Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1 | """ |
| 2 | Read and write ZIP files. |
Guido van Rossum | d6ca546 | 2007-05-22 01:29:33 +0000 | [diff] [blame] | 3 | |
| 4 | XXX references to utf-8 need further investigation. |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 5 | """ |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 6 | import io |
Barry Warsaw | 28a691b | 2010-04-17 00:19:56 +0000 | [diff] [blame] | 7 | import os |
Brett Cannon | b57a085 | 2013-06-15 17:32:30 -0400 | [diff] [blame] | 8 | import importlib.util |
Barry Warsaw | 28a691b | 2010-04-17 00:19:56 +0000 | [diff] [blame] | 9 | import sys |
| 10 | import time |
| 11 | import stat |
| 12 | import shutil |
| 13 | import struct |
| 14 | import binascii |
Antoine Pitrou | a6a4dc8 | 2017-09-07 18:56:24 +0200 | [diff] [blame] | 15 | import threading |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 16 | |
| 17 | try: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 18 | import zlib # We may need its compression method |
Christian Heimes | d5e2b6f | 2008-03-19 21:50:51 +0000 | [diff] [blame] | 19 | crc32 = zlib.crc32 |
Brett Cannon | 260fbe8 | 2013-07-04 18:16:15 -0400 | [diff] [blame] | 20 | except ImportError: |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 21 | zlib = None |
Christian Heimes | d5e2b6f | 2008-03-19 21:50:51 +0000 | [diff] [blame] | 22 | crc32 = binascii.crc32 |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 23 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 24 | try: |
| 25 | import bz2 # We may need its compression method |
Brett Cannon | 260fbe8 | 2013-07-04 18:16:15 -0400 | [diff] [blame] | 26 | except ImportError: |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 27 | bz2 = None |
| 28 | |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 29 | try: |
| 30 | import lzma # We may need its compression method |
Brett Cannon | 260fbe8 | 2013-07-04 18:16:15 -0400 | [diff] [blame] | 31 | except ImportError: |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 32 | lzma = None |
| 33 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 34 | __all__ = ["BadZipFile", "BadZipfile", "error", |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 35 | "ZIP_STORED", "ZIP_DEFLATED", "ZIP_BZIP2", "ZIP_LZMA", |
Georg Brandl | 4d54088 | 2010-10-28 06:42:33 +0000 | [diff] [blame] | 36 | "is_zipfile", "ZipInfo", "ZipFile", "PyZipFile", "LargeZipFile"] |
Skip Montanaro | 40fc160 | 2001-03-01 04:27:19 +0000 | [diff] [blame] | 37 | |
Georg Brandl | 4d54088 | 2010-10-28 06:42:33 +0000 | [diff] [blame] | 38 | class BadZipFile(Exception): |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 39 | pass |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 40 | |
| 41 | |
| 42 | class LargeZipFile(Exception): |
| 43 | """ |
| 44 | Raised when writing a zipfile, the zipfile requires ZIP64 extensions |
| 45 | and those extensions are disabled. |
| 46 | """ |
| 47 | |
Georg Brandl | 4d54088 | 2010-10-28 06:42:33 +0000 | [diff] [blame] | 48 | error = BadZipfile = BadZipFile # Pre-3.2 compatibility names |
| 49 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 50 | |
Amaury Forgeot d'Arc | 0c3f8a4 | 2009-01-17 16:42:26 +0000 | [diff] [blame] | 51 | ZIP64_LIMIT = (1 << 31) - 1 |
Serhiy Storchaka | cfbb394 | 2014-09-23 21:34:24 +0300 | [diff] [blame] | 52 | ZIP_FILECOUNT_LIMIT = (1 << 16) - 1 |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 53 | ZIP_MAX_COMMENT = (1 << 16) - 1 |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 54 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 55 | # constants for Zip file compression methods |
| 56 | ZIP_STORED = 0 |
| 57 | ZIP_DEFLATED = 8 |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 58 | ZIP_BZIP2 = 12 |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 59 | ZIP_LZMA = 14 |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 60 | # Other ZIP compression methods not supported |
| 61 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 62 | DEFAULT_VERSION = 20 |
| 63 | ZIP64_VERSION = 45 |
| 64 | BZIP2_VERSION = 46 |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 65 | LZMA_VERSION = 63 |
Martin v. Löwis | d099b56 | 2012-05-01 14:08:22 +0200 | [diff] [blame] | 66 | # we recognize (but not necessarily support) all features up to that version |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 67 | MAX_EXTRACT_VERSION = 63 |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 68 | |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 69 | # Below are some formats and associated data for reading/writing headers using |
| 70 | # the struct module. The names and structures of headers/records are those used |
| 71 | # in the PKWARE description of the ZIP file format: |
| 72 | # http://www.pkware.com/documents/casestudies/APPNOTE.TXT |
| 73 | # (URL valid as of January 2008) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 74 | |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 75 | # The "end of central directory" structure, magic number, size, and indices |
| 76 | # (section V.I in the format document) |
Georg Brandl | 2ee470f | 2008-07-16 12:55:28 +0000 | [diff] [blame] | 77 | structEndArchive = b"<4s4H2LH" |
| 78 | stringEndArchive = b"PK\005\006" |
| 79 | sizeEndCentDir = struct.calcsize(structEndArchive) |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 80 | |
| 81 | _ECD_SIGNATURE = 0 |
| 82 | _ECD_DISK_NUMBER = 1 |
| 83 | _ECD_DISK_START = 2 |
| 84 | _ECD_ENTRIES_THIS_DISK = 3 |
| 85 | _ECD_ENTRIES_TOTAL = 4 |
| 86 | _ECD_SIZE = 5 |
| 87 | _ECD_OFFSET = 6 |
| 88 | _ECD_COMMENT_SIZE = 7 |
| 89 | # These last two indices are not part of the structure as defined in the |
| 90 | # spec, but they are used internally by this module as a convenience |
| 91 | _ECD_COMMENT = 8 |
| 92 | _ECD_LOCATION = 9 |
| 93 | |
| 94 | # The "central directory" structure, magic number, size, and indices |
| 95 | # of entries in the structure (section V.F in the format document) |
| 96 | structCentralDir = "<4s4B4HL2L5H2L" |
Georg Brandl | 2ee470f | 2008-07-16 12:55:28 +0000 | [diff] [blame] | 97 | stringCentralDir = b"PK\001\002" |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 98 | sizeCentralDir = struct.calcsize(structCentralDir) |
| 99 | |
Fred Drake | 3e038e5 | 2001-02-28 17:56:26 +0000 | [diff] [blame] | 100 | # indexes of entries in the central directory structure |
| 101 | _CD_SIGNATURE = 0 |
| 102 | _CD_CREATE_VERSION = 1 |
| 103 | _CD_CREATE_SYSTEM = 2 |
| 104 | _CD_EXTRACT_VERSION = 3 |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 105 | _CD_EXTRACT_SYSTEM = 4 |
Fred Drake | 3e038e5 | 2001-02-28 17:56:26 +0000 | [diff] [blame] | 106 | _CD_FLAG_BITS = 5 |
| 107 | _CD_COMPRESS_TYPE = 6 |
| 108 | _CD_TIME = 7 |
| 109 | _CD_DATE = 8 |
| 110 | _CD_CRC = 9 |
| 111 | _CD_COMPRESSED_SIZE = 10 |
| 112 | _CD_UNCOMPRESSED_SIZE = 11 |
| 113 | _CD_FILENAME_LENGTH = 12 |
| 114 | _CD_EXTRA_FIELD_LENGTH = 13 |
| 115 | _CD_COMMENT_LENGTH = 14 |
| 116 | _CD_DISK_NUMBER_START = 15 |
| 117 | _CD_INTERNAL_FILE_ATTRIBUTES = 16 |
| 118 | _CD_EXTERNAL_FILE_ATTRIBUTES = 17 |
| 119 | _CD_LOCAL_HEADER_OFFSET = 18 |
| 120 | |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 121 | # The "local file header" structure, magic number, size, and indices |
| 122 | # (section V.A in the format document) |
| 123 | structFileHeader = "<4s2B4HL2L2H" |
Georg Brandl | 2ee470f | 2008-07-16 12:55:28 +0000 | [diff] [blame] | 124 | stringFileHeader = b"PK\003\004" |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 125 | sizeFileHeader = struct.calcsize(structFileHeader) |
| 126 | |
Fred Drake | 3e038e5 | 2001-02-28 17:56:26 +0000 | [diff] [blame] | 127 | _FH_SIGNATURE = 0 |
| 128 | _FH_EXTRACT_VERSION = 1 |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 129 | _FH_EXTRACT_SYSTEM = 2 |
Fred Drake | 3e038e5 | 2001-02-28 17:56:26 +0000 | [diff] [blame] | 130 | _FH_GENERAL_PURPOSE_FLAG_BITS = 3 |
| 131 | _FH_COMPRESSION_METHOD = 4 |
| 132 | _FH_LAST_MOD_TIME = 5 |
| 133 | _FH_LAST_MOD_DATE = 6 |
| 134 | _FH_CRC = 7 |
| 135 | _FH_COMPRESSED_SIZE = 8 |
| 136 | _FH_UNCOMPRESSED_SIZE = 9 |
| 137 | _FH_FILENAME_LENGTH = 10 |
| 138 | _FH_EXTRA_FIELD_LENGTH = 11 |
| 139 | |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 140 | # The "Zip64 end of central directory locator" structure, magic number, and size |
Georg Brandl | 2ee470f | 2008-07-16 12:55:28 +0000 | [diff] [blame] | 141 | structEndArchive64Locator = "<4sLQL" |
| 142 | stringEndArchive64Locator = b"PK\x06\x07" |
| 143 | sizeEndCentDir64Locator = struct.calcsize(structEndArchive64Locator) |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 144 | |
| 145 | # The "Zip64 end of central directory" record, magic number, size, and indices |
| 146 | # (section V.G in the format document) |
Georg Brandl | 2ee470f | 2008-07-16 12:55:28 +0000 | [diff] [blame] | 147 | structEndArchive64 = "<4sQ2H2L4Q" |
| 148 | stringEndArchive64 = b"PK\x06\x06" |
| 149 | sizeEndCentDir64 = struct.calcsize(structEndArchive64) |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 150 | |
| 151 | _CD64_SIGNATURE = 0 |
| 152 | _CD64_DIRECTORY_RECSIZE = 1 |
| 153 | _CD64_CREATE_VERSION = 2 |
| 154 | _CD64_EXTRACT_VERSION = 3 |
| 155 | _CD64_DISK_NUMBER = 4 |
| 156 | _CD64_DISK_NUMBER_START = 5 |
| 157 | _CD64_NUMBER_ENTRIES_THIS_DISK = 6 |
| 158 | _CD64_NUMBER_ENTRIES_TOTAL = 7 |
| 159 | _CD64_DIRECTORY_SIZE = 8 |
| 160 | _CD64_OFFSET_START_CENTDIR = 9 |
| 161 | |
Silas Sewell | 4ba3b50 | 2018-09-18 13:00:05 -0400 | [diff] [blame] | 162 | _DD_SIGNATURE = 0x08074b50 |
| 163 | |
Serhiy Storchaka | 9bdb7be | 2018-09-17 15:36:40 +0300 | [diff] [blame] | 164 | _EXTRA_FIELD_STRUCT = struct.Struct('<HH') |
| 165 | |
| 166 | def _strip_extra(extra, xids): |
| 167 | # Remove Extra Fields with specified IDs. |
| 168 | unpack = _EXTRA_FIELD_STRUCT.unpack |
| 169 | modified = False |
| 170 | buffer = [] |
| 171 | start = i = 0 |
| 172 | while i + 4 <= len(extra): |
| 173 | xid, xlen = unpack(extra[i : i + 4]) |
| 174 | j = i + 4 + xlen |
| 175 | if xid in xids: |
| 176 | if i != start: |
| 177 | buffer.append(extra[start : i]) |
| 178 | start = j |
| 179 | modified = True |
| 180 | i = j |
| 181 | if not modified: |
| 182 | return extra |
| 183 | return b''.join(buffer) |
| 184 | |
Antoine Pitrou | db5fe66 | 2008-12-27 15:50:40 +0000 | [diff] [blame] | 185 | def _check_zipfile(fp): |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 186 | try: |
Antoine Pitrou | db5fe66 | 2008-12-27 15:50:40 +0000 | [diff] [blame] | 187 | if _EndRecData(fp): |
| 188 | return True # file has correct magic number |
Andrew Svetlov | f7a17b4 | 2012-12-25 16:47:37 +0200 | [diff] [blame] | 189 | except OSError: |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 190 | pass |
Guido van Rossum | 8ca162f | 2002-04-07 06:36:23 +0000 | [diff] [blame] | 191 | return False |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 192 | |
Antoine Pitrou | db5fe66 | 2008-12-27 15:50:40 +0000 | [diff] [blame] | 193 | def is_zipfile(filename): |
| 194 | """Quickly see if a file is a ZIP file by checking the magic number. |
| 195 | |
| 196 | The filename argument may be a file or file-like object too. |
| 197 | """ |
| 198 | result = False |
| 199 | try: |
| 200 | if hasattr(filename, "read"): |
| 201 | result = _check_zipfile(fp=filename) |
| 202 | else: |
| 203 | with open(filename, "rb") as fp: |
| 204 | result = _check_zipfile(fp) |
Andrew Svetlov | f7a17b4 | 2012-12-25 16:47:37 +0200 | [diff] [blame] | 205 | except OSError: |
Antoine Pitrou | db5fe66 | 2008-12-27 15:50:40 +0000 | [diff] [blame] | 206 | pass |
| 207 | return result |
| 208 | |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 209 | def _EndRecData64(fpin, offset, endrec): |
| 210 | """ |
| 211 | Read the ZIP64 end-of-archive records and use that to update endrec |
| 212 | """ |
Georg Brandl | 268e4d4 | 2010-10-14 06:59:45 +0000 | [diff] [blame] | 213 | try: |
| 214 | fpin.seek(offset - sizeEndCentDir64Locator, 2) |
Andrew Svetlov | f7a17b4 | 2012-12-25 16:47:37 +0200 | [diff] [blame] | 215 | except OSError: |
Georg Brandl | 268e4d4 | 2010-10-14 06:59:45 +0000 | [diff] [blame] | 216 | # If the seek fails, the file is not large enough to contain a ZIP64 |
| 217 | # end-of-archive record, so just return the end record we were given. |
| 218 | return endrec |
| 219 | |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 220 | data = fpin.read(sizeEndCentDir64Locator) |
Serhiy Storchaka | d2b1527 | 2013-01-31 15:27:07 +0200 | [diff] [blame] | 221 | if len(data) != sizeEndCentDir64Locator: |
| 222 | return endrec |
Georg Brandl | 2ee470f | 2008-07-16 12:55:28 +0000 | [diff] [blame] | 223 | sig, diskno, reloff, disks = struct.unpack(structEndArchive64Locator, data) |
| 224 | if sig != stringEndArchive64Locator: |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 225 | return endrec |
| 226 | |
| 227 | if diskno != 0 or disks != 1: |
Éric Araujo | ae2d832 | 2010-10-28 13:49:17 +0000 | [diff] [blame] | 228 | raise BadZipFile("zipfiles that span multiple disks are not supported") |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 229 | |
| 230 | # Assume no 'zip64 extensible data' |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 231 | fpin.seek(offset - sizeEndCentDir64Locator - sizeEndCentDir64, 2) |
| 232 | data = fpin.read(sizeEndCentDir64) |
Serhiy Storchaka | d2b1527 | 2013-01-31 15:27:07 +0200 | [diff] [blame] | 233 | if len(data) != sizeEndCentDir64: |
| 234 | return endrec |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 235 | sig, sz, create_version, read_version, disk_num, disk_dir, \ |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 236 | dircount, dircount2, dirsize, diroffset = \ |
| 237 | struct.unpack(structEndArchive64, data) |
Georg Brandl | 2ee470f | 2008-07-16 12:55:28 +0000 | [diff] [blame] | 238 | if sig != stringEndArchive64: |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 239 | return endrec |
| 240 | |
| 241 | # Update the original endrec using data from the ZIP64 record |
Antoine Pitrou | 9e4fdf4 | 2008-09-05 23:43:02 +0000 | [diff] [blame] | 242 | endrec[_ECD_SIGNATURE] = sig |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 243 | endrec[_ECD_DISK_NUMBER] = disk_num |
| 244 | endrec[_ECD_DISK_START] = disk_dir |
| 245 | endrec[_ECD_ENTRIES_THIS_DISK] = dircount |
| 246 | endrec[_ECD_ENTRIES_TOTAL] = dircount2 |
| 247 | endrec[_ECD_SIZE] = dirsize |
| 248 | endrec[_ECD_OFFSET] = diroffset |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 249 | return endrec |
| 250 | |
| 251 | |
Martin v. Löwis | 6f6873b | 2002-10-13 13:54:50 +0000 | [diff] [blame] | 252 | def _EndRecData(fpin): |
| 253 | """Return data from the "End of Central Directory" record, or None. |
| 254 | |
| 255 | The data is a list of the nine items in the ZIP "End of central dir" |
| 256 | record followed by a tenth item, the file seek offset of this record.""" |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 257 | |
| 258 | # Determine file size |
| 259 | fpin.seek(0, 2) |
| 260 | filesize = fpin.tell() |
| 261 | |
| 262 | # Check to see if this is ZIP file with no archive comment (the |
| 263 | # "end of central directory" structure should be the last item in the |
| 264 | # file if this is the case). |
Amaury Forgeot d'Arc | bc34780 | 2009-07-28 22:18:57 +0000 | [diff] [blame] | 265 | try: |
| 266 | fpin.seek(-sizeEndCentDir, 2) |
Andrew Svetlov | f7a17b4 | 2012-12-25 16:47:37 +0200 | [diff] [blame] | 267 | except OSError: |
Amaury Forgeot d'Arc | bc34780 | 2009-07-28 22:18:57 +0000 | [diff] [blame] | 268 | return None |
Martin v. Löwis | 6f6873b | 2002-10-13 13:54:50 +0000 | [diff] [blame] | 269 | data = fpin.read() |
Serhiy Storchaka | d2b1527 | 2013-01-31 15:27:07 +0200 | [diff] [blame] | 270 | if (len(data) == sizeEndCentDir and |
| 271 | data[0:4] == stringEndArchive and |
| 272 | data[-2:] == b"\000\000"): |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 273 | # the signature is correct and there's no comment, unpack structure |
Georg Brandl | 2ee470f | 2008-07-16 12:55:28 +0000 | [diff] [blame] | 274 | endrec = struct.unpack(structEndArchive, data) |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 275 | endrec=list(endrec) |
| 276 | |
| 277 | # Append a blank comment and record start offset |
