Commit | Line | Data |
---|---|---|
be60ffd0 | 1 | #!/usr/bin/env python3 |
7584f5c9 ERE |
2 | #------------------------------------------------------------------- |
3 | # tarfile.py | |
4 | #------------------------------------------------------------------- | |
5 | # Copyright (C) 2002 Lars Gustäbel <lars@gustaebel.de> | |
6 | # All rights reserved. | |
7 | # | |
8 | # Permission is hereby granted, free of charge, to any person | |
9 | # obtaining a copy of this software and associated documentation | |
10 | # files (the "Software"), to deal in the Software without | |
11 | # restriction, including without limitation the rights to use, | |
12 | # copy, modify, merge, publish, distribute, sublicense, and/or sell | |
13 | # copies of the Software, and to permit persons to whom the | |
14 | # Software is furnished to do so, subject to the following | |
15 | # conditions: | |
16 | # | |
17 | # The above copyright notice and this permission notice shall be | |
18 | # included in all copies or substantial portions of the Software. | |
19 | # | |
20 | # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, | |
21 | # EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES | |
22 | # OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND | |
23 | # NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT | |
24 | # HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, | |
25 | # WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING | |
26 | # FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR | |
27 | # OTHER DEALINGS IN THE SOFTWARE. | |
28 | # | |
29 | """Read from and write to tar format archives. | |
30 | """ | |
31 | ||
32 | __version__ = "$Revision: 85213 $" | |
33 | # $Source$ | |
34 | ||
35 | version = "0.9.0" | |
36 | __author__ = "Lars Gustäbel (lars@gustaebel.de)" | |
37 | __date__ = "$Date$" | |
38 | __cvsid__ = "$Id$" | |
5fdff89f | 39 | __credits__ = "Gustavo Niemeyer, Niels Gustäbel, Richard Townsend, Eduardo Robles." |
7584f5c9 ERE |
40 | |
41 | #--------- | |
42 | # Imports | |
43 | #--------- | |
c7c736b6 | 44 | import binascii |
dfd7865e PG |
45 | import copy |
46 | import errno | |
5bd2d4b5 | 47 | import functools |
be60ffd0 | 48 | import io |
dfd7865e PG |
49 | import mmap |
50 | import operator | |
51 | import os | |
52 | import re | |
7584f5c9 ERE |
53 | import shutil |
54 | import stat | |
7584f5c9 | 55 | import struct |
dfd7865e PG |
56 | import sys |
57 | import time | |
7584f5c9 | 58 | |
c7c736b6 PG |
59 | import traceback # XXX |
60 | ||
8ab8fac5 | 61 | from . import crypto |
6e812ad9 | 62 | |
7584f5c9 ERE |
63 | try: |
64 | import grp, pwd | |
65 | except ImportError: | |
66 | grp = pwd = None | |
67 | ||
be60ffd0 ERE |
68 | # os.symlink on Windows prior to 6.0 raises NotImplementedError |
69 | symlink_exception = (AttributeError, NotImplementedError) | |
70 | try: | |
71 | # OSError (winerror=1314) will be raised if the caller does not hold the | |
72 | # SeCreateSymbolicLinkPrivilege privilege | |
73 | symlink_exception += (OSError,) | |
74 | except NameError: | |
75 | pass | |
76 | ||
7584f5c9 ERE |
77 | # from tarfile import * |
78 | __all__ = ["TarFile", "TarInfo", "is_tarfile", "TarError"] | |
79 | ||
be60ffd0 ERE |
80 | from builtins import open as _open # Since 'open' is TarFile.open |
81 | ||
7584f5c9 ERE |
82 | #--------------------------------------------------------- |
83 | # tar constants | |
84 | #--------------------------------------------------------- | |
be60ffd0 | 85 | NUL = b"\0" # the null character |
7584f5c9 ERE |
86 | BLOCKSIZE = 512 # length of processing blocks |
87 | RECORDSIZE = BLOCKSIZE * 20 # length of records | |
be60ffd0 ERE |
88 | GNU_MAGIC = b"ustar \0" # magic gnu tar string |
89 | POSIX_MAGIC = b"ustar\x0000" # magic posix tar string | |
7584f5c9 ERE |
90 | |
91 | LENGTH_NAME = 100 # maximum length of a filename | |
92 | LENGTH_LINK = 100 # maximum length of a linkname | |
93 | LENGTH_PREFIX = 155 # maximum length of the prefix field | |
94 | ||
be60ffd0 ERE |
95 | REGTYPE = b"0" # regular file |
96 | AREGTYPE = b"\0" # regular file | |
97 | LNKTYPE = b"1" # link (inside tarfile) | |
98 | SYMTYPE = b"2" # symbolic link | |
99 | CHRTYPE = b"3" # character special device | |
100 | BLKTYPE = b"4" # block special device | |
101 | DIRTYPE = b"5" # directory | |
102 | FIFOTYPE = b"6" # fifo special device | |
103 | CONTTYPE = b"7" # contiguous file | |
104 | ||
105 | GNUTYPE_LONGNAME = b"L" # GNU tar longname | |
106 | GNUTYPE_LONGLINK = b"K" # GNU tar longlink | |
107 | GNUTYPE_SPARSE = b"S" # GNU tar sparse file | |
108 | GNUTYPE_MULTIVOL = b"M" # GNU tar continuation of a file that began on | |
68ddf955 | 109 | # another volume |
7584f5c9 | 110 | |
be60ffd0 ERE |
111 | XHDTYPE = b"x" # POSIX.1-2001 extended header |
112 | XGLTYPE = b"g" # POSIX.1-2001 global header | |
113 | SOLARIS_XHDTYPE = b"X" # Solaris extended header | |
7584f5c9 ERE |
114 | |
115 | USTAR_FORMAT = 0 # POSIX.1-1988 (ustar) format | |
116 | GNU_FORMAT = 1 # GNU tar format | |
117 | PAX_FORMAT = 2 # POSIX.1-2001 (pax) format | |
118 | DEFAULT_FORMAT = GNU_FORMAT | |
119 | ||
15a81fc0 | 120 | GZ_FMT_HEADER = b"<BBBBLBB" |
203cb25e | 121 | GZ_HEADER_SIZE = 10 # not including the name |
15a81fc0 PG |
122 | GZ_MAGIC = (0x1f, 0x8b) # 0o37, 0o213 |
123 | GZ_METHOD_DEFLATE = 0x08 # 0o10 | |
dfd7865e PG |
124 | GZ_FLAG_FTEXT = 1 << 0 # ASCII payload |
125 | GZ_FLAG_FHCRC = 1 << 1 # CRC16 | |
126 | GZ_FLAG_FEXTRA = 1 << 2 # extra field | |
127 | GZ_FLAG_FNAME = 1 << 3 # set by default in gzip | |
128 | GZ_FLAG_FCOMMENT = 1 << 4 # NUL-terminated comment | |
129 | GZ_FLAG_RESERVED = 7 << 5 # unassigned | |
15a81fc0 PG |
130 | GZ_DEFLATE_FLAGS = 0x00 # 0o00, never read (deflate.c) |
131 | GZ_OS_CODE = 0x03 # 0o03, default in gzip (tailor.h) | |
d601d33b PG |
132 | GZ_MAGIC_BYTES = struct.pack ("<BB", GZ_MAGIC [0], GZ_MAGIC [1]) |
133 | GZ_MAGIC_DEFLATE = struct.pack ("<BBB", GZ_MAGIC [0], GZ_MAGIC [1], | |
134 | GZ_METHOD_DEFLATE) | |
15a81fc0 | 135 | |
04f4c7ab PG |
136 | TOLERANCE_STRICT = 0 |
137 | TOLERANCE_RECOVER = 1 # rely on offsets in index | |
138 | TOLERANCE_RESCUE = 2 # deduce metadata from archive contents | |
139 | ||
dfd7865e PG |
140 | BUFSIZE = 16 * 1024 |
141 | ||
7584f5c9 | 142 | #--------------------------------------------------------- |
d1c38f40 PG |
143 | # archive handling mode |
144 | #--------------------------------------------------------- | |
145 | ||
146 | ARCMODE_PLAIN = 0 | |
147 | ARCMODE_ENCRYPT = 1 << 0 | |
148 | ARCMODE_COMPRESS = 1 << 1 | |
149 | ARCMODE_CONCAT = 1 << 2 | |
150 | ||
151 | def arcmode_fmt (m): | |
152 | if m == ARCMODE_PLAIN: | |
153 | return "PLAIN" | |
154 | first = True | |
155 | ret = "[" | |
156 | def chkappend (b, s): | |
157 | nonlocal m | |
158 | nonlocal ret | |
159 | nonlocal first | |
160 | if m & b: | |
161 | if first is True: first = False | |
162 | else: ret += " |" | |
163 | ret += " " + s | |
164 | chkappend (ARCMODE_ENCRYPT, "ENCRYPT") | |
165 | chkappend (ARCMODE_COMPRESS, "COMPRESS") | |
166 | chkappend (ARCMODE_CONCAT, "CONCAT") | |
167 | return ret + " ]" | |
168 | ||
169 | ||
170 | def arcmode_set (concat=False, encryption=None, comptype=None, init=ARCMODE_PLAIN): | |
171 | ret = init | |
172 | if bool (concat) is True: | |
173 | ret |= ARCMODE_CONCAT | |
174 | if encryption is not None: | |
175 | ret |= ARCMODE_ENCRYPT | |
176 | if comptype == "gz": | |
177 | ret |= ARCMODE_COMPRESS | |
178 | return ret | |
179 | ||
180 | #--------------------------------------------------------- | |
7584f5c9 ERE |
181 | # tarfile constants |
182 | #--------------------------------------------------------- | |
183 | # File types that tarfile supports: | |
184 | SUPPORTED_TYPES = (REGTYPE, AREGTYPE, LNKTYPE, | |
185 | SYMTYPE, DIRTYPE, FIFOTYPE, | |
186 | CONTTYPE, CHRTYPE, BLKTYPE, | |
187 | GNUTYPE_LONGNAME, GNUTYPE_LONGLINK, | |
68ddf955 | 188 | GNUTYPE_SPARSE, GNUTYPE_MULTIVOL) |
7584f5c9 ERE |
189 | |
190 | # File types that will be treated as a regular file. | |
191 | REGULAR_TYPES = (REGTYPE, AREGTYPE, | |
192 | CONTTYPE, GNUTYPE_SPARSE) | |
193 | ||
194 | # File types that are part of the GNU tar format. | |
195 | GNU_TYPES = (GNUTYPE_LONGNAME, GNUTYPE_LONGLINK, | |
68ddf955 | 196 | GNUTYPE_SPARSE, GNUTYPE_MULTIVOL) |
7584f5c9 ERE |
197 | |
198 | # Fields from a pax header that override a TarInfo attribute. | |
199 | PAX_FIELDS = ("path", "linkpath", "size", "mtime", | |
200 | "uid", "gid", "uname", "gname") | |
201 | ||
be60ffd0 ERE |
202 | # Fields from a pax header that are affected by hdrcharset. |
203 | PAX_NAME_FIELDS = {"path", "linkpath", "uname", "gname"} | |
204 | ||
7584f5c9 ERE |
205 | # Fields in a pax header that are numbers, all other fields |
206 | # are treated as strings. | |
207 | PAX_NUMBER_FIELDS = { | |
208 | "atime": float, | |
209 | "ctime": float, | |
210 | "mtime": float, | |
211 | "uid": int, | |
212 | "gid": int, | |
213 | "size": int | |
214 | } | |
215 | ||
216 | #--------------------------------------------------------- | |
7584f5c9 ERE |
217 | # initialization |
218 | #--------------------------------------------------------- | |
be60ffd0 ERE |
219 | |
220 | if os.name in ("nt", "ce"): | |
221 | ENCODING = "utf-8" | |
222 | else: | |
223 | ENCODING = sys.getfilesystemencoding() | |
7584f5c9 ERE |
224 | |
225 | #--------------------------------------------------------- | |
226 | # Some useful functions | |
227 | #--------------------------------------------------------- | |
228 | ||
be60ffd0 ERE |
229 | def stn(s, length, encoding, errors): |
230 | """Convert a string to a null-terminated bytes object. | |
7584f5c9 | 231 | """ |
be60ffd0 | 232 | s = s.encode(encoding, errors) |
7584f5c9 ERE |
233 | return s[:length] + (length - len(s)) * NUL |
234 | ||
be60ffd0 ERE |
235 | def nts(s, encoding, errors): |
236 | """Convert a null-terminated bytes object to a string. | |
7584f5c9 | 237 | """ |
be60ffd0 ERE |
238 | p = s.find(b"\0") |
239 | if p != -1: | |
240 | s = s[:p] | |
241 | return s.decode(encoding, errors) | |
242 | ||
243 | def sbtn(s, length, encoding, errors): | |
244 | """Convert a string or a bunch of bytes to a null-terminated bytes object | |
245 | of specific size. | |
246 | """ | |
247 | if isinstance(s, str): | |
248 | s = s.encode(encoding, errors) | |
249 | return s[:length] + (length - len(s)) * NUL | |
7584f5c9 ERE |
250 | |
251 | def nti(s): | |
252 | """Convert a number field to a python number. | |
253 | """ | |
254 | # There are two possible encodings for a number field, see | |
255 | # itn() below. | |
be60ffd0 ERE |
256 | if s[0] in (0o200, 0o377): |
257 | n = 0 | |
258 | for i in range(len(s) - 1): | |
259 | n <<= 8 | |
260 | n += s[i + 1] | |
261 | if s[0] == 0o377: | |
262 | n = -(256 ** (len(s) - 1) - n) | |
263 | else: | |
7584f5c9 | 264 | try: |
be60ffd0 | 265 | n = int(nts(s, "ascii", "strict") or "0", 8) |
7584f5c9 ERE |
266 | except ValueError: |
267 | raise InvalidHeaderError("invalid header") | |
7584f5c9 ERE |
268 | return n |
269 | ||
270 | def itn(n, digits=8, format=DEFAULT_FORMAT): | |
271 | """Convert a python number to a number field. | |
272 | """ | |
273 | # POSIX 1003.1-1988 requires numbers to be encoded as a string of | |
274 | # octal digits followed by a null-byte, this allows values up to | |
275 | # (8**(digits-1))-1. GNU tar allows storing numbers greater than | |
be60ffd0 ERE |
276 | # that if necessary. A leading 0o200 or 0o377 byte indicate this |
277 | # particular encoding, the following digits-1 bytes are a big-endian | |
278 | # base-256 representation. This allows values up to (256**(digits-1))-1. | |
279 | # A 0o200 byte indicates a positive number, a 0o377 byte a negative | |
280 | # number. | |
b95e7da5 | 281 | n = int(n) |
7584f5c9 | 282 | if 0 <= n < 8 ** (digits - 1): |
b95e7da5 | 283 | s = bytes("%0*o" % (digits - 1, n), "ascii") + NUL |
be60ffd0 ERE |
284 | elif format == GNU_FORMAT and -256 ** (digits - 1) <= n < 256 ** (digits - 1): |
285 | if n >= 0: | |
286 | s = bytearray([0o200]) | |
287 | else: | |
288 | s = bytearray([0o377]) | |
289 | n = 256 ** digits + n | |
7584f5c9 | 290 | |
be60ffd0 ERE |
291 | for i in range(digits - 1): |
292 | s.insert(1, n & 0o377) | |
7584f5c9 | 293 | n >>= 8 |
7584f5c9 | 294 | else: |
be60ffd0 ERE |
295 | raise ValueError("overflow in number field") |
296 | ||
297 | return s | |
7584f5c9 ERE |
298 | |
299 | def calc_chksums(buf): | |
300 | """Calculate the checksum for a member's header by summing up all | |
301 | characters except for the chksum field which is treated as if | |
302 | it was filled with spaces. According to the GNU tar sources, | |
303 | some tars (Sun and NeXT) calculate chksum with signed char, | |
304 | which will be different if there are chars in the buffer with | |
305 | the high bit set. So we calculate two checksums, unsigned and | |
306 | signed. | |
307 | """ | |
be60ffd0 ERE |
308 | unsigned_chksum = 256 + sum(struct.unpack_from("148B8x356B", buf)) |
309 | signed_chksum = 256 + sum(struct.unpack_from("148b8x356b", buf)) | |
7584f5c9 ERE |
310 | return unsigned_chksum, signed_chksum |
311 | ||
312 | def copyfileobj(src, dst, length=None): | |
313 | """Copy length bytes from fileobj src to fileobj dst. | |
314 | If length is None, copy the entire content. | |
315 | """ | |
316 | if length == 0: | |
317 | return | |
318 | if length is None: | |
319 | shutil.copyfileobj(src, dst) | |
320 | return | |
321 | ||
7584f5c9 | 322 | blocks, remainder = divmod(length, BUFSIZE) |
be60ffd0 | 323 | for b in range(blocks): |
7584f5c9 | 324 | buf = src.read(BUFSIZE) |
c474439c | 325 | dst.write(buf) |
7584f5c9 | 326 | if len(buf) < BUFSIZE: |
be60ffd0 | 327 | raise OSError("end of file reached") |
7584f5c9 ERE |
328 | if remainder != 0: |
329 | buf = src.read(remainder) | |
c474439c | 330 | dst.write(buf) |
7584f5c9 | 331 | if len(buf) < remainder: |
be60ffd0 | 332 | raise OSError("end of file reached") |
c7c736b6 | 333 | |
7584f5c9 | 334 | |
7584f5c9 | 335 | def filemode(mode): |
be60ffd0 ERE |
336 | """Deprecated in this location; use stat.filemode.""" |
337 | import warnings | |
338 | warnings.warn("deprecated in favor of stat.filemode", | |
339 | DeprecationWarning, 2) | |
340 | return stat.filemode(mode) | |
7584f5c9 ERE |
341 | |
342 | class TarError(Exception): | |
343 | """Base exception.""" | |
344 | pass | |
345 | class ExtractError(TarError): | |
346 | """General exception for extract errors.""" | |
347 | pass | |
348 | class ReadError(TarError): | |
be60ffd0 | 349 | """Exception for unreadable tar archives.""" |
7584f5c9 ERE |
350 | pass |
351 | class CompressionError(TarError): | |
352 | """Exception for unavailable compression methods.""" | |
353 | pass | |
354 | class StreamError(TarError): | |
355 | """Exception for unsupported operations on stream-like TarFiles.""" | |
356 | pass | |
357 | class HeaderError(TarError): | |
358 | """Base exception for header errors.""" | |
359 | pass | |
360 | class EmptyHeaderError(HeaderError): | |
361 | """Exception for empty headers.""" | |
362 | pass | |
363 | class TruncatedHeaderError(HeaderError): | |
364 | """Exception for truncated headers.""" | |
365 | pass | |
366 | class EOFHeaderError(HeaderError): | |
367 | """Exception for end of file headers.""" | |
368 | pass | |
369 | class InvalidHeaderError(HeaderError): | |
370 | """Exception for invalid headers.""" | |
371 | pass | |
372 | class SubsequentHeaderError(HeaderError): | |
373 | """Exception for missing and invalid extended headers.""" | |
374 | pass | |
8ab8fac5 PG |
375 | class InvalidEncryptionError(TarError): |
376 | """Exception for undefined crypto modes and combinations.""" | |
377 | pass | |
e4e5d0b8 PG |
378 | class DecryptionError(TarError): |
379 | """Exception for error during decryption.""" | |
380 | pass | |
c7c736b6 | 381 | class EncryptionError(TarError): |
e93f83f1 | 382 | """Exception for error during encryption.""" |
c7c736b6 | 383 | pass |
e50fa574 PG |
384 | class EndOfFile(Exception): |
385 | """Signal end of file condition when they’re not an error.""" | |
65b35c42 | 386 | pass |
7584f5c9 ERE |
387 | |
388 | #--------------------------- | |
389 | # internal stream interface | |
390 | #--------------------------- | |
391 | class _LowLevelFile: | |
392 | """Low-level file object. Supports reading and writing. | |
393 | It is used instead of a regular file object for streaming | |
394 | access. | |
395 | """ | |
396 | ||
397 | def __init__(self, name, mode): | |
ad4402e8 | 398 | _mode = { |
7584f5c9 | 399 | "r": os.O_RDONLY, |
c7c736b6 | 400 | "w": os.O_RDWR | os.O_CREAT | os.O_TRUNC, |
7584f5c9 ERE |
401 | }[mode] |
402 | if hasattr(os, "O_BINARY"): | |
f0287fb7 | 403 | _mode |= os.O_BINARY # pylint: disable=no-member |
be60ffd0 | 404 | self.fd = os.open(name, _mode, 0o666) |
ad4402e8 | 405 | self.offset = 0 |
7584f5c9 ERE |
406 | |
407 | def close(self): | |
408 | os.close(self.fd) | |
409 | ||
410 | def read(self, size): | |
ad4402e8 ERE |
411 | ret = os.read(self.fd, size) |
412 | self.offset += len(ret) | |
413 | return ret | |
7584f5c9 | 414 | |
867f75f7 PG |
415 | def write(self, s, pos=None): |
416 | if pos is not None: | |
417 | p0 = self.offset | |
418 | os.lseek (self.fd, pos, os.SEEK_SET) | |
419 | n = os.write(self.fd, s) | |
420 | if pos is None: | |
421 | self.offset += len(s) | |
422 | else: | |
423 | append = pos + n - p0 | |
424 | if append > 0: | |
425 | self.offset += append | |
426 | os.lseek (self.fd, p0, os.SEEK_SET) | |
7584f5c9 | 427 | |
ad4402e8 ERE |
428 | def tell(self): |
429 | return self.offset | |
430 | ||
c7c736b6 PG |
431 | def seek_set (self, pos): |
432 | os.lseek (self.fd, pos, os.SEEK_SET) | |
433 | self.offset = pos | |
434 | ||
8ab8fac5 | 435 | |
15a81fc0 PG |
436 | def gz_header (name=None): |
437 | timestamp = int(time.time()) | |
438 | flags = 0x0 | |
439 | ||
440 | if name is None: | |
441 | name = b"" | |
442 | else: | |
dfd7865e | 443 | flags |= GZ_FLAG_FNAME |
15a81fc0 PG |
444 | if type(name) is str: |
445 | name = name.encode("iso-8859-1", "replace") | |
6e99d23a PG |
446 | if name.endswith(b".pdtcrypt"): |
447 | name = name[:-9] | |
15a81fc0 PG |
448 | if name.endswith(b".gz"): |
449 | name = name[:-3] | |
450 | # RFC1952 says we must use ISO-8859-1 for the FNAME field. | |
451 | name += NUL | |
452 | ||
453 | hdr = struct.pack (GZ_FMT_HEADER, | |
454 | GZ_MAGIC [0], GZ_MAGIC [1], | |
455 | GZ_METHOD_DEFLATE, flags, | |
456 | timestamp, | |
457 | GZ_DEFLATE_FLAGS, GZ_OS_CODE) | |
458 | ||
459 | return hdr + name | |
460 | ||
d601d33b | 461 | |
7584f5c9 ERE |
462 | class _Stream: |
463 | """Class that serves as an adapter between TarFile and | |
464 | a stream-like object. The stream-like object only | |
465 | needs to have a read() or write() method and is accessed | |
466 | blockwise. Use of gzip or bzip2 compression is possible. | |
467 | A stream-like object could be for example: sys.stdin, | |
468 | sys.stdout, a socket, a tape device etc. | |
469 | ||
3031b7ae PG |
470 | _Stream is intended to be used only internally but is |
471 | nevertherless used externally by Deltatar. | |
472 | ||
473 | When encrypting, the ``enccounter`` will be used for | |
474 | initializing the first cryptographic context. When | |
475 | decrypting, its value will be compared to the decrypted | |
476 | object. Decryption fails if the value does not match. | |
477 | In effect, this means that a ``_Stream`` whose ctor was | |
478 | passed ``enccounter`` can only be used to encrypt or | |
479 | decrypt a single object. | |
7584f5c9 ERE |
480 | """ |
481 | ||
c7c736b6 | 482 | remainder = -1 # track size in encrypted entries |
04f4c7ab | 483 | tolerance = TOLERANCE_STRICT |
c7c736b6 | 484 | |
6e812ad9 | 485 | def __init__(self, name, mode, comptype, fileobj, bufsize, |
