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. | |
7584f5c9 | 281 | if 0 <= n < 8 ** (digits - 1): |
8112b0ed | 282 | s = bytes("%0*o" % (digits - 1, int(n)), "ascii") + NUL |
be60ffd0 ERE |
283 | elif format == GNU_FORMAT and -256 ** (digits - 1) <= n < 256 ** (digits - 1): |
284 | if n >= 0: | |
285 | s = bytearray([0o200]) | |
286 | else: | |
287 | s = bytearray([0o377]) | |
288 | n = 256 ** digits + n | |
7584f5c9 | 289 | |
be60ffd0 ERE |
290 | for i in range(digits - 1): |
291 | s.insert(1, n & 0o377) | |
7584f5c9 | 292 | n >>= 8 |
7584f5c9 | 293 | else: |
be60ffd0 ERE |
294 | raise ValueError("overflow in number field") |
295 | ||
296 | return s | |
7584f5c9 ERE |
297 | |
298 | def calc_chksums(buf): | |
299 | """Calculate the checksum for a member's header by summing up all | |
300 | characters except for the chksum field which is treated as if | |
301 | it was filled with spaces. According to the GNU tar sources, | |
302 | some tars (Sun and NeXT) calculate chksum with signed char, | |
303 | which will be different if there are chars in the buffer with | |
304 | the high bit set. So we calculate two checksums, unsigned and | |
305 | signed. | |
306 | """ | |
be60ffd0 ERE |
307 | unsigned_chksum = 256 + sum(struct.unpack_from("148B8x356B", buf)) |
308 | signed_chksum = 256 + sum(struct.unpack_from("148b8x356b", buf)) | |
7584f5c9 ERE |
309 | return unsigned_chksum, signed_chksum |
310 | ||
311 | def copyfileobj(src, dst, length=None): | |
312 | """Copy length bytes from fileobj src to fileobj dst. | |
313 | If length is None, copy the entire content. | |
314 | """ | |
315 | if length == 0: | |
316 | return | |
317 | if length is None: | |
318 | shutil.copyfileobj(src, dst) | |
319 | return | |
320 | ||
7584f5c9 | 321 | blocks, remainder = divmod(length, BUFSIZE) |
be60ffd0 | 322 | for b in range(blocks): |
7584f5c9 | 323 | buf = src.read(BUFSIZE) |
c474439c | 324 | dst.write(buf) |
7584f5c9 | 325 | if len(buf) < BUFSIZE: |
be60ffd0 | 326 | raise OSError("end of file reached") |
7584f5c9 ERE |
327 | if remainder != 0: |
328 | buf = src.read(remainder) | |
c474439c | 329 | dst.write(buf) |
7584f5c9 | 330 | if len(buf) < remainder: |
be60ffd0 | 331 | raise OSError("end of file reached") |
c7c736b6 | 332 | |
7584f5c9 | 333 | |
7584f5c9 | 334 | def filemode(mode): |
be60ffd0 ERE |
335 | """Deprecated in this location; use stat.filemode.""" |
336 | import warnings | |
337 | warnings.warn("deprecated in favor of stat.filemode", | |
338 | DeprecationWarning, 2) | |
339 | return stat.filemode(mode) | |
7584f5c9 ERE |
340 | |
341 | class TarError(Exception): | |
342 | """Base exception.""" | |
343 | pass | |
344 | class ExtractError(TarError): | |
345 | """General exception for extract errors.""" | |
346 | pass | |
347 | class ReadError(TarError): | |
be60ffd0 | 348 | """Exception for unreadable tar archives.""" |
7584f5c9 ERE |
349 | pass |
350 | class CompressionError(TarError): | |
351 | """Exception for unavailable compression methods.""" | |
352 | pass | |
353 | class StreamError(TarError): | |
354 | """Exception for unsupported operations on stream-like TarFiles.""" | |
355 | pass | |
356 | class HeaderError(TarError): | |
357 | """Base exception for header errors.""" | |
358 | pass | |
359 | class EmptyHeaderError(HeaderError): | |
360 | """Exception for empty headers.""" | |
361 | pass | |
362 | class TruncatedHeaderError(HeaderError): | |
363 | """Exception for truncated headers.""" | |
364 | pass | |
365 | class EOFHeaderError(HeaderError): | |
366 | """Exception for end of file headers.""" | |
367 | pass | |
368 | class InvalidHeaderError(HeaderError): | |
369 | """Exception for invalid headers.""" | |
370 | pass | |
371 | class SubsequentHeaderError(HeaderError): | |
372 | """Exception for missing and invalid extended headers.""" | |
373 | pass | |
8ab8fac5 PG |
374 | class InvalidEncryptionError(TarError): |
375 | """Exception for undefined crypto modes and combinations.""" | |
376 | pass | |
e4e5d0b8 PG |
377 | class DecryptionError(TarError): |
378 | """Exception for error during decryption.""" | |
379 | pass | |
c7c736b6 | 380 | class EncryptionError(TarError): |
e93f83f1 | 381 | """Exception for error during encryption.""" |
c7c736b6 | 382 | pass |
e50fa574 PG |
383 | class EndOfFile(Exception): |
384 | """Signal end of file condition when they’re not an error.""" | |
65b35c42 | 385 | pass |
7584f5c9 ERE |
386 | |
387 | #--------------------------- | |
388 | # internal stream interface | |
389 | #--------------------------- | |
390 | class _LowLevelFile: | |
391 | """Low-level file object. Supports reading and writing. | |
392 | It is used instead of a regular file object for streaming | |
393 | access. | |
394 | """ | |
395 | ||
396 | def __init__(self, name, mode): | |
ad4402e8 | 397 | _mode = { |
7584f5c9 | 398 | "r": os.O_RDONLY, |
c7c736b6 | 399 | "w": os.O_RDWR | os.O_CREAT | os.O_TRUNC, |
7584f5c9 ERE |
400 | }[mode] |
401 | if hasattr(os, "O_BINARY"): | |
f0287fb7 | 402 | _mode |= os.O_BINARY # pylint: disable=no-member |
be60ffd0 | 403 | self.fd = os.open(name, _mode, 0o666) |
ad4402e8 | 404 | self.offset = 0 |
7584f5c9 ERE |
405 | |
406 | def close(self): | |
407 | os.close(self.fd) | |
408 | ||
409 | def read(self, size): | |
ad4402e8 ERE |
410 | ret = os.read(self.fd, size) |
411 | self.offset += len(ret) | |
412 | return ret | |
7584f5c9 | 413 | |
867f75f7 PG |
414 | def write(self, s, pos=None): |
415 | if pos is not None: | |
416 | p0 = self.offset | |
417 | os.lseek (self.fd, pos, os.SEEK_SET) | |
418 | n = os.write(self.fd, s) | |
419 | if pos is None: | |
420 | self.offset += len(s) | |
421 | else: | |
422 | append = pos + n - p0 | |
423 | if append > 0: | |
424 | self.offset += append | |
425 | os.lseek (self.fd, p0, os.SEEK_SET) | |
7584f5c9 | 426 | |
ad4402e8 ERE |
427 | def tell(self): |
428 | return self.offset | |
429 | ||
c7c736b6 PG |
430 | def seek_set (self, pos): |
431 | os.lseek (self.fd, pos, os.SEEK_SET) | |
432 | self.offset = pos | |
433 | ||
8ab8fac5 | 434 | |
15a81fc0 PG |
435 | def gz_header (name=None): |
436 | timestamp = int(time.time()) | |
437 | flags = 0x0 | |
438 | ||
439 | if name is None: | |
440 | name = b"" | |
441 | else: | |
dfd7865e | 442 | flags |= GZ_FLAG_FNAME |
15a81fc0 PG |
443 | if type(name) is str: |
444 | name = name.encode("iso-8859-1", "replace") | |
6e99d23a PG |
445 | if name.endswith(b".pdtcrypt"): |
446 | name = name[:-9] | |
15a81fc0 PG |
447 | if name.endswith(b".gz"): |
448 | name = name[:-3] | |
449 | # RFC1952 says we must use ISO-8859-1 for the FNAME field. | |
450 | name += NUL | |
451 | ||
452 | hdr = struct.pack (GZ_FMT_HEADER, | |
453 | GZ_MAGIC [0], GZ_MAGIC [1], | |
454 | GZ_METHOD_DEFLATE, flags, | |
455 | timestamp, | |
456 | GZ_DEFLATE_FLAGS, GZ_OS_CODE) | |
457 | ||
458 | return hdr + name | |
459 | ||
d601d33b | 460 | |
7584f5c9 ERE |
461 | class _Stream: |
462 | """Class that serves as an adapter between TarFile and | |
463 | a stream-like object. The stream-like object only | |
464 | needs to have a read() or write() method and is accessed | |
465 | blockwise. Use of gzip or bzip2 compression is possible. | |
466 | A stream-like object could be for example: sys.stdin, | |
467 | sys.stdout, a socket, a tape device etc. | |
468 | ||
3031b7ae PG |
469 | _Stream is intended to be used only internally but is |
470 | nevertherless used externally by Deltatar. | |
471 | ||
472 | When encrypting, the ``enccounter`` will be used for | |
473 | initializing the first cryptographic context. When | |
474 | decrypting, its value will be compared to the decrypted | |
475 | object. Decryption fails if the value does not match. | |
476 | In effect, this means that a ``_Stream`` whose ctor was | |
477 | passed ``enccounter`` can only be used to encrypt or | |
478 | decrypt a single object. | |
7584f5c9 ERE |
479 | """ |
480 | ||
c7c736b6 | 481 | remainder = -1 # track size in encrypted entries |
04f4c7ab | 482 | tolerance = TOLERANCE_STRICT |
c7c736b6 | 483 | |
6e812ad9 | 484 | def __init__(self, name, mode, comptype, fileobj, bufsize, |
d1c38f40 | 485 | concat=False, encryption=None, enccounter=None, |
04f4c7ab | 486 | compresslevel=9, tolerance=TOLERANCE_STRICT): |
7584f5c9 ERE |