| 278 | endrec.append(b"") |
| 279 | endrec.append(filesize - sizeEndCentDir) |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 280 | |
Amaury Forgeot d'Arc | d3fb4bb | 2009-01-18 00:29:02 +0000 | [diff] [blame] | 281 | # Try to read the "Zip64 end of central directory" structure |
| 282 | return _EndRecData64(fpin, -sizeEndCentDir, endrec) |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 283 | |
| 284 | # Either this is not a ZIP file, or it is a ZIP file with an archive |
| 285 | # comment. Search the end of the file for the "end of central directory" |
| 286 | # record signature. The comment is the last item in the ZIP file and may be |
| 287 | # up to 64K long. It is assumed that the "end of central directory" magic |
| 288 | # number does not appear in the comment. |
| 289 | maxCommentStart = max(filesize - (1 << 16) - sizeEndCentDir, 0) |
| 290 | fpin.seek(maxCommentStart, 0) |
Martin v. Löwis | 6f6873b | 2002-10-13 13:54:50 +0000 | [diff] [blame] | 291 | data = fpin.read() |
Georg Brandl | 2ee470f | 2008-07-16 12:55:28 +0000 | [diff] [blame] | 292 | start = data.rfind(stringEndArchive) |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 293 | if start >= 0: |
| 294 | # found the magic number; attempt to unpack and interpret |
| 295 | recData = data[start:start+sizeEndCentDir] |
Serhiy Storchaka | d2b1527 | 2013-01-31 15:27:07 +0200 | [diff] [blame] | 296 | if len(recData) != sizeEndCentDir: |
| 297 | # Zip file is corrupted. |
| 298 | return None |
Georg Brandl | 2ee470f | 2008-07-16 12:55:28 +0000 | [diff] [blame] | 299 | endrec = list(struct.unpack(structEndArchive, recData)) |
R David Murray | 4fbb9db | 2011-06-09 15:50:51 -0400 | [diff] [blame] | 300 | commentSize = endrec[_ECD_COMMENT_SIZE] #as claimed by the zip file |
| 301 | comment = data[start+sizeEndCentDir:start+sizeEndCentDir+commentSize] |
| 302 | endrec.append(comment) |
| 303 | endrec.append(maxCommentStart + start) |
Amaury Forgeot d'Arc | d3fb4bb | 2009-01-18 00:29:02 +0000 | [diff] [blame] | 304 | |
R David Murray | 4fbb9db | 2011-06-09 15:50:51 -0400 | [diff] [blame] | 305 | # Try to read the "Zip64 end of central directory" structure |
| 306 | return _EndRecData64(fpin, maxCommentStart + start - filesize, |
| 307 | endrec) |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 308 | |
| 309 | # Unable to find a valid end of central directory structure |
Serhiy Storchaka | d2b1527 | 2013-01-31 15:27:07 +0200 | [diff] [blame] | 310 | return None |
Martin v. Löwis | 6f6873b | 2002-10-13 13:54:50 +0000 | [diff] [blame] | 311 | |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 312 | |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 313 | class ZipInfo (object): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 314 | """Class with attributes describing each file in the ZIP archive.""" |
| 315 | |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 316 | __slots__ = ( |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 317 | 'orig_filename', |
| 318 | 'filename', |
| 319 | 'date_time', |
| 320 | 'compress_type', |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 321 | '_compresslevel', |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 322 | 'comment', |
| 323 | 'extra', |
| 324 | 'create_system', |
| 325 | 'create_version', |
| 326 | 'extract_version', |
| 327 | 'reserved', |
| 328 | 'flag_bits', |
| 329 | 'volume', |
| 330 | 'internal_attr', |
| 331 | 'external_attr', |
| 332 | 'header_offset', |
| 333 | 'CRC', |
| 334 | 'compress_size', |
| 335 | 'file_size', |
| 336 | '_raw_time', |
| 337 | ) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 338 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 339 | 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] | 340 | self.orig_filename = filename # Original file name in archive |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 341 | |
| 342 | # Terminate the file name at the first null byte. Null bytes in file |
| 343 | # names are used as tricks by viruses in archives. |
Greg Ward | 8e36d28 | 2003-06-18 00:53:06 +0000 | [diff] [blame] | 344 | null_byte = filename.find(chr(0)) |
| 345 | if null_byte >= 0: |
| 346 | filename = filename[0:null_byte] |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 347 | # This is used to ensure paths in generated ZIP files always use |
| 348 | # forward slashes as the directory separator, as required by the |
| 349 | # ZIP format specification. |
| 350 | if os.sep != "/" and os.sep in filename: |
Greg Ward | 8e36d28 | 2003-06-18 00:53:06 +0000 | [diff] [blame] | 351 | filename = filename.replace(os.sep, "/") |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 352 | |
Greg Ward | 8e36d28 | 2003-06-18 00:53:06 +0000 | [diff] [blame] | 353 | self.filename = filename # Normalized file name |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 354 | self.date_time = date_time # year, month, day, hour, min, sec |
Senthil Kumaran | 29fa9d4 | 2011-10-20 01:46:00 +0800 | [diff] [blame] | 355 | |
| 356 | if date_time[0] < 1980: |
| 357 | raise ValueError('ZIP does not support timestamps before 1980') |
| 358 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 359 | # Standard values: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 360 | self.compress_type = ZIP_STORED # Type of compression for the file |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 361 | self._compresslevel = None # Level for the compressor |
Guido van Rossum | d6ca546 | 2007-05-22 01:29:33 +0000 | [diff] [blame] | 362 | self.comment = b"" # Comment for each file |
| 363 | self.extra = b"" # ZIP extra data |
Martin v. Löwis | 0075690 | 2006-02-05 17:09:41 +0000 | [diff] [blame] | 364 | if sys.platform == 'win32': |
| 365 | self.create_system = 0 # System which created ZIP archive |
| 366 | else: |
| 367 | # Assume everything else is unix-y |
| 368 | self.create_system = 3 # System which created ZIP archive |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 369 | self.create_version = DEFAULT_VERSION # Version which created ZIP archive |
| 370 | self.extract_version = DEFAULT_VERSION # Version needed to extract archive |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 371 | self.reserved = 0 # Must be zero |
| 372 | self.flag_bits = 0 # ZIP flag bits |
| 373 | self.volume = 0 # Volume number of file header |
| 374 | self.internal_attr = 0 # Internal attributes |
| 375 | self.external_attr = 0 # External file attributes |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 376 | # Other attributes are set by class ZipFile: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 377 | # header_offset Byte offset to the file header |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 378 | # CRC CRC-32 of the uncompressed file |
| 379 | # compress_size Size of the compressed file |
| 380 | # file_size Size of the uncompressed file |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 381 | |
Serhiy Storchaka | 51a4370 | 2014-10-29 22:42:06 +0200 | [diff] [blame] | 382 | def __repr__(self): |
| 383 | result = ['<%s filename=%r' % (self.__class__.__name__, self.filename)] |
| 384 | if self.compress_type != ZIP_STORED: |
| 385 | result.append(' compress_type=%s' % |
| 386 | compressor_names.get(self.compress_type, |
| 387 | self.compress_type)) |
| 388 | hi = self.external_attr >> 16 |
| 389 | lo = self.external_attr & 0xFFFF |
| 390 | if hi: |
| 391 | result.append(' filemode=%r' % stat.filemode(hi)) |
| 392 | if lo: |
| 393 | result.append(' external_attr=%#x' % lo) |
Serhiy Storchaka | 503f908 | 2016-02-08 00:02:25 +0200 | [diff] [blame] | 394 | isdir = self.is_dir() |
Serhiy Storchaka | 51a4370 | 2014-10-29 22:42:06 +0200 | [diff] [blame] | 395 | if not isdir or self.file_size: |
| 396 | result.append(' file_size=%r' % self.file_size) |
| 397 | if ((not isdir or self.compress_size) and |
| 398 | (self.compress_type != ZIP_STORED or |
| 399 | self.file_size != self.compress_size)): |
| 400 | result.append(' compress_size=%r' % self.compress_size) |
| 401 | result.append('>') |
| 402 | return ''.join(result) |
| 403 | |
Serhiy Storchaka | 182d7cd | 2013-01-15 00:31:39 +0200 | [diff] [blame] | 404 | def FileHeader(self, zip64=None): |
Serhiy Storchaka | 4bb186d | 2018-11-25 09:51:14 +0200 | [diff] [blame] | 405 | """Return the per-file header as a bytes object.""" |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 406 | dt = self.date_time |
| 407 | dosdate = (dt[0] - 1980) << 9 | dt[1] << 5 | dt[2] |
Tim Peters | 3caca23 | 2001-12-06 06:23:26 +0000 | [diff] [blame] | 408 | dostime = dt[3] << 11 | dt[4] << 5 | (dt[5] // 2) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 409 | if self.flag_bits & 0x08: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 410 | # Set these to zero because we write them after the file data |
| 411 | CRC = compress_size = file_size = 0 |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 412 | else: |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 413 | CRC = self.CRC |
| 414 | compress_size = self.compress_size |
| 415 | file_size = self.file_size |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 416 | |
| 417 | extra = self.extra |
| 418 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 419 | min_version = 0 |
Serhiy Storchaka | 182d7cd | 2013-01-15 00:31:39 +0200 | [diff] [blame] | 420 | if zip64 is None: |
| 421 | zip64 = file_size > ZIP64_LIMIT or compress_size > ZIP64_LIMIT |
| 422 | if zip64: |
Christian Heimes | d5e2b6f | 2008-03-19 21:50:51 +0000 | [diff] [blame] | 423 | fmt = '<HHQQ' |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 424 | extra = extra + struct.pack(fmt, |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 425 | 1, struct.calcsize(fmt)-4, file_size, compress_size) |
Serhiy Storchaka | 182d7cd | 2013-01-15 00:31:39 +0200 | [diff] [blame] | 426 | if file_size > ZIP64_LIMIT or compress_size > ZIP64_LIMIT: |
| 427 | if not zip64: |
| 428 | raise LargeZipFile("Filesize would require ZIP64 extensions") |
| 429 | # File is larger than what fits into a 4 byte integer, |
| 430 | # fall back to the ZIP64 extension |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 431 | file_size = 0xffffffff |
| 432 | compress_size = 0xffffffff |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 433 | min_version = ZIP64_VERSION |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 434 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 435 | if self.compress_type == ZIP_BZIP2: |
| 436 | min_version = max(BZIP2_VERSION, min_version) |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 437 | elif self.compress_type == ZIP_LZMA: |
| 438 | min_version = max(LZMA_VERSION, min_version) |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 439 | |
| 440 | self.extract_version = max(min_version, self.extract_version) |
| 441 | self.create_version = max(min_version, self.create_version) |
Martin v. Löwis | 8570f6a | 2008-05-05 17:44:38 +0000 | [diff] [blame] | 442 | filename, flag_bits = self._encodeFilenameFlags() |
Georg Brandl | 2ee470f | 2008-07-16 12:55:28 +0000 | [diff] [blame] | 443 | header = struct.pack(structFileHeader, stringFileHeader, |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 444 | self.extract_version, self.reserved, flag_bits, |
| 445 | self.compress_type, dostime, dosdate, CRC, |
| 446 | compress_size, file_size, |
| 447 | len(filename), len(extra)) |
Martin v. Löwis | 8570f6a | 2008-05-05 17:44:38 +0000 | [diff] [blame] | 448 | return header + filename + extra |
| 449 | |
| 450 | def _encodeFilenameFlags(self): |
| 451 | try: |
| 452 | return self.filename.encode('ascii'), self.flag_bits |
| 453 | except UnicodeEncodeError: |
| 454 | return self.filename.encode('utf-8'), self.flag_bits | 0x800 |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 455 | |
| 456 | def _decodeExtra(self): |
| 457 | # Try to decode the extra field. |
| 458 | extra = self.extra |
| 459 | unpack = struct.unpack |
Gregory P. Smith | 0af8a86 | 2014-05-29 23:42:14 -0700 | [diff] [blame] | 460 | while len(extra) >= 4: |
Christian Heimes | d5e2b6f | 2008-03-19 21:50:51 +0000 | [diff] [blame] | 461 | tp, ln = unpack('<HH', extra[:4]) |
Serhiy Storchaka | feccdb2 | 2017-03-09 18:34:03 +0200 | [diff] [blame] | 462 | if ln+4 > len(extra): |
| 463 | raise BadZipFile("Corrupt extra field %04x (size=%d)" % (tp, ln)) |
| 464 | if tp == 0x0001: |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 465 | if ln >= 24: |
Christian Heimes | d5e2b6f | 2008-03-19 21:50:51 +0000 | [diff] [blame] | 466 | counts = unpack('<QQQ', extra[4:28]) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 467 | elif ln == 16: |
Christian Heimes | d5e2b6f | 2008-03-19 21:50:51 +0000 | [diff] [blame] | 468 | counts = unpack('<QQ', extra[4:20]) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 469 | elif ln == 8: |
Christian Heimes | d5e2b6f | 2008-03-19 21:50:51 +0000 | [diff] [blame] | 470 | counts = unpack('<Q', extra[4:12]) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 471 | elif ln == 0: |
| 472 | counts = () |
| 473 | else: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 474 | raise BadZipFile("Corrupt extra field %04x (size=%d)" % (tp, ln)) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 475 | |
| 476 | idx = 0 |
| 477 | |
| 478 | # ZIP64 extension (large files and/or large archives) |
Christian Heimes | d5e2b6f | 2008-03-19 21:50:51 +0000 | [diff] [blame] | 479 | if self.file_size in (0xffffffffffffffff, 0xffffffff): |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 480 | self.file_size = counts[idx] |
| 481 | idx += 1 |
| 482 | |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 483 | if self.compress_size == 0xFFFFFFFF: |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 484 | self.compress_size = counts[idx] |
| 485 | idx += 1 |
| 486 | |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 487 | if self.header_offset == 0xffffffff: |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 488 | old = self.header_offset |
| 489 | self.header_offset = counts[idx] |
| 490 | idx+=1 |
| 491 | |
| 492 | extra = extra[ln+4:] |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 493 | |
Serhiy Storchaka | 503f908 | 2016-02-08 00:02:25 +0200 | [diff] [blame] | 494 | @classmethod |
Marcel Plch | a2fe1e5 | 2018-08-02 15:04:52 +0200 | [diff] [blame] | 495 | def from_file(cls, filename, arcname=None, *, strict_timestamps=True): |
Serhiy Storchaka | 503f908 | 2016-02-08 00:02:25 +0200 | [diff] [blame] | 496 | """Construct an appropriate ZipInfo for a file on the filesystem. |
| 497 | |
| 498 | filename should be the path to a file or directory on the filesystem. |
| 499 | |
| 500 | arcname is the name which it will have within the archive (by default, |
| 501 | this will be the same as filename, but without a drive letter and with |
| 502 | leading path separators removed). |
| 503 | """ |
Serhiy Storchaka | 8606e95 | 2017-03-08 14:37:51 +0200 | [diff] [blame] | 504 | if isinstance(filename, os.PathLike): |
| 505 | filename = os.fspath(filename) |
Serhiy Storchaka | 503f908 | 2016-02-08 00:02:25 +0200 | [diff] [blame] | 506 | st = os.stat(filename) |
| 507 | isdir = stat.S_ISDIR(st.st_mode) |
| 508 | mtime = time.localtime(st.st_mtime) |
| 509 | date_time = mtime[0:6] |
Marcel Plch | a2fe1e5 | 2018-08-02 15:04:52 +0200 | [diff] [blame] | 510 | if not strict_timestamps and date_time[0] < 1980: |
| 511 | date_time = (1980, 1, 1, 0, 0, 0) |
| 512 | elif not strict_timestamps and date_time[0] > 2107: |
| 513 | date_time = (2107, 12, 31, 23, 59, 59) |
Serhiy Storchaka | 503f908 | 2016-02-08 00:02:25 +0200 | [diff] [blame] | 514 | # Create ZipInfo instance to store file information |
| 515 | if arcname is None: |
| 516 | arcname = filename |
| 517 | arcname = os.path.normpath(os.path.splitdrive(arcname)[1]) |
| 518 | while arcname[0] in (os.sep, os.altsep): |
| 519 | arcname = arcname[1:] |
| 520 | if isdir: |
| 521 | arcname += '/' |
| 522 | zinfo = cls(arcname, date_time) |
| 523 | zinfo.external_attr = (st.st_mode & 0xFFFF) << 16 # Unix attributes |
| 524 | if isdir: |
| 525 | zinfo.file_size = 0 |
| 526 | zinfo.external_attr |= 0x10 # MS-DOS directory flag |
| 527 | else: |
| 528 | zinfo.file_size = st.st_size |
| 529 | |
| 530 | return zinfo |
| 531 | |
| 532 | def is_dir(self): |