d1c38f40 | 486 | concat=False, encryption=None, enccounter=None, |
04f4c7ab | 487 | compresslevel=9, tolerance=TOLERANCE_STRICT): |
7584f5c9 ERE |
488 | """Construct a _Stream object. |
489 | """ | |
d1c38f40 | 490 | self.arcmode = arcmode_set (concat, encryption, comptype) |
04f4c7ab | 491 | self.tolerance = tolerance |
d1c38f40 | 492 | |
7584f5c9 ERE |
493 | self._extfileobj = True |
494 | if fileobj is None: | |
495 | fileobj = _LowLevelFile(name, mode) | |
496 | self._extfileobj = False | |
497 | ||
498 | if comptype == '*': | |
499 | # Enable transparent compression detection for the | |
500 | # stream interface | |
501 | fileobj = _StreamProxy(fileobj) | |
502 | comptype = fileobj.getcomptype() | |
d1c38f40 PG |
503 | if comptype == '': |
504 | comptype = "tar" | |
7584f5c9 | 505 | |
3031b7ae PG |
506 | self.enccounter = None |
507 | if self.arcmode & ARCMODE_ENCRYPT: | |
508 | self.enccounter = enccounter | |
509 | ||
7584f5c9 ERE |
510 | self.name = name or "" |
511 | self.mode = mode | |
512 | self.comptype = comptype | |
53732900 | 513 | self.cmp = None |
7584f5c9 ERE |
514 | self.fileobj = fileobj |
515 | self.bufsize = bufsize | |
be60ffd0 ERE |
516 | self.buf = b"" |
517 | self.pos = 0 | |
518 | self.concat_pos = 0 | |
7584f5c9 | 519 | self.closed = False |
be60ffd0 | 520 | self.flags = 0 |
be60ffd0 | 521 | self.last_block_offset = 0 |
e4e5d0b8 | 522 | self.dbuf = b"" # ??? |
46c03c02 | 523 | self.exception = None # communicate decompression failure |
2b82f50c | 524 | self.compresslevel = compresslevel |
784175ba | 525 | self.bytes_written = 0 |
c7c736b6 | 526 | # crypto parameters |
2ae46844 | 527 | self.encryption = encryption |
c7c736b6 | 528 | self.lasthdr = None |
7584f5c9 | 529 | |
b750b280 PG |
530 | if encryption is not None: |
531 | encryption.reset_last_iv () | |
532 | ||
be60ffd0 ERE |
533 | try: |
534 | if comptype == "gz": | |
535 | try: | |
536 | import zlib | |
537 | except ImportError: | |
538 | raise CompressionError("zlib module is not available") | |
539 | self.zlib = zlib | |
bec34b42 PG |
540 | if mode == "r": |
541 | self.exception = zlib.error | |
8ae983c4 | 542 | self._init_read_gz() |
bec34b42 | 543 | elif mode == "w": |
d1c38f40 PG |
544 | if not (self.arcmode & ARCMODE_CONCAT): |
545 | if self.arcmode & ARCMODE_ENCRYPT: | |
3031b7ae | 546 | self._init_write_encrypt (name) |
a0873dcc | 547 | self._init_write_gz () |
c2ffe2ec | 548 | self.crc = zlib.crc32(b"") & 0xFFFFffff |
7584f5c9 | 549 | |
be60ffd0 | 550 | elif comptype == "bz2": |
d1c38f40 | 551 | if self.arcmode & ARCMODE_ENCRYPT: |
2ae46844 | 552 | raise InvalidEncryptionError("encryption not available for " |
d1c38f40 | 553 | "compression “%s”" % comptype) |
be60ffd0 ERE |
554 | try: |
555 | import bz2 | |
556 | except ImportError: | |
557 | raise CompressionError("bz2 module is not available") | |
558 | if mode == "r": | |
559 | self.dbuf = b"" | |
560 | self.cmp = bz2.BZ2Decompressor() | |
561 | self.exception = OSError | |
562 | else: | |
563 | self.cmp = bz2.BZ2Compressor() | |
7584f5c9 | 564 | |
be60ffd0 | 565 | elif comptype == 'xz': |
d1c38f40 | 566 | if self.arcmode & ARCMODE_ENCRYPT: |
2ae46844 | 567 | raise InvalidEncryptionError("encryption not available for " |
d1c38f40 | 568 | "compression “%s”" % comptype) |
c7c736b6 PG |
569 | try: |
570 | import lzma | |
571 | except ImportError: | |
572 | raise CompressionError("lzma module is not available") | |
573 | if mode == "r": | |
574 | self.dbuf = b"" | |
575 | self.cmp = lzma.LZMADecompressor() | |
576 | self.exception = lzma.LZMAError | |
577 | else: | |
578 | self.cmp = lzma.LZMACompressor() | |
579 | ||
6de9444a | 580 | elif comptype == "tar": |
d1c38f40 | 581 | if not (self.arcmode & ARCMODE_CONCAT) \ |
6de9444a | 582 | and mode == "w" \ |
d1c38f40 | 583 | and self.arcmode & ARCMODE_ENCRYPT: |
3031b7ae | 584 | self._init_write_encrypt (name) |
6de9444a PG |
585 | |
586 | else: | |
d1c38f40 | 587 | if self.arcmode & ARCMODE_ENCRYPT: |
2ae46844 | 588 | raise InvalidEncryptionError("encryption not available for " |
d1c38f40 | 589 | "compression “%s”" % comptype) |
c7c736b6 | 590 | raise CompressionError("unknown compression type %r" % comptype) |
be60ffd0 | 591 | |
200d4866 | 592 | except: |
be60ffd0 ERE |
593 | if not self._extfileobj: |
594 | self.fileobj.close() | |
595 | self.closed = True | |
596 | raise | |
ac5e4184 | 597 | |
7584f5c9 ERE |
598 | def __del__(self): |
599 | if hasattr(self, "closed") and not self.closed: | |
fac2cfe1 PG |
600 | try: |
601 | self.close() | |
602 | except crypto.InternalError: | |
603 | # context already finalized due to abort but close() tried | |
604 | # to use it | |
605 | pass | |
7584f5c9 | 606 | |
c7c736b6 | 607 | |
d1c38f40 PG |
608 | def next (self, name): |
609 | if self.arcmode & ARCMODE_COMPRESS: | |
610 | if getattr (self, "cmp", None) is not None: | |
611 | self._finalize_write_gz () | |
0349168a PG |
612 | self.__sync() |
613 | if self.arcmode & ~(ARCMODE_ENCRYPT | ARCMODE_COMPRESS): | |
614 | self.last_block_offset = self.fileobj.tell() | |
d1c38f40 PG |
615 | if self.arcmode & ARCMODE_ENCRYPT: |
616 | self._finalize_write_encrypt () | |
617 | self._init_write_encrypt (name, set_last_block_offset=True) | |
618 | if self.arcmode & ARCMODE_COMPRESS: | |
619 | self._init_write_gz (set_last_block_offset = | |
0349168a | 620 | not (self.arcmode & ARCMODE_ENCRYPT)) |
d1c38f40 PG |
621 | return self.last_block_offset |
622 | ||
623 | ||
624 | def next_volume (self, name): | |
625 | # with non-concat modes, this is taken care by the _Stream | |
626 | # ctor as invoked by the newvol handler | |
627 | if self.arcmode & ARCMODE_COMPRESS: | |
628 | if getattr (self, "cmp", None) is not None: | |
629 | # e. g. compressed PAX header written | |
630 | self._finalize_write_gz () | |
631 | if self.arcmode & ARCMODE_ENCRYPT: | |
632 | self._init_write_encrypt (name) | |
633 | if self.arcmode & ARCMODE_COMPRESS: | |
634 | self._init_write_gz () | |
635 | ||
c7c736b6 | 636 | |
d1c38f40 PG |
637 | def _init_write_encrypt (self, entry=None, set_last_block_offset=False): |
638 | """ | |
639 | Save position for delayed write of header; fill the header location | |
640 | with dummy bytes. | |
641 | """ | |
642 | # first thing, proclaim new object to the encryption context | |
643 | # secondly, assemble the header with the updated parameters | |
644 | # and commit it directly to the underlying stream, bypassing the | |
645 | # encryption layer in .__write(). | |
646 | dummyhdr = self.encryption.next (entry, counter=self.enccounter) | |
647 | if dummyhdr is None: | |
648 | raise EncryptionError ("Crypto.next(): bad dummy header") # XXX | |
649 | self.lasthdr = self.fileobj.tell() | |
650 | self.__write_to_file(dummyhdr) | |
651 | if set_last_block_offset is True: | |
652 | self.last_block_offset = self.lasthdr | |
c7c736b6 PG |
653 | |
654 | ||
655 | def _finalize_write_encrypt (self): | |
656 | """ | |
657 | Seek back to header position, read dummy bytes, finalize crypto | |
658 | obtaining the actual header, write header, seek back to current | |
659 | position. | |
963d0db4 PG |
660 | |
661 | Returns the list of IV fixed parts as used during encryption. | |
c7c736b6 | 662 | """ |
d1c38f40 | 663 | if self.lasthdr is not None: |
c7c736b6 PG |
664 | pos0 = self.fileobj.tell () |
665 | self.fileobj.seek_set (self.lasthdr) | |
dd47d6a2 | 666 | dummy = self.fileobj.read (crypto.PDTCRYPT_HDR_SIZE) |
c7c736b6 PG |
667 | pos1 = self.fileobj.tell () |
668 | dpos = pos1 - self.lasthdr | |
dd47d6a2 | 669 | assert dpos == crypto.PDTCRYPT_HDR_SIZE |
c7c736b6 | 670 | self.fileobj.seek_set (pos0) |
c8c72fe1 | 671 | data, hdr, _ = self.encryption.done (dummy) |
5f38bff6 | 672 | self.__write_to_file(hdr, pos=self.lasthdr) |
c7c736b6 PG |
673 | self.__write_to_file(data) # append remainder of data |
674 | self.lasthdr = -1 | |
675 | ||
676 | ||
57db1546 PG |
677 | def _finalize_write_gz (self): |
678 | if self.cmp is not None: | |
679 | chunk = self.buf + self.cmp.flush() | |
680 | if chunk: | |
681 | if self.comptype == "gz": | |
682 | # The native zlib crc is an unsigned 32-bit integer, but | |
683 | # the Python wrapper implicitly casts that to a signed C | |
684 | # long. So, on a 32-bit box self.crc may "look negative", | |
685 | # while the same crc on a 64-bit box may "look positive". | |
686 | # To avoid irksome warnings from the `struct` module, force | |
687 | # it to look positive on all boxes. | |
688 | chunk += struct.pack("<L", self.crc & 0xffffffff) | |
689 | chunk += struct.pack("<L", self.concat_pos & 0xffffFFFF) | |
690 | self.__enc_write (chunk) | |
15a81fc0 | 691 | self.buf = b"" |
57db1546 PG |
692 | |
693 | ||
a0873dcc | 694 | def _init_write_gz (self, set_last_block_offset=False): |
5fdff89f ERE |
695 | ''' |
696 | Add a new gzip block, closing last one | |
697 | ''' | |
be60ffd0 | 698 | self.concat_pos = 0 |
c2ffe2ec | 699 | self.crc = self.zlib.crc32(b"") & 0xFFFFffff |
6de9444a | 700 | first = self.cmp is None |
2b82f50c ERE |
701 | self.cmp = self.zlib.compressobj(self.compresslevel, |
702 | self.zlib.DEFLATED, | |
703 | -self.zlib.MAX_WBITS, | |
704 | self.zlib.DEF_MEM_LEVEL, | |
705 | 0) | |
6e812ad9 DGM |
706 | |
707 | # if aes, we encrypt after compression | |
6de9444a | 708 | if set_last_block_offset is True: |
ad4402e8 | 709 | self.last_block_offset = self.fileobj.tell() |
6e812ad9 | 710 | |
15a81fc0 | 711 | self.__write(gz_header (self.name if first is True else None)) |
5fdff89f | 712 | |
ac5e4184 | 713 | |
7584f5c9 ERE |
714 | def write(self, s): |
715 | """Write string s to the stream. | |
716 | """ | |
717 | if self.comptype == "gz": | |
c2ffe2ec | 718 | self.crc = self.zlib.crc32(s, self.crc) & 0xFFFFffff |
7584f5c9 | 719 | self.pos += len(s) |
5fdff89f | 720 | self.concat_pos += len(s) |
53732900 | 721 | if self.cmp is not None: |
7584f5c9 ERE |
722 | s = self.cmp.compress(s) |
723 | self.__write(s) | |
724 | ||
c7c736b6 | 725 | def __sync(self): |
cb7a3911 | 726 | """Write what’s left in the buffer to the stream.""" |
c7c736b6 PG |
727 | self.__write (b"") # → len (buf) <= bufsiz |
728 | self.__enc_write (self.buf) | |
729 | self.buf = b"" | |
730 | ||
7584f5c9 | 731 | def __write(self, s): |
548bb8d5 CH |
732 | """Writes (and encodes) string s to the stream blockwise |
733 | ||
734 | will wait with encoding/writing until block is complete | |
7584f5c9 ERE |
735 | """ |
736 | self.buf += s | |
737 | while len(self.buf) > self.bufsize: | |
6e812ad9 | 738 | self.__enc_write(self.buf[:self.bufsize]) |
7584f5c9 ERE |
739 | self.buf = self.buf[self.bufsize:] |
740 | ||
867f75f7 | 741 | |
5f38bff6 | 742 | def __write_to_file(self, s, pos=None): |
6e812ad9 | 743 | ''' |
5f38bff6 | 744 | Writes directly to the fileobj; updates self.bytes_written. If “pos” is |
cb7a3911 | 745 | given, the stream will seek to that position first and back afterwards, |
5f38bff6 | 746 | and the total of bytes written is not updated. |
6e812ad9 | 747 | ''' |
867f75f7 | 748 | self.fileobj.write(s, pos) |
5f38bff6 PG |
749 | if pos is None: |
750 | self.bytes_written += len(s) | |
867f75f7 | 751 | |
6e812ad9 DGM |
752 | |
753 | def __enc_write(self, s): | |
cb7a3911 PG |
754 | """ |
755 | If encryption is active, the string s is encrypted before being written | |
756 | to the file. | |
757 | """ | |
758 | if len (s) == 0: | |
759 | return | |
d1c38f40 | 760 | if self.arcmode & ARCMODE_ENCRYPT: |
cb7a3911 PG |
761 | buf = s |
762 | while len (buf) > 0: | |
763 | n, ct = self.encryption.process(buf) | |
764 | self.__write_to_file(ct) | |
765 | buf = buf [n:] | |
766 | if len (buf) > 0: | |
767 | # The entire plaintext was not consumed: The size limit | |
768 | # for encrypted objects was reached. Transparently create | |
769 | # a new encrypted object and continue processing the input. | |
770 | self._finalize_write_encrypt () | |
771 | self._init_write_encrypt () | |
772 | else: | |
773 | self.__write_to_file(s) | |
774 | ||
6e812ad9 | 775 | |
784175ba CH |
776 | def estim_file_size(self): |
777 | """ estimates size of file if closing it now | |
778 | ||
779 | The result may differ greatly from the amount of data sent to write() | |
780 | due to compression, encryption and buffering. | |
781 | ||
782 | In tests the result (before calling close()) was up to 12k smaller than | |
783 | the final file size if compression is being used because zlib/bz2 | |
784 | compressors do not allow inspection of their buffered data :-( | |
785 | ||
ba5a449e CH |
786 | Still, we add what close() would add: 8 bytes for gz checksum, one |
787 | encryption block size if encryption is used and the size of our own | |
788 | buffer | |
784175ba CH |
789 | """ |
790 | if self.closed: | |
791 | return self.bytes_written | |
792 | ||
793 | result = self.bytes_written | |
794 | if self.buf: | |
795 | result += len(self.buf) | |
796 | if self.comptype == 'gz': | |
ba5a449e | 797 | result += 8 # 2 longs = 8 byte (no extra info written for bzip2) |
784175ba CH |
798 | return result |
799 | ||
5fdff89f | 800 | def close(self, close_fileobj=True): |
7584f5c9 ERE |
801 | """Close the _Stream object. No operation should be |
802 | done on it afterwards. | |
803 | """ | |
963d0db4 | 804 | |
7584f5c9 ERE |
805 | if self.closed: |
806 | return | |
807 | ||
963d0db4 | 808 | if close_fileobj is True: |
a0873dcc | 809 | |
ae3d0f2a | 810 | if self.mode == "w": |
d1c38f40 | 811 | if self.arcmode & ARCMODE_COMPRESS: |
a0873dcc | 812 | self._finalize_write_gz () |
ae3d0f2a | 813 | # end of Tar archive marker (two empty blocks) was written |
267bc643 PG |
814 | # finalize encryption last; no writes may be performed after |
815 | # this point | |
cb7a3911 | 816 | self.__sync () |
d1c38f40 PG |
817 | if self.arcmode & ARCMODE_ENCRYPT: |
818 | self._finalize_write_encrypt () | |
267bc643 | 819 | |
963d0db4 PG |
820 | if not self._extfileobj: |
821 | self.fileobj.close() | |
822 | else: | |
823 | # read the zlib crc and length and check them | |
824 | if self.mode == "r" and self.comptype == "gz": | |
825 | read_crc = self.__read(4) | |
826 | read_length = self.__read(4) | |
827 | calculated_crc = self.crc | |
828 | if struct.unpack("<L", read_crc)[0] != calculated_crc: | |
829 | raise CompressionError("bad gzip crc") | |
7584f5c9 ERE |
830 | self.closed = True |
831 | ||
54128a00 | 832 | |
7584f5c9 ERE |
833 | def _init_read_gz(self): |
834 | """Initialize for reading a gzip compressed fileobj. | |
835 | """ | |
836 | self.cmp = self.zlib.decompressobj(-self.zlib.MAX_WBITS) | |
7584f5c9 | 837 | |
85737f48 | 838 | read2 = self.__read(2) |
e50fa574 PG |
839 | if read2 == b"": |
840 | raise EndOfFile ("_init_read_gz(): read returned zero bytes at pos " | |
841 | "%d" % self.fileobj.tell()) | |
842 | # taken from gzip.GzipFile with some alterations | |
d601d33b | 843 | if read2 != GZ_MAGIC_BYTES: |
7584f5c9 | 844 | raise ReadError("not a gzip file") |
85737f48 | 845 | |
5bd2d4b5 PG |
846 | read1 = self.__read(1) |
847 | if read1 == b"": | |
848 | raise EndOfFile ("_init_read_gz(): read returned zero bytes inside " | |
849 | "gzip header at pos %d" % self.fileobj.tell()) | |
850 | if ord (read1) != GZ_METHOD_DEFLATE: | |
7584f5c9 ERE |
851 | raise CompressionError("unsupported compression method") |
852 | ||
85737f48 | 853 | self.flags = flag = ord(self.__read(1)) |
dfd7865e | 854 | self.__read(6) # discard timestamp[4], deflate flags, os code |
7584f5c9 | 855 | |
dfd7865e | 856 | if flag & GZ_FLAG_FEXTRA: |
7584f5c9 ERE |
857 | xlen = ord(self.__read(1)) + 256 * ord(self.__read(1)) |
858 | self.read(xlen) | |
dfd7865e | 859 | if flag & GZ_FLAG_FNAME: |
7584f5c9 ERE |
860 | while True: |
861 | s = self.__read(1) | |
862 | if not s or s == NUL: | |
863 | break | |
dfd7865e | 864 | if flag & GZ_FLAG_FCOMMENT: |
7584f5c9 ERE |
865 | while True: |
866 | s = self.__read(1) | |
867 | if not s or s == NUL: | |
868 | break | |
dfd7865e | 869 | if flag & GZ_FLAG_FHCRC: |
7584f5c9 ERE |
870 | self.__read(2) |
871 | ||
c7c736b6 PG |
872 | def _init_read_encrypt (self): |
873 | """Initialize encryption for next entry in archive. Read a header and | |
874 | notify the crypto context.""" | |
d1c38f40 | 875 | if self.arcmode & ARCMODE_ENCRYPT: |
6e99d23a | 876 | lasthdr = self.fileobj.tell () |
15d3eefd PG |
877 | try: |
878 | hdr = crypto.hdr_read_stream (self.fileobj) | |
8a8ac469 PG |
879 | except crypto.EndOfFile: |
880 | return False | |
6e99d23a | 881 | except crypto.InvalidHeader as exn: |
c7c736b6 | 882 | raise DecryptionError ("Crypto.hdr_read_stream(): error “%s” " |
6e99d23a PG |
883 | "processing %r at pos %d" |
884 | % (exn, self.fileobj, lasthdr)) \ | |
ae3d0f2a | 885 | from exn |
3031b7ae PG |
886 | if self.enccounter is not None: |
887 | # enforce that the iv counter in the header matches an | |
888 | # explicitly requested one | |
889 | iv = crypto.hdr_iv_counter (hdr) | |
890 | if iv != self.enccounter: | |
891 | raise DecryptionError ("expected IV counter %d, got %d" | |
892 | % (self.enccounter, iv)) | |
6e99d23a | 893 | self.lasthdr = lasthdr |
c7c736b6 | 894 | self.remainder = hdr ["ctsize"] # distance to next header |
1ed44e7b PG |
895 | try: |
896 | self.encryption.next (hdr) | |
897 | except crypto.InvalidParameter as exn: | |
898 | raise DecryptionError ("Crypto.next(): error “%s” " | |
899 | "processing %r at pos %d" | |
900 | % (exn, self.fileobj, lasthdr)) \ | |
901 | from exn | |
8a8ac469 PG |
902 | |
903 | return True | |
c7c736b6 PG |
904 | |
905 | ||
8de91f4f PG |
906 | def _read_encrypt (self, buf): |
907 | """ | |
908 | Demote a program error to a decryption error in tolerant mode. This | |
909 | allows recovery from corrupted headers and invalid data. | |
910 | """ | |
911 | try: | |
912 | return self.encryption.process (buf) | |
913 | except RuntimeError as exn: | |
04f4c7ab | 914 | if self.tolerance != TOLERANCE_STRICT: |
8de91f4f PG |
915 | raise DecryptionError (exn) |
916 | raise | |
917 | ||
918 | ||
c7c736b6 PG |
919 | def _finalize_read_encrypt (self): |
920 | """ | |
921 | Finalize decryption. | |
922 | """ | |
d1c38f40 PG |
923 | if self.arcmode & ARCMODE_ENCRYPT \ |
924 | and self.lasthdr is not None : | |
c7c736b6 PG |
925 | assert self.remainder >= 0 |
926 | if self.remainder > 0: | |
c7c736b6 | 927 | self.remainder = 0 |
b0078f26 PG |
928 | try: |
929 | data = self.encryption.done () | |
930 | except crypto.InvalidGCMTag as exn: | |
931 | raise DecryptionError ("decryption failed: %s" % exn) | |
c7c736b6 PG |
932 | return data |
933 | ||
934 | ||
7584f5c9 ERE |
935 | def tell(self): |
936 | """Return the stream's file pointer position. | |
937 | """ | |
938 | return self.pos | |
939 | ||
940 | def seek(self, pos=0): | |
941 | """Set the stream's file pointer to pos. Negative seeking | |
942 | is forbidden. | |
943 | """ | |
b750b280 PG |
944 | if pos == self.pos: |
945 | pass # nothing to do | |
946 | elif pos - self.pos >= 0: | |
7584f5c9 | 947 | blocks, remainder = divmod(pos - self.pos, self.bufsize) |
b750b280 PG |
948 | if self.encryption is not None: |
949 | # IV succession is only preserved between successive objects. | |
950 | self.encryption.reset_last_iv () | |
be60ffd0 | 951 | for i in range(blocks): |
7584f5c9 ERE |
952 | self.read(self.bufsize) |
953 | self.read(remainder) | |
954 | else: | |
955 | raise StreamError("seeking backwards is not allowed") | |
956 | return self.pos | |
957 | ||
958 | def read(self, size=None): | |
959 | """Return the next size number of bytes from the stream. | |
960 | If size is not defined, return all bytes of the stream | |
961 | up to EOF. | |
962 | """ | |
963 | if size is None: | |
964 | t = [] | |
965 | while True: | |
966 | buf = self._read(self.bufsize) | |
967 | if not buf: | |
968 | break | |
969 | t.append(buf) | |