487 | """Construct a _Stream object. |
488 | """ | |
d1c38f40 | 489 | self.arcmode = arcmode_set (concat, encryption, comptype) |
04f4c7ab | 490 | self.tolerance = tolerance |
d1c38f40 | 491 | |
7584f5c9 ERE |
492 | self._extfileobj = True |
493 | if fileobj is None: | |
494 | fileobj = _LowLevelFile(name, mode) | |
495 | self._extfileobj = False | |
496 | ||
497 | if comptype == '*': | |
498 | # Enable transparent compression detection for the | |
499 | # stream interface | |
500 | fileobj = _StreamProxy(fileobj) | |
501 | comptype = fileobj.getcomptype() | |
d1c38f40 PG |
502 | if comptype == '': |
503 | comptype = "tar" | |
7584f5c9 | 504 | |
3031b7ae PG |
505 | self.enccounter = None |
506 | if self.arcmode & ARCMODE_ENCRYPT: | |
507 | self.enccounter = enccounter | |
508 | ||
7584f5c9 ERE |
509 | self.name = name or "" |
510 | self.mode = mode | |
511 | self.comptype = comptype | |
53732900 | 512 | self.cmp = None |
7584f5c9 ERE |
513 | self.fileobj = fileobj |
514 | self.bufsize = bufsize | |
be60ffd0 ERE |
515 | self.buf = b"" |
516 | self.pos = 0 | |
517 | self.concat_pos = 0 | |
7584f5c9 | 518 | self.closed = False |
be60ffd0 | 519 | self.flags = 0 |
be60ffd0 | 520 | self.last_block_offset = 0 |
e4e5d0b8 | 521 | self.dbuf = b"" # ??? |
46c03c02 | 522 | self.exception = None # communicate decompression failure |
2b82f50c | 523 | self.compresslevel = compresslevel |
784175ba | 524 | self.bytes_written = 0 |
c7c736b6 | 525 | # crypto parameters |
2ae46844 | 526 | self.encryption = encryption |
c7c736b6 | 527 | self.lasthdr = None |
7584f5c9 | 528 | |
be60ffd0 ERE |
529 | try: |
530 | if comptype == "gz": | |
531 | try: | |
532 | import zlib | |
533 | except ImportError: | |
534 | raise CompressionError("zlib module is not available") | |
535 | self.zlib = zlib | |
bec34b42 PG |
536 | if mode == "r": |
537 | self.exception = zlib.error | |
8ae983c4 | 538 | self._init_read_gz() |
bec34b42 | 539 | elif mode == "w": |
d1c38f40 PG |
540 | if not (self.arcmode & ARCMODE_CONCAT): |
541 | if self.arcmode & ARCMODE_ENCRYPT: | |
3031b7ae | 542 | self._init_write_encrypt (name) |
a0873dcc | 543 | self._init_write_gz () |
c2ffe2ec | 544 | self.crc = zlib.crc32(b"") & 0xFFFFffff |
7584f5c9 | 545 | |
be60ffd0 | 546 | elif comptype == "bz2": |
d1c38f40 | 547 | if self.arcmode & ARCMODE_ENCRYPT: |
2ae46844 | 548 | raise InvalidEncryptionError("encryption not available for " |
d1c38f40 | 549 | "compression “%s”" % comptype) |
be60ffd0 ERE |
550 | try: |
551 | import bz2 | |
552 | except ImportError: | |
553 | raise CompressionError("bz2 module is not available") | |
554 | if mode == "r": | |
555 | self.dbuf = b"" | |
556 | self.cmp = bz2.BZ2Decompressor() | |
557 | self.exception = OSError | |
558 | else: | |
559 | self.cmp = bz2.BZ2Compressor() | |
7584f5c9 | 560 | |
be60ffd0 | 561 | elif comptype == 'xz': |
d1c38f40 | 562 | if self.arcmode & ARCMODE_ENCRYPT: |
2ae46844 | 563 | raise InvalidEncryptionError("encryption not available for " |
d1c38f40 | 564 | "compression “%s”" % comptype) |
c7c736b6 PG |
565 | try: |
566 | import lzma | |
567 | except ImportError: | |
568 | raise CompressionError("lzma module is not available") | |
569 | if mode == "r": | |
570 | self.dbuf = b"" | |
571 | self.cmp = lzma.LZMADecompressor() | |
572 | self.exception = lzma.LZMAError | |
573 | else: | |
574 | self.cmp = lzma.LZMACompressor() | |
575 | ||
6de9444a | 576 | elif comptype == "tar": |
d1c38f40 | 577 | if not (self.arcmode & ARCMODE_CONCAT) \ |
6de9444a | 578 | and mode == "w" \ |
d1c38f40 | 579 | and self.arcmode & ARCMODE_ENCRYPT: |
3031b7ae | 580 | self._init_write_encrypt (name) |
6de9444a PG |
581 | |
582 | else: | |
d1c38f40 | 583 | if self.arcmode & ARCMODE_ENCRYPT: |
2ae46844 | 584 | raise InvalidEncryptionError("encryption not available for " |
d1c38f40 | 585 | "compression “%s”" % comptype) |
c7c736b6 | 586 | raise CompressionError("unknown compression type %r" % comptype) |
be60ffd0 | 587 | |
200d4866 | 588 | except: |
be60ffd0 ERE |
589 | if not self._extfileobj: |
590 | self.fileobj.close() | |
591 | self.closed = True | |
592 | raise | |
ac5e4184 | 593 | |
7584f5c9 ERE |
594 | def __del__(self): |
595 | if hasattr(self, "closed") and not self.closed: | |
fac2cfe1 PG |
596 | try: |
597 | self.close() | |
598 | except crypto.InternalError: | |
599 | # context already finalized due to abort but close() tried | |
600 | # to use it | |
601 | pass | |
7584f5c9 | 602 | |
c7c736b6 | 603 | |
d1c38f40 PG |
604 | def next (self, name): |
605 | if self.arcmode & ARCMODE_COMPRESS: | |
606 | if getattr (self, "cmp", None) is not None: | |
607 | self._finalize_write_gz () | |
0349168a PG |
608 | self.__sync() |
609 | if self.arcmode & ~(ARCMODE_ENCRYPT | ARCMODE_COMPRESS): | |
610 | self.last_block_offset = self.fileobj.tell() | |
d1c38f40 PG |
611 | if self.arcmode & ARCMODE_ENCRYPT: |
612 | self._finalize_write_encrypt () | |
613 | self._init_write_encrypt (name, set_last_block_offset=True) | |
614 | if self.arcmode & ARCMODE_COMPRESS: | |
615 | self._init_write_gz (set_last_block_offset = | |
0349168a | 616 | not (self.arcmode & ARCMODE_ENCRYPT)) |
d1c38f40 PG |
617 | return self.last_block_offset |
618 | ||
619 | ||
620 | def next_volume (self, name): | |
621 | # with non-concat modes, this is taken care by the _Stream | |
622 | # ctor as invoked by the newvol handler | |
623 | if self.arcmode & ARCMODE_COMPRESS: | |
624 | if getattr (self, "cmp", None) is not None: | |
625 | # e. g. compressed PAX header written | |
626 | self._finalize_write_gz () | |
627 | if self.arcmode & ARCMODE_ENCRYPT: | |
628 | self._init_write_encrypt (name) | |
629 | if self.arcmode & ARCMODE_COMPRESS: | |
630 | self._init_write_gz () | |
631 | ||
c7c736b6 | 632 | |
d1c38f40 PG |
633 | def _init_write_encrypt (self, entry=None, set_last_block_offset=False): |
634 | """ | |
635 | Save position for delayed write of header; fill the header location | |
636 | with dummy bytes. | |
637 | """ | |
638 | # first thing, proclaim new object to the encryption context | |
639 | # secondly, assemble the header with the updated parameters | |
640 | # and commit it directly to the underlying stream, bypassing the | |
641 | # encryption layer in .__write(). | |
642 | dummyhdr = self.encryption.next (entry, counter=self.enccounter) | |
643 | if dummyhdr is None: | |
644 | raise EncryptionError ("Crypto.next(): bad dummy header") # XXX | |
645 | self.lasthdr = self.fileobj.tell() | |
646 | self.__write_to_file(dummyhdr) | |
647 | if set_last_block_offset is True: | |
648 | self.last_block_offset = self.lasthdr | |
c7c736b6 PG |
649 | |
650 | ||
651 | def _finalize_write_encrypt (self): | |
652 | """ | |
653 | Seek back to header position, read dummy bytes, finalize crypto | |
654 | obtaining the actual header, write header, seek back to current | |
655 | position. | |
963d0db4 PG |
656 | |
657 | Returns the list of IV fixed parts as used during encryption. | |
c7c736b6 | 658 | """ |
d1c38f40 | 659 | if self.lasthdr is not None: |
c7c736b6 PG |
660 | pos0 = self.fileobj.tell () |
661 | self.fileobj.seek_set (self.lasthdr) | |
dd47d6a2 | 662 | dummy = self.fileobj.read (crypto.PDTCRYPT_HDR_SIZE) |
c7c736b6 PG |
663 | pos1 = self.fileobj.tell () |
664 | dpos = pos1 - self.lasthdr | |
dd47d6a2 | 665 | assert dpos == crypto.PDTCRYPT_HDR_SIZE |
c7c736b6 | 666 | self.fileobj.seek_set (pos0) |
c8c72fe1 | 667 | data, hdr, _ = self.encryption.done (dummy) |
5f38bff6 | 668 | self.__write_to_file(hdr, pos=self.lasthdr) |
c7c736b6 PG |
669 | self.__write_to_file(data) # append remainder of data |
670 | self.lasthdr = -1 | |
671 | ||
672 | ||
57db1546 PG |
673 | def _finalize_write_gz (self): |
674 | if self.cmp is not None: | |
675 | chunk = self.buf + self.cmp.flush() | |
676 | if chunk: | |
677 | if self.comptype == "gz": | |
678 | # The native zlib crc is an unsigned 32-bit integer, but | |
679 | # the Python wrapper implicitly casts that to a signed C | |
680 | # long. So, on a 32-bit box self.crc may "look negative", | |
681 | # while the same crc on a 64-bit box may "look positive". | |
682 | # To avoid irksome warnings from the `struct` module, force | |
683 | # it to look positive on all boxes. | |
684 | chunk += struct.pack("<L", self.crc & 0xffffffff) | |