Serhiy Storchaka | f47fc55 | 2016-05-15 12:27:16 +0300 | [diff] [blame] | 533 | """Return True if this archive member is a directory.""" |
Serhiy Storchaka | 503f908 | 2016-02-08 00:02:25 +0200 | [diff] [blame] | 534 | return self.filename[-1] == '/' |
| 535 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 536 | |
Serhiy Storchaka | 06e5225 | 2017-03-30 19:09:08 +0300 | [diff] [blame] | 537 | # ZIP encryption uses the CRC32 one-byte primitive for scrambling some |
| 538 | # internal keys. We noticed that a direct implementation is faster than |
| 539 | # relying on binascii.crc32(). |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 540 | |
Serhiy Storchaka | 06e5225 | 2017-03-30 19:09:08 +0300 | [diff] [blame] | 541 | _crctable = None |
| 542 | def _gen_crc(crc): |
| 543 | for j in range(8): |
| 544 | if crc & 1: |
| 545 | crc = (crc >> 1) ^ 0xEDB88320 |
| 546 | else: |
| 547 | crc >>= 1 |
| 548 | return crc |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 549 | |
Serhiy Storchaka | 06e5225 | 2017-03-30 19:09:08 +0300 | [diff] [blame] | 550 | # ZIP supports a password-based form of encryption. Even though known |
| 551 | # plaintext attacks have been found against it, it is still useful |
| 552 | # to be able to get data out of such a file. |
| 553 | # |
| 554 | # Usage: |
| 555 | # zd = _ZipDecrypter(mypwd) |
| 556 | # plain_bytes = zd(cypher_bytes) |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 557 | |
Serhiy Storchaka | 06e5225 | 2017-03-30 19:09:08 +0300 | [diff] [blame] | 558 | def _ZipDecrypter(pwd): |
| 559 | key0 = 305419896 |
| 560 | key1 = 591751049 |
| 561 | key2 = 878082192 |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 562 | |
Serhiy Storchaka | 06e5225 | 2017-03-30 19:09:08 +0300 | [diff] [blame] | 563 | global _crctable |
| 564 | if _crctable is None: |
| 565 | _crctable = list(map(_gen_crc, range(256))) |
| 566 | crctable = _crctable |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 567 | |
Serhiy Storchaka | 06e5225 | 2017-03-30 19:09:08 +0300 | [diff] [blame] | 568 | def crc32(ch, crc): |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 569 | """Compute the CRC32 primitive on one byte.""" |
Serhiy Storchaka | 06e5225 | 2017-03-30 19:09:08 +0300 | [diff] [blame] | 570 | return (crc >> 8) ^ crctable[(crc ^ ch) & 0xFF] |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 571 | |
Serhiy Storchaka | 06e5225 | 2017-03-30 19:09:08 +0300 | [diff] [blame] | 572 | def update_keys(c): |
| 573 | nonlocal key0, key1, key2 |
| 574 | key0 = crc32(c, key0) |
| 575 | key1 = (key1 + (key0 & 0xFF)) & 0xFFFFFFFF |
| 576 | key1 = (key1 * 134775813 + 1) & 0xFFFFFFFF |
| 577 | key2 = crc32(key1 >> 24, key2) |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 578 | |
Serhiy Storchaka | 06e5225 | 2017-03-30 19:09:08 +0300 | [diff] [blame] | 579 | for p in pwd: |
| 580 | update_keys(p) |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 581 | |
Serhiy Storchaka | 06e5225 | 2017-03-30 19:09:08 +0300 | [diff] [blame] | 582 | def decrypter(data): |
| 583 | """Decrypt a bytes object.""" |
| 584 | result = bytearray() |
| 585 | append = result.append |
| 586 | for c in data: |
| 587 | k = key2 | 2 |
| 588 | c ^= ((k * (k^1)) >> 8) & 0xFF |
| 589 | update_keys(c) |
| 590 | append(c) |
| 591 | return bytes(result) |
| 592 | |
| 593 | return decrypter |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 594 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 595 | |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 596 | class LZMACompressor: |
| 597 | |
| 598 | def __init__(self): |
| 599 | self._comp = None |
| 600 | |
| 601 | def _init(self): |
Nadeem Vawda | a425c3d | 2012-06-21 23:36:48 +0200 | [diff] [blame] | 602 | props = lzma._encode_filter_properties({'id': lzma.FILTER_LZMA1}) |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 603 | self._comp = lzma.LZMACompressor(lzma.FORMAT_RAW, filters=[ |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 604 | lzma._decode_filter_properties(lzma.FILTER_LZMA1, props) |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 605 | ]) |
| 606 | return struct.pack('<BBH', 9, 4, len(props)) + props |
| 607 | |
| 608 | def compress(self, data): |
| 609 | if self._comp is None: |
| 610 | return self._init() + self._comp.compress(data) |
| 611 | return self._comp.compress(data) |
| 612 | |
| 613 | def flush(self): |
| 614 | if self._comp is None: |
| 615 | return self._init() + self._comp.flush() |
| 616 | return self._comp.flush() |
| 617 | |
| 618 | |
| 619 | class LZMADecompressor: |
| 620 | |
| 621 | def __init__(self): |
| 622 | self._decomp = None |
| 623 | self._unconsumed = b'' |
| 624 | self.eof = False |
| 625 | |
| 626 | def decompress(self, data): |
| 627 | if self._decomp is None: |
| 628 | self._unconsumed += data |
| 629 | if len(self._unconsumed) <= 4: |
| 630 | return b'' |
| 631 | psize, = struct.unpack('<H', self._unconsumed[2:4]) |
| 632 | if len(self._unconsumed) <= 4 + psize: |
| 633 | return b'' |
| 634 | |
| 635 | self._decomp = lzma.LZMADecompressor(lzma.FORMAT_RAW, filters=[ |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 636 | lzma._decode_filter_properties(lzma.FILTER_LZMA1, |
| 637 | self._unconsumed[4:4 + psize]) |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 638 | ]) |
| 639 | data = self._unconsumed[4 + psize:] |
| 640 | del self._unconsumed |
| 641 | |
| 642 | result = self._decomp.decompress(data) |
| 643 | self.eof = self._decomp.eof |
| 644 | return result |
| 645 | |
| 646 | |
| 647 | compressor_names = { |
| 648 | 0: 'store', |
| 649 | 1: 'shrink', |
| 650 | 2: 'reduce', |
| 651 | 3: 'reduce', |
| 652 | 4: 'reduce', |
| 653 | 5: 'reduce', |
| 654 | 6: 'implode', |
| 655 | 7: 'tokenize', |
| 656 | 8: 'deflate', |
| 657 | 9: 'deflate64', |
| 658 | 10: 'implode', |
| 659 | 12: 'bzip2', |
| 660 | 14: 'lzma', |
| 661 | 18: 'terse', |
| 662 | 19: 'lz77', |
| 663 | 97: 'wavpack', |
| 664 | 98: 'ppmd', |
| 665 | } |
| 666 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 667 | def _check_compression(compression): |
| 668 | if compression == ZIP_STORED: |
| 669 | pass |
| 670 | elif compression == ZIP_DEFLATED: |
| 671 | if not zlib: |
| 672 | raise RuntimeError( |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 673 | "Compression requires the (missing) zlib module") |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 674 | elif compression == ZIP_BZIP2: |
| 675 | if not bz2: |
| 676 | raise RuntimeError( |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 677 | "Compression requires the (missing) bz2 module") |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 678 | elif compression == ZIP_LZMA: |
| 679 | if not lzma: |
| 680 | raise RuntimeError( |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 681 | "Compression requires the (missing) lzma module") |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 682 | else: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 683 | raise NotImplementedError("That compression method is not supported") |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 684 | |
| 685 | |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 686 | def _get_compressor(compress_type, compresslevel=None): |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 687 | if compress_type == ZIP_DEFLATED: |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 688 | if compresslevel is not None: |
| 689 | return zlib.compressobj(compresslevel, zlib.DEFLATED, -15) |
| 690 | return zlib.compressobj(zlib.Z_DEFAULT_COMPRESSION, zlib.DEFLATED, -15) |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 691 | elif compress_type == ZIP_BZIP2: |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 692 | if compresslevel is not None: |
| 693 | return bz2.BZ2Compressor(compresslevel) |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 694 | return bz2.BZ2Compressor() |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 695 | # compresslevel is ignored for ZIP_LZMA |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 696 | elif compress_type == ZIP_LZMA: |
| 697 | return LZMACompressor() |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 698 | else: |
| 699 | return None |
| 700 | |
| 701 | |
| 702 | def _get_decompressor(compress_type): |
Martin v. Löwis | b3260f0 | 2012-05-01 08:38:01 +0200 | [diff] [blame] | 703 | if compress_type == ZIP_STORED: |
| 704 | return None |
| 705 | elif compress_type == ZIP_DEFLATED: |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 706 | return zlib.decompressobj(-15) |
| 707 | elif compress_type == ZIP_BZIP2: |
| 708 | return bz2.BZ2Decompressor() |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 709 | elif compress_type == ZIP_LZMA: |
| 710 | return LZMADecompressor() |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 711 | else: |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 712 | descr = compressor_names.get(compress_type) |
Martin v. Löwis | b3260f0 | 2012-05-01 08:38:01 +0200 | [diff] [blame] | 713 | if descr: |
| 714 | raise NotImplementedError("compression type %d (%s)" % (compress_type, descr)) |
| 715 | else: |
| 716 | raise NotImplementedError("compression type %d" % (compress_type,)) |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 717 | |
| 718 | |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 719 | class _SharedFile: |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 720 | def __init__(self, file, pos, close, lock, writing): |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 721 | self._file = file |
| 722 | self._pos = pos |
| 723 | self._close = close |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 724 | self._lock = lock |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 725 | self._writing = writing |
John Jolly | 066df4f | 2018-01-30 01:51:35 -0700 | [diff] [blame] | 726 | self.seekable = file.seekable |
| 727 | self.tell = file.tell |
| 728 | |
| 729 | def seek(self, offset, whence=0): |
| 730 | with self._lock: |
Mickaël Schoentgen | 3f8c691 | 2018-07-29 20:26:52 +0200 | [diff] [blame] | 731 | if self._writing(): |
John Jolly | 066df4f | 2018-01-30 01:51:35 -0700 | [diff] [blame] | 732 | raise ValueError("Can't reposition in the ZIP file while " |
| 733 | "there is an open writing handle on it. " |
| 734 | "Close the writing handle before trying to read.") |
Mickaël Schoentgen | 3f8c691 | 2018-07-29 20:26:52 +0200 | [diff] [blame] | 735 | self._file.seek(offset, whence) |
John Jolly | 066df4f | 2018-01-30 01:51:35 -0700 | [diff] [blame] | 736 | self._pos = self._file.tell() |
| 737 | return self._pos |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 738 | |
| 739 | def read(self, n=-1): |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 740 | with self._lock: |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 741 | if self._writing(): |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 742 | raise ValueError("Can't read from the ZIP file while there " |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 743 | "is an open writing handle on it. " |
| 744 | "Close the writing handle before trying to read.") |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 745 | self._file.seek(self._pos) |
| 746 | data = self._file.read(n) |
| 747 | self._pos = self._file.tell() |
| 748 | return data |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 749 | |
| 750 | def close(self): |
| 751 | if self._file is not None: |
| 752 | fileobj = self._file |
| 753 | self._file = None |
| 754 | self._close(fileobj) |
| 755 | |
Serhiy Storchaka | 77d8997 | 2015-03-23 01:09:35 +0200 | [diff] [blame] | 756 | # Provide the tell method for unseekable stream |
| 757 | class _Tellable: |
| 758 | def __init__(self, fp): |
| 759 | self.fp = fp |
| 760 | self.offset = 0 |
| 761 | |
| 762 | def write(self, data): |
| 763 | n = self.fp.write(data) |
| 764 | self.offset += n |
| 765 | return n |
| 766 | |
| 767 | def tell(self): |
| 768 | return self.offset |
| 769 | |
| 770 | def flush(self): |
| 771 | self.fp.flush() |
| 772 | |
| 773 | def close(self): |
| 774 | self.fp.close() |
| 775 | |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 776 | |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 777 | class ZipExtFile(io.BufferedIOBase): |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 778 | """File-like object for reading an archive member. |
| 779 | Is returned by ZipFile.open(). |
| 780 | """ |
| 781 | |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 782 | # Max size supported by decompressor. |
| 783 | MAX_N = 1 << 31 - 1 |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 784 | |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 785 | # Read from compressed files in 4k blocks. |
| 786 | MIN_READ_SIZE = 4096 |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 787 | |
John Jolly | 066df4f | 2018-01-30 01:51:35 -0700 | [diff] [blame] | 788 | # Chunk size to read during seek |
| 789 | MAX_SEEK_READ = 1 << 24 |
| 790 | |
Łukasz Langa | e94980a | 2010-11-22 23:31:26 +0000 | [diff] [blame] | 791 | def __init__(self, fileobj, mode, zipinfo, decrypter=None, |
| 792 | close_fileobj=False): |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 793 | self._fileobj = fileobj |
| 794 | self._decrypter = decrypter |
Łukasz Langa | e94980a | 2010-11-22 23:31:26 +0000 | [diff] [blame] | 795 | self._close_fileobj = close_fileobj |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 796 | |
Ezio Melotti | 92b4743 | 2010-01-28 01:44:41 +0000 | [diff] [blame] | 797 | self._compress_type = zipinfo.compress_type |
Ezio Melotti | 92b4743 | 2010-01-28 01:44:41 +0000 | [diff] [blame] | 798 | self._compress_left = zipinfo.compress_size |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 799 | self._left = zipinfo.file_size |
Ezio Melotti | 92b4743 | 2010-01-28 01:44:41 +0000 | [diff] [blame] | 800 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 801 | self._decompressor = _get_decompressor(self._compress_type) |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 802 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 803 | self._eof = False |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 804 | self._readbuffer = b'' |
| 805 | self._offset = 0 |
| 806 | |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 807 | self.newlines = None |
| 808 | |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 809 | # Adjust read size for encrypted files since the first 12 bytes |
| 810 | # are for the encryption/password information. |
| 811 | if self._decrypter is not None: |
| 812 | self._compress_left -= 12 |
| 813 | |
| 814 | self.mode = mode |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 815 | self.name = zipinfo.filename |
| 816 | |
Antoine Pitrou | 7c8bcb6 | 2010-08-12 15:11:50 +0000 | [diff] [blame] | 817 | if hasattr(zipinfo, 'CRC'): |
| 818 | self._expected_crc = zipinfo.CRC |
Martin Panter | b82032f | 2015-12-11 05:19:29 +0000 | [diff] [blame] | 819 | self._running_crc = crc32(b'') |
Antoine Pitrou | 7c8bcb6 | 2010-08-12 15:11:50 +0000 | [diff] [blame] | 820 | else: |
| 821 | self._expected_crc = None |
| 822 | |
John Jolly | 066df4f | 2018-01-30 01:51:35 -0700 | [diff] [blame] | 823 | self._seekable = False |
| 824 | try: |
| 825 | if fileobj.seekable(): |
| 826 | self._orig_compress_start = fileobj.tell() |
| 827 | self._orig_compress_size = zipinfo.compress_size |
| 828 | self._orig_file_size = zipinfo.file_size |
| 829 | self._orig_start_crc = self._running_crc |
| 830 | self._seekable = True |
| 831 | except AttributeError: |
| 832 | pass |
| 833 | |
Serhiy Storchaka | 51a4370 | 2014-10-29 22:42:06 +0200 | [diff] [blame] | 834 | def __repr__(self): |
| 835 | result = ['<%s.%s' % (self.__class__.__module__, |
| 836 | self.__class__.__qualname__)] |
| 837 | if not self.closed: |
| 838 | result.append(' name=%r mode=%r' % (self.name, self.mode)) |
| 839 | if self._compress_type != ZIP_STORED: |
| 840 | result.append(' compress_type=%s' % |
| 841 | compressor_names.get(self._compress_type, |
| 842 | self._compress_type)) |
| 843 | else: |
| 844 | result.append(' [closed]') |
| 845 | result.append('>') |
| 846 | return ''.join(result) |
| 847 | |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 848 | def readline(self, limit=-1): |