9dc7ac5c | 970 | buf = b"".join(t) |
7584f5c9 ERE |
971 | else: |
972 | buf = self._read(size) | |
973 | self.pos += len(buf) | |
974 | return buf | |
975 | ||
3a7e1a50 ERE |
976 | def readline(self): |
977 | """Reads just one line, new line character included | |
978 | """ | |
f0fd5e3a | 979 | # if \n in dbuf, no read neads to be done |
be60ffd0 ERE |
980 | if b'\n' in self.dbuf: |
981 | pos = self.dbuf.index(b'\n') + 1 | |
f0fd5e3a ERE |
982 | ret = self.dbuf[:pos] |
983 | self.dbuf = self.dbuf[pos:] | |
984 | return ret | |
985 | ||
1215b602 | 986 | buf = [] |
3a7e1a50 ERE |
987 | while True: |
988 | chunk = self._read(self.bufsize) | |
989 | ||
f0fd5e3a | 990 | # nothing more to read, so return the buffer |
3a7e1a50 | 991 | if not chunk: |
be60ffd0 | 992 | return b''.join(buf) |
3a7e1a50 ERE |
993 | |
994 | buf.append(chunk) | |
f0fd5e3a ERE |
995 | |
996 | # if \n found, return the new line | |
be60ffd0 ERE |
997 | if b'\n' in chunk: |
998 | dbuf = b''.join(buf) | |
999 | pos = dbuf.index(b'\n') + 1 | |
1215b602 | 1000 | self.dbuf = dbuf[pos:] + self.dbuf |
3a7e1a50 ERE |
1001 | return dbuf[:pos] |
1002 | ||
7584f5c9 ERE |
1003 | def _read(self, size): |
1004 | """Return size bytes from the stream. | |
1005 | """ | |
7584f5c9 ERE |
1006 | c = len(self.dbuf) |
1007 | t = [self.dbuf] | |
e4e5d0b8 | 1008 | |
7584f5c9 | 1009 | while c < size: |
867f75f7 | 1010 | buf = self.__read(self.bufsize) |
7584f5c9 ERE |
1011 | if not buf: |
1012 | break | |
3a7e1a50 | 1013 | |
53732900 | 1014 | if self.cmp is not None: |
85737f48 | 1015 | try: |
3a7e1a50 | 1016 | buf = self.cmp.decompress(buf) |
54128a00 PG |
1017 | except self.exception as exn: |
1018 | raise ReadError("invalid compressed data (%r)" % exn) | |
be60ffd0 | 1019 | except Exception as e: |
04fb06f4 DGM |
1020 | # happens at the end of the file |
1021 | # _init_read_gz failed in the previous iteration so | |
e4e5d0b8 | 1022 | # self.cmp.decompress fails here |
d1c38f40 | 1023 | if self.arcmode & ARCMODE_CONCAT: |
be60ffd0 ERE |
1024 | pass |
1025 | else: | |
1026 | raise ReadError("invalid compressed data") | |
d1c38f40 | 1027 | if self.arcmode & ARCMODE_COMPRESS and hasattr(self, "crc"): |
c2ffe2ec | 1028 | self.crc = self.zlib.crc32(buf, self.crc) & 0xFFFFffff |
d1c38f40 PG |
1029 | if self.arcmode & ARCMODE_CONCAT \ |
1030 | and len(self.cmp.unused_data) != 0: | |
3a7e1a50 ERE |
1031 | self.buf = self.cmp.unused_data + self.buf |
1032 | self.close(close_fileobj=False) | |
1033 | try: | |
1034 | self._init_read_gz() | |
8de91f4f | 1035 | except DecryptionError: |
04f4c7ab | 1036 | if self.tolerance != TOLERANCE_STRICT: |
8de91f4f PG |
1037 | # return whatever data was processed successfully |
1038 | if len (buf) > 0: | |
1039 | t.append (buf) | |
1040 | if len (t) > 0: | |
1041 | break | |
24afaf18 PG |
1042 | raise |
1043 | except ReadError: # gzip troubles | |
1044 | if self.tolerance == TOLERANCE_RESCUE: | |
1045 | if len (buf) > 0: | |
1046 | t.append (buf) | |
1047 | if len (t) > 0: | |
1048 | break | |
1049 | raise | |
e50fa574 | 1050 | except EndOfFile: |
3a7e1a50 ERE |
1051 | # happens at the end of the file |
1052 | pass | |
c2ffe2ec | 1053 | self.crc = self.zlib.crc32(b"") & 0xFFFFffff |
3a7e1a50 | 1054 | self.closed = False |
7584f5c9 ERE |
1055 | t.append(buf) |
1056 | c += len(buf) | |
be60ffd0 | 1057 | t = b"".join(t) |
7584f5c9 ERE |
1058 | self.dbuf = t[size:] |
1059 | return t[:size] | |
1060 | ||
e4e5d0b8 | 1061 | |
7584f5c9 | 1062 | def __read(self, size): |
ef3b4499 PG |
1063 | """ |
1064 | Return size bytes from stream. If internal buffer is empty, read | |
1065 | another block from the stream. | |
1066 | ||
1067 | The function returns up to size bytes of data. When an error occurs | |
1068 | during decryption, everything until the end of the last successfully | |
1069 | finalized object is returned. | |
7584f5c9 ERE |
1070 | """ |
1071 | c = len(self.buf) | |
8de91f4f | 1072 | t = [self.buf] if c > 0 else [] |
1ed44e7b | 1073 | good_crypto = len (t) |
8de91f4f | 1074 | |
7584f5c9 | 1075 | while c < size: |
c7c736b6 | 1076 | todo = size |
8de91f4f PG |
1077 | try: |
1078 | if self.arcmode & ARCMODE_ENCRYPT: | |
1079 | if self.remainder <= 0: | |
1080 | # prepare next object | |
044585c6 PG |
1081 | if self._init_read_encrypt () is False: # EOF |
1082 | buf = None | |
1083 | break # while | |
8de91f4f PG |
1084 | |
1085 | # only read up to the end of the encrypted object | |
1086 | todo = min (size, self.remainder) | |
1087 | buf = self.fileobj.read(todo) | |
1088 | if self.arcmode & ARCMODE_ENCRYPT: | |
1089 | # decrypt the thing | |
1090 | buf = self._read_encrypt (buf) | |
1091 | if todo == self.remainder: | |
1092 | # at the end of a crypto object; finalization will fail if | |
1093 | # the GCM tag does not match | |
ef3b4499 | 1094 | trailing = self._finalize_read_encrypt () |
8de91f4f PG |
1095 | good_crypto = len (t) + 1 |
1096 | if len (trailing) > 0: | |
1097 | buf += trailing | |
1098 | self.remainder = 0 | |
1099 | else: | |
1100 | self.remainder -= todo | |
1101 | except DecryptionError: | |
04f4c7ab | 1102 | if self.tolerance == TOLERANCE_STRICT: |
8de91f4f PG |
1103 | raise |
1104 | self.encryption.drop () | |
24afaf18 PG |
1105 | if self.tolerance == TOLERANCE_RECOVER: |
1106 | if good_crypto == 0: | |
1107 | raise | |
1108 | # this may occur at any of the three crypto operations above. | |
1109 | # some objects did validate; discard all data after it; next | |
1110 | # call will start with the bad object and error out immediately | |
1111 | self.buf = b"".join (t [good_crypto:]) | |
1112 | return b"".join (t [:good_crypto]) | |
1113 | elif self.tolerance == TOLERANCE_RESCUE: | |
1114 | # keep what we have so far despite the finalization issue | |
1115 | t.append (buf) | |
1116 | c += len (buf) | |
1117 | break | |
1118 | else: | |
1119 | raise RuntimeError("internal error: bad tolerance level") | |
c7c736b6 PG |
1120 | |
1121 | if not buf: ## XXX stream terminated prematurely; this should be an error | |
7584f5c9 | 1122 | break |
c7c736b6 | 1123 | |
7584f5c9 ERE |
1124 | t.append(buf) |
1125 | c += len(buf) | |
be60ffd0 | 1126 | t = b"".join(t) |
7584f5c9 | 1127 | self.buf = t[size:] |
fb27c6e8 | 1128 | |
7584f5c9 | 1129 | return t[:size] |
7d372216 | 1130 | |
7584f5c9 ERE |
1131 | |
1132 | class _StreamProxy(object): | |
1133 | """Small proxy class that enables transparent compression | |
1134 | detection for the Stream interface (mode 'r|*'). | |
1135 | """ | |
1136 | ||
1137 | def __init__(self, fileobj): | |
1138 | self.fileobj = fileobj | |
1139 | self.buf = self.fileobj.read(BLOCKSIZE) | |
1140 | ||
f0287fb7 | 1141 | def read(self, size): # pylint: disable=method-hidden |
7584f5c9 ERE |
1142 | self.read = self.fileobj.read |
1143 | return self.buf | |
1144 | ||
1145 | def getcomptype(self): | |
d601d33b | 1146 | if self.buf.startswith(GZ_MAGIC_DEFLATE): |
7584f5c9 | 1147 | return "gz" |
be60ffd0 | 1148 | elif self.buf[0:3] == b"BZh" and self.buf[4:10] == b"1AY&SY": |
7584f5c9 | 1149 | return "bz2" |
be60ffd0 ERE |
1150 | elif self.buf.startswith((b"\x5d\x00\x00\x80", b"\xfd7zXZ")): |
1151 | return "xz" | |
1152 | else: | |
1153 | return "tar" | |
7584f5c9 ERE |
1154 | |
1155 | def close(self): | |
1156 | self.fileobj.close() | |
1157 | # class StreamProxy | |
1158 | ||
7584f5c9 ERE |
1159 | #------------------------ |
1160 | # Extraction file object | |
1161 | #------------------------ | |
1162 | class _FileInFile(object): | |
1163 | """A thin wrapper around an existing file object that | |
1164 | provides a part of its data as an individual file | |
1165 | object. | |
1166 | """ | |
1167 | ||
be60ffd0 | 1168 | def __init__(self, fileobj, offset, size, blockinfo=None): |
7584f5c9 ERE |
1169 | self.fileobj = fileobj |
1170 | self.offset = offset | |
1171 | self.size = size | |
7584f5c9 | 1172 | self.position = 0 |
be60ffd0 ERE |
1173 | self.name = getattr(fileobj, "name", None) |
1174 | self.closed = False | |
1175 | ||
1176 | if blockinfo is None: | |
1177 | blockinfo = [(0, size)] | |
1178 | ||
1179 | # Construct a map with data and zero blocks. | |
1180 | self.map_index = 0 | |
1181 | self.map = [] | |
1182 | lastpos = 0 | |
1183 | realpos = self.offset | |
1184 | for offset, size in blockinfo: | |
1185 | if offset > lastpos: | |
1186 | self.map.append((False, lastpos, offset, None)) | |
1187 | self.map.append((True, offset, offset + size, realpos)) | |
1188 | realpos += size | |
1189 | lastpos = offset + size | |
1190 | if lastpos < self.size: | |
1191 | self.map.append((False, lastpos, self.size, None)) | |
1192 | ||
1193 | def flush(self): | |
1194 | pass | |
1195 | ||
1196 | def readable(self): | |
1197 | return True | |
1198 | ||
1199 | def writable(self): | |
1200 | return False | |
1201 | ||
1202 | def seekable(self): | |
1203 | return self.fileobj.seekable() | |
7584f5c9 ERE |
1204 | |
1205 | def tell(self): | |
1206 | """Return the current file position. | |
1207 | """ | |
1208 | return self.position | |
1209 | ||
be60ffd0 | 1210 | def seek(self, position, whence=io.SEEK_SET): |
7584f5c9 ERE |
1211 | """Seek to a position in the file. |
1212 | """ | |
be60ffd0 ERE |
1213 | if whence == io.SEEK_SET: |
1214 | self.position = min(max(position, 0), self.size) | |
1215 | elif whence == io.SEEK_CUR: | |
1216 | if position < 0: | |
1217 | self.position = max(self.position + position, 0) | |
1218 | else: | |
1219 | self.position = min(self.position + position, self.size) | |
1220 | elif whence == io.SEEK_END: | |
1221 | self.position = max(min(self.size + position, self.size), 0) | |
1222 | else: | |
1223 | raise ValueError("Invalid argument") | |
1224 | return self.position | |
7584f5c9 ERE |
1225 | |
1226 | def read(self, size=None): | |
1227 | """Read data from the file. | |
1228 | """ | |
1229 | if size is None: | |
1230 | size = self.size - self.position | |
1231 | else: | |
1232 | size = min(size, self.size - self.position) | |
1233 | ||
be60ffd0 | 1234 | buf = b"" |
7584f5c9 | 1235 | while size > 0: |
7584f5c9 | 1236 | while True: |
be60ffd0 ERE |
1237 | data, start, stop, offset = self.map[self.map_index] |
1238 | if start <= self.position < stop: | |
7584f5c9 | 1239 | break |
be60ffd0 ERE |
1240 | else: |
1241 | self.map_index += 1 | |
1242 | if self.map_index == len(self.map): | |
1243 | self.map_index = 0 | |
1244 | length = min(size, stop - self.position) | |
1245 | if data: | |
1246 | self.fileobj.seek(offset + (self.position - start)) | |
1247 | buf += self.fileobj.read(length) | |
7584f5c9 | 1248 | else: |
be60ffd0 ERE |
1249 | buf += NUL * length |
1250 | size -= length | |
1251 | self.position += length | |
1252 | return buf | |
7584f5c9 | 1253 | |
be60ffd0 ERE |
1254 | def readinto(self, b): |
1255 | buf = self.read(len(b)) | |
1256 | b[:len(buf)] = buf | |
1257 | return len(buf) | |
7584f5c9 ERE |
1258 | |
1259 | def close(self): | |
7584f5c9 | 1260 | self.closed = True |
be60ffd0 | 1261 | #class _FileInFile |
7584f5c9 | 1262 | |
be60ffd0 ERE |
1263 | |
1264 | class ExFileObject(io.BufferedReader): | |
1265 | ||
1266 | def __init__(self, tarfile, tarinfo): | |
1267 | fileobj = _FileInFile(tarfile.fileobj, tarinfo.offset_data, | |
1268 | tarinfo.size, tarinfo.sparse) | |
1269 | super().__init__(fileobj) | |
7584f5c9 ERE |
1270 | #class ExFileObject |
1271 | ||
1272 | #------------------ | |
1273 | # Exported Classes | |
1274 | #------------------ | |
1275 | class TarInfo(object): | |
1276 | """Informational class which holds the details about an | |
1277 | archive member given by a tar header block. | |
1278 | TarInfo objects are returned by TarFile.getmember(), | |
1279 | TarFile.getmembers() and TarFile.gettarinfo() and are | |
1280 | usually created internally. | |
1281 | """ | |
1282 | ||
be60ffd0 ERE |
1283 | __slots__ = ("name", "mode", "uid", "gid", "size", "mtime", |
1284 | "chksum", "type", "linkname", "uname", "gname", | |
1285 | "devmajor", "devminor", "volume_offset", | |
1286 | "offset", "offset_data", "pax_headers", "sparse", | |
1287 | "tarfile", "_sparse_structs", "_link_target") | |
1288 | ||
7584f5c9 ERE |
1289 | def __init__(self, name=""): |
1290 | """Construct a TarInfo object. name is the optional name | |
1291 | of the member. | |
1292 | """ | |
1293 | self.name = name # member name | |
be60ffd0 | 1294 | self.mode = 0o644 # file permissions |
7584f5c9 ERE |
1295 | self.uid = 0 # user id |
1296 | self.gid = 0 # group id | |
1297 | self.size = 0 # file size | |
1298 | self.mtime = 0 # modification time | |
1299 | self.chksum = 0 # header checksum | |
1300 | self.type = REGTYPE # member type | |
1301 | self.linkname = "" # link name | |
1302 | self.uname = "" # user name | |
1303 | self.gname = "" # group name | |
1304 | self.devmajor = 0 # device major number | |
1305 | self.devminor = 0 # device minor number | |
1306 | ||
1307 | self.offset = 0 # the tar header starts here | |
1308 | self.offset_data = 0 # the file's data starts here | |
0eb5048f ERE |
1309 | self.volume_offset = 0 # the file's data corresponds with the data |
1310 | # starting at this position | |
7584f5c9 | 1311 | |
be60ffd0 | 1312 | self.sparse = None # sparse member information |
7584f5c9 ERE |
1313 | self.pax_headers = {} # pax header information |
1314 | ||
1315 | # In pax headers the "name" and "linkname" field are called | |
1316 | # "path" and "linkpath". | |
1317 | def _getpath(self): | |
1318 | return self.name | |
1319 | def _setpath(self, name): | |
1320 | self.name = name | |
1321 | path = property(_getpath, _setpath) | |
1322 | ||
1323 | def _getlinkpath(self): | |
1324 | return self.linkname | |
1325 | def _setlinkpath(self, linkname): | |
1326 | self.linkname = linkname | |
1327 | linkpath = property(_getlinkpath, _setlinkpath) | |
1328 | ||
1329 | def __repr__(self): | |
1330 | return "<%s %r at %#x>" % (self.__class__.__name__,self.name,id(self)) | |
1331 | ||
be60ffd0 | 1332 | def get_info(self, encoding=None, errors=None): |
7584f5c9 ERE |
1333 | """Return the TarInfo's attributes as a dictionary. |
1334 | """ | |
1335 | info = { | |
1336 | "name": self.name, | |
be60ffd0 | 1337 | "mode": self.mode & 0o7777, |
7584f5c9 ERE |
1338 | "uid": self.uid, |
1339 | "gid": self.gid, | |
1340 | "size": self.size, | |
1341 | "mtime": self.mtime, | |
1342 | "chksum": self.chksum, | |
1343 | "type": self.type, | |
1344 | "linkname": self.linkname, | |
1345 | "uname": self.uname, | |
1346 | "gname": self.gname, | |
1347 | "devmajor": self.devmajor, | |
36a315a0 | 1348 | "devminor": self.devminor, |
0eb5048f ERE |
1349 | "offset_data": self.offset_data, |
1350 | "volume_offset": self.volume_offset | |
7584f5c9 ERE |
1351 | } |
1352 | ||
1353 | if info["type"] == DIRTYPE and not info["name"].endswith("/"): | |
1354 | info["name"] += "/" | |
1355 | ||
7584f5c9 ERE |
1356 | return info |
1357 | ||
be60ffd0 ERE |
1358 | def tobuf(self, format=DEFAULT_FORMAT, encoding=ENCODING, |
1359 | errors="surrogateescape"): | |
7584f5c9 ERE |
1360 | """Return a tar header as a string of 512 byte blocks. |
1361 | """ | |
1362 | info = self.get_info(encoding, errors) | |
1363 | ||
1364 | if format == USTAR_FORMAT: | |
be60ffd0 | 1365 | return self.create_ustar_header(info, encoding, errors) |
7584f5c9 | 1366 | elif format == GNU_FORMAT: |
be60ffd0 | 1367 | return self.create_gnu_header(info, encoding, errors) |
7584f5c9 ERE |
1368 | elif format == PAX_FORMAT: |
1369 | return self.create_pax_header(info, encoding, errors) | |
1370 | else: | |
1371 | raise ValueError("invalid format") | |
1372 | ||
be60ffd0 | 1373 | def create_ustar_header(self, info, encoding, errors): |
7584f5c9 ERE |
1374 | """Return the object as a ustar header block. |
1375 | """ | |
1376 | info["magic"] = POSIX_MAGIC | |
1377 | ||
1378 | if len(info["linkname"]) > LENGTH_LINK: | |
1379 | raise ValueError("linkname is too long") | |
1380 | ||
1381 | if len(info["name"]) > LENGTH_NAME: | |
1382 | info["prefix"], info["name"] = self._posix_split_name(info["name"]) | |
1383 | ||
be60ffd0 | 1384 | return self._create_header(info, USTAR_FORMAT, encoding, errors) |
7584f5c9 | 1385 | |
be60ffd0 | 1386 | def create_gnu_header(self, info, encoding, errors): |
7584f5c9 ERE |
1387 | """Return the object as a GNU header block sequence. |
1388 | """ | |
1389 | info["magic"] = GNU_MAGIC | |
1390 | ||
2f854e77 ERE |
1391 | if self.ismultivol(): |
1392 | prefix = [ | |
1393 | itn(info.get("atime", 0), 12, GNU_FORMAT), | |
1394 | itn(info.get("ctime", 0), 12, GNU_FORMAT), | |
0eb5048f | 1395 | itn(self.volume_offset, 12, GNU_FORMAT), |
2f854e77 ERE |
1396 | itn(0, 119, GNU_FORMAT), # stuff unused in this tar implementation, set to zero |
1397 | ] | |
be60ffd0 | 1398 | info['prefix'] = b"".join(prefix) |
0eb5048f | 1399 | info['size'] = info['size'] - self.volume_offset |
2f854e77 | 1400 | |
be60ffd0 | 1401 | buf = b"" |
7584f5c9 | 1402 | if len(info["linkname"]) > LENGTH_LINK: |
be60ffd0 ERE |
1403 | buf += self._create_gnu_long_header(info["linkname"], |
1404 | GNUTYPE_LONGLINK, encoding, errors) | |
7584f5c9 ERE |
1405 | |
1406 | if len(info["name"]) > LENGTH_NAME: | |
be60ffd0 ERE |
1407 | buf += self._create_gnu_long_header(info["name"], GNUTYPE_LONGNAME, |
1408 | encoding, errors) | |
7584f5c9 | 1409 | |
be60ffd0 | 1410 | return buf + self._create_header(info, GNU_FORMAT, encoding, errors) |
7584f5c9 ERE |
1411 | |
1412 | def create_pax_header(self, info, encoding, errors): | |
1413 | """Return the object as a ustar header block. If it cannot be | |
1414 | represented this way, prepend a pax extended header sequence | |
1415 | with supplement information. | |
1416 | """ | |
1417 | info["magic"] = POSIX_MAGIC | |
1418 | pax_headers = self.pax_headers.copy() | |
c04e0751 ERE |
1419 | if self.ismultivol(): |
1420 | info['size'] = info['size'] - self.volume_offset | |
7584f5c9 ERE |
1421 | |
1422 | # Test string fields for values that exceed the field length or cannot | |
1423 | # be represented in ASCII encoding. | |
1424 | for name, hname, length in ( | |
36a315a0 ERE |
1425 | ("name", "path", LENGTH_NAME), |
1426 | ("linkname", "linkpath", LENGTH_LINK), | |
1427 | ("uname", "uname", 32), | |
1428 | ("gname", "gname", 32)): | |
7584f5c9 ERE |
1429 | |
1430 | if hname in pax_headers: | |
1431 | # The pax header has priority. | |
1432 | continue | |
1433 | ||
7584f5c9 ERE |
1434 | # Try to encode the string as ASCII. |
1435 | try: | |
be60ffd0 | 1436 | info[name].encode("ascii", "strict") |
7584f5c9 | 1437 | except UnicodeEncodeError: |
be60ffd0 | 1438 | pax_headers[hname] = info[name] |
7584f5c9 ERE |
1439 | continue |
1440 | ||
1441 | if len(info[name]) > length: | |
be60ffd0 | 1442 | pax_headers[hname] = info[name] |
7584f5c9 ERE |
1443 | |
1444 | # Test number fields for values that exceed the field limit or values | |
1445 | # that like to be stored as float. | |
1446 | for name, digits in (("uid", 8), ("gid", 8), ("size", 12), ("mtime", 12)): | |
1447 | if name in pax_headers: | |
1448 | # The pax header has priority. Avoid overflow. | |
1449 | info[name] = 0 | |
1450 | continue | |
1451 | ||
1452 | val = info[name] | |
1453 | if not 0 <= val < 8 ** (digits - 1) or isinstance(val, float): | |
be60ffd0 | 1454 | pax_headers[name] = str(val) |
7584f5c9 ERE |
1455 | info[name] = 0 |
1456 | ||
1457 | # Create a pax extended header if necessary. | |
1458 | if pax_headers: | |
be60ffd0 | 1459 | buf = self._create_pax_generic_header(pax_headers, XHDTYPE, encoding) |
7584f5c9 | 1460 | else: |
be60ffd0 | 1461 | buf = b"" |
7584f5c9 | 1462 | |
be60ffd0 | 1463 | return buf + self._create_header(info, USTAR_FORMAT, "ascii", "replace") |
7584f5c9 ERE |
1464 | |
1465 | @classmethod | |
1466 | def create_pax_global_header(cls, pax_headers): | |
1467 | """Return the object as a pax global header block sequence. | |