685 | chunk += struct.pack("<L", self.concat_pos & 0xffffFFFF) | |
686 | self.__enc_write (chunk) | |
15a81fc0 | 687 | self.buf = b"" |
57db1546 PG |
688 | |
689 | ||
a0873dcc | 690 | def _init_write_gz (self, set_last_block_offset=False): |
5fdff89f ERE |
691 | ''' |
692 | Add a new gzip block, closing last one | |
693 | ''' | |
be60ffd0 | 694 | self.concat_pos = 0 |
c2ffe2ec | 695 | self.crc = self.zlib.crc32(b"") & 0xFFFFffff |
6de9444a | 696 | first = self.cmp is None |
2b82f50c ERE |
697 | self.cmp = self.zlib.compressobj(self.compresslevel, |
698 | self.zlib.DEFLATED, | |
699 | -self.zlib.MAX_WBITS, | |
700 | self.zlib.DEF_MEM_LEVEL, | |
701 | 0) | |
6e812ad9 DGM |
702 | |
703 | # if aes, we encrypt after compression | |
6de9444a | 704 | if set_last_block_offset is True: |
ad4402e8 | 705 | self.last_block_offset = self.fileobj.tell() |
6e812ad9 | 706 | |
15a81fc0 | 707 | self.__write(gz_header (self.name if first is True else None)) |
5fdff89f | 708 | |
ac5e4184 | 709 | |
7584f5c9 ERE |
710 | def write(self, s): |
711 | """Write string s to the stream. | |
712 | """ | |
713 | if self.comptype == "gz": | |
c2ffe2ec | 714 | self.crc = self.zlib.crc32(s, self.crc) & 0xFFFFffff |
7584f5c9 | 715 | self.pos += len(s) |
5fdff89f | 716 | self.concat_pos += len(s) |
53732900 | 717 | if self.cmp is not None: |
7584f5c9 ERE |
718 | s = self.cmp.compress(s) |
719 | self.__write(s) | |
720 | ||
c7c736b6 | 721 | def __sync(self): |
cb7a3911 | 722 | """Write what’s left in the buffer to the stream.""" |
c7c736b6 PG |
723 | self.__write (b"") # → len (buf) <= bufsiz |
724 | self.__enc_write (self.buf) | |
725 | self.buf = b"" | |
726 | ||
7584f5c9 | 727 | def __write(self, s): |
548bb8d5 CH |
728 | """Writes (and encodes) string s to the stream blockwise |
729 | ||
730 | will wait with encoding/writing until block is complete | |
7584f5c9 ERE |
731 | """ |
732 | self.buf += s | |
733 | while len(self.buf) > self.bufsize: | |
6e812ad9 | 734 | self.__enc_write(self.buf[:self.bufsize]) |
7584f5c9 ERE |
735 | self.buf = self.buf[self.bufsize:] |
736 | ||
867f75f7 | 737 | |
5f38bff6 | 738 | def __write_to_file(self, s, pos=None): |
6e812ad9 | 739 | ''' |
5f38bff6 | 740 | Writes directly to the fileobj; updates self.bytes_written. If “pos” is |
cb7a3911 | 741 | given, the stream will seek to that position first and back afterwards, |
5f38bff6 | 742 | and the total of bytes written is not updated. |
6e812ad9 | 743 | ''' |
867f75f7 | 744 | self.fileobj.write(s, pos) |
5f38bff6 PG |
745 | if pos is None: |
746 | self.bytes_written += len(s) | |
867f75f7 | 747 | |
6e812ad9 DGM |
748 | |
749 | def __enc_write(self, s): | |
cb7a3911 PG |
750 | """ |
751 | If encryption is active, the string s is encrypted before being written | |
752 | to the file. | |
753 | """ | |
754 | if len (s) == 0: | |
755 | return | |
d1c38f40 | 756 | if self.arcmode & ARCMODE_ENCRYPT: |
cb7a3911 PG |
757 | buf = s |
758 | while len (buf) > 0: | |
759 | n, ct = self.encryption.process(buf) | |
760 | self.__write_to_file(ct) | |
761 | buf = buf [n:] | |
762 | if len (buf) > 0: | |
763 | # The entire plaintext was not consumed: The size limit | |
764 | # for encrypted objects was reached. Transparently create | |
765 | # a new encrypted object and continue processing the input. | |
766 | self._finalize_write_encrypt () | |
767 | self._init_write_encrypt () | |
768 | else: | |
769 | self.__write_to_file(s) | |
770 | ||
6e812ad9 | 771 | |
784175ba CH |
772 | def estim_file_size(self): |
773 | """ estimates size of file if closing it now | |
774 | ||
775 | The result may differ greatly from the amount of data sent to write() | |
776 | due to compression, encryption and buffering. | |
777 | ||
778 | In tests the result (before calling close()) was up to 12k smaller than | |
779 | the final file size if compression is being used because zlib/bz2 | |
780 | compressors do not allow inspection of their buffered data :-( | |
781 | ||
ba5a449e CH |
782 | Still, we add what close() would add: 8 bytes for gz checksum, one |
783 | encryption block size if encryption is used and the size of our own | |
784 | buffer | |
784175ba CH |
785 | """ |
786 | if self.closed: | |
787 | return self.bytes_written | |
788 | ||
789 | result = self.bytes_written | |
790 | if self.buf: | |
791 | result += len(self.buf) | |
792 | if self.comptype == 'gz': | |
ba5a449e | 793 | result += 8 # 2 longs = 8 byte (no extra info written for bzip2) |
784175ba CH |
794 | return result |
795 | ||
5fdff89f | 796 | def close(self, close_fileobj=True): |
7584f5c9 ERE |
797 | """Close the _Stream object. No operation should be |
798 | done on it afterwards. | |
799 | """ | |
963d0db4 | 800 | |
7584f5c9 ERE |
801 | if self.closed: |
802 | return | |
803 | ||
963d0db4 | 804 | if close_fileobj is True: |
a0873dcc | 805 | |
ae3d0f2a | 806 | if self.mode == "w": |
d1c38f40 | 807 | if self.arcmode & ARCMODE_COMPRESS: |
a0873dcc | 808 | self._finalize_write_gz () |
ae3d0f2a | 809 | # end of Tar archive marker (two empty blocks) was written |
267bc643 PG |
810 | # finalize encryption last; no writes may be performed after |
811 | # this point | |
cb7a3911 | 812 | self.__sync () |
d1c38f40 PG |
813 | if self.arcmode & ARCMODE_ENCRYPT: |
814 | self._finalize_write_encrypt () | |
267bc643 | 815 | |
963d0db4 PG |
816 | if not self._extfileobj: |
817 | self.fileobj.close() | |
818 | else: | |
819 | # read the zlib crc and length and check them | |
820 | if self.mode == "r" and self.comptype == "gz": | |
821 | read_crc = self.__read(4) | |
822 | read_length = self.__read(4) | |
823 | calculated_crc = self.crc | |
824 | if struct.unpack("<L", read_crc)[0] != calculated_crc: | |
825 | raise CompressionError("bad gzip crc") | |
7584f5c9 ERE |
826 | self.closed = True |
827 | ||
54128a00 | 828 | |
7584f5c9 ERE |
829 | def _init_read_gz(self): |
830 | """Initialize for reading a gzip compressed fileobj. | |
831 | """ | |
832 | self.cmp = self.zlib.decompressobj(-self.zlib.MAX_WBITS) | |
7584f5c9 | 833 | |
85737f48 | 834 | read2 = self.__read(2) |
e50fa574 PG |
835 | if read2 == b"": |
836 | raise EndOfFile ("_init_read_gz(): read returned zero bytes at pos " | |
837 | "%d" % self.fileobj.tell()) | |
838 | # taken from gzip.GzipFile with some alterations | |
d601d33b | 839 | if read2 != GZ_MAGIC_BYTES: |
7584f5c9 | 840 | raise ReadError("not a gzip file") |
85737f48 | 841 | |
5bd2d4b5 PG |
842 | read1 = self.__read(1) |
843 | if read1 == b"": | |
844 | raise EndOfFile ("_init_read_gz(): read returned zero bytes inside " | |
845 | "gzip header at pos %d" % self.fileobj.tell()) | |
846 | if ord (read1) != GZ_METHOD_DEFLATE: | |
7584f5c9 ERE |
847 | raise CompressionError("unsupported compression method") |
848 | ||
85737f48 | 849 | self.flags = flag = ord(self.__read(1)) |
dfd7865e | 850 | self.__read(6) # discard timestamp[4], deflate flags, os code |
7584f5c9 | 851 | |
dfd7865e | 852 | if flag & GZ_FLAG_FEXTRA: |
7584f5c9 ERE |
853 | xlen = ord(self.__read(1)) + 256 * ord(self.__read(1)) |
854 | self.read(xlen) | |
dfd7865e | 855 | if flag & GZ_FLAG_FNAME: |
7584f5c9 ERE |
856 | while True: |
857 | s = self.__read(1) | |
858 | if not s or s == NUL: | |
859 | break | |
dfd7865e | 860 | if flag & GZ_FLAG_FCOMMENT: |
7584f5c9 ERE |
861 | while True: |
862 | s = self.__read(1) | |
863 | if not s or s == NUL: | |
864 | break | |
dfd7865e | 865 | if flag & GZ_FLAG_FHCRC: |
7584f5c9 ERE |
866 | self.__read(2) |
867 | ||
c7c736b6 PG |
868 | def _init_read_encrypt (self): |
869 | """Initialize encryption for next entry in archive. Read a header and | |
870 | notify the crypto context.""" | |
d1c38f40 | 871 | if self.arcmode & ARCMODE_ENCRYPT: |
6e99d23a | 872 | lasthdr = self.fileobj.tell () |
15d3eefd PG |
873 | try: |
874 | hdr = crypto.hdr_read_stream (self.fileobj) | |
8a8ac469 PG |
875 | except crypto.EndOfFile: |
876 | return False | |
6e99d23a | 877 | except crypto.InvalidHeader as exn: |
c7c736b6 | 878 | raise DecryptionError ("Crypto.hdr_read_stream(): error “%s” " |
6e99d23a PG |
879 | "processing %r at pos %d" |
880 | % (exn, self.fileobj, lasthdr)) \ | |
ae3d0f2a | 881 | from exn |
3031b7ae PG |
882 | if self.enccounter is not None: |