| 849 | """Read and return a line from the stream. |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 850 | |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 851 | If limit is specified, at most limit bytes will be read. |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 852 | """ |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 853 | |
Serhiy Storchaka | e670be2 | 2016-06-11 19:32:44 +0300 | [diff] [blame] | 854 | if limit < 0: |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 855 | # Shortcut common case - newline found in buffer. |
| 856 | i = self._readbuffer.find(b'\n', self._offset) + 1 |
| 857 | if i > 0: |
| 858 | line = self._readbuffer[self._offset: i] |
| 859 | self._offset = i |
| 860 | return line |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 861 | |
Serhiy Storchaka | e670be2 | 2016-06-11 19:32:44 +0300 | [diff] [blame] | 862 | return io.BufferedIOBase.readline(self, limit) |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 863 | |
| 864 | def peek(self, n=1): |
| 865 | """Returns buffered bytes without advancing the position.""" |
| 866 | if n > len(self._readbuffer) - self._offset: |
| 867 | chunk = self.read(n) |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 868 | if len(chunk) > self._offset: |
| 869 | self._readbuffer = chunk + self._readbuffer[self._offset:] |
| 870 | self._offset = 0 |
| 871 | else: |
| 872 | self._offset -= len(chunk) |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 873 | |
| 874 | # Return up to 512 bytes to reduce allocation overhead for tight loops. |
| 875 | return self._readbuffer[self._offset: self._offset + 512] |
| 876 | |
| 877 | def readable(self): |
| 878 | return True |
| 879 | |
| 880 | def read(self, n=-1): |
| 881 | """Read and return up to n bytes. |
nick sung | 53c2935 | 2019-03-15 03:26:25 +0800 | [diff] [blame^] | 882 | If the argument is omitted, None, or negative, data is read and returned until EOF is reached. |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 883 | """ |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 884 | if n is None or n < 0: |
| 885 | buf = self._readbuffer[self._offset:] |
| 886 | self._readbuffer = b'' |
| 887 | self._offset = 0 |
| 888 | while not self._eof: |
| 889 | buf += self._read1(self.MAX_N) |
| 890 | return buf |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 891 | |
Antoine Pitrou | 78157b3 | 2012-06-23 16:44:48 +0200 | [diff] [blame] | 892 | end = n + self._offset |
| 893 | if end < len(self._readbuffer): |
| 894 | buf = self._readbuffer[self._offset:end] |
| 895 | self._offset = end |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 896 | return buf |
| 897 | |
Antoine Pitrou | 78157b3 | 2012-06-23 16:44:48 +0200 | [diff] [blame] | 898 | n = end - len(self._readbuffer) |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 899 | buf = self._readbuffer[self._offset:] |
| 900 | self._readbuffer = b'' |
| 901 | self._offset = 0 |
| 902 | while n > 0 and not self._eof: |
| 903 | data = self._read1(n) |
| 904 | if n < len(data): |
| 905 | self._readbuffer = data |
| 906 | self._offset = n |
| 907 | buf += data[:n] |
| 908 | break |
| 909 | buf += data |
| 910 | n -= len(data) |
| 911 | return buf |
| 912 | |
| 913 | def _update_crc(self, newdata): |
Antoine Pitrou | 7c8bcb6 | 2010-08-12 15:11:50 +0000 | [diff] [blame] | 914 | # Update the CRC using the given data. |
| 915 | if self._expected_crc is None: |
| 916 | # No need to compute the CRC if we don't have a reference value |
| 917 | return |
Martin Panter | b82032f | 2015-12-11 05:19:29 +0000 | [diff] [blame] | 918 | self._running_crc = crc32(newdata, self._running_crc) |
Antoine Pitrou | 7c8bcb6 | 2010-08-12 15:11:50 +0000 | [diff] [blame] | 919 | # Check the CRC if we're at the end of the file |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 920 | if self._eof and self._running_crc != self._expected_crc: |
Georg Brandl | 4d54088 | 2010-10-28 06:42:33 +0000 | [diff] [blame] | 921 | raise BadZipFile("Bad CRC-32 for file %r" % self.name) |
Antoine Pitrou | 7c8bcb6 | 2010-08-12 15:11:50 +0000 | [diff] [blame] | 922 | |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 923 | def read1(self, n): |
| 924 | """Read up to n bytes with at most one read() system call.""" |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 925 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 926 | if n is None or n < 0: |
| 927 | buf = self._readbuffer[self._offset:] |
| 928 | self._readbuffer = b'' |
| 929 | self._offset = 0 |
Serhiy Storchaka | d2c07a5 | 2013-09-27 22:11:57 +0300 | [diff] [blame] | 930 | while not self._eof: |
| 931 | data = self._read1(self.MAX_N) |
| 932 | if data: |
| 933 | buf += data |
| 934 | break |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 935 | return buf |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 936 | |
Antoine Pitrou | 78157b3 | 2012-06-23 16:44:48 +0200 | [diff] [blame] | 937 | end = n + self._offset |
| 938 | if end < len(self._readbuffer): |
| 939 | buf = self._readbuffer[self._offset:end] |
| 940 | self._offset = end |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 941 | return buf |
| 942 | |
Antoine Pitrou | 78157b3 | 2012-06-23 16:44:48 +0200 | [diff] [blame] | 943 | n = end - len(self._readbuffer) |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 944 | buf = self._readbuffer[self._offset:] |
| 945 | self._readbuffer = b'' |
| 946 | self._offset = 0 |
| 947 | if n > 0: |
Serhiy Storchaka | d2c07a5 | 2013-09-27 22:11:57 +0300 | [diff] [blame] | 948 | while not self._eof: |
| 949 | data = self._read1(n) |
| 950 | if n < len(data): |
| 951 | self._readbuffer = data |
| 952 | self._offset = n |
| 953 | buf += data[:n] |
| 954 | break |
| 955 | if data: |
| 956 | buf += data |
| 957 | break |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 958 | return buf |
| 959 | |
| 960 | def _read1(self, n): |
| 961 | # Read up to n compressed bytes with at most one read() system call, |
| 962 | # decrypt and decompress them. |
| 963 | if self._eof or n <= 0: |
| 964 | return b'' |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 965 | |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 966 | # Read from file. |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 967 | if self._compress_type == ZIP_DEFLATED: |
| 968 | ## Handle unconsumed data. |
| 969 | data = self._decompressor.unconsumed_tail |
| 970 | if n > len(data): |
| 971 | data += self._read2(n - len(data)) |
| 972 | else: |
| 973 | data = self._read2(n) |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 974 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 975 | if self._compress_type == ZIP_STORED: |
| 976 | self._eof = self._compress_left <= 0 |
| 977 | elif self._compress_type == ZIP_DEFLATED: |
| 978 | n = max(n, self.MIN_READ_SIZE) |
| 979 | data = self._decompressor.decompress(data, n) |
| 980 | self._eof = (self._decompressor.eof or |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 981 | self._compress_left <= 0 and |
| 982 | not self._decompressor.unconsumed_tail) |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 983 | if self._eof: |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 984 | data += self._decompressor.flush() |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 985 | else: |
| 986 | data = self._decompressor.decompress(data) |
| 987 | self._eof = self._decompressor.eof or self._compress_left <= 0 |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 988 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 989 | data = data[:self._left] |
| 990 | self._left -= len(data) |
| 991 | if self._left <= 0: |
| 992 | self._eof = True |
| 993 | self._update_crc(data) |
| 994 | return data |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 995 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 996 | def _read2(self, n): |
| 997 | if self._compress_left <= 0: |
| 998 | return b'' |
| 999 | |
| 1000 | n = max(n, self.MIN_READ_SIZE) |
| 1001 | n = min(n, self._compress_left) |
| 1002 | |
| 1003 | data = self._fileobj.read(n) |
| 1004 | self._compress_left -= len(data) |
Serhiy Storchaka | 5ce3f10 | 2014-01-09 14:50:20 +0200 | [diff] [blame] | 1005 | if not data: |
| 1006 | raise EOFError |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 1007 | |
| 1008 | if self._decrypter is not None: |
Serhiy Storchaka | 06e5225 | 2017-03-30 19:09:08 +0300 | [diff] [blame] | 1009 | data = self._decrypter(data) |
Guido van Rossum | d6ca546 | 2007-05-22 01:29:33 +0000 | [diff] [blame] | 1010 | return data |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 1011 | |
Łukasz Langa | e94980a | 2010-11-22 23:31:26 +0000 | [diff] [blame] | 1012 | def close(self): |
| 1013 | try: |
| 1014 | if self._close_fileobj: |
| 1015 | self._fileobj.close() |
| 1016 | finally: |
| 1017 | super().close() |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 1018 | |
John Jolly | 066df4f | 2018-01-30 01:51:35 -0700 | [diff] [blame] | 1019 | def seekable(self): |
| 1020 | return self._seekable |
| 1021 | |
| 1022 | def seek(self, offset, whence=0): |
| 1023 | if not self._seekable: |
| 1024 | raise io.UnsupportedOperation("underlying stream is not seekable") |
| 1025 | curr_pos = self.tell() |
| 1026 | if whence == 0: # Seek from start of file |
| 1027 | new_pos = offset |
| 1028 | elif whence == 1: # Seek from current position |
| 1029 | new_pos = curr_pos + offset |
| 1030 | elif whence == 2: # Seek from EOF |
| 1031 | new_pos = self._orig_file_size + offset |
| 1032 | else: |
| 1033 | raise ValueError("whence must be os.SEEK_SET (0), " |
| 1034 | "os.SEEK_CUR (1), or os.SEEK_END (2)") |
| 1035 | |
| 1036 | if new_pos > self._orig_file_size: |
| 1037 | new_pos = self._orig_file_size |
| 1038 | |
| 1039 | if new_pos < 0: |
| 1040 | new_pos = 0 |
| 1041 | |
| 1042 | read_offset = new_pos - curr_pos |
| 1043 | buff_offset = read_offset + self._offset |
| 1044 | |
| 1045 | if buff_offset >= 0 and buff_offset < len(self._readbuffer): |
| 1046 | # Just move the _offset index if the new position is in the _readbuffer |
| 1047 | self._offset = buff_offset |
| 1048 | read_offset = 0 |
| 1049 | elif read_offset < 0: |
| 1050 | # Position is before the current position. Reset the ZipExtFile |
John Jolly | 066df4f | 2018-01-30 01:51:35 -0700 | [diff] [blame] | 1051 | self._fileobj.seek(self._orig_compress_start) |
| 1052 | self._running_crc = self._orig_start_crc |
| 1053 | self._compress_left = self._orig_compress_size |
| 1054 | self._left = self._orig_file_size |
| 1055 | self._readbuffer = b'' |
| 1056 | self._offset = 0 |
Mickaël Schoentgen | 3f8c691 | 2018-07-29 20:26:52 +0200 | [diff] [blame] | 1057 | self._decompressor = _get_decompressor(self._compress_type) |
John Jolly | 066df4f | 2018-01-30 01:51:35 -0700 | [diff] [blame] | 1058 | self._eof = False |
| 1059 | read_offset = new_pos |
| 1060 | |
| 1061 | while read_offset > 0: |
| 1062 | read_len = min(self.MAX_SEEK_READ, read_offset) |
| 1063 | self.read(read_len) |
| 1064 | read_offset -= read_len |
| 1065 | |
| 1066 | return self.tell() |
| 1067 | |
| 1068 | def tell(self): |
| 1069 | if not self._seekable: |
| 1070 | raise io.UnsupportedOperation("underlying stream is not seekable") |
| 1071 | filepos = self._orig_file_size - self._left - len(self._readbuffer) + self._offset |
| 1072 | return filepos |
| 1073 | |
Antoine Pitrou | a32f9a2 | 2010-01-27 21:18:57 +0000 | [diff] [blame] | 1074 | |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1075 | class _ZipWriteFile(io.BufferedIOBase): |
| 1076 | def __init__(self, zf, zinfo, zip64): |
| 1077 | self._zinfo = zinfo |
| 1078 | self._zip64 = zip64 |
| 1079 | self._zipfile = zf |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 1080 | self._compressor = _get_compressor(zinfo.compress_type, |
| 1081 | zinfo._compresslevel) |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1082 | self._file_size = 0 |
| 1083 | self._compress_size = 0 |
| 1084 | self._crc = 0 |
| 1085 | |
| 1086 | @property |
| 1087 | def _fileobj(self): |
| 1088 | return self._zipfile.fp |
| 1089 | |
| 1090 | def writable(self): |
| 1091 | return True |
| 1092 | |
| 1093 | def write(self, data): |
Serhiy Storchaka | 4c0d9ea | 2017-04-12 16:03:23 +0300 | [diff] [blame] | 1094 | if self.closed: |
| 1095 | raise ValueError('I/O operation on closed file.') |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1096 | nbytes = len(data) |
| 1097 | self._file_size += nbytes |
| 1098 | self._crc = crc32(data, self._crc) |
| 1099 | if self._compressor: |
| 1100 | data = self._compressor.compress(data) |
| 1101 | self._compress_size += len(data) |
| 1102 | self._fileobj.write(data) |
| 1103 | return nbytes |
| 1104 | |
| 1105 | def close(self): |
Serhiy Storchaka | 4c0d9ea | 2017-04-12 16:03:23 +0300 | [diff] [blame] | 1106 | if self.closed: |
| 1107 | return |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1108 | super().close() |
| 1109 | # Flush any data from the compressor, and update header info |
| 1110 | if self._compressor: |
| 1111 | buf = self._compressor.flush() |
| 1112 | self._compress_size += len(buf) |
| 1113 | self._fileobj.write(buf) |
| 1114 | self._zinfo.compress_size = self._compress_size |
| 1115 | else: |
| 1116 | self._zinfo.compress_size = self._file_size |
| 1117 | self._zinfo.CRC = self._crc |
| 1118 | self._zinfo.file_size = self._file_size |
| 1119 | |
| 1120 | # Write updated header info |
| 1121 | if self._zinfo.flag_bits & 0x08: |
| 1122 | # Write CRC and file sizes after the file data |
Silas Sewell | 4ba3b50 | 2018-09-18 13:00:05 -0400 | [diff] [blame] | 1123 | fmt = '<LLQQ' if self._zip64 else '<LLLL' |
| 1124 | self._fileobj.write(struct.pack(fmt, _DD_SIGNATURE, self._zinfo.CRC, |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1125 | self._zinfo.compress_size, self._zinfo.file_size)) |
| 1126 | self._zipfile.start_dir = self._fileobj.tell() |
| 1127 | else: |
| 1128 | if not self._zip64: |
| 1129 | if self._file_size > ZIP64_LIMIT: |
| 1130 | raise RuntimeError('File size unexpectedly exceeded ZIP64 ' |
| 1131 | 'limit') |
| 1132 | if self._compress_size > ZIP64_LIMIT: |
| 1133 | raise RuntimeError('Compressed size unexpectedly exceeded ' |
| 1134 | 'ZIP64 limit') |
| 1135 | # Seek backwards and write file header (which will now include |
| 1136 | # correct CRC and file sizes) |
| 1137 | |
| 1138 | # Preserve current position in file |
| 1139 | self._zipfile.start_dir = self._fileobj.tell() |
| 1140 | self._fileobj.seek(self._zinfo.header_offset) |
| 1141 | self._fileobj.write(self._zinfo.FileHeader(self._zip64)) |
| 1142 | self._fileobj.seek(self._zipfile.start_dir) |
| 1143 | |
| 1144 | self._zipfile._writing = False |
| 1145 | |
| 1146 | # Successfully written: Add file to our caches |
| 1147 | self._zipfile.filelist.append(self._zinfo) |
| 1148 | self._zipfile.NameToInfo[self._zinfo.filename] = self._zinfo |
| 1149 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1150 | class ZipFile: |
Tim Peters | a19a168 | 2001-03-29 04:36:09 +0000 | [diff] [blame] | 1151 | """ Class with methods to open, read, write, close, list zip files. |
| 1152 | |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 1153 | z = ZipFile(file, mode="r", compression=ZIP_STORED, allowZip64=True, |
| 1154 | compresslevel=None) |
Tim Peters | a19a168 | 2001-03-29 04:36:09 +0000 | [diff] [blame] | 1155 | |
Fred Drake | 3d9091e | 2001-03-26 15:49:24 +0000 | [diff] [blame] | 1156 | file: Either the path to the file, or a file-like object. |
| 1157 | If it is a path, the file will be opened and closed by ZipFile. |
Serhiy Storchaka | 764fc9b | 2015-03-25 10:09:41 +0200 | [diff] [blame] | 1158 | mode: The mode can be either read 'r', write 'w', exclusive create 'x', |
| 1159 | or append 'a'. |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 1160 | compression: ZIP_STORED (no compression), ZIP_DEFLATED (requires zlib), |
| 1161 | ZIP_BZIP2 (requires bz2) or ZIP_LZMA (requires lzma). |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1162 | allowZip64: if True ZipFile will create files with ZIP64 extensions when |
| 1163 | needed, otherwise it will raise an exception when this would |
| 1164 | be necessary. |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 1165 | compresslevel: None (default for the given compression type) or an integer |
| 1166 | specifying the level to pass to the compressor. |