1468 | """ | |
be60ffd0 | 1469 | return cls._create_pax_generic_header(pax_headers, XGLTYPE, "utf-8") |
7584f5c9 ERE |
1470 | |
1471 | def _posix_split_name(self, name): | |
1472 | """Split a name longer than 100 chars into a prefix | |
1473 | and a name part. | |
1474 | """ | |
1475 | prefix = name[:LENGTH_PREFIX + 1] | |
1476 | while prefix and prefix[-1] != "/": | |
1477 | prefix = prefix[:-1] | |
1478 | ||
1479 | name = name[len(prefix):] | |
1480 | prefix = prefix[:-1] | |
1481 | ||
1482 | if not prefix or len(name) > LENGTH_NAME: | |
1483 | raise ValueError("name is too long") | |
1484 | return prefix, name | |
1485 | ||
1486 | @staticmethod | |
be60ffd0 | 1487 | def _create_header(info, format, encoding, errors): |
7584f5c9 ERE |
1488 | """Return a header block. info is a dictionary with file |
1489 | information, format must be one of the *_FORMAT constants. | |
1490 | """ | |
1491 | parts = [ | |
be60ffd0 ERE |
1492 | stn(info.get("name", ""), 100, encoding, errors), |
1493 | itn(info.get("mode", 0) & 0o7777, 8, format), | |
7584f5c9 ERE |
1494 | itn(info.get("uid", 0), 8, format), |
1495 | itn(info.get("gid", 0), 8, format), | |
1496 | itn(info.get("size", 0), 12, format), | |
1497 | itn(info.get("mtime", 0), 12, format), | |
be60ffd0 | 1498 | b" ", # checksum field |
2f854e77 | 1499 | info.get("type", REGTYPE), |
be60ffd0 ERE |
1500 | stn(info.get("linkname", ""), 100, encoding, errors), |
1501 | info.get("magic", POSIX_MAGIC), | |
1502 | stn(info.get("uname", ""), 32, encoding, errors), | |
1503 | stn(info.get("gname", ""), 32, encoding, errors), | |
7584f5c9 ERE |
1504 | itn(info.get("devmajor", 0), 8, format), |
1505 | itn(info.get("devminor", 0), 8, format), | |
be60ffd0 | 1506 | sbtn(info.get("prefix", ""), 155, encoding, errors) |
7584f5c9 ERE |
1507 | ] |
1508 | ||
be60ffd0 | 1509 | buf = struct.pack("%ds" % BLOCKSIZE, b"".join(parts)) |
7584f5c9 | 1510 | chksum = calc_chksums(buf[-BLOCKSIZE:])[0] |
be60ffd0 | 1511 | buf = buf[:-364] + bytes("%06o\0" % chksum, "ascii") + buf[-357:] |
7584f5c9 ERE |
1512 | return buf |
1513 | ||
1514 | @staticmethod | |
1515 | def _create_payload(payload): | |
1516 | """Return the string payload filled with zero bytes | |
1517 | up to the next 512 byte border. | |
1518 | """ | |
1519 | blocks, remainder = divmod(len(payload), BLOCKSIZE) | |
1520 | if remainder > 0: | |
1521 | payload += (BLOCKSIZE - remainder) * NUL | |
1522 | return payload | |
1523 | ||
1524 | @classmethod | |
be60ffd0 | 1525 | def _create_gnu_long_header(cls, name, type, encoding, errors): |
7584f5c9 ERE |
1526 | """Return a GNUTYPE_LONGNAME or GNUTYPE_LONGLINK sequence |
1527 | for name. | |
1528 | """ | |
be60ffd0 | 1529 | name = name.encode(encoding, errors) + NUL |
7584f5c9 ERE |
1530 | |
1531 | info = {} | |
1532 | info["name"] = "././@LongLink" | |
1533 | info["type"] = type | |
1534 | info["size"] = len(name) | |
1535 | info["magic"] = GNU_MAGIC | |
1536 | ||
1537 | # create extended header + name blocks. | |
be60ffd0 | 1538 | return cls._create_header(info, USTAR_FORMAT, encoding, errors) + \ |
7584f5c9 ERE |
1539 | cls._create_payload(name) |
1540 | ||
1541 | @classmethod | |
be60ffd0 ERE |
1542 | def _create_pax_generic_header(cls, pax_headers, type, encoding): |
1543 | """Return a POSIX.1-2008 extended or global header sequence | |
7584f5c9 | 1544 | that contains a list of keyword, value pairs. The values |
be60ffd0 | 1545 | must be strings. |
7584f5c9 | 1546 | """ |
be60ffd0 ERE |
1547 | # Check if one of the fields contains surrogate characters and thereby |
1548 | # forces hdrcharset=BINARY, see _proc_pax() for more information. | |
1549 | binary = False | |
1550 | for keyword, value in pax_headers.items(): | |
1551 | try: | |
1552 | value.encode("utf-8", "strict") | |
1553 | except UnicodeEncodeError: | |
1554 | binary = True | |
1555 | break | |
1556 | ||
1557 | records = b"" | |
1558 | if binary: | |
1559 | # Put the hdrcharset field at the beginning of the header. | |
1560 | records += b"21 hdrcharset=BINARY\n" | |
1561 | ||
1562 | for keyword, value in pax_headers.items(): | |
1563 | keyword = keyword.encode("utf-8") | |
1564 | if binary: | |
1565 | # Try to restore the original byte representation of `value'. | |
1566 | # Needless to say, that the encoding must match the string. | |
1567 | value = value.encode(encoding, "surrogateescape") | |
1568 | else: | |
1569 | value = value.encode("utf-8") | |
1570 | ||
7584f5c9 ERE |
1571 | l = len(keyword) + len(value) + 3 # ' ' + '=' + '\n' |
1572 | n = p = 0 | |
1573 | while True: | |
1574 | n = l + len(str(p)) | |
1575 | if n == p: | |
1576 | break | |
1577 | p = n | |
be60ffd0 | 1578 | records += bytes(str(p), "ascii") + b" " + keyword + b"=" + value + b"\n" |
7584f5c9 ERE |
1579 | |
1580 | # We use a hardcoded "././@PaxHeader" name like star does | |
1581 | # instead of the one that POSIX recommends. | |
1582 | info = {} | |
1583 | info["name"] = "././@PaxHeader" | |
1584 | info["type"] = type | |
1585 | info["size"] = len(records) | |
1586 | info["magic"] = POSIX_MAGIC | |
1587 | ||
1588 | # Create pax header + record blocks. | |
be60ffd0 | 1589 | return cls._create_header(info, USTAR_FORMAT, "ascii", "replace") + \ |
7584f5c9 ERE |
1590 | cls._create_payload(records) |
1591 | ||
1592 | @classmethod | |
be60ffd0 ERE |
1593 | def frombuf(cls, buf, encoding, errors): |
1594 | """Construct a TarInfo object from a 512 byte bytes object. | |
7584f5c9 ERE |
1595 | """ |
1596 | if len(buf) == 0: | |
1597 | raise EmptyHeaderError("empty header") | |
1598 | if len(buf) != BLOCKSIZE: | |
1599 | raise TruncatedHeaderError("truncated header") | |
1600 | if buf.count(NUL) == BLOCKSIZE: | |
1601 | raise EOFHeaderError("end of file header") | |
1602 | ||
1603 | chksum = nti(buf[148:156]) | |
1604 | if chksum not in calc_chksums(buf): | |
1605 | raise InvalidHeaderError("bad checksum") | |
1606 | ||
1607 | obj = cls() | |
be60ffd0 | 1608 | obj.name = nts(buf[0:100], encoding, errors) |
7584f5c9 ERE |
1609 | obj.mode = nti(buf[100:108]) |
1610 | obj.uid = nti(buf[108:116]) | |
1611 | obj.gid = nti(buf[116:124]) | |
1612 | obj.size = nti(buf[124:136]) | |
1613 | obj.mtime = nti(buf[136:148]) | |
1614 | obj.chksum = chksum | |
1615 | obj.type = buf[156:157] | |
be60ffd0 ERE |
1616 | obj.linkname = nts(buf[157:257], encoding, errors) |
1617 | obj.uname = nts(buf[265:297], encoding, errors) | |
1618 | obj.gname = nts(buf[297:329], encoding, errors) | |
7584f5c9 ERE |
1619 | obj.devmajor = nti(buf[329:337]) |
1620 | obj.devminor = nti(buf[337:345]) | |
be60ffd0 ERE |
1621 | prefix = nts(buf[345:500], encoding, errors) |
1622 | ||
1623 | # The old GNU sparse format occupies some of the unused | |
1624 | # space in the buffer for up to 4 sparse structures. | |
1625 | # Save the them for later processing in _proc_sparse(). | |
1626 | if obj.type == GNUTYPE_SPARSE: | |
1627 | pos = 386 | |
1628 | structs = [] | |
1629 | for i in range(4): | |
1630 | try: | |
1631 | offset = nti(buf[pos:pos + 12]) | |
1632 | numbytes = nti(buf[pos + 12:pos + 24]) | |
1633 | except ValueError: | |
1634 | break | |
1635 | structs.append((offset, numbytes)) | |
1636 | pos += 24 | |
1637 | isextended = bool(buf[482]) | |
1638 | origsize = nti(buf[483:495]) | |
1639 | obj._sparse_structs = (structs, isextended, origsize) | |
7584f5c9 ERE |
1640 | |
1641 | # Old V7 tar format represents a directory as a regular | |
1642 | # file with a trailing slash. | |
1643 | if obj.type == AREGTYPE and obj.name.endswith("/"): | |
1644 | obj.type = DIRTYPE | |
1645 | ||
1646 | # Remove redundant slashes from directories. | |
1647 | if obj.isdir(): | |
1648 | obj.name = obj.name.rstrip("/") | |
1649 | ||
1650 | # Reconstruct a ustar longname. | |
1651 | if prefix and obj.type not in GNU_TYPES: | |
1652 | obj.name = prefix + "/" + obj.name | |
c474439c ERE |
1653 | else: |
1654 | obj.offset_data = nti(buf[369:381]) | |
7584f5c9 ERE |
1655 | return obj |
1656 | ||
1657 | @classmethod | |
1658 | def fromtarfile(cls, tarfile): | |
1659 | """Return the next TarInfo object from TarFile object | |
1660 | tarfile. | |
1661 | """ | |
1662 | buf = tarfile.fileobj.read(BLOCKSIZE) | |
be60ffd0 | 1663 | obj = cls.frombuf(buf, tarfile.encoding, tarfile.errors) |
7584f5c9 ERE |
1664 | obj.offset = tarfile.fileobj.tell() - BLOCKSIZE |
1665 | return obj._proc_member(tarfile) | |
1666 | ||
1667 | #-------------------------------------------------------------------------- | |
1668 | # The following are methods that are called depending on the type of a | |
1669 | # member. The entry point is _proc_member() which can be overridden in a | |
1670 | # subclass to add custom _proc_*() methods. A _proc_*() method MUST | |
1671 | # implement the following | |
1672 | # operations: | |
1673 | # 1. Set self.offset_data to the position where the data blocks begin, | |
1674 | # if there is data that follows. | |
1675 | # 2. Set tarfile.offset to the position where the next member's header will | |
1676 | # begin. | |
1677 | # 3. Return self or another valid TarInfo object. | |
1678 | def _proc_member(self, tarfile): | |
1679 | """Choose the right processing method depending on | |
1680 | the type and call it. | |
1681 | """ | |
1682 | if self.type in (GNUTYPE_LONGNAME, GNUTYPE_LONGLINK): | |
1683 | return self._proc_gnulong(tarfile) | |
1684 | elif self.type == GNUTYPE_SPARSE: | |
1685 | return self._proc_sparse(tarfile) | |
1686 | elif self.type in (XHDTYPE, XGLTYPE, SOLARIS_XHDTYPE): | |
1687 | return self._proc_pax(tarfile) | |
1688 | else: | |
1689 | return self._proc_builtin(tarfile) | |
1690 | ||
1691 | def _proc_builtin(self, tarfile): | |
1692 | """Process a builtin type or an unknown type which | |
1693 | will be treated as a regular file. | |
1694 | """ | |
1695 | self.offset_data = tarfile.fileobj.tell() | |
1696 | offset = self.offset_data | |
00c34a12 | 1697 | if self.isreg() or self.ismultivol() or self.type not in SUPPORTED_TYPES: |
7584f5c9 ERE |
1698 | # Skip the following data blocks. |
1699 | offset += self._block(self.size) | |
1700 | tarfile.offset = offset | |
1701 | ||
1702 | # Patch the TarInfo object with saved global | |
1703 | # header information. | |
1704 | self._apply_pax_info(tarfile.pax_headers, tarfile.encoding, tarfile.errors) | |
1705 | ||
1706 | return self | |
1707 | ||
1708 | def _proc_gnulong(self, tarfile): | |
1709 | """Process the blocks that hold a GNU longname | |
1710 | or longlink member. | |
1711 | """ | |
1712 | buf = tarfile.fileobj.read(self._block(self.size)) | |
1713 | ||
1714 | # Fetch the next header and process it. | |
1715 | try: | |
1716 | next = self.fromtarfile(tarfile) | |
1717 | except HeaderError: | |
1718 | raise SubsequentHeaderError("missing or bad subsequent header") | |
1719 | ||
1720 | # Patch the TarInfo object from the next header with | |
1721 | # the longname information. | |
1722 | next.offset = self.offset | |
1723 | if self.type == GNUTYPE_LONGNAME: | |
be60ffd0 | 1724 | next.name = nts(buf, tarfile.encoding, tarfile.errors) |
7584f5c9 | 1725 | elif self.type == GNUTYPE_LONGLINK: |
be60ffd0 | 1726 | next.linkname = nts(buf, tarfile.encoding, tarfile.errors) |
7584f5c9 ERE |
1727 | |
1728 | return next | |
1729 | ||
1730 | def _proc_sparse(self, tarfile): | |
1731 | """Process a GNU sparse header plus extra headers. | |
1732 | """ | |
be60ffd0 ERE |
1733 | # We already collected some sparse structures in frombuf(). |
1734 | structs, isextended, origsize = self._sparse_structs | |
1735 | del self._sparse_structs | |
1736 | ||
1737 | # Collect sparse structures from extended header blocks. | |
1738 | while isextended: | |
7584f5c9 ERE |
1739 | buf = tarfile.fileobj.read(BLOCKSIZE) |
1740 | pos = 0 | |
be60ffd0 | 1741 | for i in range(21): |
7584f5c9 ERE |
1742 | try: |
1743 | offset = nti(buf[pos:pos + 12]) | |
1744 | numbytes = nti(buf[pos + 12:pos + 24]) | |
1745 | except ValueError: | |
1746 | break | |
be60ffd0 ERE |
1747 | if offset and numbytes: |
1748 | structs.append((offset, numbytes)) | |
7584f5c9 | 1749 | pos += 24 |
be60ffd0 ERE |
1750 | isextended = bool(buf[504]) |
1751 | self.sparse = structs | |
7584f5c9 ERE |
1752 | |
1753 | self.offset_data = tarfile.fileobj.tell() | |
1754 | tarfile.offset = self.offset_data + self._block(self.size) | |
1755 | self.size = origsize | |
7584f5c9 ERE |
1756 | return self |
1757 | ||
1758 | def _proc_pax(self, tarfile): | |
1759 | """Process an extended or global header as described in | |
be60ffd0 | 1760 | POSIX.1-2008. |
7584f5c9 ERE |
1761 | """ |
1762 | # Read the header information. | |
1763 | buf = tarfile.fileobj.read(self._block(self.size)) | |
1764 | ||
1765 | # A pax header stores supplemental information for either | |
1766 | # the following file (extended) or all following files | |
1767 | # (global). | |
1768 | if self.type == XGLTYPE: | |
1769 | pax_headers = tarfile.pax_headers | |
1770 | else: | |
1771 | pax_headers = tarfile.pax_headers.copy() | |
1772 | ||
be60ffd0 ERE |
1773 | # Check if the pax header contains a hdrcharset field. This tells us |
1774 | # the encoding of the path, linkpath, uname and gname fields. Normally, | |
1775 | # these fields are UTF-8 encoded but since POSIX.1-2008 tar | |
1776 | # implementations are allowed to store them as raw binary strings if | |
1777 | # the translation to UTF-8 fails. | |
1778 | match = re.search(br"\d+ hdrcharset=([^\n]+)\n", buf) | |
1779 | if match is not None: | |
1780 | pax_headers["hdrcharset"] = match.group(1).decode("utf-8") | |
1781 | ||
1782 | # For the time being, we don't care about anything other than "BINARY". | |
1783 | # The only other value that is currently allowed by the standard is | |
1784 | # "ISO-IR 10646 2000 UTF-8" in other words UTF-8. | |
1785 | hdrcharset = pax_headers.get("hdrcharset") | |
1786 | if hdrcharset == "BINARY": | |
1787 | encoding = tarfile.encoding | |
1788 | else: | |
1789 | encoding = "utf-8" | |
1790 | ||
7584f5c9 ERE |
1791 | # Parse pax header information. A record looks like that: |
1792 | # "%d %s=%s\n" % (length, keyword, value). length is the size | |
1793 | # of the complete record including the length field itself and | |
1794 | # the newline. keyword and value are both UTF-8 encoded strings. | |
be60ffd0 | 1795 | regex = re.compile(br"(\d+) ([^=]+)=") |
7584f5c9 ERE |
1796 | pos = 0 |
1797 | while True: | |
1798 | match = regex.match(buf, pos) | |
1799 | if not match: | |
1800 | break | |
1801 | ||
1802 | length, keyword = match.groups() | |
1803 | length = int(length) | |
1804 | value = buf[match.end(2) + 1:match.start(1) + length - 1] | |
1805 | ||
be60ffd0 ERE |
1806 | # Normally, we could just use "utf-8" as the encoding and "strict" |
1807 | # as the error handler, but we better not take the risk. For | |
1808 | # example, GNU tar <= 1.23 is known to store filenames it cannot | |
1809 | # translate to UTF-8 as raw strings (unfortunately without a | |
1810 | # hdrcharset=BINARY header). | |
1811 | # We first try the strict standard encoding, and if that fails we | |
1812 | # fall back on the user's encoding and error handler. | |
1813 | keyword = self._decode_pax_field(keyword, "utf-8", "utf-8", | |
1814 | tarfile.errors) | |
1815 | if keyword in PAX_NAME_FIELDS: | |
1816 | value = self._decode_pax_field(value, encoding, tarfile.encoding, | |
1817 | tarfile.errors) | |
1818 | else: | |
1819 | value = self._decode_pax_field(value, "utf-8", "utf-8", | |
1820 | tarfile.errors) | |
7584f5c9 ERE |
1821 | |
1822 | pax_headers[keyword] = value | |
1823 | pos += length | |
1824 | ||
36a315a0 | 1825 | |
7584f5c9 ERE |
1826 | # Fetch the next header. |
1827 | try: | |
1828 | next = self.fromtarfile(tarfile) | |
1829 | except HeaderError: | |
1830 | raise SubsequentHeaderError("missing or bad subsequent header") | |
1831 | ||
be60ffd0 ERE |
1832 | # Process GNU sparse information. |
1833 | if "GNU.sparse.map" in pax_headers: | |
1834 | # GNU extended sparse format version 0.1. | |
1835 | self._proc_gnusparse_01(next, pax_headers) | |
1836 | ||
1837 | elif "GNU.sparse.size" in pax_headers: | |
1838 | # GNU extended sparse format version 0.0. | |
1839 | self._proc_gnusparse_00(next, pax_headers, buf) | |
1840 | ||
1841 | elif pax_headers.get("GNU.sparse.major") == "1" and pax_headers.get("GNU.sparse.minor") == "0": | |
1842 | # GNU extended sparse format version 1.0. | |
1843 | self._proc_gnusparse_10(next, pax_headers, tarfile) | |
1844 | ||
7584f5c9 ERE |
1845 | if self.type in (XHDTYPE, SOLARIS_XHDTYPE): |
1846 | # Patch the TarInfo object with the extended header info. | |
1847 | next._apply_pax_info(pax_headers, tarfile.encoding, tarfile.errors) | |
1848 | next.offset = self.offset | |
1849 | ||
1850 | if "size" in pax_headers: | |
1851 | # If the extended header replaces the size field, | |
1852 | # we need to recalculate the offset where the next | |
1853 | # header starts. | |
1854 | offset = next.offset_data | |
1855 | if next.isreg() or next.type not in SUPPORTED_TYPES: | |
1856 | offset += next._block(next.size) | |
1857 | tarfile.offset = offset | |
1858 | ||
c04e0751 ERE |
1859 | if next is not None: |
1860 | if "GNU.volume.filename" in pax_headers: | |
1861 | if pax_headers["GNU.volume.filename"] == next.name: | |
1862 | if "GNU.volume.size" in pax_headers: | |
1863 | next.size = int(pax_headers["GNU.volume.size"]) | |
1864 | if "GNU.volume.offset" in pax_headers: | |
1865 | next.volume_offset = int(pax_headers["GNU.volume.offset"]) | |
1866 | ||
1867 | for key in pax_headers.keys(): | |
1868 | if key.startswith("GNU.volume"): | |
1869 | del tarfile.pax_headers[key] | |
0eb5048f | 1870 | |
7584f5c9 ERE |
1871 | return next |
1872 | ||
be60ffd0 ERE |
1873 | def _proc_gnusparse_00(self, next, pax_headers, buf): |
1874 | """Process a GNU tar extended sparse header, version 0.0. | |
7584f5c9 | 1875 | """ |
be60ffd0 ERE |
1876 | offsets = [] |
1877 | for match in re.finditer(br"\d+ GNU.sparse.offset=(\d+)\n", buf): | |
1878 | offsets.append(int(match.group(1))) | |
1879 | numbytes = [] | |
1880 | for match in re.finditer(br"\d+ GNU.sparse.numbytes=(\d+)\n", buf): | |
1881 | numbytes.append(int(match.group(1))) | |
1882 | next.sparse = list(zip(offsets, numbytes)) | |
7584f5c9 | 1883 | |
be60ffd0 ERE |
1884 | def _proc_gnusparse_01(self, next, pax_headers): |
1885 | """Process a GNU tar extended sparse header, version 0.1. | |
1886 | """ | |
1887 | sparse = [int(x) for x in pax_headers["GNU.sparse.map"].split(",")] | |
1888 | next.sparse = list(zip(sparse[::2], sparse[1::2])) | |
7584f5c9 | 1889 | |
be60ffd0 ERE |
1890 | def _proc_gnusparse_10(self, next, pax_headers, tarfile): |
1891 | """Process a GNU tar extended sparse header, version 1.0. | |
1892 | """ | |
1893 | fields = None | |
1894 | sparse = [] | |
1895 | buf = tarfile.fileobj.read(BLOCKSIZE) | |
1896 | fields, buf = buf.split(b"\n", 1) | |
1897 | fields = int(fields) | |
1898 | while len(sparse) < fields * 2: | |
1899 | if b"\n" not in buf: | |
1900 | buf += tarfile.fileobj.read(BLOCKSIZE) | |
1901 | number, buf = buf.split(b"\n", 1) | |
1902 | sparse.append(int(number)) | |
1903 | next.offset_data = tarfile.fileobj.tell() | |
1904 | next.sparse = list(zip(sparse[::2], sparse[1::2])) | |
7584f5c9 | 1905 | |
be60ffd0 ERE |
1906 | def _apply_pax_info(self, pax_headers, encoding, errors): |
1907 | """Replace fields with supplemental information from a previous | |
1908 | pax extended or global header. | |
1909 | """ | |
1910 | for keyword, value in pax_headers.items(): | |
1911 | if keyword == "GNU.sparse.name": | |
1912 | setattr(self, "path", value) | |
1913 | elif keyword == "GNU.sparse.size": | |
1914 | setattr(self, "size", int(value)) | |
1915 | elif keyword == "GNU.sparse.realsize": | |
1916 | setattr(self, "size", int(value)) | |