883 | # enforce that the iv counter in the header matches an | |
884 | # explicitly requested one | |
885 | iv = crypto.hdr_iv_counter (hdr) | |
886 | if iv != self.enccounter: | |
887 | raise DecryptionError ("expected IV counter %d, got %d" | |
888 | % (self.enccounter, iv)) | |
6e99d23a | 889 | self.lasthdr = lasthdr |
c7c736b6 | 890 | self.remainder = hdr ["ctsize"] # distance to next header |
1ed44e7b PG |
891 | try: |
892 | self.encryption.next (hdr) | |
893 | except crypto.InvalidParameter as exn: | |
894 | raise DecryptionError ("Crypto.next(): error “%s” " | |
895 | "processing %r at pos %d" | |
896 | % (exn, self.fileobj, lasthdr)) \ | |
897 | from exn | |
8a8ac469 PG |
898 | |
899 | return True | |
c7c736b6 PG |
900 | |
901 | ||
8de91f4f PG |
902 | def _read_encrypt (self, buf): |
903 | """ | |
904 | Demote a program error to a decryption error in tolerant mode. This | |
905 | allows recovery from corrupted headers and invalid data. | |
906 | """ | |
907 | try: | |
908 | return self.encryption.process (buf) | |
909 | except RuntimeError as exn: | |
04f4c7ab | 910 | if self.tolerance != TOLERANCE_STRICT: |
8de91f4f PG |
911 | raise DecryptionError (exn) |
912 | raise | |
913 | ||
914 | ||
c7c736b6 PG |
915 | def _finalize_read_encrypt (self): |
916 | """ | |
917 | Finalize decryption. | |
918 | """ | |
d1c38f40 PG |
919 | if self.arcmode & ARCMODE_ENCRYPT \ |
920 | and self.lasthdr is not None : | |
c7c736b6 PG |
921 | assert self.remainder >= 0 |
922 | if self.remainder > 0: | |
c7c736b6 | 923 | self.remainder = 0 |
b0078f26 PG |
924 | try: |
925 | data = self.encryption.done () | |
926 | except crypto.InvalidGCMTag as exn: | |
927 | raise DecryptionError ("decryption failed: %s" % exn) | |
c7c736b6 PG |
928 | return data |
929 | ||
930 | ||
7584f5c9 ERE |
931 | def tell(self): |
932 | """Return the stream's file pointer position. | |
933 | """ | |
934 | return self.pos | |
935 | ||
936 | def seek(self, pos=0): | |
937 | """Set the stream's file pointer to pos. Negative seeking | |
938 | is forbidden. | |
939 | """ | |
940 | if pos - self.pos >= 0: | |
941 | blocks, remainder = divmod(pos - self.pos, self.bufsize) | |
be60ffd0 | 942 | for i in range(blocks): |
7584f5c9 ERE |
943 | self.read(self.bufsize) |
944 | self.read(remainder) | |
945 | else: | |
946 | raise StreamError("seeking backwards is not allowed") | |
947 | return self.pos | |
948 | ||
949 | def read(self, size=None): | |
950 | """Return the next size number of bytes from the stream. | |
951 | If size is not defined, return all bytes of the stream | |
952 | up to EOF. | |
953 | """ | |
954 | if size is None: | |
955 | t = [] | |
956 | while True: | |
957 | buf = self._read(self.bufsize) | |
958 | if not buf: | |
959 | break | |
960 | t.append(buf) | |
9dc7ac5c | 961 | buf = b"".join(t) |
7584f5c9 ERE |
962 | else: |
963 | buf = self._read(size) | |
964 | self.pos += len(buf) | |
965 | return buf | |
966 | ||
3a7e1a50 ERE |
967 | def readline(self): |
968 | """Reads just one line, new line character included | |
969 | """ | |
f0fd5e3a | 970 | # if \n in dbuf, no read neads to be done |
be60ffd0 ERE |
971 | if b'\n' in self.dbuf: |
972 | pos = self.dbuf.index(b'\n') + 1 | |
f0fd5e3a ERE |
973 | ret = self.dbuf[:pos] |
974 | self.dbuf = self.dbuf[pos:] | |
975 | return ret | |
976 | ||
1215b602 | 977 | buf = [] |
3a7e1a50 ERE |
978 | while True: |
979 | chunk = self._read(self.bufsize) | |
980 | ||
f0fd5e3a | 981 | # nothing more to read, so return the buffer |
3a7e1a50 | 982 | if not chunk: |
be60ffd0 | 983 | return b''.join(buf) |
3a7e1a50 ERE |
984 | |
985 | buf.append(chunk) | |
f0fd5e3a ERE |
986 | |
987 | # if \n found, return the new line | |
be60ffd0 ERE |
988 | if b'\n' in chunk: |
989 | dbuf = b''.join(buf) | |
990 | pos = dbuf.index(b'\n') + 1 | |
1215b602 | 991 | self.dbuf = dbuf[pos:] + self.dbuf |
3a7e1a50 ERE |
992 | return dbuf[:pos] |
993 | ||
7584f5c9 ERE |
994 | def _read(self, size): |
995 | """Return size bytes from the stream. | |
996 | """ | |
7584f5c9 ERE |
997 | c = len(self.dbuf) |
998 | t = [self.dbuf] | |
e4e5d0b8 | 999 | |
7584f5c9 | 1000 | while c < size: |
867f75f7 | 1001 | buf = self.__read(self.bufsize) |
7584f5c9 ERE |
1002 | if not buf: |
1003 | break | |
3a7e1a50 | 1004 | |
53732900 | 1005 | if self.cmp is not None: |
85737f48 | 1006 | try: |
3a7e1a50 | 1007 | buf = self.cmp.decompress(buf) |
54128a00 PG |
1008 | except self.exception as exn: |
1009 | raise ReadError("invalid compressed data (%r)" % exn) | |
be60ffd0 | 1010 | except Exception as e: |
04fb06f4 DGM |
1011 | # happens at the end of the file |
1012 | # _init_read_gz failed in the previous iteration so | |
e4e5d0b8 | 1013 | # self.cmp.decompress fails here |
d1c38f40 | 1014 | if self.arcmode & ARCMODE_CONCAT: |
be60ffd0 ERE |
1015 | pass |
1016 | else: | |
1017 | raise ReadError("invalid compressed data") | |
d1c38f40 | 1018 | if self.arcmode & ARCMODE_COMPRESS and hasattr(self, "crc"): |
c2ffe2ec | 1019 | self.crc = self.zlib.crc32(buf, self.crc) & 0xFFFFffff |
d1c38f40 PG |
1020 | if self.arcmode & ARCMODE_CONCAT \ |
1021 | and len(self.cmp.unused_data) != 0: | |
3a7e1a50 ERE |
1022 | self.buf = self.cmp.unused_data + self.buf |
1023 | self.close(close_fileobj=False) | |
1024 | try: | |
1025 | self._init_read_gz() | |
8de91f4f | 1026 | except DecryptionError: |
04f4c7ab | 1027 | if self.tolerance != TOLERANCE_STRICT: |
8de91f4f PG |
1028 | # return whatever data was processed successfully |
1029 | if len (buf) > 0: | |
1030 | t.append (buf) | |
1031 | if len (t) > 0: | |
1032 | break | |
24afaf18 PG |
1033 | raise |
1034 | except ReadError: # gzip troubles | |
1035 | if self.tolerance == TOLERANCE_RESCUE: | |
1036 | if len (buf) > 0: | |
1037 | t.append (buf) | |
1038 | if len (t) > 0: | |
1039 | break | |
1040 | raise | |
e50fa574 | 1041 | except EndOfFile: |
3a7e1a50 ERE |
1042 | # happens at the end of the file |
1043 | pass | |
c2ffe2ec | 1044 | self.crc = self.zlib.crc32(b"") & 0xFFFFffff |
3a7e1a50 | 1045 | self.closed = False |
7584f5c9 ERE |
1046 | t.append(buf) |
1047 | c += len(buf) | |
be60ffd0 | 1048 | t = b"".join(t) |
7584f5c9 ERE |
1049 | self.dbuf = t[size:] |
1050 | return t[:size] | |
1051 | ||
e4e5d0b8 | 1052 | |
7584f5c9 | 1053 | def __read(self, size): |
ef3b4499 PG |
1054 | """ |
1055 | Return size bytes from stream. If internal buffer is empty, read | |
1056 | another block from the stream. | |
1057 | ||
1058 | The function returns up to size bytes of data. When an error occurs | |
1059 | during decryption, everything until the end of the last successfully | |
1060 | finalized object is returned. | |
7584f5c9 ERE |
1061 | """ |
1062 | c = len(self.buf) | |
8de91f4f | 1063 | t = [self.buf] if c > 0 else [] |
1ed44e7b | 1064 | good_crypto = len (t) |
8de91f4f | 1065 | |
7584f5c9 | 1066 | while c < size: |
c7c736b6 | 1067 | todo = size |
8de91f4f PG |
1068 | try: |
1069 | if self.arcmode & ARCMODE_ENCRYPT: | |
1070 | if self.remainder <= 0: | |
1071 | # prepare next object | |
044585c6 PG |
1072 | if self._init_read_encrypt () is False: # EOF |
1073 | buf = None | |
1074 | break # while | |
8de91f4f PG |
1075 | |
1076 | # only read up to the end of the encrypted object | |
1077 | todo = min (size, self.remainder) | |
1078 | buf = self.fileobj.read(todo) | |
1079 | if self.arcmode & ARCMODE_ENCRYPT: | |
1080 | # decrypt the thing | |
1081 | buf = self._read_encrypt (buf) | |
1082 | if todo == self.remainder: | |
1083 | # at the end of a crypto object; finalization will fail if | |
1084 | # the GCM tag does not match | |
ef3b4499 | 1085 | trailing = self._finalize_read_encrypt () |
8de91f4f PG |
1086 | good_crypto = len (t) + 1 |
1087 | if len (trailing) > 0: | |
1088 | buf += trailing | |
1089 | self.remainder = 0 | |
1090 | else: | |
1091 | self.remainder -= todo | |
1092 | except DecryptionError: | |
04f4c7ab | 1093 | if self.tolerance == TOLERANCE_STRICT: |
8de91f4f PG |
1094 | raise |
1095 | self.encryption.drop () | |
24afaf18 PG |
1096 | if self.tolerance == TOLERANCE_RECOVER: |
1097 | if good_crypto == 0: | |
1098 | raise | |
1099 | # this may occur at any of the three crypto operations above. | |
1100 | # some objects did validate; discard all data after it; next | |
