
| 1478 | addToZip(zf, src, os.path.basename(src)) |
Ronald Oussoren | 143cefb | 2006-06-15 08:14:18 +0000 | [diff] [blame] | 1479 | |
| 1480 | if __name__ == "__main__": |
| 1481 | main() |