1917 | elif keyword in PAX_FIELDS: | |
1918 | if keyword in PAX_NUMBER_FIELDS: | |
1919 | try: | |
1920 | value = PAX_NUMBER_FIELDS[keyword](value) | |
1921 | except ValueError: | |
1922 | value = 0 | |
1923 | if keyword == "path": | |
f0287fb7 | 1924 | value = value.rstrip("/") # pylint: disable=no-member |
be60ffd0 | 1925 | setattr(self, keyword, value) |
7584f5c9 ERE |
1926 | |
1927 | self.pax_headers = pax_headers.copy() | |
1928 | ||
be60ffd0 ERE |
1929 | def _decode_pax_field(self, value, encoding, fallback_encoding, fallback_errors): |
1930 | """Decode a single field from a pax record. | |
1931 | """ | |
1932 | try: | |
1933 | return value.decode(encoding, "strict") | |
1934 | except UnicodeDecodeError: | |
1935 | return value.decode(fallback_encoding, fallback_errors) | |
1936 | ||
7584f5c9 ERE |
1937 | def _block(self, count): |
1938 | """Round up a byte count by BLOCKSIZE and return it, | |
1939 | e.g. _block(834) => 1024. | |
1940 | """ | |
1941 | blocks, remainder = divmod(count, BLOCKSIZE) | |
1942 | if remainder: | |
1943 | blocks += 1 | |
1944 | return blocks * BLOCKSIZE | |
1945 | ||
1946 | def isreg(self): | |
1947 | return self.type in REGULAR_TYPES | |
1948 | def isfile(self): | |
1949 | return self.isreg() | |
1950 | def isdir(self): | |
1951 | return self.type == DIRTYPE | |
1952 | def issym(self): | |
1953 | return self.type == SYMTYPE | |
1954 | def islnk(self): | |
1955 | return self.type == LNKTYPE | |
1956 | def ischr(self): | |
1957 | return self.type == CHRTYPE | |
1958 | def isblk(self): | |
1959 | return self.type == BLKTYPE | |
1960 | def isfifo(self): | |
1961 | return self.type == FIFOTYPE | |
1962 | def issparse(self): | |
be60ffd0 | 1963 | return self.sparse is not None |
7584f5c9 ERE |
1964 | def isdev(self): |
1965 | return self.type in (CHRTYPE, BLKTYPE, FIFOTYPE) | |
68ddf955 | 1966 | def ismultivol(self): |
c04e0751 ERE |
1967 | return self.type == GNUTYPE_MULTIVOL or self.volume_offset > 0 or\ |
1968 | "GNU.volume.offset" in self.pax_headers | |
7584f5c9 ERE |
1969 | # class TarInfo |
1970 | ||
1971 | class TarFile(object): | |
1972 | """The TarFile Class provides an interface to tar archives. | |
1973 | """ | |
1974 | ||
1975 | debug = 0 # May be set from 0 (no msgs) to 3 (all msgs) | |
1976 | ||
1977 | dereference = False # If true, add content of linked file to the | |
1978 | # tar file, else the link. | |
1979 | ||
1980 | ignore_zeros = False # If true, skips empty or invalid blocks and | |
1981 | # continues processing. | |
1982 | ||
83f2d71e | 1983 | max_volume_size = None # If different from None, establishes maximum |
68ddf955 ERE |
1984 | # size of tar volumes |
1985 | ||
1986 | new_volume_handler = None # function handler to be executed before when | |
1987 | # a new volume is needed | |
1988 | ||
1989 | volume_number = 0 # current volume number, used for multi volume | |
1990 | # support | |
1991 | ||
7584f5c9 ERE |
1992 | errorlevel = 1 # If 0, fatal errors only appear in debug |
1993 | # messages (if debug >= 0). If > 0, errors | |
1994 | # are passed to the caller as exceptions. | |
1995 | ||
1996 | format = DEFAULT_FORMAT # The format to use when creating an archive. | |
1997 | ||
1998 | encoding = ENCODING # Encoding for 8-bit character strings. | |
1999 | ||
2000 | errors = None # Error handler for unicode conversion. | |
2001 | ||
2002 | tarinfo = TarInfo # The default TarInfo class to use. | |
2003 | ||
be60ffd0 | 2004 | fileobject = ExFileObject # The file-object for extractfile(). |
7584f5c9 | 2005 | |
d1c38f40 PG |
2006 | arcmode = ARCMODE_PLAIN # Object processing mode (“concat”, encryption, |
2007 | # compression) | |
5fdff89f | 2008 | |
ea625b04 ERE |
2009 | save_to_members = True # If new members are saved. This can be disabled |
2010 | # if you manage lots of files and don't want | |
2011 | # to have high memory usage | |
2012 | ||
9ef1fb87 TJ |
2013 | cache_uid2user = {} # cache to avoid getpwuid calls. It always parses /etc/passwd. |
2014 | cache_gid2group = {} # same cache for groups | |
2015 | ||
7584f5c9 ERE |
2016 | def __init__(self, name=None, mode="r", fileobj=None, format=None, |
2017 | tarinfo=None, dereference=None, ignore_zeros=None, encoding=None, | |
be60ffd0 | 2018 | errors="surrogateescape", pax_headers=None, debug=None, |
548bb8d5 | 2019 | errorlevel=None, max_volume_size=None, new_volume_handler=None, |
d1c38f40 | 2020 | concat=False, nacl=None, |
c7c736b6 | 2021 | save_to_members=True): |
7584f5c9 ERE |
2022 | """Open an (uncompressed) tar archive `name'. `mode' is either 'r' to |
2023 | read from an existing archive, 'a' to append data to an existing | |
2024 | file or 'w' to create a new file overwriting an existing one. `mode' | |
2025 | defaults to 'r'. | |
2026 | If `fileobj' is given, it is used for reading or writing data. If it | |
2027 | can be determined, `mode' is overridden by `fileobj's mode. | |
2028 | `fileobj' is not closed, when TarFile is closed. | |
2029 | """ | |
2030 | if len(mode) > 1 or mode not in "raw": | |
2031 | raise ValueError("mode must be 'r', 'a' or 'w'") | |
2032 | self.mode = mode | |
d1c38f40 | 2033 | self.arcmode = arcmode_set (concat) |
c7c736b6 | 2034 | self.nacl = nacl |
7584f5c9 ERE |
2035 | self._mode = {"r": "rb", "a": "r+b", "w": "wb"}[mode] |
2036 | ||
2037 | if not fileobj: | |
2038 | if self.mode == "a" and not os.path.exists(name): | |
2039 | # Create nonexistent files in append mode. | |
2040 | self.mode = "w" | |
2041 | self._mode = "wb" | |
2042 | fileobj = bltn_open(name, self._mode) | |
2043 | self._extfileobj = False | |
2044 | else: | |
2045 | if name is None and hasattr(fileobj, "name"): | |
2046 | name = fileobj.name | |
d5361dac | 2047 | # when fileobj is a gzip.GzipFile, fileobj.mode is an int (not valid for us) |
be60ffd0 | 2048 | if hasattr(fileobj, "mode") and isinstance(fileobj.mode, str): |
7584f5c9 ERE |
2049 | self._mode = fileobj.mode |
2050 | self._extfileobj = True | |
be60ffd0 | 2051 | self.name = os.path.abspath(name) if name else None |
2f854e77 | 2052 | self.base_name = self.name = os.path.abspath(name) if name else None |
7584f5c9 ERE |
2053 | self.fileobj = fileobj |
2054 | ||
2055 | # Init attributes. | |
2056 | if format is not None: | |
2057 | self.format = format | |
2058 | if tarinfo is not None: | |
2059 | self.tarinfo = tarinfo | |
2060 | if dereference is not None: | |
2061 | self.dereference = dereference | |
2062 | if ignore_zeros is not None: | |
2063 | self.ignore_zeros = ignore_zeros | |
2064 | if encoding is not None: | |
2065 | self.encoding = encoding | |
2066 | ||
be60ffd0 | 2067 | self.errors = errors |
7584f5c9 ERE |
2068 | |
2069 | if pax_headers is not None and self.format == PAX_FORMAT: | |
2070 | self.pax_headers = pax_headers | |
2071 | else: | |
2072 | self.pax_headers = {} | |
2073 | ||
2074 | if debug is not None: | |
2075 | self.debug = debug | |
2076 | if errorlevel is not None: | |
2077 | self.errorlevel = errorlevel | |
2078 | ||
2079 | # Init datastructures. | |
ae48acc8 | 2080 | if max_volume_size and max_volume_size < 3*BLOCKSIZE: |
0c818a18 | 2081 | raise ValueError("max_volume_size needs to be at least %d" % (3*BLOCKSIZE)) |
ae48acc8 ERE |
2082 | if max_volume_size and not callable(new_volume_handler): |
2083 | raise ValueError("new_volume_handler needs to be set and be callable for multivolume support") | |
5ab3f8f9 CH |
2084 | if max_volume_size: |
2085 | self.max_volume_size = int(max_volume_size) | |
2086 | else: | |
2087 | self.max_volume_size = None | |
ae48acc8 | 2088 | |
ea625b04 | 2089 | self.save_to_members = save_to_members |
68ddf955 | 2090 | self.new_volume_handler = new_volume_handler |
7584f5c9 ERE |
2091 | self.closed = False |
2092 | self.members = [] # list of members as TarInfo objects | |
2093 | self._loaded = False # flag if all members have been read | |
2094 | self.offset = self.fileobj.tell() | |
2095 | # current position in the archive file | |
2096 | self.inodes = {} # dictionary caching the inodes of | |
2097 | # archive members already added | |
2098 | ||
2099 | try: | |
2100 | if self.mode == "r": | |
2101 | self.firstmember = None | |
2102 | self.firstmember = self.next() | |
2103 | ||
2104 | if self.mode == "a": | |
2105 | # Move to the end of the archive, | |
2106 | # before the first empty block. | |
2107 | while True: | |
2108 | self.fileobj.seek(self.offset) | |
2109 | try: | |
2110 | tarinfo = self.tarinfo.fromtarfile(self) | |
2111 | self.members.append(tarinfo) | |
2112 | except EOFHeaderError: | |
2113 | self.fileobj.seek(self.offset) | |
2114 | break | |
be60ffd0 | 2115 | except HeaderError as e: |
7584f5c9 ERE |
2116 | raise ReadError(str(e)) |
2117 | ||
2118 | if self.mode in "aw": | |
2119 | self._loaded = True | |
2120 | ||
2121 | if self.pax_headers: | |
2122 | buf = self.tarinfo.create_pax_global_header(self.pax_headers.copy()) | |
2123 | self.fileobj.write(buf) | |
2124 | self.offset += len(buf) | |
2125 | except: | |
2126 | if not self._extfileobj: | |
2127 | self.fileobj.close() | |
2128 | self.closed = True | |
2129 | raise | |
2130 | ||
7584f5c9 ERE |
2131 | #-------------------------------------------------------------------------- |
2132 | # Below are the classmethods which act as alternate constructors to the | |
2133 | # TarFile class. The open() method is the only one that is needed for | |
2134 | # public use; it is the "super"-constructor and is able to select an | |
2135 | # adequate "sub"-constructor for a particular compression using the mapping | |
2136 | # from OPEN_METH. | |
2137 | # | |
2138 | # This concept allows one to subclass TarFile without losing the comfort of | |
2139 | # the super-constructor. A sub-constructor is registered and made available | |
2140 | # by adding it to the mapping in OPEN_METH. | |
2141 | ||
2142 | @classmethod | |
2b82f50c | 2143 | def open(cls, name=None, mode="r", fileobj=None, bufsize=RECORDSIZE, |
04f4c7ab PG |
2144 | encryption=None, compresslevel=9, tolerance=TOLERANCE_STRICT, |
2145 | **kwargs): | |
7584f5c9 ERE |
2146 | """Open a tar archive for reading, writing or appending. Return |
2147 | an appropriate TarFile class. | |
2148 | ||
2149 | mode: | |
2150 | 'r' or 'r:*' open for reading with transparent compression | |
2151 | 'r:' open for reading exclusively uncompressed | |
2152 | 'r:gz' open for reading with gzip compression | |
2153 | 'r:bz2' open for reading with bzip2 compression | |
be60ffd0 | 2154 | 'r:xz' open for reading with lzma compression |
7584f5c9 ERE |
2155 | 'a' or 'a:' open for appending, creating the file if necessary |
2156 | 'w' or 'w:' open for writing without compression | |
2157 | 'w:gz' open for writing with gzip compression | |
2158 | 'w:bz2' open for writing with bzip2 compression | |
be60ffd0 | 2159 | 'w:xz' open for writing with lzma compression |
7584f5c9 ERE |
2160 | |
2161 | 'r|*' open a stream of tar blocks with transparent compression | |
2162 | 'r|' open an uncompressed stream of tar blocks for reading | |
2163 | 'r|gz' open a gzip compressed stream of tar blocks | |
2164 | 'r|bz2' open a bzip2 compressed stream of tar blocks | |
be60ffd0 | 2165 | 'r|xz' open an lzma compressed stream of tar blocks |
7584f5c9 ERE |
2166 | 'w|' open an uncompressed stream for writing |
2167 | 'w|gz' open a gzip compressed stream for writing | |
2168 | 'w|bz2' open a bzip2 compressed stream for writing | |
be60ffd0 | 2169 | 'w|xz' open an lzma compressed stream for writing |
85737f48 ERE |
2170 | |
2171 | 'r#gz' open a stream of gzip compressed tar blocks for reading | |
2172 | 'w#gz' open a stream of gzip compressed tar blocks for writing | |
7584f5c9 | 2173 | """ |
7584f5c9 ERE |
2174 | if not name and not fileobj: |
2175 | raise ValueError("nothing to open") | |
2176 | ||
2177 | if mode in ("r", "r:*"): | |
2178 | # Find out which *open() is appropriate for opening the file. | |
2179 | for comptype in cls.OPEN_METH: | |
2180 | func = getattr(cls, cls.OPEN_METH[comptype]) | |
2181 | if fileobj is not None: | |
2182 | saved_pos = fileobj.tell() | |
2183 | try: | |
2184 | return func(name, "r", fileobj, **kwargs) | |
be60ffd0 | 2185 | except (ReadError, CompressionError) as e: |
c7c736b6 | 2186 | # usually nothing exceptional but sometimes is |
7584f5c9 ERE |
2187 | if fileobj is not None: |
2188 | fileobj.seek(saved_pos) | |
2189 | continue | |
2190 | raise ReadError("file could not be opened successfully") | |
2191 | ||
2192 | elif ":" in mode: | |
2193 | filemode, comptype = mode.split(":", 1) | |
2194 | filemode = filemode or "r" | |
2195 | comptype = comptype or "tar" | |
2196 | ||
2197 | # Select the *open() function according to | |
2198 | # given compression. | |
2199 | if comptype in cls.OPEN_METH: | |
2200 | func = getattr(cls, cls.OPEN_METH[comptype]) | |
2201 | else: | |
2202 | raise CompressionError("unknown compression type %r" % comptype) | |
e05f0440 TJ |
2203 | |
2204 | # Pass on compression level for gzip / bzip2. | |
2205 | if comptype == 'gz' or comptype == 'bz2': | |
2206 | kwargs['compresslevel'] = compresslevel | |
2207 | ||
7a2b9329 CH |
2208 | if 'max_volume_size' in kwargs: |
2209 | if comptype != 'tar' and filemode in 'wa' \ | |
2210 | and kwargs['max_volume_size']: | |
2211 | import warnings | |
2212 | warnings.warn('Only the first volume will be compressed ' | |
2213 | 'for modes with "w:"!') | |
2214 | ||
e05f0440 | 2215 | return func(name, filemode, fileobj, **kwargs) |
7584f5c9 ERE |
2216 | |
2217 | elif "|" in mode: | |
2218 | filemode, comptype = mode.split("|", 1) | |
2219 | filemode = filemode or "r" | |
2220 | comptype = comptype or "tar" | |
2221 | ||
2222 | if filemode not in "rw": | |
2223 | raise ValueError("mode must be 'r' or 'w'") | |
2224 | ||
2225 | t = cls(name, filemode, | |
2b82f50c ERE |
2226 | _Stream(name, filemode, comptype, fileobj, bufsize, |
2227 | compresslevel=compresslevel), | |
7584f5c9 ERE |
2228 | **kwargs) |
2229 | t._extfileobj = False | |
2230 | return t | |
2231 | ||
5fdff89f ERE |
2232 | elif "#" in mode: |
2233 | filemode, comptype = mode.split("#", 1) | |
2234 | filemode = filemode or "r" | |
5fdff89f ERE |
2235 | |
2236 | if filemode not in "rw": | |
5faea0e1 PG |
2237 | raise ValueError ("mode %s not compatible with concat " |
2238 | "archive; must be 'r' or 'w'" % mode) | |
5fdff89f | 2239 | |
be60ffd0 | 2240 | stream = _Stream(name, filemode, comptype, fileobj, bufsize, |
d1c38f40 | 2241 | concat=True, encryption=encryption, |
04f4c7ab | 2242 | compresslevel=compresslevel, tolerance=tolerance) |
d1c38f40 | 2243 | kwargs ["concat"] = True |
be60ffd0 ERE |
2244 | try: |
2245 | t = cls(name, filemode, stream, **kwargs) | |
c7c736b6 | 2246 | except: # XXX except what? |
be60ffd0 | 2247 | stream.close() |
c7c736b6 | 2248 | raise # XXX raise what? |
5fdff89f ERE |
2249 | t._extfileobj = False |
2250 | return t | |
2251 | ||
7584f5c9 ERE |
2252 | elif mode in "aw": |
2253 | return cls.taropen(name, mode, fileobj, **kwargs) | |
2254 | ||
133d30da | 2255 | raise ValueError("undiscernible mode %r" % mode) |
7584f5c9 | 2256 | |
d39d4cbf PG |
2257 | |
2258 | @classmethod | |
2259 | def open_at_offset(cls, offset, *a, **kwa): | |
2260 | """ | |
2261 | Same as ``.open()``, but start reading at the given offset. Assumes a | |
5bd2d4b5 PG |
2262 | seekable file object. Returns *None* if opening failed due to a read |
2263 | problem. | |
d39d4cbf PG |
2264 | """ |
2265 | fileobj = kwa.get ("fileobj") | |
2266 | if fileobj is not None: | |
2267 | fileobj.seek (offset) | |
5bd2d4b5 | 2268 | |
d39d4cbf PG |
2269 | return cls.open (*a, **kwa) |
2270 | ||
2271 | ||
7584f5c9 ERE |
2272 | @classmethod |
2273 | def taropen(cls, name, mode="r", fileobj=None, **kwargs): | |
2274 | """Open uncompressed tar archive name for reading or writing. | |
2275 | """ | |
2276 | if len(mode) > 1 or mode not in "raw": | |
2277 | raise ValueError("mode must be 'r', 'a' or 'w'") | |
2278 | return cls(name, mode, fileobj, **kwargs) | |
2279 | ||
2280 | @classmethod | |
2281 | def gzopen(cls, name, mode="r", fileobj=None, compresslevel=9, **kwargs): | |
2282 | """Open gzip compressed tar archive name for reading or writing. | |
2283 | Appending is not allowed. | |
2284 | """ | |
2285 | if len(mode) > 1 or mode not in "rw": | |
2286 | raise ValueError("mode must be 'r' or 'w'") | |
2287 | ||
2288 | try: | |
2289 | import gzip | |
2290 | gzip.GzipFile | |
2291 | except (ImportError, AttributeError): | |
2292 | raise CompressionError("gzip module is not available") | |
2293 | ||
be60ffd0 | 2294 | extfileobj = fileobj is not None |
7584f5c9 | 2295 | try: |
be60ffd0 ERE |
2296 | fileobj = gzip.GzipFile(name, mode + "b", compresslevel, fileobj) |
2297 | t = cls.taropen(name, mode, fileobj, **kwargs) | |
2298 | except OSError: | |
2299 | if not extfileobj and fileobj is not None: | |
2300 | fileobj.close() | |
2301 | if fileobj is None: | |
2302 | raise | |
7584f5c9 | 2303 | raise ReadError("not a gzip file") |
be60ffd0 ERE |
2304 | except: |
2305 | if not extfileobj and fileobj is not None: | |
2306 | fileobj.close() | |
2307 | raise | |
2308 | t._extfileobj = extfileobj | |
7584f5c9 ERE |
2309 | return t |
2310 | ||
2311 | @classmethod | |
2312 | def bz2open(cls, name, mode="r", fileobj=None, compresslevel=9, **kwargs): | |
2313 | """Open bzip2 compressed tar archive name for reading or writing. | |
2314 | Appending is not allowed. | |
2315 | """ | |
2316 | if len(mode) > 1 or mode not in "rw": | |
2317 | raise ValueError("mode must be 'r' or 'w'.") | |
2318 | ||
2319 | try: | |
2320 | import bz2 | |
2321 | except ImportError: | |
2322 | raise CompressionError("bz2 module is not available") | |
2323 | ||
be60ffd0 ERE |
2324 | fileobj = bz2.BZ2File(fileobj or name, mode, |
2325 | compresslevel=compresslevel) | |
7584f5c9 ERE |
2326 | |
2327 | try: | |
2328 | t = cls.taropen(name, mode, fileobj, **kwargs) | |
be60ffd0 ERE |
2329 | except (OSError, EOFError): |
2330 | fileobj.close() | |
7584f5c9 ERE |
2331 | raise ReadError("not a bzip2 file") |
2332 | t._extfileobj = False | |
2333 | return t | |
2334 | ||
be60ffd0 ERE |
2335 | @classmethod |
2336 | def xzopen(cls, name, mode="r", fileobj=None, preset=None, **kwargs): | |
2337 | """Open lzma compressed tar archive name for reading or writing. | |
2338 | Appending is not allowed. | |
2339 | """ | |
2340 | if mode not in ("r", "w"): | |
2341 | raise ValueError("mode must be 'r' or 'w'") | |
2342 | ||
2343 | try: | |
2344 | import lzma | |
2345 | except ImportError: | |
2346 | raise CompressionError("lzma module is not available") | |
2347 | ||
2348 | fileobj = lzma.LZMAFile(fileobj or name, mode, preset=preset) | |
2349 | ||
2350 | try: | |
2351 | t = cls.taropen(name, mode, fileobj, **kwargs) | |
2352 | except (lzma.LZMAError, EOFError): | |
2353 | fileobj.close() | |
2354 | raise ReadError("not an lzma file") | |
2355 | t._extfileobj = False | |
2356 | return t | |
2357 | ||
7584f5c9 ERE |
2358 | # All *open() methods are registered here. |
2359 | OPEN_METH = { | |
2360 | "tar": "taropen", # uncompressed tar | |
2361 | "gz": "gzopen", # gzip compressed tar | |
be60ffd0 ERE |
2362 | "bz2": "bz2open", # bzip2 compressed tar |
2363 | "xz": "xzopen" # lzma compressed tar | |
7584f5c9 ERE |
2364 | } |
2365 | ||
2366 | #-------------------------------------------------------------------------- | |
2367 | # The public methods which TarFile provides: | |
2368 | ||
2369 | def close(self): | |
2370 | """Close the TarFile. In write-mode, two finishing zero blocks are | |
fd2f01f2 PG |
2371 | appended to the archive. A special case are empty archives which are |
2372 | initialized accordingly so the two mandatory blocks of zeros are | |
2373 | written abiding by the requested encryption and compression settings. | |
7584f5c9 ERE |
2374 | """ |
2375 | if self.closed: | |
2376 | return | |