1101 | # call will start with the bad object and error out immediately | |
1102 | self.buf = b"".join (t [good_crypto:]) | |
1103 | return b"".join (t [:good_crypto]) | |
1104 | elif self.tolerance == TOLERANCE_RESCUE: | |
1105 | # keep what we have so far despite the finalization issue | |
1106 | t.append (buf) | |
1107 | c += len (buf) | |
1108 | break | |
1109 | else: | |
1110 | raise RuntimeError("internal error: bad tolerance level") | |
c7c736b6 PG |
1111 | |
1112 | if not buf: ## XXX stream terminated prematurely; this should be an error | |
7584f5c9 | 1113 | break |
c7c736b6 | 1114 | |
7584f5c9 ERE |
1115 | t.append(buf) |
1116 | c += len(buf) | |
be60ffd0 | 1117 | t = b"".join(t) |
7584f5c9 | 1118 | self.buf = t[size:] |
fb27c6e8 | 1119 | |
7584f5c9 | 1120 | return t[:size] |
7d372216 | 1121 | |
7584f5c9 ERE |
1122 | |
1123 | class _StreamProxy(object): | |
1124 | """Small proxy class that enables transparent compression | |
1125 | detection for the Stream interface (mode 'r|*'). | |
1126 | """ | |
1127 | ||
1128 | def __init__(self, fileobj): | |
1129 | self.fileobj = fileobj | |
1130 | self.buf = self.fileobj.read(BLOCKSIZE) | |
1131 | ||
f0287fb7 | 1132 | def read(self, size): # pylint: disable=method-hidden |
7584f5c9 ERE |
1133 | self.read = self.fileobj.read |
1134 | return self.buf | |
1135 | ||
1136 | def getcomptype(self): | |
d601d33b | 1137 | if self.buf.startswith(GZ_MAGIC_DEFLATE): |
7584f5c9 | 1138 | return "gz" |
be60ffd0 | 1139 | elif self.buf[0:3] == b"BZh" and self.buf[4:10] == b"1AY&SY": |
7584f5c9 | 1140 | return "bz2" |
be60ffd0 ERE |
1141 | elif self.buf.startswith((b"\x5d\x00\x00\x80", b"\xfd7zXZ")): |
1142 | return "xz" | |
1143 | else: | |
1144 | return "tar" | |
7584f5c9 ERE |
1145 | |
1146 | def close(self): | |
1147 | self.fileobj.close() | |
1148 | # class StreamProxy | |
1149 | ||
7584f5c9 ERE |
1150 | #------------------------ |
1151 | # Extraction file object | |
1152 | #------------------------ | |
1153 | class _FileInFile(object): | |
1154 | """A thin wrapper around an existing file object that | |
1155 | provides a part of its data as an individual file | |
1156 | object. | |
1157 | """ | |
1158 | ||
be60ffd0 | 1159 | def __init__(self, fileobj, offset, size, blockinfo=None): |
7584f5c9 ERE |
1160 | self.fileobj = fileobj |
1161 | self.offset = offset | |
1162 | self.size = size | |
7584f5c9 | 1163 | self.position = 0 |
be60ffd0 ERE |
1164 | self.name = getattr(fileobj, "name", None) |
1165 | self.closed = False | |
1166 | ||
1167 | if blockinfo is None: | |
1168 | blockinfo = [(0, size)] | |
1169 | ||
1170 | # Construct a map with data and zero blocks. | |
1171 | self.map_index = 0 | |
1172 | self.map = [] | |
1173 | lastpos = 0 | |
1174 | realpos = self.offset | |
1175 | for offset, size in blockinfo: | |
1176 | if offset > lastpos: | |
1177 | self.map.append((False, lastpos, offset, None)) | |
1178 | self.map.append((True, offset, offset + size, realpos)) | |
1179 | realpos += size | |
1180 | lastpos = offset + size | |
1181 | if lastpos < self.size: | |
1182 | self.map.append((False, lastpos, self.size, None)) | |
1183 | ||
1184 | def flush(self): | |
1185 | pass | |
1186 | ||
1187 | def readable(self): | |
1188 | return True | |
1189 | ||
1190 | def writable(self): | |
1191 | return False | |
1192 | ||
1193 | def seekable(self): | |
1194 | return self.fileobj.seekable() | |
7584f5c9 ERE |
1195 | |
1196 | def tell(self): | |
1197 | """Return the current file position. | |
1198 | """ | |
1199 | return self.position | |
1200 | ||
be60ffd0 | 1201 | def seek(self, position, whence=io.SEEK_SET): |
7584f5c9 ERE |
1202 | """Seek to a position in the file. |
1203 | """ | |
be60ffd0 ERE |
1204 | if whence == io.SEEK_SET: |
1205 | self.position = min(max(position, 0), self.size) | |
1206 | elif whence == io.SEEK_CUR: | |
1207 | if position < 0: | |
1208 | self.position = max(self.position + position, 0) | |
1209 | else: | |
1210 | self.position = min(self.position + position, self.size) | |
1211 | elif whence == io.SEEK_END: | |
1212 | self.position = max(min(self.size + position, self.size), 0) | |
1213 | else: | |
1214 | raise ValueError("Invalid argument") | |
1215 | return self.position | |
7584f5c9 ERE |
1216 | |
1217 | def read(self, size=None): | |
1218 | """Read data from the file. | |
1219 | """ | |
1220 | if size is None: | |
1221 | size = self.size - self.position | |
1222 | else: | |
1223 | size = min(size, self.size - self.position) | |
1224 | ||
be60ffd0 | 1225 | buf = b"" |
7584f5c9 | 1226 | while size > 0: |
7584f5c9 | 1227 | while True: |
be60ffd0 ERE |
1228 | data, start, stop, offset = self.map[self.map_index] |
1229 | if start <= self.position < stop: | |
7584f5c9 | 1230 | break |
be60ffd0 ERE |
1231 | else: |
1232 | self.map_index += 1 | |
1233 | if self.map_index == len(self.map): | |
1234 | self.map_index = 0 | |
1235 | length = min(size, stop - self.position) | |
1236 | if data: | |
1237 | self.fileobj.seek(offset + (self.position - start)) | |
1238 | buf += self.fileobj.read(length) | |
7584f5c9 | 1239 | else: |
be60ffd0 ERE |
1240 | buf += NUL * length |
1241 | size -= length | |
1242 | self.position += length | |
1243 | return buf | |
7584f5c9 | 1244 | |
be60ffd0 ERE |
1245 | def readinto(self, b): |
1246 | buf = self.read(len(b)) | |
1247 | b[:len(buf)] = buf | |
1248 | return len(buf) | |
7584f5c9 ERE |
1249 | |
1250 | def close(self): | |
7584f5c9 | 1251 | self.closed = True |
be60ffd0 | 1252 | #class _FileInFile |
7584f5c9 | 1253 | |
be60ffd0 ERE |
1254 | |
1255 | class ExFileObject(io.BufferedReader): | |
1256 | ||
1257 | def __init__(self, tarfile, tarinfo): | |
1258 | fileobj = _FileInFile(tarfile.fileobj, tarinfo.offset_data, | |
1259 | tarinfo.size, tarinfo.sparse) | |
1260 | super().__init__(fileobj) | |
7584f5c9 ERE |
1261 | #class ExFileObject |
1262 | ||
1263 | #------------------ | |
1264 | # Exported Classes | |
1265 | #------------------ | |
1266 | class TarInfo(object): | |
1267 | """Informational class which holds the details about an | |
1268 | archive member given by a tar header block. | |
1269 | TarInfo objects are returned by TarFile.getmember(), | |
1270 | TarFile.getmembers() and TarFile.gettarinfo() and are | |
1271 | usually created internally. | |
1272 | """ | |
1273 | ||
be60ffd0 ERE |
1274 | __slots__ = ("name", "mode", "uid", "gid", "size", "mtime", |
1275 | "chksum", "type", "linkname", "uname", "gname", | |
1276 | "devmajor", "devminor", "volume_offset", | |
1277 | "offset", "offset_data", "pax_headers", "sparse", | |
1278 | "tarfile", "_sparse_structs", "_link_target") | |
1279 | ||
7584f5c9 ERE |
1280 | def __init__(self, name=""): |
1281 | """Construct a TarInfo object. name is the optional name | |
1282 | of the member. | |
1283 | """ | |
1284 | self.name = name # member name | |
be60ffd0 | 1285 | self.mode = 0o644 # file permissions |
7584f5c9 ERE |
1286 | self.uid = 0 # user id |
1287 | self.gid = 0 # group id | |
1288 | self.size = 0 # file size | |
1289 | self.mtime = 0 # modification time | |
1290 | self.chksum = 0 # header checksum | |
1291 | self.type = REGTYPE # member type | |
1292 | self.linkname = "" # link name | |
1293 | self.uname = "" # user name | |
1294 | self.gname = "" # group name | |
1295 | self.devmajor = 0 # device major number | |
1296 | self.devminor = 0 # device minor number | |
1297 | ||
1298 | self.offset = 0 # the tar header starts here | |
1299 | self.offset_data = 0 # the file's data starts here | |
0eb5048f ERE |
1300 | self.volume_offset = 0 # the file's data corresponds with the data |
1301 | # starting at this position | |
7584f5c9 | 1302 | |
be60ffd0 | 1303 | self.sparse = None # sparse member information |
7584f5c9 ERE |
1304 | self.pax_headers = {} # pax header information |
1305 | ||
1306 | # In pax headers the "name" and "linkname" field are called | |
1307 | # "path" and "linkpath". | |
1308 | def _getpath(self): | |
1309 | return self.name | |
1310 | def _setpath(self, name): | |
1311 | self.name = name | |
1312 | path = property(_getpath, _setpath) | |
1313 | ||
1314 | def _getlinkpath(self): | |
1315 | return self.linkname | |
1316 | def _setlinkpath(self, linkname): | |
1317 | self.linkname = linkname | |
1318 | linkpath = property(_getlinkpath, _setlinkpath) | |