| 1167 | When using ZIP_STORED or ZIP_LZMA this keyword has no effect. |
| 1168 | When using ZIP_DEFLATED integers 0 through 9 are accepted. |
| 1169 | When using ZIP_BZIP2 integers 1 through 9 are accepted. |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1170 | |
Fred Drake | 3d9091e | 2001-03-26 15:49:24 +0000 | [diff] [blame] | 1171 | """ |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1172 | |
Fred Drake | 90eac28 | 2001-02-28 05:29:34 +0000 | [diff] [blame] | 1173 | fp = None # Set here since __del__ checks it |
Gregory P. Smith | 09aa752 | 2013-02-03 00:36:32 -0800 | [diff] [blame] | 1174 | _windows_illegal_name_trans_table = None |
Fred Drake | 90eac28 | 2001-02-28 05:29:34 +0000 | [diff] [blame] | 1175 | |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 1176 | def __init__(self, file, mode="r", compression=ZIP_STORED, allowZip64=True, |
Marcel Plch | 77b112c | 2018-08-31 16:43:31 +0200 | [diff] [blame] | 1177 | compresslevel=None, *, strict_timestamps=True): |
Serhiy Storchaka | 764fc9b | 2015-03-25 10:09:41 +0200 | [diff] [blame] | 1178 | """Open the ZIP file with mode read 'r', write 'w', exclusive create 'x', |
| 1179 | or append 'a'.""" |
| 1180 | if mode not in ('r', 'w', 'x', 'a'): |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1181 | raise ValueError("ZipFile requires mode 'r', 'w', 'x', or 'a'") |
Guido van Rossum | b5a755e | 2007-07-18 18:15:48 +0000 | [diff] [blame] | 1182 | |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 1183 | _check_compression(compression) |
Guido van Rossum | b5a755e | 2007-07-18 18:15:48 +0000 | [diff] [blame] | 1184 | |
| 1185 | self._allowZip64 = allowZip64 |
| 1186 | self._didModify = False |
Tim Peters | e119006 | 2001-01-15 03:34:38 +0000 | [diff] [blame] | 1187 | self.debug = 0 # Level of printing: 0 through 3 |
| 1188 | self.NameToInfo = {} # Find file info given name |
| 1189 | self.filelist = [] # List of ZipInfo instances for archive |
| 1190 | self.compression = compression # Method of compression |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 1191 | self.compresslevel = compresslevel |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1192 | self.mode = mode |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 1193 | self.pwd = None |
R David Murray | f50b38a | 2012-04-12 18:44:58 -0400 | [diff] [blame] | 1194 | self._comment = b'' |
Marcel Plch | 77b112c | 2018-08-31 16:43:31 +0200 | [diff] [blame] | 1195 | self._strict_timestamps = strict_timestamps |
Tim Peters | a19a168 | 2001-03-29 04:36:09 +0000 | [diff] [blame] | 1196 | |
Fred Drake | 3d9091e | 2001-03-26 15:49:24 +0000 | [diff] [blame] | 1197 | # Check if we were passed a file-like object |
Serhiy Storchaka | 8606e95 | 2017-03-08 14:37:51 +0200 | [diff] [blame] | 1198 | if isinstance(file, os.PathLike): |
| 1199 | file = os.fspath(file) |
Guido van Rossum | 3172c5d | 2007-10-16 18:12:55 +0000 | [diff] [blame] | 1200 | if isinstance(file, str): |
Guido van Rossum | d6ca546 | 2007-05-22 01:29:33 +0000 | [diff] [blame] | 1201 | # No, it's a filename |
Fred Drake | 3d9091e | 2001-03-26 15:49:24 +0000 | [diff] [blame] | 1202 | self._filePassed = 0 |
| 1203 | self.filename = file |
Serhiy Storchaka | 764fc9b | 2015-03-25 10:09:41 +0200 | [diff] [blame] | 1204 | modeDict = {'r' : 'rb', 'w': 'w+b', 'x': 'x+b', 'a' : 'r+b', |
| 1205 | 'r+b': 'w+b', 'w+b': 'wb', 'x+b': 'xb'} |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1206 | filemode = modeDict[mode] |
| 1207 | while True: |
| 1208 | try: |
| 1209 | self.fp = io.open(file, filemode) |
| 1210 | except OSError: |
| 1211 | if filemode in modeDict: |
| 1212 | filemode = modeDict[filemode] |
| 1213 | continue |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 1214 | raise |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1215 | break |
Fred Drake | 3d9091e | 2001-03-26 15:49:24 +0000 | [diff] [blame] | 1216 | else: |
| 1217 | self._filePassed = 1 |
| 1218 | self.fp = file |
| 1219 | self.filename = getattr(file, 'name', None) |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1220 | self._fileRefCnt = 1 |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 1221 | self._lock = threading.RLock() |
Serhiy Storchaka | 77d8997 | 2015-03-23 01:09:35 +0200 | [diff] [blame] | 1222 | self._seekable = True |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1223 | self._writing = False |
Tim Peters | a19a168 | 2001-03-29 04:36:09 +0000 | [diff] [blame] | 1224 | |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1225 | try: |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1226 | if mode == 'r': |
Martin v. Löwis | 6f6873b | 2002-10-13 13:54:50 +0000 | [diff] [blame] | 1227 | self._RealGetContents() |
Serhiy Storchaka | 764fc9b | 2015-03-25 10:09:41 +0200 | [diff] [blame] | 1228 | elif mode in ('w', 'x'): |
Georg Brandl | 268e4d4 | 2010-10-14 06:59:45 +0000 | [diff] [blame] | 1229 | # set the modified flag so central directory gets written |
| 1230 | # even if no files are added to the archive |
| 1231 | self._didModify = True |
Serhiy Storchaka | 77d8997 | 2015-03-23 01:09:35 +0200 | [diff] [blame] | 1232 | try: |
Serhiy Storchaka | 34cba33 | 2017-01-01 19:00:30 +0200 | [diff] [blame] | 1233 | self.start_dir = self.fp.tell() |
Serhiy Storchaka | 77d8997 | 2015-03-23 01:09:35 +0200 | [diff] [blame] | 1234 | except (AttributeError, OSError): |
| 1235 | self.fp = _Tellable(self.fp) |
Serhiy Storchaka | 34cba33 | 2017-01-01 19:00:30 +0200 | [diff] [blame] | 1236 | self.start_dir = 0 |
Serhiy Storchaka | 77d8997 | 2015-03-23 01:09:35 +0200 | [diff] [blame] | 1237 | self._seekable = False |
| 1238 | else: |
| 1239 | # Some file-like objects can provide tell() but not seek() |
| 1240 | try: |
| 1241 | self.fp.seek(self.start_dir) |
| 1242 | except (AttributeError, OSError): |
| 1243 | self._seekable = False |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1244 | elif mode == 'a': |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1245 | try: |
| 1246 | # See if file is a zip file |
| 1247 | self._RealGetContents() |
| 1248 | # seek to start of directory and overwrite |
Serhiy Storchaka | 77d8997 | 2015-03-23 01:09:35 +0200 | [diff] [blame] | 1249 | self.fp.seek(self.start_dir) |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1250 | except BadZipFile: |
| 1251 | # file is not a zip file, just append |
| 1252 | self.fp.seek(0, 2) |
| 1253 | |
| 1254 | # set the modified flag so central directory gets written |
| 1255 | # even if no files are added to the archive |
| 1256 | self._didModify = True |
Serhiy Storchaka | 3763ea8 | 2017-05-06 14:46:01 +0300 | [diff] [blame] | 1257 | self.start_dir = self.fp.tell() |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1258 | else: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1259 | raise ValueError("Mode must be 'r', 'w', 'x', or 'a'") |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1260 | except: |
| 1261 | fp = self.fp |
| 1262 | self.fp = None |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1263 | self._fpclose(fp) |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1264 | raise |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1265 | |
Ezio Melotti | faa6b7f | 2009-12-30 12:34:59 +0000 | [diff] [blame] | 1266 | def __enter__(self): |
| 1267 | return self |
| 1268 | |
| 1269 | def __exit__(self, type, value, traceback): |
| 1270 | self.close() |
| 1271 | |
Serhiy Storchaka | 51a4370 | 2014-10-29 22:42:06 +0200 | [diff] [blame] | 1272 | def __repr__(self): |
| 1273 | result = ['<%s.%s' % (self.__class__.__module__, |
| 1274 | self.__class__.__qualname__)] |
| 1275 | if self.fp is not None: |
| 1276 | if self._filePassed: |
| 1277 | result.append(' file=%r' % self.fp) |
| 1278 | elif self.filename is not None: |
| 1279 | result.append(' filename=%r' % self.filename) |
| 1280 | result.append(' mode=%r' % self.mode) |
| 1281 | else: |
| 1282 | result.append(' [closed]') |
| 1283 | result.append('>') |
| 1284 | return ''.join(result) |
| 1285 | |
Tim Peters | 7d3bad6 | 2001-04-04 18:56:49 +0000 | [diff] [blame] | 1286 | def _RealGetContents(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1287 | """Read in the table of contents for the ZIP file.""" |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1288 | fp = self.fp |
Georg Brandl | 268e4d4 | 2010-10-14 06:59:45 +0000 | [diff] [blame] | 1289 | try: |
| 1290 | endrec = _EndRecData(fp) |
Andrew Svetlov | f7a17b4 | 2012-12-25 16:47:37 +0200 | [diff] [blame] | 1291 | except OSError: |
Georg Brandl | 4d54088 | 2010-10-28 06:42:33 +0000 | [diff] [blame] | 1292 | raise BadZipFile("File is not a zip file") |
Martin v. Löwis | 6f6873b | 2002-10-13 13:54:50 +0000 | [diff] [blame] | 1293 | if not endrec: |
Georg Brandl | 4d54088 | 2010-10-28 06:42:33 +0000 | [diff] [blame] | 1294 | raise BadZipFile("File is not a zip file") |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1295 | if self.debug > 1: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 1296 | print(endrec) |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 1297 | size_cd = endrec[_ECD_SIZE] # bytes in central directory |
| 1298 | offset_cd = endrec[_ECD_OFFSET] # offset of central directory |
R David Murray | f50b38a | 2012-04-12 18:44:58 -0400 | [diff] [blame] | 1299 | self._comment = endrec[_ECD_COMMENT] # archive comment |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 1300 | |
Serhiy Storchaka | 3763ea8 | 2017-05-06 14:46:01 +0300 | [diff] [blame] | 1301 | # "concat" is zero, unless zip was concatenated to another file |
| 1302 | concat = endrec[_ECD_LOCATION] - size_cd - offset_cd |
Antoine Pitrou | 9e4fdf4 | 2008-09-05 23:43:02 +0000 | [diff] [blame] | 1303 | if endrec[_ECD_SIGNATURE] == stringEndArchive64: |
| 1304 | # If Zip64 extension structures are present, account for them |
Serhiy Storchaka | 3763ea8 | 2017-05-06 14:46:01 +0300 | [diff] [blame] | 1305 | concat -= (sizeEndCentDir64 + sizeEndCentDir64Locator) |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 1306 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1307 | if self.debug > 2: |
Serhiy Storchaka | 3763ea8 | 2017-05-06 14:46:01 +0300 | [diff] [blame] | 1308 | inferred = concat + offset_cd |
| 1309 | print("given, inferred, offset", offset_cd, inferred, concat) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1310 | # self.start_dir: Position of start of central directory |
Serhiy Storchaka | 3763ea8 | 2017-05-06 14:46:01 +0300 | [diff] [blame] | 1311 | self.start_dir = offset_cd + concat |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1312 | fp.seek(self.start_dir, 0) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1313 | data = fp.read(size_cd) |
Guido van Rossum | d6ca546 | 2007-05-22 01:29:33 +0000 | [diff] [blame] | 1314 | fp = io.BytesIO(data) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1315 | total = 0 |
| 1316 | while total < size_cd: |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 1317 | centdir = fp.read(sizeCentralDir) |
Serhiy Storchaka | d2b1527 | 2013-01-31 15:27:07 +0200 | [diff] [blame] | 1318 | if len(centdir) != sizeCentralDir: |
| 1319 | raise BadZipFile("Truncated central directory") |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1320 | centdir = struct.unpack(structCentralDir, centdir) |
Serhiy Storchaka | d2b1527 | 2013-01-31 15:27:07 +0200 | [diff] [blame] | 1321 | if centdir[_CD_SIGNATURE] != stringCentralDir: |
| 1322 | raise BadZipFile("Bad magic number for central directory") |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1323 | if self.debug > 2: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 1324 | print(centdir) |
Fred Drake | 3e038e5 | 2001-02-28 17:56:26 +0000 | [diff] [blame] | 1325 | filename = fp.read(centdir[_CD_FILENAME_LENGTH]) |
Martin v. Löwis | 8570f6a | 2008-05-05 17:44:38 +0000 | [diff] [blame] | 1326 | flags = centdir[5] |
| 1327 | if flags & 0x800: |
| 1328 | # UTF-8 file names extension |
| 1329 | filename = filename.decode('utf-8') |
| 1330 | else: |
| 1331 | # Historical ZIP filename encoding |
| 1332 | filename = filename.decode('cp437') |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1333 | # Create ZipInfo instance to store file information |
Martin v. Löwis | 8570f6a | 2008-05-05 17:44:38 +0000 | [diff] [blame] | 1334 | x = ZipInfo(filename) |
Fred Drake | 3e038e5 | 2001-02-28 17:56:26 +0000 | [diff] [blame] | 1335 | x.extra = fp.read(centdir[_CD_EXTRA_FIELD_LENGTH]) |
| 1336 | x.comment = fp.read(centdir[_CD_COMMENT_LENGTH]) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1337 | x.header_offset = centdir[_CD_LOCAL_HEADER_OFFSET] |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1338 | (x.create_version, x.create_system, x.extract_version, x.reserved, |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 1339 | x.flag_bits, x.compress_type, t, d, |
| 1340 | x.CRC, x.compress_size, x.file_size) = centdir[1:12] |
Martin v. Löwis | d099b56 | 2012-05-01 14:08:22 +0200 | [diff] [blame] | 1341 | if x.extract_version > MAX_EXTRACT_VERSION: |
| 1342 | raise NotImplementedError("zip file version %.1f" % |
| 1343 | (x.extract_version / 10)) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1344 | x.volume, x.internal_attr, x.external_attr = centdir[15:18] |
| 1345 | # Convert date/time code to (year, month, day, hour, min, sec) |
Christian Heimes | fdab48e | 2008-01-20 09:06:41 +0000 | [diff] [blame] | 1346 | x._raw_time = t |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1347 | x.date_time = ( (d>>9)+1980, (d>>5)&0xF, d&0x1F, |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 1348 | t>>11, (t>>5)&0x3F, (t&0x1F) * 2 ) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1349 | |
| 1350 | x._decodeExtra() |
Serhiy Storchaka | 3763ea8 | 2017-05-06 14:46:01 +0300 | [diff] [blame] | 1351 | x.header_offset = x.header_offset + concat |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1352 | self.filelist.append(x) |
| 1353 | self.NameToInfo[x.filename] = x |
Martin v. Löwis | b09b844 | 2008-07-03 14:13:42 +0000 | [diff] [blame] | 1354 | |
| 1355 | # update total bytes read from central directory |
| 1356 | total = (total + sizeCentralDir + centdir[_CD_FILENAME_LENGTH] |
| 1357 | + centdir[_CD_EXTRA_FIELD_LENGTH] |
| 1358 | + centdir[_CD_COMMENT_LENGTH]) |
| 1359 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1360 | if self.debug > 2: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 1361 | print("total", total) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1362 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1363 | |
| 1364 | def namelist(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1365 | """Return a list of file names in the archive.""" |
Ezio Melotti | 006917e | 2012-04-16 21:34:24 -0600 | [diff] [blame] | 1366 | return [data.filename for data in self.filelist] |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1367 | |
| 1368 | def infolist(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1369 | """Return a list of class ZipInfo instances for files in the |
| 1370 | archive.""" |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1371 | return self.filelist |
| 1372 | |
Guido van Rossum | d6ca546 | 2007-05-22 01:29:33 +0000 | [diff] [blame] | 1373 | def printdir(self, file=None): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1374 | """Print a table of contents for the zip file.""" |
Guido van Rossum | d6ca546 | 2007-05-22 01:29:33 +0000 | [diff] [blame] | 1375 | print("%-46s %19s %12s" % ("File Name", "Modified ", "Size"), |
| 1376 | file=file) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1377 | for zinfo in self.filelist: |
Guido van Rossum | 7736b5b | 2008-01-15 21:44:53 +0000 | [diff] [blame] | 1378 | date = "%d-%02d-%02d %02d:%02d:%02d" % zinfo.date_time[:6] |
Guido van Rossum | d6ca546 | 2007-05-22 01:29:33 +0000 | [diff] [blame] | 1379 | print("%-46s %s %12d" % (zinfo.filename, date, zinfo.file_size), |
| 1380 | file=file) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1381 | |
| 1382 | def testzip(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1383 | """Read all the files and check the CRC.""" |
Benjamin Peterson | 4cd6a95 | 2008-08-17 20:23:46 +0000 | [diff] [blame] | 1384 | chunk_size = 2 ** 20 |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1385 | for zinfo in self.filelist: |
| 1386 | try: |