2377 | ||
2378 | if self.mode in "aw": | |
fd2f01f2 PG |
2379 | if self.arcmode & ARCMODE_CONCAT and self.fileobj.tell () == 0: |
2380 | self.fileobj.next ("") | |
7584f5c9 ERE |
2381 | self.fileobj.write(NUL * (BLOCKSIZE * 2)) |
2382 | self.offset += (BLOCKSIZE * 2) | |
2383 | # fill up the end with zero-blocks | |
2384 | # (like option -b20 for tar does) | |
2385 | blocks, remainder = divmod(self.offset, RECORDSIZE) | |
2386 | if remainder > 0: | |
2387 | self.fileobj.write(NUL * (RECORDSIZE - remainder)) | |
7584f5c9 ERE |
2388 | if not self._extfileobj: |
2389 | self.fileobj.close() | |
2390 | self.closed = True | |
2391 | ||
2392 | def getmember(self, name): | |
2393 | """Return a TarInfo object for member `name'. If `name' can not be | |
2394 | found in the archive, KeyError is raised. If a member occurs more | |
2395 | than once in the archive, its last occurrence is assumed to be the | |
2396 | most up-to-date version. | |
2397 | """ | |
2398 | tarinfo = self._getmember(name) | |
2399 | if tarinfo is None: | |
2400 | raise KeyError("filename %r not found" % name) | |
2401 | return tarinfo | |
2402 | ||
2403 | def getmembers(self): | |
2404 | """Return the members of the archive as a list of TarInfo objects. The | |
2405 | list has the same order as the members in the archive. | |
2406 | """ | |
2407 | self._check() | |
2408 | if not self._loaded: # if we want to obtain a list of | |
2409 | self._load() # all members, we first have to | |
2410 | # scan the whole archive. | |
2411 | return self.members | |
2412 | ||
ad4402e8 ERE |
2413 | def get_last_member_offset(self): |
2414 | """Return the last member offset. Usually this is self.fileobj.tell(), | |
2415 | but when there's encryption or concat compression going on it's more | |
2416 | complicated than that. | |
2417 | """ | |
b8fc2f5d | 2418 | return self.last_block_offset |
ad4402e8 | 2419 | |
7584f5c9 ERE |
2420 | def getnames(self): |
2421 | """Return the members of the archive as a list of their names. It has | |
2422 | the same order as the list returned by getmembers(). | |
2423 | """ | |
2424 | return [tarinfo.name for tarinfo in self.getmembers()] | |
2425 | ||
2426 | def gettarinfo(self, name=None, arcname=None, fileobj=None): | |
2427 | """Create a TarInfo object for either the file `name' or the file | |
2428 | object `fileobj' (using os.fstat on its file descriptor). You can | |
2429 | modify some of the TarInfo's attributes before you add it using | |
2430 | addfile(). If given, `arcname' specifies an alternative name for the | |
2431 | file in the archive. | |
2432 | """ | |
2433 | self._check("aw") | |
2434 | ||
2435 | # When fileobj is given, replace name by | |
2436 | # fileobj's real name. | |
2437 | if fileobj is not None: | |
2438 | name = fileobj.name | |
2439 | ||
2440 | # Building the name of the member in the archive. | |
2441 | # Backward slashes are converted to forward slashes, | |
2442 | # Absolute paths are turned to relative paths. | |
2443 | if arcname is None: | |
2444 | arcname = name | |
2445 | drv, arcname = os.path.splitdrive(arcname) | |
be60ffd0 | 2446 | arcname = arcname.replace(os.sep, "/") |
7584f5c9 ERE |
2447 | arcname = arcname.lstrip("/") |
2448 | ||
2449 | # Now, fill the TarInfo object with | |
2450 | # information specific for the file. | |
2451 | tarinfo = self.tarinfo() | |
2452 | tarinfo.tarfile = self | |
2453 | ||
2454 | # Use os.stat or os.lstat, depending on platform | |
2455 | # and if symlinks shall be resolved. | |
2456 | if fileobj is None: | |
2457 | if hasattr(os, "lstat") and not self.dereference: | |
2458 | statres = os.lstat(name) | |
2459 | else: | |
2460 | statres = os.stat(name) | |
2461 | else: | |
2462 | statres = os.fstat(fileobj.fileno()) | |
2463 | linkname = "" | |
2464 | ||
2465 | stmd = statres.st_mode | |
2466 | if stat.S_ISREG(stmd): | |
2467 | inode = (statres.st_ino, statres.st_dev) | |
2468 | if not self.dereference and statres.st_nlink > 1 and \ | |
2469 | inode in self.inodes and arcname != self.inodes[inode]: | |
2470 | # Is it a hardlink to an already | |
2471 | # archived file? | |
2472 | type = LNKTYPE | |
2473 | linkname = self.inodes[inode] | |
2474 | else: | |
2475 | # The inode is added only if its valid. | |
2476 | # For win32 it is always 0. | |
2477 | type = REGTYPE | |
6f422b65 | 2478 | if inode[0] and self.save_to_members: |
7584f5c9 ERE |
2479 | self.inodes[inode] = arcname |
2480 | elif stat.S_ISDIR(stmd): | |
2481 | type = DIRTYPE | |
2482 | elif stat.S_ISFIFO(stmd): | |
2483 | type = FIFOTYPE | |
2484 | elif stat.S_ISLNK(stmd): | |
2485 | type = SYMTYPE | |
2486 | linkname = os.readlink(name) | |
2487 | elif stat.S_ISCHR(stmd): | |
2488 | type = CHRTYPE | |
2489 | elif stat.S_ISBLK(stmd): | |
2490 | type = BLKTYPE | |
2491 | else: | |
2492 | return None | |
2493 | ||
2494 | # Fill the TarInfo object with all | |
2495 | # information we can get. | |
2496 | tarinfo.name = arcname | |
2497 | tarinfo.mode = stmd | |
2498 | tarinfo.uid = statres.st_uid | |
2499 | tarinfo.gid = statres.st_gid | |
2500 | if type == REGTYPE: | |
2501 | tarinfo.size = statres.st_size | |
2502 | else: | |
be60ffd0 | 2503 | tarinfo.size = 0 |
7584f5c9 ERE |
2504 | tarinfo.mtime = statres.st_mtime |
2505 | tarinfo.type = type | |
2506 | tarinfo.linkname = linkname | |
2507 | if pwd: | |
9ef1fb87 TJ |
2508 | if tarinfo.uid in self.cache_uid2user: |
2509 | tarinfo.uname = self.cache_uid2user[tarinfo.uid] | |
2510 | else: | |
2511 | try: | |
2512 | tarinfo.uname = pwd.getpwuid(tarinfo.uid)[0] | |
2513 | self.cache_uid2user[tarinfo.uid] = tarinfo.uname | |
2514 | except KeyError: | |
2515 | # remember user does not exist: | |
2516 | # same default value as in tarinfo class | |
2517 | self.cache_uid2user[tarinfo.uid] = "" | |
7584f5c9 | 2518 | if grp: |
9ef1fb87 TJ |
2519 | if tarinfo.gid in self.cache_gid2group: |
2520 | tarinfo.gname = self.cache_gid2group[tarinfo.gid] | |
2521 | else: | |
2522 | try: | |
2523 | tarinfo.gname = grp.getgrgid(tarinfo.gid)[0] | |
2524 | self.cache_gid2group[tarinfo.gid] = tarinfo.gname | |
2525 | except KeyError: | |
2526 | # remember group does not exist: | |
2527 | # same default value as in tarinfo class | |
2528 | self.cache_gid2group[tarinfo.gid] = "" | |
7584f5c9 ERE |
2529 | |
2530 | if type in (CHRTYPE, BLKTYPE): | |
2531 | if hasattr(os, "major") and hasattr(os, "minor"): | |
2532 | tarinfo.devmajor = os.major(statres.st_rdev) | |
2533 | tarinfo.devminor = os.minor(statres.st_rdev) | |
2534 | return tarinfo | |
2535 | ||
2536 | def list(self, verbose=True): | |
2537 | """Print a table of contents to sys.stdout. If `verbose' is False, only | |
2538 | the names of the members are printed. If it is True, an `ls -l'-like | |
2539 | output is produced. | |
2540 | """ | |
2541 | self._check() | |
2542 | ||
2543 | for tarinfo in self: | |
2544 | if verbose: | |
be60ffd0 ERE |
2545 | print(stat.filemode(tarinfo.mode), end=' ') |
2546 | print("%s/%s" % (tarinfo.uname or tarinfo.uid, | |
2547 | tarinfo.gname or tarinfo.gid), end=' ') | |
7584f5c9 | 2548 | if tarinfo.ischr() or tarinfo.isblk(): |
be60ffd0 ERE |
2549 | print("%10s" % ("%d,%d" \ |
2550 | % (tarinfo.devmajor, tarinfo.devminor)), end=' ') | |
7584f5c9 | 2551 | else: |
be60ffd0 ERE |
2552 | print("%10d" % tarinfo.size, end=' ') |
2553 | print("%d-%02d-%02d %02d:%02d:%02d" \ | |
2554 | % time.localtime(tarinfo.mtime)[:6], end=' ') | |
7584f5c9 | 2555 | |
be60ffd0 | 2556 | print(tarinfo.name + ("/" if tarinfo.isdir() else ""), end=' ') |
7584f5c9 ERE |
2557 | |
2558 | if verbose: | |
2559 | if tarinfo.issym(): | |
be60ffd0 | 2560 | print("->", tarinfo.linkname, end=' ') |
7584f5c9 | 2561 | if tarinfo.islnk(): |
be60ffd0 ERE |
2562 | print("link to", tarinfo.linkname, end=' ') |
2563 | print() | |
7584f5c9 | 2564 | |
be60ffd0 | 2565 | def add(self, name, arcname=None, recursive=True, exclude=None, *, filter=None): |
7584f5c9 ERE |
2566 | """Add the file `name' to the archive. `name' may be any type of file |
2567 | (directory, fifo, symbolic link, etc.). If given, `arcname' | |
2568 | specifies an alternative name for the file in the archive. | |
2569 | Directories are added recursively by default. This can be avoided by | |
2570 | setting `recursive' to False. `exclude' is a function that should | |
2571 | return True for each filename to be excluded. `filter' is a function | |
2572 | that expects a TarInfo object argument and returns the changed | |
2573 | TarInfo object, if it returns None the TarInfo object will be | |
2574 | excluded from the archive. | |
2575 | """ | |
2576 | self._check("aw") | |
2577 | ||
2578 | if arcname is None: | |
2579 | arcname = name | |
2580 | ||
2581 | # Exclude pathnames. | |
2582 | if exclude is not None: | |
2583 | import warnings | |
2584 | warnings.warn("use the filter argument instead", | |
2585 | DeprecationWarning, 2) | |
2586 | if exclude(name): | |
2587 | self._dbg(2, "tarfile: Excluded %r" % name) | |
2588 | return | |
2589 | ||
2590 | # Skip if somebody tries to archive the archive... | |
2591 | if self.name is not None and os.path.abspath(name) == self.name: | |
2592 | self._dbg(2, "tarfile: Skipped %r" % name) | |
2593 | return | |
2594 | ||
2595 | self._dbg(1, name) | |
2596 | ||
2597 | # Create a TarInfo object from the file. | |
2598 | tarinfo = self.gettarinfo(name, arcname) | |
2599 | ||
2600 | if tarinfo is None: | |
2601 | self._dbg(1, "tarfile: Unsupported type %r" % name) | |
2602 | return | |
2603 | ||
2604 | # Change or exclude the TarInfo object. | |
2605 | if filter is not None: | |
2606 | tarinfo = filter(tarinfo) | |
2607 | if tarinfo is None: | |
2608 | self._dbg(2, "tarfile: Excluded %r" % name) | |
2609 | return | |
2610 | ||
2611 | # Append the tar header and data to the archive. | |
2612 | if tarinfo.isreg(): | |
2613 | with bltn_open(name, "rb") as f: | |
2614 | self.addfile(tarinfo, f) | |
2615 | ||
2616 | elif tarinfo.isdir(): | |
2617 | self.addfile(tarinfo) | |
2618 | if recursive: | |
2619 | for f in os.listdir(name): | |
2620 | self.add(os.path.join(name, f), os.path.join(arcname, f), | |
be60ffd0 | 2621 | recursive, exclude, filter=filter) |
7584f5c9 ERE |
2622 | |
2623 | else: | |
2624 | self.addfile(tarinfo) | |
2625 | ||
defc9a22 | 2626 | def _size_left_file(self): |
be60ffd0 | 2627 | """Calculates size left in a volume with a maximum volume size. |
ba5a449e | 2628 | |
be60ffd0 | 2629 | Assumes self.max_volume_size is set. |
ba5a449e | 2630 | If using compression through a _Stream, use _size_left_stream instead |
be60ffd0 | 2631 | """ |
ba5a449e | 2632 | # left-over size = max_size - offset - 2 zero-blocks written in close |
ae48acc8 ERE |
2633 | size_left = self.max_volume_size - 2*BLOCKSIZE - self.offset |
2634 | # limit size left to a discrete number of blocks, because we won't | |
be60ffd0 | 2635 | # write only half a block when writting the end of a volume |
ae48acc8 | 2636 | # and filling with zeros |
defc9a22 CH |
2637 | return BLOCKSIZE * (size_left // BLOCKSIZE) |
2638 | ||
2639 | def _size_left_stream(self): | |
ba5a449e CH |
2640 | """ Calculates size left in a volume if using comression/encryption |
2641 | ||
2642 | Assumes self.max_volume_size is set and self.fileobj is a _Stream | |
2643 | (otherwise use _size_left_file) | |
2644 | """ | |
2645 | # left-over size = max_size - bytes written - 2 zero-blocks (close) | |
defc9a22 CH |
2646 | size_left = self.max_volume_size - self.fileobj.estim_file_size() \ |
2647 | - 2*BLOCKSIZE | |
2648 | return BLOCKSIZE * (size_left // BLOCKSIZE) | |
ae48acc8 | 2649 | |
7584f5c9 ERE |
2650 | def addfile(self, tarinfo, fileobj=None): |
2651 | """Add the TarInfo object `tarinfo' to the archive. If `fileobj' is | |
2652 | given, tarinfo.size bytes are read from it and added to the archive. | |
2653 | You can create TarInfo objects using gettarinfo(). | |
2654 | On Windows platforms, `fileobj' should always be opened with mode | |
2655 | 'rb' to avoid irritation about the file size. | |
2656 | """ | |
2657 | self._check("aw") | |
2658 | ||
2659 | tarinfo = copy.copy(tarinfo) | |
cbf55ffb | 2660 | |
d1c38f40 PG |
2661 | if self.arcmode & ARCMODE_CONCAT: |
2662 | self.last_block_offset = self.fileobj.next (tarinfo.name) | |
11684b1d ERE |
2663 | else: |
2664 | self.last_block_offset = self.fileobj.tell() | |
7584f5c9 ERE |
2665 | |
2666 | buf = tarinfo.tobuf(self.format, self.encoding, self.errors) | |
2667 | self.fileobj.write(buf) | |
2668 | self.offset += len(buf) | |
2669 | ||
ae9c8de2 CH |
2670 | if self.max_volume_size: |
2671 | if isinstance(self.fileobj, _Stream): | |
2672 | _size_left = self._size_left_stream | |
2673 | else: | |
2674 | _size_left = self._size_left_file | |
2675 | else: | |
2676 | _size_left = lambda: tarinfo.size | |
68ddf955 | 2677 | |
29c354ac PG |
2678 | # If there's no data to follow, finish |
2679 | if not fileobj: | |
29c354ac PG |
2680 | if self.save_to_members: |
2681 | self.members.append(tarinfo) | |
2682 | return | |
2683 | ||
2684 | target_size_left = _size_left() | |
2685 | source_size_left = tarinfo.size | |
2686 | assert tarinfo.volume_offset == 0 | |
2687 | ||
2688 | # we only split volumes in the middle of a file, that means we have | |
2689 | # to write at least one block | |
2690 | if target_size_left < BLOCKSIZE: | |
2691 | target_size_left = BLOCKSIZE | |
2692 | ||
ae9c8de2 CH |
2693 | # loop over multiple volumes |
2694 | while source_size_left > 0: | |
ae48acc8 | 2695 | |
ae9c8de2 CH |
2696 | # Write as much data as possble from source into target. |
2697 | # When compressing data, we cannot easily predict how much data we | |
2698 | # can write until target_size_left == 0 --> need to iterate | |
2699 | size_can_write = min(target_size_left, source_size_left) | |
c04e0751 | 2700 | |
ae9c8de2 CH |
2701 | while size_can_write > 0: |
2702 | copyfileobj(fileobj, self.fileobj, size_can_write) | |
2703 | self.offset += size_can_write | |
2704 | source_size_left -= size_can_write | |
2705 | target_size_left = _size_left() | |
2706 | size_can_write = min(target_size_left, source_size_left) | |
68ddf955 | 2707 | |
ae9c8de2 CH |
2708 | # now target_size_left == 0 or source_size_left == 0 |
2709 | ||
2710 | # if there is data left to write, we need to create a new volume | |
2711 | if source_size_left > 0: | |
5f38bff6 PG |
2712 | # Only finalize the crypto entry here if we’re continuing with |
2713 | # another one; otherwise, the encryption must include the block | |
2714 | # padding below. | |
2f854e77 | 2715 | tarinfo.type = GNUTYPE_MULTIVOL |
68ddf955 ERE |
2716 | |
2717 | if not self.new_volume_handler or\ | |
2718 | not callable(self.new_volume_handler): | |
c04e0751 | 2719 | raise Exception("We need to create a new volume and you " |
ae9c8de2 | 2720 | "didn't supply a new_volume_handler") |
68ddf955 | 2721 | |
54128a00 | 2722 | |
68ddf955 ERE |
2723 | # the new volume handler should do everything needed to |
2724 | # start working in a new volume. usually, the handler calls | |
2725 | # to self.open_volume | |
2f854e77 | 2726 | self.volume_number += 1 |
0eb5048f | 2727 | |
ae9c8de2 | 2728 | # set to be used by open_volume, because in the case of a PAX |
0eb5048f ERE |
2729 | # tar it needs to write information about the volume and offset |
2730 | # in the global header | |
ae9c8de2 | 2731 | tarinfo.volume_offset = tarinfo.size - source_size_left |
0eb5048f | 2732 | self.volume_tarinfo = tarinfo |
ae9c8de2 | 2733 | |
a0873dcc PG |
2734 | # the “new_volume_handler” is supposed to call .close() on the |
2735 | # “fileobj” _Stream | |
2f854e77 ERE |
2736 | self.new_volume_handler(self, self.base_name, self.volume_number) |
2737 | ||
0eb5048f ERE |
2738 | self.volume_tarinfo = None |
2739 | ||
d1c38f40 PG |
2740 | if self.arcmode & ARCMODE_CONCAT: |
2741 | self.fileobj.next_volume (tarinfo.name) | |
5f38bff6 | 2742 | |
2f854e77 ERE |
2743 | # write new volume header |
2744 | buf = tarinfo.tobuf(self.format, self.encoding, self.errors) | |
2f854e77 | 2745 | self.fileobj.write(buf) |
ae9c8de2 CH |
2746 | self.offset += len(buf) |
2747 | ||
2748 | # adjust variables; open_volume should have reset self.offset | |
2749 | # --> _size_left should be big again | |
2750 | target_size_left = _size_left() | |
2751 | size_can_write = min(target_size_left, source_size_left) | |
e0da4709 | 2752 | self._dbg(3, 'new volume') |
ae9c8de2 CH |
2753 | |
2754 | # now, all data has been written. We may have to fill up the rest of | |
2755 | # the block in target with 0s | |
2756 | remainder = (tarinfo.size - tarinfo.volume_offset) % BLOCKSIZE | |
2757 | if remainder > 0: | |
2758 | self.fileobj.write(NUL * (BLOCKSIZE - remainder)) | |
2759 | self.offset += BLOCKSIZE - remainder | |
7584f5c9 | 2760 | |
ea625b04 ERE |
2761 | if self.save_to_members: |
2762 | self.members.append(tarinfo) | |
7584f5c9 | 2763 | |
170c6c52 | 2764 | def open_volume(self, name="", fileobj=None, encryption=None): |
68ddf955 | 2765 | ''' |
0eb5048f | 2766 | Called by the user to change this tar file to point to a new volume. |
68ddf955 | 2767 | ''' |
27ee4dd4 | 2768 | |
68ddf955 ERE |
2769 | # open the file using either fileobj or name |
2770 | if not fileobj: | |
2771 | if self.mode == "a" and not os.path.exists(name): | |
2772 | # Create nonexistent files in append mode. | |
2773 | self.mode = "w" | |
2774 | self._mode = "wb" | |
68ddf955 | 2775 | self._extfileobj = False |
26fa5ad5 ERE |
2776 | |
2777 | if isinstance(self.fileobj, _Stream): | |
e0da4709 | 2778 | self._dbg(3, 'open_volume: create a _Stream') |
26fa5ad5 ERE |
2779 | fileobj = _Stream(name=name, |
2780 | mode=self.fileobj.mode, | |
2781 | comptype=self.fileobj.comptype, | |
2782 | fileobj=None, | |
2783 | bufsize=self.fileobj.bufsize, | |
cea130ec | 2784 | encryption=encryption or self.fileobj.encryption, |
27ee4dd4 PG |
2785 | concat=self.fileobj.arcmode & ARCMODE_CONCAT, |
2786 | tolerance=self.fileobj.tolerance) | |
26fa5ad5 | 2787 | else: |
7a2b9329 | 2788 | # here, we lose information about compression/encryption! |
e0da4709 | 2789 | self._dbg(3, 'open_volume: builtin open') |
26fa5ad5 | 2790 | fileobj = bltn_open(name, self._mode) |
68ddf955 ERE |
2791 | else: |
2792 | if name is None and hasattr(fileobj, "name"): | |
2793 | name = fileobj.name | |
2794 | if hasattr(fileobj, "mode"): | |
2795 | self._mode = fileobj.mode | |
2796 | self._extfileobj = True | |
1027433a | 2797 | self._dbg(3, 'open_volume: using external fileobj {}', fileobj) |
68ddf955 | 2798 | self.name = os.path.abspath(name) if name else None |
5cf23eab | 2799 | self.fileobj.close() |
68ddf955 ERE |
2800 | self.fileobj = fileobj |
2801 | ||
2802 | # init data structures | |
2803 | self.closed = False | |
2804 | self.members = [] # list of members as TarInfo objects | |
2805 | self._loaded = False # flag if all members have been read | |
2806 | self.offset = self.fileobj.tell() | |
2807 | # current position in the archive file | |
2808 | self.inodes = {} # dictionary caching the inodes of | |
2809 | # archive members already added | |
2810 | ||
2811 | try: | |
2812 | if self.mode == "r": | |
2813 | self.firstmember = None | |
2814 | self.firstmember = self.next() | |
2815 | ||
2816 | if self.mode == "a": | |
2817 | # Move to the end of the archive, | |
2818 | # before the first empty block. | |
2819 | while True: | |
2820 | self.fileobj.seek(self.offset) | |
2821 | try: | |
2822 | tarinfo = self.tarinfo.fromtarfile(self) | |
2823 | self.members.append(tarinfo) | |
2824 | except EOFHeaderError: | |
2825 | self.fileobj.seek(self.offset) | |
2826 | break | |
be60ffd0 | 2827 | except HeaderError as e: |
68ddf955 ERE |
2828 | raise ReadError(str(e)) |
2829 | ||
2830 | if self.mode in "aw": | |
2831 | self._loaded = True | |
2832 | ||
c04e0751 ERE |
2833 | if self.format == PAX_FORMAT: |