1319 | ||
1320 | def __repr__(self): | |
1321 | return "<%s %r at %#x>" % (self.__class__.__name__,self.name,id(self)) | |
1322 | ||
be60ffd0 | 1323 | def get_info(self, encoding=None, errors=None): |
7584f5c9 ERE |
1324 | """Return the TarInfo's attributes as a dictionary. |
1325 | """ | |
1326 | info = { | |
1327 | "name": self.name, | |
be60ffd0 | 1328 | "mode": self.mode & 0o7777, |
7584f5c9 ERE |
1329 | "uid": self.uid, |
1330 | "gid": self.gid, | |
1331 | "size": self.size, | |
1332 | "mtime": self.mtime, | |
1333 | "chksum": self.chksum, | |
1334 | "type": self.type, | |
1335 | "linkname": self.linkname, | |
1336 | "uname": self.uname, | |
1337 | "gname": self.gname, | |
1338 | "devmajor": self.devmajor, | |
36a315a0 | 1339 | "devminor": self.devminor, |
0eb5048f ERE |
1340 | "offset_data": self.offset_data, |
1341 | "volume_offset": self.volume_offset | |
7584f5c9 ERE |
1342 | } |
1343 | ||
1344 | if info["type"] == DIRTYPE and not info["name"].endswith("/"): | |
1345 | info["name"] += "/" | |
1346 | ||
7584f5c9 ERE |
1347 | return info |
1348 | ||
be60ffd0 ERE |
1349 | def tobuf(self, format=DEFAULT_FORMAT, encoding=ENCODING, |
1350 | errors="surrogateescape"): | |
7584f5c9 ERE |
1351 | """Return a tar header as a string of 512 byte blocks. |
1352 | """ | |
1353 | info = self.get_info(encoding, errors) | |
1354 | ||
1355 | if format == USTAR_FORMAT: | |
be60ffd0 | 1356 | return self.create_ustar_header(info, encoding, errors) |
7584f5c9 | 1357 | elif format == GNU_FORMAT: |
be60ffd0 | 1358 | return self.create_gnu_header(info, encoding, errors) |
7584f5c9 ERE |
1359 | elif format == PAX_FORMAT: |
1360 | return self.create_pax_header(info, encoding, errors) | |
1361 | else: | |
1362 | raise ValueError("invalid format") | |
1363 | ||
be60ffd0 | 1364 | def create_ustar_header(self, info, encoding, errors): |
7584f5c9 ERE |
1365 | """Return the object as a ustar header block. |
1366 | """ | |
1367 | info["magic"] = POSIX_MAGIC | |
1368 | ||
1369 | if len(info["linkname"]) > LENGTH_LINK: | |
1370 | raise ValueError("linkname is too long") | |
1371 | ||
1372 | if len(info["name"]) > LENGTH_NAME: | |
1373 | info["prefix"], info["name"] = self._posix_split_name(info["name"]) | |
1374 | ||
be60ffd0 | 1375 | return self._create_header(info, USTAR_FORMAT, encoding, errors) |
7584f5c9 | 1376 | |
be60ffd0 | 1377 | def create_gnu_header(self, info, encoding, errors): |
7584f5c9 ERE |
1378 | """Return the object as a GNU header block sequence. |
1379 | """ | |
1380 | info["magic"] = GNU_MAGIC | |
1381 | ||
2f854e77 ERE |
1382 | if self.ismultivol(): |
1383 | prefix = [ | |
1384 | itn(info.get("atime", 0), 12, GNU_FORMAT), | |
1385 | itn(info.get("ctime", 0), 12, GNU_FORMAT), | |
0eb5048f | 1386 | itn(self.volume_offset, 12, GNU_FORMAT), |
2f854e77 ERE |
1387 | itn(0, 119, GNU_FORMAT), # stuff unused in this tar implementation, set to zero |
1388 | ] | |
be60ffd0 | 1389 | info['prefix'] = b"".join(prefix) |
0eb5048f | 1390 | info['size'] = info['size'] - self.volume_offset |
2f854e77 | 1391 | |
be60ffd0 | 1392 | buf = b"" |
7584f5c9 | 1393 | if len(info["linkname"]) > LENGTH_LINK: |
be60ffd0 ERE |
1394 | buf += self._create_gnu_long_header(info["linkname"], |
1395 | GNUTYPE_LONGLINK, encoding, errors) | |
7584f5c9 ERE |
1396 | |
1397 | if len(info["name"]) > LENGTH_NAME: | |
be60ffd0 ERE |
1398 | buf += self._create_gnu_long_header(info["name"], GNUTYPE_LONGNAME, |
1399 | encoding, errors) | |
7584f5c9 | 1400 | |
be60ffd0 | 1401 | return buf + self._create_header(info, GNU_FORMAT, encoding, errors) |
7584f5c9 ERE |
1402 | |
1403 | def create_pax_header(self, info, encoding, errors): | |
1404 | """Return the object as a ustar header block. If it cannot be | |
1405 | represented this way, prepend a pax extended header sequence | |
1406 | with supplement information. | |
1407 | """ | |
1408 | info["magic"] = POSIX_MAGIC | |
1409 | pax_headers = self.pax_headers.copy() | |
c04e0751 ERE |
1410 | if self.ismultivol(): |
1411 | info['size'] = info['size'] - self.volume_offset | |
7584f5c9 ERE |
1412 | |
1413 | # Test string fields for values that exceed the field length or cannot | |
1414 | # be represented in ASCII encoding. | |
1415 | for name, hname, length in ( | |
36a315a0 ERE |
1416 | ("name", "path", LENGTH_NAME), |
1417 | ("linkname", "linkpath", LENGTH_LINK), | |
1418 | ("uname", "uname", 32), | |
1419 | ("gname", "gname", 32)): | |
7584f5c9 ERE |
1420 | |
1421 | if hname in pax_headers: | |
1422 | # The pax header has priority. | |
1423 | continue | |
1424 | ||
7584f5c9 ERE |
1425 | # Try to encode the string as ASCII. |
1426 | try: | |
be60ffd0 | 1427 | info[name].encode("ascii", "strict") |
7584f5c9 | 1428 | except UnicodeEncodeError: |
be60ffd0 | 1429 | pax_headers[hname] = info[name] |
7584f5c9 ERE |
1430 | continue |
1431 | ||
1432 | if len(info[name]) > length: | |
be60ffd0 | 1433 | pax_headers[hname] = info[name] |
7584f5c9 ERE |
1434 | |
1435 | # Test number fields for values that exceed the field limit or values | |
1436 | # that like to be stored as float. | |
1437 | for name, digits in (("uid", 8), ("gid", 8), ("size", 12), ("mtime", 12)): | |
1438 | if name in pax_headers: | |
1439 | # The pax header has priority. Avoid overflow. | |
1440 | info[name] = 0 | |
1441 | continue | |
1442 | ||
1443 | val = info[name] | |
1444 | if not 0 <= val < 8 ** (digits - 1) or isinstance(val, float): | |
be60ffd0 | 1445 | pax_headers[name] = str(val) |
7584f5c9 ERE |
1446 | info[name] = 0 |
1447 | ||
1448 | # Create a pax extended header if necessary. | |
1449 | if pax_headers: | |
be60ffd0 | 1450 | buf = self._create_pax_generic_header(pax_headers, XHDTYPE, encoding) |
7584f5c9 | 1451 | else: |
be60ffd0 | 1452 | buf = b"" |
7584f5c9 | 1453 | |
be60ffd0 | 1454 | return buf + self._create_header(info, USTAR_FORMAT, "ascii", "replace") |
7584f5c9 ERE |
1455 | |
1456 | @classmethod | |
1457 | def create_pax_global_header(cls, pax_headers): | |
1458 | """Return the object as a pax global header block sequence. | |
1459 | """ | |
be60ffd0 | 1460 | return cls._create_pax_generic_header(pax_headers, XGLTYPE, "utf-8") |
7584f5c9 ERE |
1461 | |
1462 | def _posix_split_name(self, name): | |
1463 | """Split a name longer than 100 chars into a prefix | |
1464 | and a name part. | |
1465 | """ | |
1466 | prefix = name[:LENGTH_PREFIX + 1] | |
1467 | while prefix and prefix[-1] != "/": | |
1468 | prefix = prefix[:-1] | |
1469 | ||
1470 | name = name[len(prefix):] | |
1471 | prefix = prefix[:-1] | |
1472 | ||
1473 | if not prefix or len(name) > LENGTH_NAME: | |
1474 | raise ValueError("name is too long") | |
1475 | return prefix, name | |
1476 | ||
1477 | @staticmethod | |
be60ffd0 | 1478 | def _create_header(info, format, encoding, errors): |
7584f5c9 ERE |
1479 | """Return a header block. info is a dictionary with file |
1480 | information, format must be one of the *_FORMAT constants. | |
1481 | """ | |
1482 | parts = [ | |
be60ffd0 ERE |
1483 | stn(info.get("name", ""), 100, encoding, errors), |
1484 | itn(info.get("mode", 0) & 0o7777, 8, format), | |
7584f5c9 ERE |
1485 | itn(info.get("uid", 0), 8, format), |
1486 | itn(info.get("gid", 0), 8, format), | |
1487 | itn(info.get("size", 0), 12, format), | |
1488 | itn(info.get("mtime", 0), 12, format), | |
be60ffd0 | 1489 | b" ", # checksum field |
2f854e77 | 1490 | info.get("type", REGTYPE), |
be60ffd0 ERE |
1491 | stn(info.get("linkname", ""), 100, encoding, errors), |
1492 | info.get("magic", POSIX_MAGIC), | |
1493 | stn(info.get("uname", ""), 32, encoding, errors), | |
1494 | stn(info.get("gname", ""), 32, encoding, errors), | |
7584f5c9 ERE |
1495 | itn(info.get("devmajor", 0), 8, format), |
1496 | itn(info.get("devminor", 0), 8, format), | |
be60ffd0 | 1497 | sbtn(info.get("prefix", ""), 155, encoding, errors) |
7584f5c9 ERE |
1498 | ] |
1499 | ||
be60ffd0 | 1500 | buf = struct.pack("%ds" % BLOCKSIZE, b"".join(parts)) |
7584f5c9 | 1501 | chksum = calc_chksums(buf[-BLOCKSIZE:])[0] |
be60ffd0 | 1502 | buf = buf[:-364] + bytes("%06o\0" % chksum, "ascii") + buf[-357:] |
7584f5c9 ERE |
1503 | return buf |
1504 | ||
1505 | @staticmethod | |
1506 | def _create_payload(payload): | |
1507 | """Return the string payload filled with zero bytes | |