Benjamin Peterson | 4cd6a95 | 2008-08-17 20:23:46 +0000 | [diff] [blame] | 1387 | # Read by chunks, to avoid an OverflowError or a |
| 1388 | # MemoryError with very large embedded files. |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1389 | with self.open(zinfo.filename, "r") as f: |
| 1390 | while f.read(chunk_size): # Check CRC-32 |
| 1391 | pass |
Georg Brandl | 4d54088 | 2010-10-28 06:42:33 +0000 | [diff] [blame] | 1392 | except BadZipFile: |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1393 | return zinfo.filename |
| 1394 | |
| 1395 | def getinfo(self, name): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1396 | """Return the instance of ZipInfo given 'name'.""" |
Guido van Rossum | b5a755e | 2007-07-18 18:15:48 +0000 | [diff] [blame] | 1397 | info = self.NameToInfo.get(name) |
| 1398 | if info is None: |
| 1399 | raise KeyError( |
| 1400 | 'There is no item named %r in the archive' % name) |
| 1401 | |
| 1402 | return info |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1403 | |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 1404 | def setpassword(self, pwd): |
| 1405 | """Set default password for encrypted files.""" |
R. David Murray | 8d855d8 | 2010-12-21 21:53:37 +0000 | [diff] [blame] | 1406 | if pwd and not isinstance(pwd, bytes): |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1407 | raise TypeError("pwd: expected bytes, got %s" % type(pwd).__name__) |
R. David Murray | 8d855d8 | 2010-12-21 21:53:37 +0000 | [diff] [blame] | 1408 | if pwd: |
| 1409 | self.pwd = pwd |
| 1410 | else: |
| 1411 | self.pwd = None |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 1412 | |
R David Murray | f50b38a | 2012-04-12 18:44:58 -0400 | [diff] [blame] | 1413 | @property |
| 1414 | def comment(self): |
| 1415 | """The comment text associated with the ZIP file.""" |
| 1416 | return self._comment |
| 1417 | |
| 1418 | @comment.setter |
| 1419 | def comment(self, comment): |
| 1420 | if not isinstance(comment, bytes): |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1421 | raise TypeError("comment: expected bytes, got %s" % type(comment).__name__) |
R David Murray | f50b38a | 2012-04-12 18:44:58 -0400 | [diff] [blame] | 1422 | # check for valid comment length |
Serhiy Storchaka | 9b7a1a1 | 2014-01-20 21:57:40 +0200 | [diff] [blame] | 1423 | if len(comment) > ZIP_MAX_COMMENT: |
| 1424 | import warnings |
| 1425 | warnings.warn('Archive comment is too long; truncating to %d bytes' |
| 1426 | % ZIP_MAX_COMMENT, stacklevel=2) |
R David Murray | f50b38a | 2012-04-12 18:44:58 -0400 | [diff] [blame] | 1427 | comment = comment[:ZIP_MAX_COMMENT] |
| 1428 | self._comment = comment |
| 1429 | self._didModify = True |
| 1430 | |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 1431 | def read(self, name, pwd=None): |
Serhiy Storchaka | 4bb186d | 2018-11-25 09:51:14 +0200 | [diff] [blame] | 1432 | """Return file bytes for name.""" |
Benjamin Peterson | d285bdb | 2010-10-31 17:57:22 +0000 | [diff] [blame] | 1433 | with self.open(name, "r", pwd) as fp: |
| 1434 | return fp.read() |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 1435 | |
Serhiy Storchaka | f47fc55 | 2016-05-15 12:27:16 +0300 | [diff] [blame] | 1436 | def open(self, name, mode="r", pwd=None, *, force_zip64=False): |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1437 | """Return file-like object for 'name'. |
| 1438 | |
| 1439 | name is a string for the file name within the ZIP file, or a ZipInfo |
| 1440 | object. |
| 1441 | |
| 1442 | mode should be 'r' to read a file already in the ZIP file, or 'w' to |
| 1443 | write to a file newly added to the archive. |
| 1444 | |
| 1445 | pwd is the password to decrypt files (only used for reading). |
| 1446 | |
| 1447 | When writing, if the file size is not known in advance but may exceed |
| 1448 | 2 GiB, pass force_zip64 to use the ZIP64 format, which can handle large |
| 1449 | files. If the size is known in advance, it is best to pass a ZipInfo |
| 1450 | instance for name, with zinfo.file_size set. |
| 1451 | """ |
Serhiy Storchaka | e670be2 | 2016-06-11 19:32:44 +0300 | [diff] [blame] | 1452 | if mode not in {"r", "w"}: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1453 | raise ValueError('open() requires mode "r" or "w"') |
R. David Murray | 8d855d8 | 2010-12-21 21:53:37 +0000 | [diff] [blame] | 1454 | if pwd and not isinstance(pwd, bytes): |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1455 | raise TypeError("pwd: expected bytes, got %s" % type(pwd).__name__) |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1456 | if pwd and (mode == "w"): |
| 1457 | raise ValueError("pwd is only supported for reading files") |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1458 | if not self.fp: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1459 | raise ValueError( |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1460 | "Attempt to use ZIP archive that was already closed") |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1461 | |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1462 | # Make sure we have an info object |
| 1463 | if isinstance(name, ZipInfo): |
| 1464 | # 'name' is already an info object |
| 1465 | zinfo = name |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1466 | elif mode == 'w': |
| 1467 | zinfo = ZipInfo(name) |
| 1468 | zinfo.compress_type = self.compression |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 1469 | zinfo._compresslevel = self.compresslevel |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 1470 | else: |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1471 | # Get info object for name |
| 1472 | zinfo = self.getinfo(name) |
Guido van Rossum | d8faa36 | 2007-04-27 19:54:29 +0000 | [diff] [blame] | 1473 | |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1474 | if mode == 'w': |
| 1475 | return self._open_to_write(zinfo, force_zip64=force_zip64) |
| 1476 | |
| 1477 | if self._writing: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1478 | raise ValueError("Can't read from the ZIP file while there " |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1479 | "is an open writing handle on it. " |
| 1480 | "Close the writing handle before trying to read.") |
| 1481 | |
| 1482 | # Open for reading: |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1483 | self._fileRefCnt += 1 |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1484 | zef_file = _SharedFile(self.fp, zinfo.header_offset, |
| 1485 | self._fpclose, self._lock, lambda: self._writing) |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1486 | try: |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1487 | # Skip the file header: |
| 1488 | fheader = zef_file.read(sizeFileHeader) |
Serhiy Storchaka | d2b1527 | 2013-01-31 15:27:07 +0200 | [diff] [blame] | 1489 | if len(fheader) != sizeFileHeader: |
| 1490 | raise BadZipFile("Truncated file header") |
| 1491 | fheader = struct.unpack(structFileHeader, fheader) |
| 1492 | if fheader[_FH_SIGNATURE] != stringFileHeader: |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1493 | raise BadZipFile("Bad magic number for file header") |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1494 | |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1495 | fname = zef_file.read(fheader[_FH_FILENAME_LENGTH]) |
| 1496 | if fheader[_FH_EXTRA_FIELD_LENGTH]: |
| 1497 | zef_file.read(fheader[_FH_EXTRA_FIELD_LENGTH]) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1498 | |
Antoine Pitrou | 8572da5 | 2012-11-17 23:52:05 +0100 | [diff] [blame] | 1499 | if zinfo.flag_bits & 0x20: |
| 1500 | # Zip 2.7: compressed patched data |
| 1501 | raise NotImplementedError("compressed patched data (flag bit 5)") |
Martin v. Löwis | 2a2ce32 | 2012-05-01 08:44:08 +0200 | [diff] [blame] | 1502 | |
Antoine Pitrou | 8572da5 | 2012-11-17 23:52:05 +0100 | [diff] [blame] | 1503 | if zinfo.flag_bits & 0x40: |
| 1504 | # strong encryption |
| 1505 | raise NotImplementedError("strong encryption (flag bit 6)") |
Martin v. Löwis | 7fb79fc | 2012-05-13 10:06:36 +0200 | [diff] [blame] | 1506 | |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1507 | if zinfo.flag_bits & 0x800: |
| 1508 | # UTF-8 filename |
| 1509 | fname_str = fname.decode("utf-8") |
| 1510 | else: |
| 1511 | fname_str = fname.decode("cp437") |
Georg Brandl | 5ba11de | 2011-01-01 10:09:32 +0000 | [diff] [blame] | 1512 | |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1513 | if fname_str != zinfo.orig_filename: |
| 1514 | raise BadZipFile( |
| 1515 | 'File name in directory %r and header %r differ.' |
| 1516 | % (zinfo.orig_filename, fname)) |
| 1517 | |
| 1518 | # check for encrypted flag & handle password |
| 1519 | is_encrypted = zinfo.flag_bits & 0x1 |
| 1520 | zd = None |
| 1521 | if is_encrypted: |
| 1522 | if not pwd: |
| 1523 | pwd = self.pwd |
| 1524 | if not pwd: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1525 | raise RuntimeError("File %r is encrypted, password " |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1526 | "required for extraction" % name) |
| 1527 | |
| 1528 | zd = _ZipDecrypter(pwd) |
| 1529 | # The first 12 bytes in the cypher stream is an encryption header |
| 1530 | # used to strengthen the algorithm. The first 11 bytes are |
| 1531 | # completely random, while the 12th contains the MSB of the CRC, |
| 1532 | # or the MSB of the file time depending on the header type |
| 1533 | # and is used to check the correctness of the password. |
| 1534 | header = zef_file.read(12) |
Serhiy Storchaka | 06e5225 | 2017-03-30 19:09:08 +0300 | [diff] [blame] | 1535 | h = zd(header[0:12]) |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1536 | if zinfo.flag_bits & 0x8: |
| 1537 | # compare against the file type from extended local headers |
| 1538 | check_byte = (zinfo._raw_time >> 8) & 0xff |
| 1539 | else: |
| 1540 | # compare against the CRC otherwise |
| 1541 | check_byte = (zinfo.CRC >> 24) & 0xff |
| 1542 | if h[11] != check_byte: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1543 | raise RuntimeError("Bad password for file %r" % name) |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1544 | |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1545 | return ZipExtFile(zef_file, mode, zinfo, zd, True) |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1546 | except: |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1547 | zef_file.close() |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1548 | raise |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1549 | |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1550 | def _open_to_write(self, zinfo, force_zip64=False): |
| 1551 | if force_zip64 and not self._allowZip64: |
| 1552 | raise ValueError( |
| 1553 | "force_zip64 is True, but allowZip64 was False when opening " |
| 1554 | "the ZIP file." |
| 1555 | ) |
| 1556 | if self._writing: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1557 | raise ValueError("Can't write to the ZIP file while there is " |
| 1558 | "another write handle open on it. " |
| 1559 | "Close the first handle before opening another.") |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1560 | |
| 1561 | # Sizes and CRC are overwritten with correct data after processing the file |
| 1562 | if not hasattr(zinfo, 'file_size'): |
| 1563 | zinfo.file_size = 0 |
| 1564 | zinfo.compress_size = 0 |
| 1565 | zinfo.CRC = 0 |
| 1566 | |
| 1567 | zinfo.flag_bits = 0x00 |
| 1568 | if zinfo.compress_type == ZIP_LZMA: |
| 1569 | # Compressed data includes an end-of-stream (EOS) marker |
| 1570 | zinfo.flag_bits |= 0x02 |
| 1571 | if not self._seekable: |
| 1572 | zinfo.flag_bits |= 0x08 |
| 1573 | |
| 1574 | if not zinfo.external_attr: |
| 1575 | zinfo.external_attr = 0o600 << 16 # permissions: ?rw------- |
| 1576 | |
| 1577 | # Compressed size can be larger than uncompressed size |
| 1578 | zip64 = self._allowZip64 and \ |
| 1579 | (force_zip64 or zinfo.file_size * 1.05 > ZIP64_LIMIT) |
| 1580 | |
| 1581 | if self._seekable: |
| 1582 | self.fp.seek(self.start_dir) |
| 1583 | zinfo.header_offset = self.fp.tell() |
| 1584 | |
| 1585 | self._writecheck(zinfo) |
| 1586 | self._didModify = True |
| 1587 | |
| 1588 | self.fp.write(zinfo.FileHeader(zip64)) |
| 1589 | |
| 1590 | self._writing = True |
| 1591 | return _ZipWriteFile(self, zinfo, zip64) |
| 1592 | |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 1593 | def extract(self, member, path=None, pwd=None): |
| 1594 | """Extract a member from the archive to the current working directory, |
| 1595 | using its full name. Its file information is extracted as accurately |
| 1596 | as possible. `member' may be a filename or a ZipInfo object. You can |
| 1597 | specify a different directory using `path'. |
| 1598 | """ |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 1599 | if path is None: |
| 1600 | path = os.getcwd() |
Serhiy Storchaka | 8606e95 | 2017-03-08 14:37:51 +0200 | [diff] [blame] | 1601 | else: |
| 1602 | path = os.fspath(path) |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 1603 | |
| 1604 | return self._extract_member(member, path, pwd) |
| 1605 | |
| 1606 | def extractall(self, path=None, members=None, pwd=None): |
| 1607 | """Extract all members from the archive to the current working |
| 1608 | directory. `path' specifies a different directory to extract to. |
| 1609 | `members' is optional and must be a subset of the list returned |
| 1610 | by namelist(). |
| 1611 | """ |
| 1612 | if members is None: |
| 1613 | members = self.namelist() |
| 1614 | |
Serhiy Storchaka | 8606e95 | 2017-03-08 14:37:51 +0200 | [diff] [blame] | 1615 | if path is None: |
| 1616 | path = os.getcwd() |
| 1617 | else: |
| 1618 | path = os.fspath(path) |
| 1619 | |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 1620 | for zipinfo in members: |
Serhiy Storchaka | 8606e95 | 2017-03-08 14:37:51 +0200 | [diff] [blame] | 1621 | self._extract_member(zipinfo, path, pwd) |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 1622 | |
Gregory P. Smith | 09aa752 | 2013-02-03 00:36:32 -0800 | [diff] [blame] | 1623 | @classmethod |
| 1624 | def _sanitize_windows_name(cls, arcname, pathsep): |
| 1625 | """Replace bad characters and remove trailing dots from parts.""" |
| 1626 | table = cls._windows_illegal_name_trans_table |
| 1627 | if not table: |
| 1628 | illegal = ':<>|"?*' |
| 1629 | table = str.maketrans(illegal, '_' * len(illegal)) |
| 1630 | cls._windows_illegal_name_trans_table = table |
| 1631 | arcname = arcname.translate(table) |
| 1632 | # remove trailing dots |
| 1633 | arcname = (x.rstrip('.') for x in arcname.split(pathsep)) |
| 1634 | # rejoin, removing empty parts. |
| 1635 | arcname = pathsep.join(x for x in arcname if x) |
| 1636 | return arcname |
| 1637 | |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 1638 | def _extract_member(self, member, targetpath, pwd): |
| 1639 | """Extract the ZipInfo object 'member' to a physical |
| 1640 | file on the path targetpath. |
| 1641 | """ |
Serhiy Storchaka | 8606e95 | 2017-03-08 14:37:51 +0200 | [diff] [blame] | 1642 | if not isinstance(member, ZipInfo): |
| 1643 | member = self.getinfo(member) |
| 1644 | |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 1645 | # build the destination pathname, replacing |
| 1646 | # forward slashes to platform specific separators. |
Gregory P. Smith | b47acbf | 2013-02-01 11:22:43 -0800 | [diff] [blame] | 1647 | arcname = member.filename.replace('/', os.path.sep) |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 1648 | |
Gregory P. Smith | b47acbf | 2013-02-01 11:22:43 -0800 | [diff] [blame] | 1649 | if os.path.altsep: |
| 1650 | arcname = arcname.replace(os.path.altsep, os.path.sep) |
| 1651 | # interpret absolute pathname as relative, remove drive letter or |
| 1652 | # UNC path, redundant separators, "." and ".." components. |
| 1653 | arcname = os.path.splitdrive(arcname)[1] |
Gregory P. Smith | 09aa752 | 2013-02-03 00:36:32 -0800 | [diff] [blame] | 1654 | invalid_path_parts = ('', os.path.curdir, os.path.pardir) |
Gregory P. Smith | b47acbf | 2013-02-01 11:22:43 -0800 | [diff] [blame] | 1655 | arcname = os.path.sep.join(x for x in arcname.split(os.path.sep) |
Gregory P. Smith | 09aa752 | 2013-02-03 00:36:32 -0800 | [diff] [blame] | 1656 | if x not in invalid_path_parts) |