2834 | volume_info = { | |
be60ffd0 ERE |
2835 | "GNU.volume.filename": str(self.volume_tarinfo.name), |
2836 | "GNU.volume.size": str(self.volume_tarinfo.size - self.volume_tarinfo.volume_offset), | |
2837 | "GNU.volume.offset": str(self.volume_tarinfo.volume_offset), | |
c04e0751 | 2838 | } |
0eb5048f | 2839 | |
c04e0751 ERE |
2840 | self.pax_headers.update(volume_info) |
2841 | ||
a0873dcc PG |
2842 | if isinstance(self.fileobj, _Stream): |
2843 | self.fileobj._init_write_gz () | |
c04e0751 ERE |
2844 | buf = self.tarinfo.create_pax_global_header(volume_info.copy()) |
2845 | self.fileobj.write(buf) | |
2846 | self.offset += len(buf) | |
54128a00 | 2847 | except Exception as exn: |
68ddf955 ERE |
2848 | if not self._extfileobj: |
2849 | self.fileobj.close() | |
2850 | self.closed = True | |
2851 | raise | |
2852 | ||
c650acfa | 2853 | def extractall(self, path=".", members=None, filter=None, unlink=False): |
7584f5c9 ERE |
2854 | """Extract all members from the archive to the current working |
2855 | directory and set owner, modification time and permissions on | |
2856 | directories afterwards. `path' specifies a different directory | |
2857 | to extract to. `members' is optional and must be a subset of the | |
2858 | list returned by getmembers(). | |
2859 | """ | |
2860 | directories = [] | |
2861 | ||
2862 | if members is None: | |
2863 | members = self | |
2864 | ||
2865 | for tarinfo in members: | |
c474439c ERE |
2866 | if self.volume_number > 0 and tarinfo.ismultivol(): |
2867 | continue | |
2868 | ||
974408b5 | 2869 | if filter and not filter(tarinfo): |
e5f5681b ERE |
2870 | continue |
2871 | ||
7584f5c9 ERE |
2872 | if tarinfo.isdir(): |
2873 | # Extract directories with a safe mode. | |
2874 | directories.append(tarinfo) | |
2875 | tarinfo = copy.copy(tarinfo) | |
be60ffd0 ERE |
2876 | tarinfo.mode = 0o0700 |
2877 | # Do not set_attrs directories, as we will do that further down | |
c650acfa | 2878 | self.extract(tarinfo, path, set_attrs=not tarinfo.isdir(), unlink=unlink) |
7584f5c9 ERE |
2879 | |
2880 | # Reverse sort directories. | |
be60ffd0 | 2881 | directories.sort(key=lambda a: a.name) |
7584f5c9 ERE |
2882 | directories.reverse() |
2883 | ||
2884 | # Set correct owner, mtime and filemode on directories. | |
2885 | for tarinfo in directories: | |
2886 | dirpath = os.path.join(path, tarinfo.name) | |
2887 | try: | |
2888 | self.chown(tarinfo, dirpath) | |
2889 | self.utime(tarinfo, dirpath) | |
2890 | self.chmod(tarinfo, dirpath) | |
be60ffd0 | 2891 | except ExtractError as e: |
7584f5c9 ERE |
2892 | if self.errorlevel > 1: |
2893 | raise | |
2894 | else: | |
2895 | self._dbg(1, "tarfile: %s" % e) | |
2896 | ||
c650acfa PG |
2897 | def extract(self, member, path="", set_attrs=True, symlink_cb=None, |
2898 | unlink=False): | |
7584f5c9 ERE |
2899 | """Extract a member from the archive to the current working directory, |
2900 | using its full name. Its file information is extracted as accurately | |
2901 | as possible. `member' may be a filename or a TarInfo object. You can | |
be60ffd0 ERE |
2902 | specify a different directory using `path'. File attributes (owner, |
2903 | mtime, mode) are set unless `set_attrs' is False. | |
786addd6 PG |
2904 | ``symlink_cb`` is a hook accepting a function that is passed the |
2905 | ``member``, ``path``, and ``set_attrs`` arguments if the tarinfo for | |
2906 | ``member`` indicates a symlink in which case only the callback | |
9b13f5c4 PG |
2907 | passed will be applied, skipping the actual extraction. In case the |
2908 | callback is invoked, its return value is passed on to the caller. | |
7584f5c9 ERE |
2909 | """ |
2910 | self._check("r") | |
2911 | ||
be60ffd0 | 2912 | if isinstance(member, str): |
7584f5c9 ERE |
2913 | tarinfo = self.getmember(member) |
2914 | else: | |
2915 | tarinfo = member | |
2916 | ||
2917 | # Prepare the link target for makelink(). | |
2918 | if tarinfo.islnk(): | |
2919 | tarinfo._link_target = os.path.join(path, tarinfo.linkname) | |
2920 | ||
9b13f5c4 | 2921 | if symlink_cb is not None and tarinfo.issym(): |
83f5fd71 | 2922 | return symlink_cb(member, path, set_attrs) |
786addd6 | 2923 | |
7584f5c9 | 2924 | try: |
be60ffd0 | 2925 | self._extract_member(tarinfo, os.path.join(path, tarinfo.name), |
c650acfa | 2926 | set_attrs=set_attrs, unlink=unlink) |
be60ffd0 | 2927 | except EnvironmentError as e: |
7584f5c9 ERE |
2928 | if self.errorlevel > 0: |
2929 | raise | |
2930 | else: | |
2931 | if e.filename is None: | |
2932 | self._dbg(1, "tarfile: %s" % e.strerror) | |
2933 | else: | |
2934 | self._dbg(1, "tarfile: %s %r" % (e.strerror, e.filename)) | |
be60ffd0 | 2935 | except ExtractError as e: |
7584f5c9 ERE |
2936 | if self.errorlevel > 1: |
2937 | raise | |
2938 | else: | |
2939 | self._dbg(1, "tarfile: %s" % e) | |
2940 | ||
2941 | def extractfile(self, member): | |
2942 | """Extract a member from the archive as a file object. `member' may be | |
be60ffd0 ERE |
2943 | a filename or a TarInfo object. If `member' is a regular file or a |
2944 | link, an io.BufferedReader object is returned. Otherwise, None is | |
2945 | returned. | |
7584f5c9 ERE |
2946 | """ |
2947 | self._check("r") | |
2948 | ||
be60ffd0 | 2949 | if isinstance(member, str): |
7584f5c9 ERE |
2950 | tarinfo = self.getmember(member) |
2951 | else: | |
2952 | tarinfo = member | |
2953 | ||
be60ffd0 ERE |
2954 | if tarinfo.isreg() or tarinfo.ismultivol() or\ |
2955 | tarinfo.type not in SUPPORTED_TYPES: | |
7584f5c9 ERE |
2956 | # If a member's type is unknown, it is treated as a |
2957 | # regular file. | |
2958 | return self.fileobject(self, tarinfo) | |
2959 | ||
2960 | elif tarinfo.islnk() or tarinfo.issym(): | |
2961 | if isinstance(self.fileobj, _Stream): | |
2962 | # A small but ugly workaround for the case that someone tries | |
2963 | # to extract a (sym)link as a file-object from a non-seekable | |
2964 | # stream of tar blocks. | |
2965 | raise StreamError("cannot extract (sym)link as file object") | |
2966 | else: | |
2967 | # A (sym)link's file object is its target's file object. | |
2968 | return self.extractfile(self._find_link_target(tarinfo)) | |
2969 | else: | |
2970 | # If there's no data associated with the member (directory, chrdev, | |
2971 | # blkdev, etc.), return None instead of a file object. | |
2972 | return None | |
2973 | ||
c650acfa | 2974 | def _extract_member(self, tarinfo, targetpath, set_attrs=True, unlink=False): |
7584f5c9 ERE |
2975 | """Extract the TarInfo object tarinfo to a physical |
2976 | file called targetpath. | |
2977 | """ | |
2978 | # Fetch the TarInfo object for the given name | |
2979 | # and build the destination pathname, replacing | |
2980 | # forward slashes to platform specific separators. | |
2981 | targetpath = targetpath.rstrip("/") | |
2982 | targetpath = targetpath.replace("/", os.sep) | |
2983 | ||
2984 | # Create all upper directories. | |
2985 | upperdirs = os.path.dirname(targetpath) | |
2986 | if upperdirs and not os.path.exists(upperdirs): | |
2987 | # Create directories that are not part of the archive with | |
2988 | # default permissions. | |
2989 | os.makedirs(upperdirs) | |
2990 | ||
2991 | if tarinfo.islnk() or tarinfo.issym(): | |
2992 | self._dbg(1, "%s -> %s" % (tarinfo.name, tarinfo.linkname)) | |
2993 | else: | |
2994 | self._dbg(1, tarinfo.name) | |
2995 | ||
c650acfa PG |
2996 | if unlink is True: |
2997 | _unlinkfirst(targetpath) | |
2998 | ||
7584f5c9 ERE |
2999 | if tarinfo.isreg(): |
3000 | self.makefile(tarinfo, targetpath) | |
3001 | elif tarinfo.isdir(): | |
3002 | self.makedir(tarinfo, targetpath) | |
3003 | elif tarinfo.isfifo(): | |
3004 | self.makefifo(tarinfo, targetpath) | |
3005 | elif tarinfo.ischr() or tarinfo.isblk(): | |
3006 | self.makedev(tarinfo, targetpath) | |
3007 | elif tarinfo.islnk() or tarinfo.issym(): | |
3008 | self.makelink(tarinfo, targetpath) | |
3009 | elif tarinfo.type not in SUPPORTED_TYPES: | |
3010 | self.makeunknown(tarinfo, targetpath) | |
3011 | else: | |
3012 | self.makefile(tarinfo, targetpath) | |
3013 | ||
be60ffd0 ERE |
3014 | if set_attrs: |
3015 | self.chown(tarinfo, targetpath) | |
3016 | if not tarinfo.issym(): | |
3017 | self.chmod(tarinfo, targetpath) | |
3018 | self.utime(tarinfo, targetpath) | |
7584f5c9 ERE |
3019 | |
3020 | #-------------------------------------------------------------------------- | |
3021 | # Below are the different file methods. They are called via | |
3022 | # _extract_member() when extract() is called. They can be replaced in a | |
3023 | # subclass to implement other functionality. | |
3024 | ||
3025 | def makedir(self, tarinfo, targetpath): | |
3026 | """Make a directory called targetpath. | |
3027 | """ | |
3028 | try: | |
3029 | # Use a safe mode for the directory, the real mode is set | |
3030 | # later in _extract_member(). | |
be60ffd0 ERE |
3031 | os.mkdir(targetpath, 0o0700) |
3032 | except FileExistsError: | |
3033 | pass | |
7584f5c9 ERE |
3034 | |
3035 | def makefile(self, tarinfo, targetpath): | |
3036 | """Make a file called targetpath. | |
3037 | """ | |
be60ffd0 ERE |
3038 | source = self.fileobj |
3039 | source.seek(tarinfo.offset_data) | |
c7c736b6 | 3040 | decrypt = False |
c474439c ERE |
3041 | iterate = True |
3042 | target = bltn_open(targetpath, "wb") | |
3043 | ||
be60ffd0 ERE |
3044 | if tarinfo.sparse is not None: |
3045 | try: | |
3046 | for offset, size in tarinfo.sparse: | |
3047 | target.seek(offset) | |
3048 | copyfileobj(source, target, size) | |
3049 | target.seek(tarinfo.size) | |
3050 | target.truncate() | |
3051 | finally: | |
3052 | target.close() | |
3053 | return | |
3054 | ||
c474439c ERE |
3055 | while iterate: |
3056 | iterate = False | |
3057 | try: | |
3058 | copyfileobj(source, target, tarinfo.size) | |
aa828cd1 | 3059 | except OSError: |
c474439c ERE |
3060 | source.close() |
3061 | # only if we are extracting a multivolume this can be treated | |
3062 | if not self.new_volume_handler: | |
c474439c ERE |
3063 | raise Exception("We need to read a new volume and you" |
3064 | " didn't supply a new_volume_handler") | |
3065 | ||
3066 | # the new volume handler should do everything needed to | |
3067 | # start working in a new volume. usually, the handler calls | |
3068 | # to self.open_volume | |
3069 | self.volume_number += 1 | |
3070 | self.new_volume_handler(self, self.base_name, self.volume_number) | |
be60ffd0 ERE |
3071 | tarinfo = self.firstmember |
3072 | source = self.fileobj | |
c474439c | 3073 | iterate = True |
bcc8b174 PG |
3074 | finally: |
3075 | if iterate is False: target.close() | |
c474439c | 3076 | |
7584f5c9 ERE |
3077 | |
3078 | def makeunknown(self, tarinfo, targetpath): | |
3079 | """Make a file from a TarInfo object with an unknown type | |
3080 | at targetpath. | |
3081 | """ | |
3082 | self.makefile(tarinfo, targetpath) | |
3083 | self._dbg(1, "tarfile: Unknown file type %r, " \ | |
3084 | "extracted as regular file." % tarinfo.type) | |
3085 | ||
3086 | def makefifo(self, tarinfo, targetpath): | |
3087 | """Make a fifo called targetpath. | |
3088 | """ | |
3089 | if hasattr(os, "mkfifo"): | |
3090 | os.mkfifo(targetpath) | |
3091 | else: | |
3092 | raise ExtractError("fifo not supported by system") | |
3093 | ||
3094 | def makedev(self, tarinfo, targetpath): | |
3095 | """Make a character or block device called targetpath. | |
3096 | """ | |
3097 | if not hasattr(os, "mknod") or not hasattr(os, "makedev"): | |
3098 | raise ExtractError("special devices not supported by system") | |
3099 | ||
3100 | mode = tarinfo.mode | |
3101 | if tarinfo.isblk(): | |
3102 | mode |= stat.S_IFBLK | |
3103 | else: | |
3104 | mode |= stat.S_IFCHR | |
3105 | ||
3106 | os.mknod(targetpath, mode, | |
3107 | os.makedev(tarinfo.devmajor, tarinfo.devminor)) | |
3108 | ||
3109 | def makelink(self, tarinfo, targetpath): | |
3110 | """Make a (symbolic) link called targetpath. If it cannot be created | |
3111 | (platform limitation), we try to make a copy of the referenced file | |
3112 | instead of a link. | |
3113 | """ | |
be60ffd0 | 3114 | try: |
7584f5c9 ERE |
3115 | # For systems that support symbolic and hard links. |
3116 | if tarinfo.issym(): | |
7584f5c9 ERE |
3117 | os.symlink(tarinfo.linkname, targetpath) |
3118 | else: | |
3119 | # See extract(). | |
3120 | if os.path.exists(tarinfo._link_target): | |
7584f5c9 ERE |
3121 | os.link(tarinfo._link_target, targetpath) |
3122 | else: | |
be60ffd0 ERE |
3123 | self._extract_member(self._find_link_target(tarinfo), |
3124 | targetpath) | |
3125 | except symlink_exception: | |
7584f5c9 | 3126 | try: |
be60ffd0 ERE |
3127 | self._extract_member(self._find_link_target(tarinfo), |
3128 | targetpath) | |
7584f5c9 ERE |
3129 | except KeyError: |
3130 | raise ExtractError("unable to resolve link inside archive") | |
3131 | ||
3132 | def chown(self, tarinfo, targetpath): | |
3133 | """Set owner of targetpath according to tarinfo. | |
3134 | """ | |
3135 | if pwd and hasattr(os, "geteuid") and os.geteuid() == 0: | |
3136 | # We have to be root to do so. | |
3137 | try: | |
3138 | g = grp.getgrnam(tarinfo.gname)[2] | |
3139 | except KeyError: | |
3140 | g = tarinfo.gid | |
3141 | try: | |
3142 | u = pwd.getpwnam(tarinfo.uname)[2] | |
3143 | except KeyError: | |
3144 | u = tarinfo.uid | |
3145 | try: | |
3146 | if tarinfo.issym() and hasattr(os, "lchown"): | |
3147 | os.lchown(targetpath, u, g) | |
3148 | else: | |
be60ffd0 ERE |
3149 | os.chown(targetpath, u, g) |
3150 | except OSError as e: | |
7584f5c9 ERE |
3151 | raise ExtractError("could not change owner") |
3152 | ||
3153 | def chmod(self, tarinfo, targetpath): | |
3154 | """Set file permissions of targetpath according to tarinfo. | |
3155 | """ | |
3156 | if hasattr(os, 'chmod'): | |
3157 | try: | |
3158 | os.chmod(targetpath, tarinfo.mode) | |
be60ffd0 | 3159 | except OSError as e: |
7584f5c9 ERE |
3160 | raise ExtractError("could not change mode") |
3161 | ||
3162 | def utime(self, tarinfo, targetpath): | |
3163 | """Set modification time of targetpath according to tarinfo. | |
3164 | """ | |
3165 | if not hasattr(os, 'utime'): | |
3166 | return | |
3167 | try: | |
3168 | os.utime(targetpath, (tarinfo.mtime, tarinfo.mtime)) | |
be60ffd0 | 3169 | except OSError as e: |
7584f5c9 ERE |
3170 | raise ExtractError("could not change modification time") |
3171 | ||
3172 | #-------------------------------------------------------------------------- | |
3173 | def next(self): | |
3174 | """Return the next member of the archive as a TarInfo object, when | |
3175 | TarFile is opened for reading. Return None if there is no more | |
3176 | available. | |
3177 | """ | |
3178 | self._check("ra") | |
3179 | if self.firstmember is not None: | |
3180 | m = self.firstmember | |
3181 | self.firstmember = None | |
3182 | return m | |
3183 | ||
be60ffd0 ERE |
3184 | # Read the next block. |
3185 | self.fileobj.seek(self.offset) | |
7584f5c9 ERE |
3186 | tarinfo = None |
3187 | while True: | |
3188 | try: | |
3189 | tarinfo = self.tarinfo.fromtarfile(self) | |
be60ffd0 | 3190 | except EOFHeaderError as e: |
7584f5c9 ERE |
3191 | if self.ignore_zeros: |
3192 | self._dbg(2, "0x%X: %s" % (self.offset, e)) | |
3193 | self.offset += BLOCKSIZE | |
3194 | continue | |
be60ffd0 | 3195 | except InvalidHeaderError as e: |
7584f5c9 ERE |
3196 | if self.ignore_zeros: |
3197 | self._dbg(2, "0x%X: %s" % (self.offset, e)) | |
3198 | self.offset += BLOCKSIZE | |
3199 | continue | |
3200 | elif self.offset == 0: | |
3201 | raise ReadError(str(e)) | |
3202 | except EmptyHeaderError: | |
3203 | if self.offset == 0: | |
3204 | raise ReadError("empty file") | |
be60ffd0 | 3205 | except TruncatedHeaderError as e: |
7584f5c9 ERE |
3206 | if self.offset == 0: |
3207 | raise ReadError(str(e)) | |
be60ffd0 | 3208 | except SubsequentHeaderError as e: |
7584f5c9 ERE |
3209 | raise ReadError(str(e)) |
3210 | break | |
3211 | ||
3212 | if tarinfo is not None: | |
ea625b04 ERE |
3213 | if self.save_to_members: |
3214 | self.members.append(tarinfo) | |
7584f5c9 ERE |
3215 | else: |
3216 | self._loaded = True | |
3217 | ||
3218 | return tarinfo | |
3219 | ||
3220 | #-------------------------------------------------------------------------- | |
3221 | # Little helper methods: | |
3222 | ||
3223 | def _getmember(self, name, tarinfo=None, normalize=False): | |
3224 | """Find an archive member by name from bottom to top. | |
3225 | If tarinfo is given, it is used as the starting point. | |
3226 | """ | |
3227 | # Ensure that all members have been loaded. | |
3228 | members = self.getmembers() | |
3229 | ||
3230 | # Limit the member search list up to tarinfo. | |
3231 | if tarinfo is not None: | |
3232 | members = members[:members.index(tarinfo)] | |
3233 | ||
3234 | if normalize: | |
3235 | name = os.path.normpath(name) | |
3236 | ||
3237 | for member in reversed(members): | |
3238 | if normalize: | |
3239 | member_name = os.path.normpath(member.name) | |
3240 | else: | |
3241 | member_name = member.name | |
3242 | ||
3243 | if name == member_name: | |
3244 | return member | |
3245 | ||
3246 | def _load(self): | |
3247 | """Read through the entire archive file and look for readable | |
3248 | members. | |
3249 | """ | |
3250 | while True: | |
3251 | tarinfo = self.next() | |
3252 | if tarinfo is None: | |
3253 | break | |
3254 | self._loaded = True | |
3255 | ||
3256 | def _check(self, mode=None): | |
3257 | """Check if TarFile is still open, and if the operation's mode | |
3258 | corresponds to TarFile's mode. | |
3259 | """ | |
3260 | if self.closed: | |
be60ffd0 | 3261 | raise OSError("%s is closed" % self.__class__.__name__) |
7584f5c9 | 3262 | if mode is not None and self.mode not in mode: |
be60ffd0 | 3263 | raise OSError("bad operation for mode %r" % self.mode) |
7584f5c9 ERE |
3264 | |
3265 | def _find_link_target(self, tarinfo): | |
3266 | """Find the target member of a symlink or hardlink member in the | |
3267 | archive. | |
3268 | """ | |
3269 | if tarinfo.issym(): | |
3270 | # Always search the entire archive. | |
3271 | linkname = "/".join(filter(None, (os.path.dirname(tarinfo.name), tarinfo.linkname))) | |
3272 | limit = None | |
3273 | else: | |
3274 | # Search the archive before the link, because a hard link is | |
3275 | # just a reference to an already archived file. | |
3276 | linkname = tarinfo.linkname | |
3277 | limit = tarinfo | |
3278 | ||
3279 | member = self._getmember(linkname, tarinfo=limit, normalize=True) | |
3280 | if member is None: | |
3281 | raise KeyError("linkname %r not found" % linkname) | |
3282 | return member | |
3283 | ||
3284 | def __iter__(self): | |
3285 | """Provide an iterator object. | |
3286 | """ | |
3287 | if self._loaded: | |
3288 | return iter(self.members) | |
3289 | else: | |
3290 | return TarIter(self) | |
3291 | ||
1027433a | 3292 | def _dbg(self, level, msg, *args): |
7584f5c9 ERE |
3293 | """Write debugging output to sys.stderr. |
3294 | """ | |
3295 | if level <= self.debug: | |
1027433a | 3296 | print(msg.format(*args), file=sys.stderr) |
7584f5c9 ERE |
3297 | |
3298 | def __enter__(self): | |
3299 | self._check() | |
3300 | return self | |
3301 | ||
3302 | def __exit__(self, type, value, traceback): | |
3303 | if type is None: | |
3304 | self.close() | |
3305 | else: | |
3306 | # An exception occurred. We must not call close() because | |
3307 | # it would try to write end-of-archive blocks and padding. | |
3308 | if not self._extfileobj: | |
3309 | self.fileobj.close() | |
3310 | self.closed = True | |
c650acfa PG |
3311 | |
3312 | def _unlinkfirst(targetpath): | |
3313 | try: | |
3314 | os.unlink(targetpath) | |
3315 | except OSError as e: | |
3316 | if e.errno == errno.ENOENT or e.errno == errno.EISDIR: | |
3317 | pass | |
3318 | ||
3319 | ||
7584f5c9 ERE |
3320 | # class TarFile |
3321 | ||