1508 | up to the next 512 byte border. | |
1509 | """ | |
1510 | blocks, remainder = divmod(len(payload), BLOCKSIZE) | |
1511 | if remainder > 0: | |
1512 | payload += (BLOCKSIZE - remainder) * NUL | |
1513 | return payload | |
1514 | ||
1515 | @classmethod | |
be60ffd0 | 1516 | def _create_gnu_long_header(cls, name, type, encoding, errors): |
7584f5c9 ERE |
1517 | """Return a GNUTYPE_LONGNAME or GNUTYPE_LONGLINK sequence |
1518 | for name. | |
1519 | """ | |
be60ffd0 | 1520 | name = name.encode(encoding, errors) + NUL |
7584f5c9 ERE |
1521 | |
1522 | info = {} | |
1523 | info["name"] = "././@LongLink" | |
1524 | info["type"] = type | |
1525 | info["size"] = len(name) | |
1526 | info["magic"] = GNU_MAGIC | |
1527 | ||
1528 | # create extended header + name blocks. | |
be60ffd0 | 1529 | return cls._create_header(info, USTAR_FORMAT, encoding, errors) + \ |
7584f5c9 ERE |
1530 | cls._create_payload(name) |
1531 | ||
1532 | @classmethod | |
be60ffd0 ERE |
1533 | def _create_pax_generic_header(cls, pax_headers, type, encoding): |
1534 | """Return a POSIX.1-2008 extended or global header sequence | |
7584f5c9 | 1535 | that contains a list of keyword, value pairs. The values |
be60ffd0 | 1536 | must be strings. |
7584f5c9 | 1537 | """ |
be60ffd0 ERE |
1538 | # Check if one of the fields contains surrogate characters and thereby |
1539 | # forces hdrcharset=BINARY, see _proc_pax() for more information. | |
1540 | binary = False | |
1541 | for keyword, value in pax_headers.items(): | |
1542 | try: | |
1543 | value.encode("utf-8", "strict") | |
1544 | except UnicodeEncodeError: | |
1545 | binary = True | |
1546 | break | |
1547 | ||
1548 | records = b"" | |
1549 | if binary: | |
1550 | # Put the hdrcharset field at the beginning of the header. | |
1551 | records += b"21 hdrcharset=BINARY\n" | |
1552 | ||
1553 | for keyword, value in pax_headers.items(): | |
1554 | keyword = keyword.encode("utf-8") | |
1555 | if binary: | |
1556 | # Try to restore the original byte representation of `value'. | |
1557 | # Needless to say, that the encoding must match the string. | |
1558 | value = value.encode(encoding, "surrogateescape") | |
1559 | else: | |
1560 | value = value.encode("utf-8") | |
1561 | ||
7584f5c9 ERE |
1562 | l = len(keyword) + len(value) + 3 # ' ' + '=' + '\n' |
1563 | n = p = 0 | |
1564 | while True: | |
1565 | n = l + len(str(p)) | |
1566 | if n == p: | |
1567 | break | |
1568 | p = n | |
be60ffd0 | 1569 | records += bytes(str(p), "ascii") + b" " + keyword + b"=" + value + b"\n" |
7584f5c9 ERE |
1570 | |
1571 | # We use a hardcoded "././@PaxHeader" name like star does | |
1572 | # instead of the one that POSIX recommends. | |
1573 | info = {} | |
1574 | info["name"] = "././@PaxHeader" | |
1575 | info["type"] = type | |
1576 | info["size"] = len(records) | |
1577 | info["magic"] = POSIX_MAGIC | |
1578 | ||
1579 | # Create pax header + record blocks. | |
be60ffd0 | 1580 | return cls._create_header(info, USTAR_FORMAT, "ascii", "replace") + \ |
7584f5c9 ERE |
1581 | cls._create_payload(records) |
1582 | ||
1583 | @classmethod | |
be60ffd0 ERE |
1584 | def frombuf(cls, buf, encoding, errors): |
1585 | """Construct a TarInfo object from a 512 byte bytes object. | |
7584f5c9 ERE |
1586 | """ |
1587 | if len(buf) == 0: | |
1588 | raise EmptyHeaderError("empty header") | |
1589 | if len(buf) != BLOCKSIZE: | |
1590 | raise TruncatedHeaderError("truncated header") | |
1591 | if buf.count(NUL) == BLOCKSIZE: | |
1592 | raise EOFHeaderError("end of file header") | |
1593 | ||
1594 | chksum = nti(buf[148:156]) | |
1595 | if chksum not in calc_chksums(buf): | |
1596 | raise InvalidHeaderError("bad checksum") | |
1597 | ||
1598 | obj = cls() | |
be60ffd0 | 1599 | obj.name = nts(buf[0:100], encoding, errors) |
7584f5c9 ERE |
1600 | obj.mode = nti(buf[100:108]) |
1601 | obj.uid = nti(buf[108:116]) | |
1602 | obj.gid = nti(buf[116:124]) | |
1603 | obj.size = nti(buf[124:136]) | |
1604 | obj.mtime = nti(buf[136:148]) | |
1605 | obj.chksum = chksum | |
1606 | obj.type = buf[156:157] | |
be60ffd0 ERE |
1607 | obj.linkname = nts(buf[157:257], encoding, errors) |
1608 | obj.uname = nts(buf[265:297], encoding, errors) | |
1609 | obj.gname = nts(buf[297:329], encoding, errors) | |
7584f5c9 ERE |
1610 | obj.devmajor = nti(buf[329:337]) |
1611 | obj.devminor = nti(buf[337:345]) | |
be60ffd0 ERE |
1612 | prefix = nts(buf[345:500], encoding, errors) |
1613 | ||
1614 | # The old GNU sparse format occupies some of the unused | |
1615 | # space in the buffer for up to 4 sparse structures. | |
1616 | # Save the them for later processing in _proc_sparse(). | |
1617 | if obj.type == GNUTYPE_SPARSE: | |
1618 | pos = 386 | |
1619 | structs = [] | |
1620 | for i in range(4): | |
1621 | try: | |
1622 | offset = nti(buf[pos:pos + 12]) | |
1623 | numbytes = nti(buf[pos + 12:pos + 24]) | |
1624 | except ValueError: | |
1625 | break | |
1626 | structs.append((offset, numbytes)) | |
1627 | pos += 24 | |
1628 | isextended = bool(buf[482]) | |
1629 | origsize = nti(buf[483:495]) | |
1630 | obj._sparse_structs = (structs, isextended, origsize) | |
7584f5c9 ERE |
1631 | |
1632 | # Old V7 tar format represents a directory as a regular | |
1633 | # file with a trailing slash. | |
1634 | if obj.type == AREGTYPE and obj.name.endswith("/"): | |
1635 | obj.type = DIRTYPE | |
1636 | ||
1637 | # Remove redundant slashes from directories. | |
1638 | if obj.isdir(): | |
1639 | obj.name = obj.name.rstrip("/") | |
1640 | ||
1641 | # Reconstruct a ustar longname. | |
1642 | if prefix and obj.type not in GNU_TYPES: | |
1643 | obj.name = prefix + "/" + obj.name | |
c474439c ERE |
1644 | else: |
1645 | obj.offset_data = nti(buf[369:381]) | |
7584f5c9 ERE |
1646 | return obj |
1647 | ||
1648 | @classmethod | |
1649 | def fromtarfile(cls, tarfile): | |
1650 | """Return the next TarInfo object from TarFile object | |
1651 | tarfile. | |
1652 | """ | |
1653 | buf = tarfile.fileobj.read(BLOCKSIZE) | |
be60ffd0 | 1654 | obj = cls.frombuf(buf, tarfile.encoding, tarfile.errors) |
7584f5c9 ERE |
1655 | obj.offset = tarfile.fileobj.tell() - BLOCKSIZE |
1656 | return obj._proc_member(tarfile) | |
1657 | ||
1658 | #-------------------------------------------------------------------------- | |
1659 | # The following are methods that are called depending on the type of a | |
1660 | # member. The entry point is _proc_member() which can be overridden in a | |
1661 | # subclass to add custom _proc_*() methods. A _proc_*() method MUST | |
1662 | # implement the following | |
1663 | # operations: | |
1664 | # 1. Set self.offset_data to the position where the data blocks begin, | |
1665 | # if there is data that follows. | |
1666 | # 2. Set tarfile.offset to the position where the next member's header will | |
1667 | # begin. | |
1668 | # 3. Return self or another valid TarInfo object. | |
1669 | def _proc_member(self, tarfile): | |
1670 | """Choose the right processing method depending on | |
1671 | the type and call it. | |
1672 | """ | |
1673 | if self.type in (GNUTYPE_LONGNAME, GNUTYPE_LONGLINK): | |
1674 | return self._proc_gnulong(tarfile) | |
1675 | elif self.type == GNUTYPE_SPARSE: | |
1676 | return self._proc_sparse(tarfile) | |
1677 | elif self.type in (XHDTYPE, XGLTYPE, SOLARIS_XHDTYPE): | |
1678 | return self._proc_pax(tarfile) | |
1679 | else: | |
1680 | return self._proc_builtin(tarfile) | |
1681 | ||
1682 | def _proc_builtin(self, tarfile): | |
1683 | """Process a builtin type or an unknown type which | |
1684 | will be treated as a regular file. | |
1685 | """ | |
1686 | self.offset_data = tarfile.fileobj.tell() | |
1687 | offset = self.offset_data | |
00c34a12 | 1688 | if self.isreg() or self.ismultivol() or self.type not in SUPPORTED_TYPES: |
7584f5c9 ERE |
1689 | # Skip the following data blocks. |
1690 | offset += self._block(self.size) | |
1691 | tarfile.offset = offset | |
1692 | ||
1693 | # Patch the TarInfo object with saved global | |
1694 | # header information. | |
1695 | self._apply_pax_info(tarfile.pax_headers, tarfile.encoding, tarfile.errors) | |
1696 | ||
1697 | return self | |
1698 | ||
1699 | def _proc_gnulong(self, tarfile): | |
1700 | """Process the blocks that hold a GNU longname | |
1701 | or longlink member. | |