Gregory P. Smith | b47acbf | 2013-02-01 11:22:43 -0800 | [diff] [blame] | 1657 | if os.path.sep == '\\': |
Serhiy Storchaka | e5e6444 | 2013-02-02 19:50:59 +0200 | [diff] [blame] | 1658 | # filter illegal characters on Windows |
Gregory P. Smith | 09aa752 | 2013-02-03 00:36:32 -0800 | [diff] [blame] | 1659 | arcname = self._sanitize_windows_name(arcname, os.path.sep) |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 1660 | |
Gregory P. Smith | b47acbf | 2013-02-01 11:22:43 -0800 | [diff] [blame] | 1661 | targetpath = os.path.join(targetpath, arcname) |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 1662 | targetpath = os.path.normpath(targetpath) |
| 1663 | |
| 1664 | # Create all upper directories if necessary. |
| 1665 | upperdirs = os.path.dirname(targetpath) |
| 1666 | if upperdirs and not os.path.exists(upperdirs): |
| 1667 | os.makedirs(upperdirs) |
| 1668 | |
Serhiy Storchaka | 503f908 | 2016-02-08 00:02:25 +0200 | [diff] [blame] | 1669 | if member.is_dir(): |
Martin v. Löwis | 70ccd16 | 2009-05-24 19:47:22 +0000 | [diff] [blame] | 1670 | if not os.path.isdir(targetpath): |
| 1671 | os.mkdir(targetpath) |
Martin v. Löwis | 59e4779 | 2009-01-24 14:10:07 +0000 | [diff] [blame] | 1672 | return targetpath |
| 1673 | |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1674 | with self.open(member, pwd=pwd) as source, \ |
| 1675 | open(targetpath, "wb") as target: |
| 1676 | shutil.copyfileobj(source, target) |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 1677 | |
| 1678 | return targetpath |
| 1679 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1680 | def _writecheck(self, zinfo): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1681 | """Check for errors before writing a file to the archive.""" |
Raymond Hettinger | 54f0222 | 2002-06-01 14:18:47 +0000 | [diff] [blame] | 1682 | if zinfo.filename in self.NameToInfo: |
Serhiy Storchaka | 9b7a1a1 | 2014-01-20 21:57:40 +0200 | [diff] [blame] | 1683 | import warnings |
| 1684 | warnings.warn('Duplicate name: %r' % zinfo.filename, stacklevel=3) |
Serhiy Storchaka | 764fc9b | 2015-03-25 10:09:41 +0200 | [diff] [blame] | 1685 | if self.mode not in ('w', 'x', 'a'): |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1686 | raise ValueError("write() requires mode 'w', 'x', or 'a'") |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1687 | if not self.fp: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1688 | raise ValueError( |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 1689 | "Attempt to write ZIP archive that was already closed") |
Martin v. Löwis | f6b16a4 | 2012-05-01 07:58:44 +0200 | [diff] [blame] | 1690 | _check_compression(zinfo.compress_type) |
Serhiy Storchaka | cfbb394 | 2014-09-23 21:34:24 +0300 | [diff] [blame] | 1691 | if not self._allowZip64: |
| 1692 | requires_zip64 = None |
| 1693 | if len(self.filelist) >= ZIP_FILECOUNT_LIMIT: |
| 1694 | requires_zip64 = "Files count" |
| 1695 | elif zinfo.file_size > ZIP64_LIMIT: |
| 1696 | requires_zip64 = "Filesize" |
| 1697 | elif zinfo.header_offset > ZIP64_LIMIT: |
| 1698 | requires_zip64 = "Zipfile size" |
| 1699 | if requires_zip64: |
| 1700 | raise LargeZipFile(requires_zip64 + |
| 1701 | " would require ZIP64 extensions") |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1702 | |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 1703 | def write(self, filename, arcname=None, |
Marcel Plch | 77b112c | 2018-08-31 16:43:31 +0200 | [diff] [blame] | 1704 | compress_type=None, compresslevel=None): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1705 | """Put the bytes from filename into the archive under the name |
| 1706 | arcname.""" |
Guido van Rossum | b5a755e | 2007-07-18 18:15:48 +0000 | [diff] [blame] | 1707 | if not self.fp: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1708 | raise ValueError( |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 1709 | "Attempt to write to ZIP archive that was already closed") |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1710 | if self._writing: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1711 | raise ValueError( |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1712 | "Can't write to ZIP archive while an open writing handle exists" |
| 1713 | ) |
Guido van Rossum | b5a755e | 2007-07-18 18:15:48 +0000 | [diff] [blame] | 1714 | |
Marcel Plch | a2fe1e5 | 2018-08-02 15:04:52 +0200 | [diff] [blame] | 1715 | zinfo = ZipInfo.from_file(filename, arcname, |
Marcel Plch | 77b112c | 2018-08-31 16:43:31 +0200 | [diff] [blame] | 1716 | strict_timestamps=self._strict_timestamps) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1717 | |
Serhiy Storchaka | 503f908 | 2016-02-08 00:02:25 +0200 | [diff] [blame] | 1718 | if zinfo.is_dir(): |
| 1719 | zinfo.compress_size = 0 |
| 1720 | zinfo.CRC = 0 |
| 1721 | else: |
| 1722 | if compress_type is not None: |
| 1723 | zinfo.compress_type = compress_type |
| 1724 | else: |
| 1725 | zinfo.compress_type = self.compression |
| 1726 | |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 1727 | if compresslevel is not None: |
| 1728 | zinfo._compresslevel = compresslevel |
| 1729 | else: |
| 1730 | zinfo._compresslevel = self.compresslevel |
| 1731 | |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1732 | if zinfo.is_dir(): |
| 1733 | with self._lock: |
| 1734 | if self._seekable: |
| 1735 | self.fp.seek(self.start_dir) |
| 1736 | zinfo.header_offset = self.fp.tell() # Start of header bytes |
| 1737 | if zinfo.compress_type == ZIP_LZMA: |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 1738 | # Compressed data includes an end-of-stream (EOS) marker |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1739 | zinfo.flag_bits |= 0x02 |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1740 | |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1741 | self._writecheck(zinfo) |
| 1742 | self._didModify = True |
Martin v. Löwis | 59e4779 | 2009-01-24 14:10:07 +0000 | [diff] [blame] | 1743 | |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 1744 | self.filelist.append(zinfo) |
| 1745 | self.NameToInfo[zinfo.filename] = zinfo |
| 1746 | self.fp.write(zinfo.FileHeader(False)) |
| 1747 | self.start_dir = self.fp.tell() |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1748 | else: |
| 1749 | with open(filename, "rb") as src, self.open(zinfo, 'w') as dest: |
| 1750 | shutil.copyfileobj(src, dest, 1024*8) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1751 | |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 1752 | def writestr(self, zinfo_or_arcname, data, |
| 1753 | compress_type=None, compresslevel=None): |
Guido van Rossum | 85825dc | 2007-08-27 17:03:28 +0000 | [diff] [blame] | 1754 | """Write a file into the archive. The contents is 'data', which |
| 1755 | may be either a 'str' or a 'bytes' instance; if it is a 'str', |
| 1756 | it is encoded as UTF-8 first. |
| 1757 | 'zinfo_or_arcname' is either a ZipInfo instance or |
Just van Rossum | b083cb3 | 2002-12-12 12:23:32 +0000 | [diff] [blame] | 1758 | the name of the file in the archive.""" |
Guido van Rossum | 85825dc | 2007-08-27 17:03:28 +0000 | [diff] [blame] | 1759 | if isinstance(data, str): |
| 1760 | data = data.encode("utf-8") |
Just van Rossum | b083cb3 | 2002-12-12 12:23:32 +0000 | [diff] [blame] | 1761 | if not isinstance(zinfo_or_arcname, ZipInfo): |
| 1762 | zinfo = ZipInfo(filename=zinfo_or_arcname, |
Guido van Rossum | 7736b5b | 2008-01-15 21:44:53 +0000 | [diff] [blame] | 1763 | date_time=time.localtime(time.time())[:6]) |
Just van Rossum | b083cb3 | 2002-12-12 12:23:32 +0000 | [diff] [blame] | 1764 | zinfo.compress_type = self.compression |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 1765 | zinfo._compresslevel = self.compresslevel |
Serhiy Storchaka | 46a3492 | 2014-09-23 22:40:23 +0300 | [diff] [blame] | 1766 | if zinfo.filename[-1] == '/': |
| 1767 | zinfo.external_attr = 0o40775 << 16 # drwxrwxr-x |
| 1768 | zinfo.external_attr |= 0x10 # MS-DOS directory flag |
| 1769 | else: |
| 1770 | zinfo.external_attr = 0o600 << 16 # ?rw------- |
Just van Rossum | b083cb3 | 2002-12-12 12:23:32 +0000 | [diff] [blame] | 1771 | else: |
| 1772 | zinfo = zinfo_or_arcname |
Guido van Rossum | b5a755e | 2007-07-18 18:15:48 +0000 | [diff] [blame] | 1773 | |
| 1774 | if not self.fp: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1775 | raise ValueError( |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 1776 | "Attempt to write to ZIP archive that was already closed") |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1777 | if self._writing: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1778 | raise ValueError( |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1779 | "Can't write to ZIP archive while an open writing handle exists." |
| 1780 | ) |
| 1781 | |
| 1782 | if compress_type is not None: |
| 1783 | zinfo.compress_type = compress_type |
Guido van Rossum | b5a755e | 2007-07-18 18:15:48 +0000 | [diff] [blame] | 1784 | |
Bo Bayles | ce237c7 | 2018-01-29 23:54:07 -0600 | [diff] [blame] | 1785 | if compresslevel is not None: |
| 1786 | zinfo._compresslevel = compresslevel |
| 1787 | |
Guido van Rossum | 85825dc | 2007-08-27 17:03:28 +0000 | [diff] [blame] | 1788 | zinfo.file_size = len(data) # Uncompressed size |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 1789 | with self._lock: |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1790 | with self.open(zinfo, mode='w') as dest: |
| 1791 | dest.write(data) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1792 | |
| 1793 | def __del__(self): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1794 | """Call the "close()" method in case the user forgot.""" |
Tim Peters | d15f8bb | 2001-11-28 23:16:40 +0000 | [diff] [blame] | 1795 | self.close() |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1796 | |
| 1797 | def close(self): |
Serhiy Storchaka | 764fc9b | 2015-03-25 10:09:41 +0200 | [diff] [blame] | 1798 | """Close the file, and for mode 'w', 'x' and 'a' write the ending |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1799 | records.""" |
Tim Peters | d15f8bb | 2001-11-28 23:16:40 +0000 | [diff] [blame] | 1800 | if self.fp is None: |
| 1801 | return |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 1802 | |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1803 | if self._writing: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1804 | raise ValueError("Can't close the ZIP file while there is " |
| 1805 | "an open writing handle on it. " |
| 1806 | "Close the writing handle before closing the zip.") |
Serhiy Storchaka | 18ee29d | 2016-05-13 13:52:49 +0300 | [diff] [blame] | 1807 | |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1808 | try: |
Serhiy Storchaka | 764fc9b | 2015-03-25 10:09:41 +0200 | [diff] [blame] | 1809 | if self.mode in ('w', 'x', 'a') and self._didModify: # write ending records |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 1810 | with self._lock: |
Serhiy Storchaka | 77d8997 | 2015-03-23 01:09:35 +0200 | [diff] [blame] | 1811 | if self._seekable: |
Serhiy Storchaka | a14f7d2 | 2015-01-26 14:01:27 +0200 | [diff] [blame] | 1812 | self.fp.seek(self.start_dir) |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 1813 | self._write_end_record() |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 1814 | finally: |
| 1815 | fp = self.fp |
| 1816 | self.fp = None |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1817 | self._fpclose(fp) |
| 1818 | |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 1819 | def _write_end_record(self): |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 1820 | for zinfo in self.filelist: # write central directory |
| 1821 | dt = zinfo.date_time |
| 1822 | dosdate = (dt[0] - 1980) << 9 | dt[1] << 5 | dt[2] |
| 1823 | dostime = dt[3] << 11 | dt[4] << 5 | (dt[5] // 2) |
| 1824 | extra = [] |
| 1825 | if zinfo.file_size > ZIP64_LIMIT \ |
| 1826 | or zinfo.compress_size > ZIP64_LIMIT: |
| 1827 | extra.append(zinfo.file_size) |
| 1828 | extra.append(zinfo.compress_size) |
| 1829 | file_size = 0xffffffff |
| 1830 | compress_size = 0xffffffff |
| 1831 | else: |
| 1832 | file_size = zinfo.file_size |
| 1833 | compress_size = zinfo.compress_size |
| 1834 | |
Serhiy Storchaka | 3763ea8 | 2017-05-06 14:46:01 +0300 | [diff] [blame] | 1835 | if zinfo.header_offset > ZIP64_LIMIT: |
| 1836 | extra.append(zinfo.header_offset) |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 1837 | header_offset = 0xffffffff |
Serhiy Storchaka | 3763ea8 | 2017-05-06 14:46:01 +0300 | [diff] [blame] | 1838 | else: |
| 1839 | header_offset = zinfo.header_offset |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 1840 | |
| 1841 | extra_data = zinfo.extra |
| 1842 | min_version = 0 |
| 1843 | if extra: |
| 1844 | # Append a ZIP64 field to the extra's |
Serhiy Storchaka | 9bdb7be | 2018-09-17 15:36:40 +0300 | [diff] [blame] | 1845 | extra_data = _strip_extra(extra_data, (1,)) |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 1846 | extra_data = struct.pack( |
| 1847 | '<HH' + 'Q'*len(extra), |
| 1848 | 1, 8*len(extra), *extra) + extra_data |
| 1849 | |
| 1850 | min_version = ZIP64_VERSION |
| 1851 | |
| 1852 | if zinfo.compress_type == ZIP_BZIP2: |
| 1853 | min_version = max(BZIP2_VERSION, min_version) |
| 1854 | elif zinfo.compress_type == ZIP_LZMA: |
| 1855 | min_version = max(LZMA_VERSION, min_version) |
| 1856 | |
| 1857 | extract_version = max(min_version, zinfo.extract_version) |
| 1858 | create_version = max(min_version, zinfo.create_version) |
| 1859 | try: |
| 1860 | filename, flag_bits = zinfo._encodeFilenameFlags() |
| 1861 | centdir = struct.pack(structCentralDir, |
| 1862 | stringCentralDir, create_version, |
| 1863 | zinfo.create_system, extract_version, zinfo.reserved, |
| 1864 | flag_bits, zinfo.compress_type, dostime, dosdate, |
| 1865 | zinfo.CRC, compress_size, file_size, |
| 1866 | len(filename), len(extra_data), len(zinfo.comment), |
| 1867 | 0, zinfo.internal_attr, zinfo.external_attr, |
| 1868 | header_offset) |
| 1869 | except DeprecationWarning: |
| 1870 | print((structCentralDir, stringCentralDir, create_version, |
| 1871 | zinfo.create_system, extract_version, zinfo.reserved, |
| 1872 | zinfo.flag_bits, zinfo.compress_type, dostime, dosdate, |
| 1873 | zinfo.CRC, compress_size, file_size, |
| 1874 | len(zinfo.filename), len(extra_data), len(zinfo.comment), |
| 1875 | 0, zinfo.internal_attr, zinfo.external_attr, |
| 1876 | header_offset), file=sys.stderr) |
| 1877 | raise |
| 1878 | self.fp.write(centdir) |
| 1879 | self.fp.write(filename) |
| 1880 | self.fp.write(extra_data) |
| 1881 | self.fp.write(zinfo.comment) |
| 1882 | |
| 1883 | pos2 = self.fp.tell() |
| 1884 | # Write end-of-zip-archive record |
| 1885 | centDirCount = len(self.filelist) |
| 1886 | centDirSize = pos2 - self.start_dir |
Serhiy Storchaka | 3763ea8 | 2017-05-06 14:46:01 +0300 | [diff] [blame] | 1887 | centDirOffset = self.start_dir |
Serhiy Storchaka | f15e524 | 2015-01-26 13:53:38 +0200 | [diff] [blame] | 1888 | requires_zip64 = None |
| 1889 | if centDirCount > ZIP_FILECOUNT_LIMIT: |
| 1890 | requires_zip64 = "Files count" |
| 1891 | elif centDirOffset > ZIP64_LIMIT: |
| 1892 | requires_zip64 = "Central directory offset" |
| 1893 | elif centDirSize > ZIP64_LIMIT: |
| 1894 | requires_zip64 = "Central directory size" |
| 1895 | if requires_zip64: |
| 1896 | # Need to write the ZIP64 end-of-archive records |
| 1897 | if not self._allowZip64: |
| 1898 | raise LargeZipFile(requires_zip64 + |
| 1899 | " would require ZIP64 extensions") |
| 1900 | zip64endrec = struct.pack( |
| 1901 | structEndArchive64, stringEndArchive64, |
| 1902 | 44, 45, 45, 0, 0, centDirCount, centDirCount, |
| 1903 | centDirSize, centDirOffset) |
| 1904 | self.fp.write(zip64endrec) |
| 1905 | |
| 1906 | zip64locrec = struct.pack( |
| 1907 | structEndArchive64Locator, |
| 1908 | stringEndArchive64Locator, 0, pos2, 1) |
| 1909 | self.fp.write(zip64locrec) |
| 1910 | centDirCount = min(centDirCount, 0xFFFF) |
| 1911 | centDirSize = min(centDirSize, 0xFFFFFFFF) |
| 1912 | centDirOffset = min(centDirOffset, 0xFFFFFFFF) |
| 1913 | |
| 1914 | endrec = struct.pack(structEndArchive, stringEndArchive, |
| 1915 | 0, 0, centDirCount, centDirCount, |
| 1916 | centDirSize, centDirOffset, len(self._comment)) |
| 1917 | self.fp.write(endrec) |