3322 | class TarIter: | |
3323 | """Iterator Class. | |
3324 | ||
3325 | for tarinfo in TarFile(...): | |
3326 | suite... | |
3327 | """ | |
3328 | ||
3329 | def __init__(self, tarfile): | |
3330 | """Construct a TarIter object. | |
3331 | """ | |
3332 | self.tarfile = tarfile | |
3333 | self.index = 0 | |
3334 | def __iter__(self): | |
3335 | """Return iterator object. | |
3336 | """ | |
3337 | return self | |
be60ffd0 | 3338 | def __next__(self): |
7584f5c9 ERE |
3339 | """Return the next item using TarFile's next() method. |
3340 | When all members have been read, set TarFile as _loaded. | |
3341 | """ | |
3342 | # Fix for SF #1100429: Under rare circumstances it can | |
3343 | # happen that getmembers() is called during iteration, | |
3344 | # which will cause TarIter to stop prematurely. | |
3345 | ||
3346 | if self.index == 0 and self.tarfile.firstmember is not None: | |
3347 | tarinfo = self.tarfile.next() | |
3348 | elif self.index < len(self.tarfile.members): | |
3349 | tarinfo = self.tarfile.members[self.index] | |
3350 | elif not self.tarfile._loaded: | |
3351 | tarinfo = self.tarfile.next() | |
3352 | if not tarinfo: | |
3353 | self.tarfile._loaded = True | |
3354 | raise StopIteration | |
3355 | else: | |
3356 | raise StopIteration | |
3357 | self.index += 1 | |
fb27c6e8 | 3358 | |
7584f5c9 ERE |
3359 | return tarinfo |
3360 | ||
6690f5e0 PG |
3361 | #--------------------------------------------------------- |
3362 | # support functionality for rescue mode | |
3363 | #--------------------------------------------------------- | |
3364 | ||
8fc6040c PG |
3365 | TAR_FMT_HDR = (# See tar(5): |
3366 | "<" | |
3367 | "100s" # ← char name[100]; /* 100 */ | |
3368 | "8s" # ← char mode[8]; /* 108 */ | |
3369 | "8s" # ← char uid[8]; /* 116 */ | |
3370 | "8s" # ← char gid[8]; /* 124 */ | |
3371 | "12s" # ← char size[12]; /* 136 */ | |
3372 | "12s" # ← char mtime[12]; /* 148 */ | |
3373 | "8s" # ← char checksum[8]; /* 156 */ | |
3374 | "B" # ← char typeflag[1]; /* 157 */ | |
3375 | "100s" # ← char linkname[100]; /* 257 */ | |
3376 | "6s" # ← char magic[6]; /* 263 */ | |
3377 | "2s" # ← char version[2]; /* 265 */ | |
3378 | "32s" # ← char uname[32]; /* 297 */ | |
3379 | "32s" # ← char gname[32]; /* 329 */ | |
3380 | "8s" # ← char devmajor[8]; /* 337 */ | |
3381 | "8s" # ← char devminor[8]; /* 345 */ | |
3382 | "12s" # ← char atime[12]; /* 357 */ | |
3383 | "12s" # ← char ctime[12]; /* 369 */ | |
3384 | "12s" # ← char offset[12]; /* 381 */ | |
3385 | "4s" # ← char longnames[4]; /* 385 */ | |
3386 | "B" # ← char unused[1]; /* 386 */ | |
3387 | "" # struct { | |
3388 | "12s" # ← char offset[12]; | |
3389 | "12s" # ← char numbytes[12]; | |
3390 | "12s" # ← char offset[12]; | |
3391 | "12s" # ← char numbytes[12]; | |
3392 | "12s" # ← char offset[12]; | |
3393 | "12s" # ← char numbytes[12]; | |
3394 | "12s" # ← char offset[12]; | |
3395 | "12s" # ← char numbytes[12]; | |
3396 | "" # } sparse[4]; /* 482 */ | |
3397 | "B" # ← char isextended[1]; /* 483 */ | |
3398 | "12s" # ← char realsize[12]; /* 495 */ | |
3399 | "17s" # ← char pad[17]; /* 512 */ | |
3400 | ) | |
3401 | ||
3402 | # The “magic” and “version” fields are special: | |
3403 | # | |
3404 | # tar(5) | |
3405 | # magic The magic field holds the five characters “ustar” followed by a | |
3406 | # space. Note that POSIX ustar archives have a trailing null. | |
3407 | # | |
3408 | # however, “tar.h”: | |
3409 | # | |
3410 | # /* OLDGNU_MAGIC uses both magic and version fields, which are contiguous. | |
3411 | # Found in an archive, it indicates an old GNU header format, which will be | |
3412 | # hopefully become obsolescent. With OLDGNU_MAGIC, uname and gname are | |
3413 | # valid, though the header is not truly POSIX conforming. */ | |
3414 | # | |
3415 | # | |
a793ee30 | 3416 | TAR_HDR_OFF_MAGIC = 257 |
8fc6040c PG |
3417 | TAR_FMT_OLDGNU_MAGIC = b"ustar " |
3418 | ||
3419 | def read_gnu_tar_hdr (data): | |
3420 | if len (data) != BLOCKSIZE: # header requires one complete block | |
3421 | return None | |
65b35c42 | 3422 | |
8fc6040c PG |
3423 | try: |
3424 | name, mode, \ | |
3425 | uid, gid, \ | |
3426 | size, mtime, \ | |
3427 | checksum, \ | |
3428 | typeflag, \ | |
3429 | linkname, \ | |
3430 | magic, \ | |
3431 | version, \ | |
3432 | uname, \ | |
3433 | gname, \ | |
3434 | devmajor, \ | |
3435 | devminor, \ | |
3436 | atime, \ | |
3437 | ctime, \ | |
3438 | offset, \ | |
3439 | longnames, \ | |
3440 | unused, \ | |
3441 | offset1, numbytes1, \ | |
3442 | offset2, numbytes2, \ | |
3443 | offset3, numbytes3, \ | |
3444 | offset4, numbytes4, \ | |
3445 | isextended, \ | |
3446 | realsize, \ | |
3447 | pad = struct.unpack (TAR_FMT_HDR, data) | |
3448 | except struct.error: | |
3449 | return None | |
3450 | ||
3451 | if magic != TAR_FMT_OLDGNU_MAGIC: | |
3452 | return None | |
3453 | ||
3454 | # return all except “unused” and “pad” | |
3455 | return \ | |
3456 | { "name" : name, "mode" : mode | |
3457 | , "uid" : uid , "gid" : gid | |
3458 | , "size" : size, "mtime" : mtime | |
3459 | , "checksum" : checksum | |
3460 | , "typeflag" : typeflag | |
3461 | , "linkname" : linkname | |
3462 | , "magic" : magic | |
3463 | , "version" : version | |
3464 | , "uname" : uname, "gname" : gname | |
3465 | , "devmajor" : devmajor, "devminor" : devminor | |
3466 | , "atime" : atime, "ctime" : ctime | |
3467 | , "offset" : offset | |
3468 | , "longnames" : longnames | |
3469 | , "offset1" : offset1, "numbytes1" : numbytes1 | |
3470 | , "offset2" : offset2, "numbytes2" : numbytes2 | |
3471 | , "offset3" : offset3, "numbytes3" : numbytes3 | |
3472 | , "offset4" : offset4, "numbytes4" : numbytes4 | |
3473 | , "isextended" : isextended | |
3474 | , "realsize" : realsize | |
3475 | } | |
3476 | ||
3477 | ||
a793ee30 PG |
3478 | def tar_hdr_check_chksum (data): |
3479 | hdr = read_gnu_tar_hdr (data) | |
3480 | if hdr is None: | |
3481 | return False | |
3482 | s = calc_chksums (data) | |
3483 | return nti (hdr ["checksum"]) in s | |
3484 | ||
3485 | ||
8fc6040c PG |
3486 | def readable_tar_objects_offsets (ifd): |
3487 | """ | |
3488 | Traverse blocks in file, trying to extract tar headers. | |
3489 | """ | |
3490 | pos = 0 | |
3491 | offsets = [] | |
3492 | ||
a793ee30 PG |
3493 | mm = mmap.mmap(ifd, 0, mmap.MAP_SHARED, mmap.PROT_READ) |
3494 | pos = TAR_HDR_OFF_MAGIC | |
3495 | ||
8fc6040c | 3496 | while True: |
a793ee30 PG |
3497 | pos = mm.find (TAR_FMT_OLDGNU_MAGIC, pos) |
3498 | if pos == -1: | |
8fc6040c | 3499 | break |
a793ee30 PG |
3500 | off = pos - TAR_HDR_OFF_MAGIC |
3501 | mm.seek (off) | |
3502 | blk = mm.read (BLOCKSIZE) | |
3503 | if tar_hdr_check_chksum (blk) is True: | |
3504 | offsets.append (off) | |
3505 | pos += 1 | |
65b35c42 | 3506 | |
8fc6040c | 3507 | return offsets |
65b35c42 PG |
3508 | |
3509 | ||
dfd7865e PG |
3510 | def locate_gz_hdr_candidates (fd): |
3511 | """ | |
3512 | Walk over instances of the GZ magic in the payload, collecting their | |
3513 | positions. If the offset of the first found instance is not zero, the file | |
3514 | begins with leading garbage. | |
3515 | ||
3516 | Note that since the GZ magic consists of only two bytes, we expect a lot of | |
3517 | false positives inside binary data. | |
3518 | ||
3519 | :return: The list of offsets in the file. | |
3520 | """ | |
3521 | pos = 0 | |
3522 | cands = [] | |
3523 | mm = mmap.mmap(fd, 0, mmap.MAP_SHARED, mmap.PROT_READ) | |
3524 | ||
3525 | while True: | |
3526 | pos = mm.find (GZ_MAGIC_BYTES, pos) | |
3527 | if pos == -1: | |
3528 | break | |
3529 | cands.append (pos) | |
3530 | pos += len (GZ_MAGIC_BYTES) | |
3531 | ||
3532 | return cands | |
3533 | ||
3534 | ||
3535 | HDR_CAND_GOOD = 0 # header marks begin of valid object | |
3536 | HDR_CAND_FISHY = 1 # inconclusive | |
3537 | HDR_CAND_JUNK = 2 # not a header / object unreadable | |
3538 | ||
3539 | ||
3540 | def read_cstring (fd, max=-1, encoding=None): | |
3541 | """ | |
3542 | Read one NUL-terminated string from *fd* into a Python string. If *max* is | |
3543 | non-negative, reading will terminate after the specified number of bytes. | |
3544 | ||
3545 | Optionally, an *encoding* may be specified to interpret the data as. | |
3546 | ||
3547 | :returns: *None* if parsing failed or the maximum number of bytes has been | |
3548 | exceeded; a Python string with the data otherwise. | |
3549 | """ | |
3550 | buf = b"" | |
3551 | l = 0 | |
3552 | ||
3553 | while True: | |
3554 | c = os.read (fd, 1) | |
3555 | if c == NUL: | |
3556 | break | |
3557 | if max >= 0 and l > max: | |
3558 | return None | |
3559 | buf += c | |
3560 | l += 1 | |
3561 | if encoding is not None: | |
3562 | buf = buf.decode (encoding) | |
3563 | ||
3564 | return buf | |
3565 | ||
3566 | ||
3567 | def inspect_gz_hdr (fd, off): | |
3568 | """ | |
3569 | Attempt to parse a Gzip header in *fd* at position *off*. The format is | |
3570 | documented as RFC1952. | |
3571 | ||
3572 | Returns a verdict about the quality of that header plus the parsed header | |
3573 | when readable. Problematic sizes such as fields running past the EOF are | |
3574 | treated as garbage. Properties in which the header merely doesn’t conform | |
3575 | to the spec (garbage flag bits, bogus timestamp) are considered “fishy”. No | |
3576 | validation is possible on embedded strings because they are single-byte | |
3577 | encoded. | |
3578 | """ | |
3579 | fname = None | |
3580 | flags = 0x00 | |
3581 | dflags = 0x00 | |
3582 | mtime = 0x00000000 | |
3583 | oscode = 0x00 | |
3584 | verdict = HDR_CAND_GOOD | |
3585 | ||
3586 | os.lseek (fd, off, os.SEEK_SET) | |
3587 | if os.lseek (fd, 0, os.SEEK_CUR) != off: | |
3588 | return HDR_CAND_JUNK, None | |
3589 | ||
3590 | raw = os.read (fd, GZ_HEADER_SIZE) | |
3591 | if len (raw) != GZ_HEADER_SIZE: | |
3592 | return HDR_CAND_JUNK, None | |
3593 | ||
3594 | flags = 0x0 | |
3595 | try: | |
3596 | _m1, _m2, meth, flags, mtime, dflags, oscode = \ | |
3597 | struct.unpack (GZ_FMT_HEADER, raw) | |
3598 | if meth != GZ_METHOD_DEFLATE: # only deflate is supported | |
3599 | return HDR_CAND_JUNK, None | |
3600 | except struct.error as exn: | |
3601 | return HDR_CAND_JUNK, None | |
3602 | ||
3603 | if mtime > int (time.time ()): | |
3604 | verdict = HDR_CAND_FISHY | |
3605 | ||
3606 | if dflags != GZ_DEFLATE_FLAGS: | |
3607 | verdict = HDR_CAND_FISHY | |
3608 | ||
3609 | if oscode != GZ_OS_CODE: | |
3610 | verdict = HDR_CAND_FISHY | |
3611 | ||
3612 | if flags & GZ_FLAG_FTEXT: # created by some contrarian | |
3613 | verdict = HDR_CAND_FISHY | |
3614 | if flags & GZ_FLAG_FEXTRA: | |
04263795 | 3615 | xlen = struct.unpack ("<H", os.read (fd, 2))[0] |
dfd7865e PG |
3616 | xtra = os.read (fd, xlen) |
3617 | if len (xtra) != xlen: # eof inside header | |
3618 | return HDR_CAND_JUNK, None | |
3619 | if flags & GZ_FLAG_FNAME: | |
3620 | # read up to the next NUL byte, not exceeding the maximum path length | |
3621 | # allowed by tar(5) | |
3622 | fname = read_cstring (fd, max=(LENGTH_NAME + LENGTH_PREFIX), | |
3623 | encoding="iso-8859-1") | |
3624 | if fname is None: | |
3625 | return HDR_CAND_JUNK, None | |
3626 | if flags & GZ_FLAG_FCOMMENT: | |
3627 | fname = read_cstring (fd, max=(LENGTH_NAME + LENGTH_PREFIX), | |
3628 | encoding="iso-8859-1") | |
3629 | if fname is None: | |
3630 | return HDR_CAND_JUNK, None | |
3631 | if flags & GZ_FLAG_FHCRC: # half a CRC32; discarded | |
3632 | crc16 = os.read (fd, 2) | |
3633 | if len (crc16) != 2: # eof inside header | |
3634 | return HDR_CAND_JUNK, None | |
3635 | if flags & GZ_FLAG_RESERVED: | |
3636 | # according to the RFC, these must not be set | |
3637 | verdict = HDR_CAND_FISHY | |
3638 | ||
3639 | hlen = os.lseek (fd, 0, os.SEEK_CUR) - off | |
3640 | ||
3641 | return verdict, \ | |
3642 | { "fname" : fname | |
3643 | , "flags" : flags | |
3644 | , "dflags" : dflags | |
3645 | , "mtime" : mtime | |
3646 | , "oscode" : oscode | |
3647 | , "hlen" : hlen | |
3648 | } | |
3649 | ||
3650 | ||
3651 | def try_decompress (ifd, off, hdr): | |
3652 | """ | |
3653 | Attempt to process the object starting at *off* with gzip. | |
3654 | ||
3655 | :returns: A pair containing the values of the decompressed data and | |
3656 | the length of the input consumed. Note that the latter value | |
3657 | may exceed the length of the compressed data because the | |
3658 | *zlib* module does not provide a means to query how much | |
3659 | of the input it processed before the end of an object. | |
3660 | """ | |
3661 | import zlib | |
3662 | decmp = zlib.decompressobj (-zlib.MAX_WBITS) | |
3663 | pos = off | |
3664 | dlen = 0 # size of decompressed data | |
3665 | ||
3666 | os.lseek (ifd, pos, os.SEEK_SET) | |
3667 | while True: | |
3668 | cnk = os.read (ifd, BUFSIZE) | |
3669 | pos += len (cnk) | |
3670 | try: | |
3671 | data = decmp.decompress (cnk) | |
3672 | except zlib.error as exn: # probably CRC32 mismatch; terminate softly | |
3673 | break # fishy | |
3674 | dlen += len (data) | |
3675 | if decmp.eof is True: | |
3676 | break | |
3677 | if len (cnk) != BUFSIZE: # eof, but not end of decompressed object! | |
3678 | break # fishy | |
3679 | ||
3680 | return dlen, pos - off | |
3681 | ||
3682 | def readable_gz_objects_offsets (ifd, cands): | |
3683 | """ | |
3684 | Inspect header candidates for parseable *ifd* gzipped objects. | |
3685 | """ | |
3686 | good = [] | |
3687 | nobj = 0 | |
3688 | ||
3689 | for cand in cands: | |
3690 | nobj += 1 | |
3691 | vdt, hdr = inspect_gz_hdr (ifd, cand) | |
3692 | if vdt == HDR_CAND_JUNK: | |
3693 | pass # ignore unreadable ones | |
3694 | elif vdt in [HDR_CAND_GOOD, HDR_CAND_FISHY]: | |
3695 | off0 = cand + hdr ["hlen"] | |
3696 | dlen, clen = try_decompress (ifd, off0, hdr) | |
3697 | if dlen > 0 and clen > 0: | |
3698 | good.append (cand) | |
3699 | ||
3700 | return good | |
3701 | ||
3702 | ||
3703 | def reconstruct_offsets_gz (fname): | |
3704 | """ | |
3705 | From the given file, retrieve all GZ header-like offsets (“candidates”). | |
3706 | Then check each of those locations whether they can be processed as | |
3707 | compressed data. | |
3708 | """ | |
3709 | ifd = os.open (fname, os.O_RDONLY) | |
3710 | ||
3711 | try: | |
3712 | cands = locate_gz_hdr_candidates (ifd) | |
3713 | return readable_gz_objects_offsets (ifd, cands) | |
3714 | finally: | |
3715 | os.close (ifd) | |
3716 | ||
3717 | ||
65b35c42 PG |
3718 | def reconstruct_offsets_tar (fname): |
3719 | """ | |
3720 | From the given file, retrieve all tar header-like offsets (“candidates”). | |
3721 | Then check each of those locations whether they can be processed as tar | |
3722 | data. | |
3723 | """ | |
3724 | ifd = os.open (fname, os.O_RDONLY) | |
3725 | ||
3726 | try: | |
8fc6040c | 3727 | return readable_tar_objects_offsets (ifd) |
65b35c42 PG |
3728 | finally: |
3729 | os.close (ifd) | |
3730 | ||
3731 | ||
b750b280 PG |
3732 | def read_tarobj_at_offset (fileobj, offset, mode, secret=None, |
3733 | strict_validation=True): | |
3734 | """ | |
3735 | :type strict_validation: bool | |
3736 | :param strict_validation: Enable strict IV checking in the crypto | |
3737 | layer. Should be disabled when dealing with | |
3738 | potentially corrupted data. | |
3739 | """ | |
d39d4cbf | 3740 | decr = None |
d39d4cbf | 3741 | |
dfd7865e PG |
3742 | if secret is not None: |
3743 | ks = secret [0] | |
3744 | ||
3745 | if ks == crypto.PDTCRYPT_SECRET_PW: | |
b750b280 PG |
3746 | decr = crypto.Decrypt (password=secret [1], |
3747 | strict_ivs=strict_validation) | |
dfd7865e PG |
3748 | elif ks == crypto.PDTCRYPT_SECRET_KEY: |
3749 | key = binascii.unhexlify (secret [1]) | |
b750b280 PG |
3750 | decr = crypto.Decrypt (key=key, |
3751 | strict_ivs=strict_validation) | |
dfd7865e PG |
3752 | else: |
3753 | raise RuntimeError | |
d39d4cbf | 3754 | |
5bd2d4b5 PG |
3755 | try: |
3756 | tarobj = \ | |
3757 | TarFile.open_at_offset (offset, | |
3758 | mode=mode, | |
3759 | fileobj=fileobj, | |
3760 | format=GNU_FORMAT, | |
3761 | concat='#' in mode, | |
3762 | encryption=decr, | |
3763 | save_to_members=False, | |
3764 | tolerance=TOLERANCE_RESCUE) | |
3765 | except (ReadError, EndOfFile): | |
3766 | return None | |
d39d4cbf PG |
3767 | |
3768 | return tarobj.next () | |
3769 | ||
3770 | ||
2d50b7f7 PG |
3771 | def idxent_of_tarinfo (tarinfo): |
3772 | """ | |
3773 | Scrape the information relevant for the index from a *TarInfo* object. | |
3774 | Keys like the inode number that lack a corresponding field in a TarInfo | |
3775 | will be set to some neutral value. | |
3776 | Example output: | |
3777 | ||
3778 | { "inode" : 0 | |
3779 | , "uid" : 0 | |
3780 | , "path" : "snapshot://annotations.db" | |
3781 | , "offset" : 0 | |
3782 | , "volume" : 0 | |
3783 | , "mode" : 33152 | |
3784 | , "ctime" : 1502798115 | |
3785 | , "mtime" : 1502196423 | |
3786 | , "size" : 144 | |
3787 | , "type" : "file" | |
3788 | , "gid" : 0 | |
3789 | } | |
3790 | ||
3791 | """ | |
3792 | ||
3793 | return \ | |
3794 | { "inode" : 0 # ignored when reading the index | |
3795 | , "uid" : tarinfo.uid | |
3796 | , "gid" : tarinfo.gid | |
3797 | , "path" : tarinfo.name # keeping URI scheme | |
3798 | , "offset" : 0 # to be added by the caller | |
3799 | , "volume" : tarinfo.volume_offset | |
3800 | , "mode" : tarinfo.mode | |
3801 | , "ctime" : tarinfo.mtime | |
3802 | , "mtime" : tarinfo.mtime | |
3803 | , "size" : tarinfo.size | |
3804 | , "type" : tarinfo.type | |
3805 | } | |
3806 | ||
3807 | ||
27ee4dd4 PG |
3808 | def gen_rescue_index (gen_volume_name, mode, maxvol=None, password=None, key=None): |
3809 | infos = [] | |
6690f5e0 PG |
3810 | psidx = [] # pseudo index, return value |
3811 | offsets = None | |
addcec42 | 3812 | secret = crypto.make_secret (password=password, key=key) |
6690f5e0 | 3813 | |
27ee4dd4 | 3814 | nvol = 0 |
dfd7865e | 3815 | |
27ee4dd4 PG |
3816 | while True: |
3817 | vpath = gen_volume_name (nvol) | |
3818 | try: | |
3819 | if secret is not None: | |
3820 | offsets = crypto.reconstruct_offsets (vpath, secret) | |
3821 | elif mode == "#gz": | |
3822 | offsets = reconstruct_offsets_gz (vpath) | |
3823 | elif mode == "#": | |
3824 | offsets = reconstruct_offsets_tar (vpath) | |
3825 | else: | |
3826 | raise TarError ("no rescue handling for mode “%s”" % mode) | |
3827 | except FileNotFoundError as exn: | |
3828 | # volume does not exist | |
611c5d03 | 3829 | if maxvol is not None and nvol < maxvol: |
27ee4dd4 PG |
3830 | continue # explicit volume number specified, ignore missing ones |
3831 | else: | |
3832 | break | |
3833 | ||
3834 | fileobj = bltn_open (vpath, "rb") | |
5bd2d4b5 PG |
3835 | |
3836 | def aux (acc, off): | |
b750b280 PG |
3837 | obj = read_tarobj_at_offset (fileobj, off, mode, secret=secret, |
3838 | strict_validation=False) | |
5bd2d4b5 PG |
3839 | if obj is not None: |
3840 | acc.append ((off, nvol, obj)) | |
3841 | return acc | |
3842 | infos += functools.reduce (aux, offsets, []) | |
3843 | ||
bcc8b174 PG |
3844 | fileobj.close() |
3845 | ||
27ee4dd4 PG |
3846 | nvol += 1 |
3847 | ||
5bd2d4b5 PG |
3848 | def aux (o, nvol, ti): |
3849 | ie = idxent_of_tarinfo (ti) | |
3850 | ie ["offset"] = o | |
3851 | ie ["volume"] = nvol | |
3852 | return ie | |
3853 | ||
27ee4dd4 | 3854 | psidx = [ aux (o, nvol, ti) for o, nvol, ti in infos ] |
6690f5e0 PG |
3855 | |
3856 | return psidx | |
7584f5c9 ERE |
3857 | |
3858 | #-------------------- | |
3859 | # exported functions | |
3860 | #-------------------- | |
3861 | def is_tarfile(name): | |
3862 | """Return True if name points to a tar archive that we | |
3863 | are able to handle, else return False. | |
3864 | """ | |
3865 | try: | |
3866 | t = open(name) | |
3867 | t.close() | |
3868 | return True | |
3869 | except TarError: | |
3870 | return False | |
3871 | ||
3872 | bltn_open = open | |
3873 | open = TarFile.open |