| 1918 | self.fp.write(self._comment) |
| 1919 | self.fp.flush() |
| 1920 | |
Serhiy Storchaka | 1ad088f | 2014-12-03 09:11:57 +0200 | [diff] [blame] | 1921 | def _fpclose(self, fp): |
| 1922 | assert self._fileRefCnt > 0 |
| 1923 | self._fileRefCnt -= 1 |
| 1924 | if not self._fileRefCnt and not self._filePassed: |
| 1925 | fp.close() |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1926 | |
| 1927 | |
| 1928 | class PyZipFile(ZipFile): |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1929 | """Class to create ZIP archives with Python library files and packages.""" |
| 1930 | |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 1931 | def __init__(self, file, mode="r", compression=ZIP_STORED, |
Serhiy Storchaka | 235c5e0 | 2013-11-23 15:55:38 +0200 | [diff] [blame] | 1932 | allowZip64=True, optimize=-1): |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 1933 | ZipFile.__init__(self, file, mode=mode, compression=compression, |
| 1934 | allowZip64=allowZip64) |
| 1935 | self._optimize = optimize |
| 1936 | |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 1937 | def writepy(self, pathname, basename="", filterfunc=None): |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1938 | """Add all files from "pathname" to the ZIP archive. |
| 1939 | |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1940 | If pathname is a package directory, search the directory and |
| 1941 | all package subdirectories recursively for all *.py and enter |
| 1942 | the modules into the archive. If pathname is a plain |
| 1943 | directory, listdir *.py and enter all modules. Else, pathname |
| 1944 | must be a Python *.py file and the module will be put into the |
Brett Cannon | f299abd | 2015-04-13 14:21:02 -0400 | [diff] [blame] | 1945 | archive. Added modules are always module.pyc. |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1946 | This method will compile the module.py into module.pyc if |
| 1947 | necessary. |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 1948 | If filterfunc(pathname) is given, it is called with every argument. |
| 1949 | When it is False, the file or directory is skipped. |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 1950 | """ |
Serhiy Storchaka | 8606e95 | 2017-03-08 14:37:51 +0200 | [diff] [blame] | 1951 | pathname = os.fspath(pathname) |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 1952 | if filterfunc and not filterfunc(pathname): |
| 1953 | if self.debug: |
Christian Tismer | 410d931 | 2013-10-22 04:09:28 +0200 | [diff] [blame] | 1954 | label = 'path' if os.path.isdir(pathname) else 'file' |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1955 | print('%s %r skipped by filterfunc' % (label, pathname)) |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 1956 | return |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1957 | dir, name = os.path.split(pathname) |
| 1958 | if os.path.isdir(pathname): |
| 1959 | initname = os.path.join(pathname, "__init__.py") |
| 1960 | if os.path.isfile(initname): |
| 1961 | # This is a package directory, add it |
| 1962 | if basename: |
| 1963 | basename = "%s/%s" % (basename, name) |
| 1964 | else: |
| 1965 | basename = name |
| 1966 | if self.debug: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 1967 | print("Adding package in", pathname, "as", basename) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1968 | fname, arcname = self._get_codename(initname[0:-3], basename) |
| 1969 | if self.debug: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 1970 | print("Adding", arcname) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1971 | self.write(fname, arcname) |
Bernhard M. Wiedemann | 8452104 | 2018-01-31 11:17:10 +0100 | [diff] [blame] | 1972 | dirlist = sorted(os.listdir(pathname)) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1973 | dirlist.remove("__init__.py") |
| 1974 | # Add all *.py files and package subdirectories |
| 1975 | for filename in dirlist: |
| 1976 | path = os.path.join(pathname, filename) |
| 1977 | root, ext = os.path.splitext(filename) |
| 1978 | if os.path.isdir(path): |
| 1979 | if os.path.isfile(os.path.join(path, "__init__.py")): |
| 1980 | # This is a package directory, add it |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 1981 | self.writepy(path, basename, |
| 1982 | filterfunc=filterfunc) # Recursive call |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1983 | elif ext == ".py": |
Christian Tismer | 410d931 | 2013-10-22 04:09:28 +0200 | [diff] [blame] | 1984 | if filterfunc and not filterfunc(path): |
| 1985 | if self.debug: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 1986 | print('file %r skipped by filterfunc' % path) |
Christian Tismer | 410d931 | 2013-10-22 04:09:28 +0200 | [diff] [blame] | 1987 | continue |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1988 | fname, arcname = self._get_codename(path[0:-3], |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 1989 | basename) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1990 | if self.debug: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 1991 | print("Adding", arcname) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1992 | self.write(fname, arcname) |
| 1993 | else: |
| 1994 | # This is NOT a package directory, add its files at top level |
| 1995 | if self.debug: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 1996 | print("Adding files from directory", pathname) |
Bernhard M. Wiedemann | 8452104 | 2018-01-31 11:17:10 +0100 | [diff] [blame] | 1997 | for filename in sorted(os.listdir(pathname)): |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 1998 | path = os.path.join(pathname, filename) |
| 1999 | root, ext = os.path.splitext(filename) |
| 2000 | if ext == ".py": |
Christian Tismer | 410d931 | 2013-10-22 04:09:28 +0200 | [diff] [blame] | 2001 | if filterfunc and not filterfunc(path): |
| 2002 | if self.debug: |
Serhiy Storchaka | b0d497c | 2016-09-10 21:28:07 +0300 | [diff] [blame] | 2003 | print('file %r skipped by filterfunc' % path) |
Christian Tismer | 410d931 | 2013-10-22 04:09:28 +0200 | [diff] [blame] | 2004 | continue |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 2005 | fname, arcname = self._get_codename(path[0:-3], |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 2006 | basename) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 2007 | if self.debug: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 2008 | print("Adding", arcname) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 2009 | self.write(fname, arcname) |
| 2010 | else: |
| 2011 | if pathname[-3:] != ".py": |
Collin Winter | ce36ad8 | 2007-08-30 01:19:48 +0000 | [diff] [blame] | 2012 | raise RuntimeError( |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 2013 | 'Files added with writepy() must end with ".py"') |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 2014 | fname, arcname = self._get_codename(pathname[0:-3], basename) |
| 2015 | if self.debug: |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 2016 | print("Adding file", arcname) |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 2017 | self.write(fname, arcname) |
| 2018 | |
| 2019 | def _get_codename(self, pathname, basename): |
| 2020 | """Return (filename, archivename) for the path. |
| 2021 | |
Fred Drake | 484d735 | 2000-10-02 21:14:52 +0000 | [diff] [blame] | 2022 | Given a module name path, return the correct file path and |
| 2023 | archive name, compiling if necessary. For example, given |
| 2024 | /python/lib/string, return (/python/lib/string.pyc, string). |
| 2025 | """ |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 2026 | def _compile(file, optimize=-1): |
| 2027 | import py_compile |
| 2028 | if self.debug: |
| 2029 | print("Compiling", file) |
| 2030 | try: |
| 2031 | py_compile.compile(file, doraise=True, optimize=optimize) |
Serhiy Storchaka | 45c4375 | 2013-01-29 20:10:28 +0200 | [diff] [blame] | 2032 | except py_compile.PyCompileError as err: |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 2033 | print(err.msg) |
| 2034 | return False |
| 2035 | return True |
| 2036 | |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 2037 | file_py = pathname + ".py" |
| 2038 | file_pyc = pathname + ".pyc" |
Brett Cannon | f299abd | 2015-04-13 14:21:02 -0400 | [diff] [blame] | 2039 | pycache_opt0 = importlib.util.cache_from_source(file_py, optimization='') |
| 2040 | pycache_opt1 = importlib.util.cache_from_source(file_py, optimization=1) |
| 2041 | pycache_opt2 = importlib.util.cache_from_source(file_py, optimization=2) |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 2042 | if self._optimize == -1: |
| 2043 | # legacy mode: use whatever file is present |
Brett Cannon | f299abd | 2015-04-13 14:21:02 -0400 | [diff] [blame] | 2044 | if (os.path.isfile(file_pyc) and |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 2045 | os.stat(file_pyc).st_mtime >= os.stat(file_py).st_mtime): |
| 2046 | # Use .pyc file. |
| 2047 | arcname = fname = file_pyc |
Brett Cannon | f299abd | 2015-04-13 14:21:02 -0400 | [diff] [blame] | 2048 | elif (os.path.isfile(pycache_opt0) and |
| 2049 | os.stat(pycache_opt0).st_mtime >= os.stat(file_py).st_mtime): |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 2050 | # Use the __pycache__/*.pyc file, but write it to the legacy pyc |
| 2051 | # file name in the archive. |
Brett Cannon | f299abd | 2015-04-13 14:21:02 -0400 | [diff] [blame] | 2052 | fname = pycache_opt0 |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 2053 | arcname = file_pyc |
Brett Cannon | f299abd | 2015-04-13 14:21:02 -0400 | [diff] [blame] | 2054 | elif (os.path.isfile(pycache_opt1) and |
| 2055 | os.stat(pycache_opt1).st_mtime >= os.stat(file_py).st_mtime): |
| 2056 | # Use the __pycache__/*.pyc file, but write it to the legacy pyc |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 2057 | # file name in the archive. |
Brett Cannon | f299abd | 2015-04-13 14:21:02 -0400 | [diff] [blame] | 2058 | fname = pycache_opt1 |
| 2059 | arcname = file_pyc |
| 2060 | elif (os.path.isfile(pycache_opt2) and |
| 2061 | os.stat(pycache_opt2).st_mtime >= os.stat(file_py).st_mtime): |
| 2062 | # Use the __pycache__/*.pyc file, but write it to the legacy pyc |
| 2063 | # file name in the archive. |
| 2064 | fname = pycache_opt2 |
| 2065 | arcname = file_pyc |
Barry Warsaw | 28a691b | 2010-04-17 00:19:56 +0000 | [diff] [blame] | 2066 | else: |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 2067 | # Compile py into PEP 3147 pyc file. |
| 2068 | if _compile(file_py): |
Brett Cannon | f299abd | 2015-04-13 14:21:02 -0400 | [diff] [blame] | 2069 | if sys.flags.optimize == 0: |
| 2070 | fname = pycache_opt0 |
| 2071 | elif sys.flags.optimize == 1: |
| 2072 | fname = pycache_opt1 |
| 2073 | else: |
| 2074 | fname = pycache_opt2 |
| 2075 | arcname = file_pyc |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 2076 | else: |
| 2077 | fname = arcname = file_py |
| 2078 | else: |
| 2079 | # new mode: use given optimization level |
| 2080 | if self._optimize == 0: |
Brett Cannon | f299abd | 2015-04-13 14:21:02 -0400 | [diff] [blame] | 2081 | fname = pycache_opt0 |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 2082 | arcname = file_pyc |
| 2083 | else: |
Brett Cannon | f299abd | 2015-04-13 14:21:02 -0400 | [diff] [blame] | 2084 | arcname = file_pyc |
| 2085 | if self._optimize == 1: |
| 2086 | fname = pycache_opt1 |
| 2087 | elif self._optimize == 2: |
| 2088 | fname = pycache_opt2 |
| 2089 | else: |
| 2090 | msg = "invalid value for 'optimize': {!r}".format(self._optimize) |
| 2091 | raise ValueError(msg) |
Georg Brandl | 8334fd9 | 2010-12-04 10:26:46 +0000 | [diff] [blame] | 2092 | if not (os.path.isfile(fname) and |
| 2093 | os.stat(fname).st_mtime >= os.stat(file_py).st_mtime): |
| 2094 | if not _compile(file_py, optimize=self._optimize): |
| 2095 | fname = arcname = file_py |
Barry Warsaw | 28a691b | 2010-04-17 00:19:56 +0000 | [diff] [blame] | 2096 | archivename = os.path.split(arcname)[1] |
Guido van Rossum | 32abe6f | 2000-03-31 17:30:02 +0000 | [diff] [blame] | 2097 | if basename: |
| 2098 | archivename = "%s/%s" % (basename, archivename) |
| 2099 | return (fname, archivename) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 2100 | |
| 2101 | |
Serhiy Storchaka | 8c93310 | 2016-10-23 13:32:12 +0300 | [diff] [blame] | 2102 | def main(args=None): |
| 2103 | import argparse |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 2104 | |
Serhiy Storchaka | 150cd19 | 2017-04-07 18:56:12 +0300 | [diff] [blame] | 2105 | description = 'A simple command-line interface for zipfile module.' |
Serhiy Storchaka | 8c93310 | 2016-10-23 13:32:12 +0300 | [diff] [blame] | 2106 | parser = argparse.ArgumentParser(description=description) |
Serhiy Storchaka | 150cd19 | 2017-04-07 18:56:12 +0300 | [diff] [blame] | 2107 | group = parser.add_mutually_exclusive_group(required=True) |
Serhiy Storchaka | 8c93310 | 2016-10-23 13:32:12 +0300 | [diff] [blame] | 2108 | group.add_argument('-l', '--list', metavar='<zipfile>', |
| 2109 | help='Show listing of a zipfile') |
| 2110 | group.add_argument('-e', '--extract', nargs=2, |
| 2111 | metavar=('<zipfile>', '<output_dir>'), |
| 2112 | help='Extract zipfile into target dir') |
| 2113 | group.add_argument('-c', '--create', nargs='+', |
| 2114 | metavar=('<name>', '<file>'), |
| 2115 | help='Create zipfile from sources') |
| 2116 | group.add_argument('-t', '--test', metavar='<zipfile>', |
| 2117 | help='Test if a zipfile is valid') |
| 2118 | args = parser.parse_args(args) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 2119 | |
Serhiy Storchaka | 8c93310 | 2016-10-23 13:32:12 +0300 | [diff] [blame] | 2120 | if args.test is not None: |
| 2121 | src = args.test |
| 2122 | with ZipFile(src, 'r') as zf: |
Antoine Pitrou | 17babc5 | 2012-11-17 23:50:08 +0100 | [diff] [blame] | 2123 | badfile = zf.testzip() |
Antoine Pitrou | 7c8bcb6 | 2010-08-12 15:11:50 +0000 | [diff] [blame] | 2124 | if badfile: |
| 2125 | print("The following enclosed file is corrupted: {!r}".format(badfile)) |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 2126 | print("Done testing") |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 2127 | |
Serhiy Storchaka | 8c93310 | 2016-10-23 13:32:12 +0300 | [diff] [blame] | 2128 | elif args.list is not None: |
| 2129 | src = args.list |
| 2130 | with ZipFile(src, 'r') as zf: |
| 2131 | zf.printdir() |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 2132 | |
Serhiy Storchaka | 8c93310 | 2016-10-23 13:32:12 +0300 | [diff] [blame] | 2133 | elif args.extract is not None: |
| 2134 | src, curdir = args.extract |
| 2135 | with ZipFile(src, 'r') as zf: |
| 2136 | zf.extractall(curdir) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 2137 | |
Serhiy Storchaka | 8c93310 | 2016-10-23 13:32:12 +0300 | [diff] [blame] | 2138 | elif args.create is not None: |
| 2139 | zip_name = args.create.pop(0) |
| 2140 | files = args.create |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 2141 | |
| 2142 | def addToZip(zf, path, zippath): |
| 2143 | if os.path.isfile(path): |
| 2144 | zf.write(path, zippath, ZIP_DEFLATED) |
| 2145 | elif os.path.isdir(path): |
Serhiy Storchaka | 518e71b | 2014-10-04 13:39:34 +0300 | [diff] [blame] | 2146 | if zippath: |
| 2147 | zf.write(path, zippath) |
Bernhard M. Wiedemann | 8452104 | 2018-01-31 11:17:10 +0100 | [diff] [blame] | 2148 | for nm in sorted(os.listdir(path)): |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 2149 | addToZip(zf, |
Christian Tismer | 59202e5 | 2013-10-21 03:59:23 +0200 | [diff] [blame] | 2150 | os.path.join(path, nm), os.path.join(zippath, nm)) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 2151 | # else: ignore |
| 2152 | |
Serhiy Storchaka | 8c93310 | 2016-10-23 13:32:12 +0300 | [diff] [blame] | 2153 | with ZipFile(zip_name, 'w') as zf: |
| 2154 | for path in files: |
Serhiy Storchaka | 518e71b | 2014-10-04 13:39:34 +0300 | [diff] [blame] | 2155 | zippath = os.path.basename(path) |
| 2156 | if not zippath: |
| 2157 | zippath = os.path.basename(os.path.dirname(path)) |
| 2158 | if zippath in ('', os.curdir, os.pardir): |
| 2159 | zippath = '' |
| 2160 | addToZip(zf, path, zippath) |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 2161 | |
| 2162 | if __name__ == "__main__": |
| 2163 | main() |