1702 | """ | |
1703 | buf = tarfile.fileobj.read(self._block(self.size)) | |
1704 | ||
1705 | # Fetch the next header and process it. | |
1706 | try: | |
1707 | next = self.fromtarfile(tarfile) | |
1708 | except HeaderError: | |
1709 | raise SubsequentHeaderError("missing or bad subsequent header") | |
1710 | ||
1711 | # Patch the TarInfo object from the next header with | |
1712 | # the longname information. | |
1713 | next.offset = self.offset | |
1714 | if self.type == GNUTYPE_LONGNAME: | |
be60ffd0 | 1715 | next.name = nts(buf, tarfile.encoding, tarfile.errors) |
7584f5c9 | 1716 | elif self.type == GNUTYPE_LONGLINK: |
be60ffd0 | 1717 | next.linkname = nts(buf, tarfile.encoding, tarfile.errors) |
7584f5c9 ERE |
1718 | |
1719 | return next | |
1720 | ||
1721 | def _proc_sparse(self, tarfile): | |
1722 | """Process a GNU sparse header plus extra headers. | |
1723 | """ | |
be60ffd0 ERE |
1724 | # We already collected some sparse structures in frombuf(). |
1725 | structs, isextended, origsize = self._sparse_structs | |
1726 | del self._sparse_structs | |
1727 | ||
1728 | # Collect sparse structures from extended header blocks. | |
1729 | while isextended: | |
7584f5c9 ERE |
1730 | buf = tarfile.fileobj.read(BLOCKSIZE) |
1731 | pos = 0 | |
be60ffd0 | 1732 | for i in range(21): |
7584f5c9 ERE |
1733 | try: |
1734 | offset = nti(buf[pos:pos + 12]) | |
1735 | numbytes = nti(buf[pos + 12:pos + 24]) | |
1736 | except ValueError: | |
1737 | break | |
be60ffd0 ERE |
1738 | if offset and numbytes: |
1739 | structs.append((offset, numbytes)) | |
7584f5c9 | 1740 | pos += 24 |
be60ffd0 ERE |
1741 | isextended = bool(buf[504]) |
1742 | self.sparse = structs | |
7584f5c9 ERE |
1743 | |
1744 | self.offset_data = tarfile.fileobj.tell() | |
1745 | tarfile.offset = self.offset_data + self._block(self.size) | |
1746 | self.size = origsize | |
7584f5c9 ERE |
1747 | return self |
1748 | ||
1749 | def _proc_pax(self, tarfile): | |
1750 | """Process an extended or global header as described in | |
be60ffd0 | 1751 | POSIX.1-2008. |
7584f5c9 ERE |
1752 | """ |
1753 | # Read the header information. | |
1754 | buf = tarfile.fileobj.read(self._block(self.size)) | |
1755 | ||
1756 | # A pax header stores supplemental information for either | |
1757 | # the following file (extended) or all following files | |
1758 | # (global). | |
1759 | if self.type == XGLTYPE: | |
1760 | pax_headers = tarfile.pax_headers | |
1761 | else: | |
1762 | pax_headers = tarfile.pax_headers.copy() | |
1763 | ||
be60ffd0 ERE |
1764 | # Check if the pax header contains a hdrcharset field. This tells us |
1765 | # the encoding of the path, linkpath, uname and gname fields. Normally, | |
1766 | # these fields are UTF-8 encoded but since POSIX.1-2008 tar | |
1767 | # implementations are allowed to store them as raw binary strings if | |
1768 | # the translation to UTF-8 fails. | |
1769 | match = re.search(br"\d+ hdrcharset=([^\n]+)\n", buf) | |
1770 | if match is not None: | |
1771 | pax_headers["hdrcharset"] = match.group(1).decode("utf-8") | |
1772 | ||
1773 | # For the time being, we don't care about anything other than "BINARY". | |
1774 | # The only other value that is currently allowed by the standard is | |
1775 | # "ISO-IR 10646 2000 UTF-8" in other words UTF-8. | |
1776 | hdrcharset = pax_headers.get("hdrcharset") | |
1777 | if hdrcharset == "BINARY": | |
1778 | encoding = tarfile.encoding | |
1779 | else: | |
1780 | encoding = "utf-8" | |
1781 | ||
7584f5c9 ERE |
1782 | # Parse pax header information. A record looks like that: |
1783 | # "%d %s=%s\n" % (length, keyword, value). length is the size | |
1784 | # of the complete record including the length field itself and | |
1785 | # the newline. keyword and value are both UTF-8 encoded strings. | |
be60ffd0 | 1786 | regex = re.compile(br"(\d+) ([^=]+)=") |
7584f5c9 ERE |
1787 | pos = 0 |
1788 | while True: | |
1789 | match = regex.match(buf, pos) | |
1790 | if not match: | |
1791 | break | |
1792 | ||
1793 | length, keyword = match.groups() | |
1794 | length = int(length) | |
1795 | value = buf[match.end(2) + 1:match.start(1) + length - 1] | |
1796 | ||
be60ffd0 ERE |
1797 | # Normally, we could just use "utf-8" as the encoding and "strict" |
1798 | # as the error handler, but we better not take the risk. For | |
1799 | # example, GNU tar <= 1.23 is known to store filenames it cannot | |
1800 | # translate to UTF-8 as raw strings (unfortunately without a | |
1801 | # hdrcharset=BINARY header). | |
1802 | # We first try the strict standard encoding, and if that fails we | |
1803 | # fall back on the user's encoding and error handler. | |
1804 | keyword = self._decode_pax_field(keyword, "utf-8", "utf-8", | |
1805 | tarfile.errors) | |
1806 | if keyword in PAX_NAME_FIELDS: | |
1807 | value = self._decode_pax_field(value, encoding, tarfile.encoding, | |
1808 | tarfile.errors) | |
1809 | else: | |
1810 | value = self._decode_pax_field(value, "utf-8", "utf-8", | |
1811 | tarfile.errors) | |
7584f5c9 ERE |
1812 | |
1813 | pax_headers[keyword] = value | |
1814 | pos += length | |
1815 | ||
36a315a0 | 1816 | |
7584f5c9 ERE |
1817 | # Fetch the next header. |
1818 | try: | |
1819 | next = self.fromtarfile(tarfile) | |
1820 | except HeaderError: | |
1821 | raise SubsequentHeaderError("missing or bad subsequent header") | |
1822 | ||
be60ffd0 ERE |
1823 | # Process GNU sparse information. |
1824 | if "GNU.sparse.map" in pax_headers: | |
1825 | # GNU extended sparse format version 0.1. | |
1826 | self._proc_gnusparse_01(next, pax_headers) | |
1827 | ||
1828 | elif "GNU.sparse.size" in pax_headers: | |
1829 | # GNU extended sparse format version 0.0. | |
1830 | self._proc_gnusparse_00(next, pax_headers, buf) | |
1831 | ||
1832 | elif pax_headers.get("GNU.sparse.major") == "1" and pax_headers.get("GNU.sparse.minor") == "0": | |
1833 | # GNU extended sparse format version 1.0. | |
1834 | self._proc_gnusparse_10(next, pax_headers, tarfile) | |
1835 | ||
7584f5c9 ERE |
1836 | if self.type in (XHDTYPE, SOLARIS_XHDTYPE): |
1837 | # Patch the TarInfo object with the extended header info. | |
1838 | next._apply_pax_info(pax_headers, tarfile.encoding, tarfile.errors) | |
1839 | next.offset = self.offset | |
1840 | ||
1841 | if "size" in pax_headers: | |
1842 | # If the extended header replaces the size field, | |
1843 | # we need to recalculate the offset where the next | |
1844 | # header starts. | |
1845 | offset = next.offset_data | |
1846 | if next.isreg() or next.type not in SUPPORTED_TYPES: | |
1847 | offset += next._block(next.size) | |
1848 | tarfile.offset = offset | |
1849 | ||
c04e0751 ERE |
1850 | if next is not None: |
1851 | if "GNU.volume.filename" in pax_headers: | |
1852 | if pax_headers["GNU.volume.filename"] == next.name: | |
1853 | if "GNU.volume.size" in pax_headers: | |
1854 | next.size = int(pax_headers["GNU.volume.size"]) | |
1855 | if "GNU.volume.offset" in pax_headers: | |
1856 | next.volume_offset = int(pax_headers["GNU.volume.offset"]) | |
1857 | ||
1858 | for key in pax_headers.keys(): | |
1859 | if key.startswith("GNU.volume"): | |
1860 | del tarfile.pax_headers[key] | |
0eb5048f | 1861 | |
7584f5c9 ERE |
1862 | return next |
1863 | ||
be60ffd0 ERE |
1864 | def _proc_gnusparse_00(self, next, pax_headers, buf): |
1865 | """Process a GNU tar extended sparse header, version 0.0. | |
7584f5c9 | 1866 | """ |
be60ffd0 ERE |
1867 | offsets = [] |
1868 | for match in re.finditer(br"\d+ GNU.sparse.offset=(\d+)\n", buf): | |
1869 | offsets.append(int(match.group(1))) | |
1870 | numbytes = [] | |
1871 | for match in re.finditer(br"\d+ GNU.sparse.numbytes=(\d+)\n", buf): | |
1872 | numbytes.append(int(match.group(1))) | |
1873 | next.sparse = list(zip(offsets, numbytes)) | |
7584f5c9 | 1874 | |
be60ffd0 ERE |
1875 | def _proc_gnusparse_01(self, next, pax_headers): |
1876 | """Process a GNU tar extended sparse header, version 0.1. | |
1877 | """ | |
1878 | sparse = [int(x) for x in pax_headers["GNU.sparse.map"].split(",")] | |
1879 | next.sparse = list(zip(sparse[::2], sparse[1::2])) | |
7584f5c9 | 1880 | |
be60ffd0 ERE |
1881 | def _proc_gnusparse_10(self, next, pax_headers, tarfile): |
1882 | """Process a GNU tar extended sparse header, version 1.0. | |
1883 | """ | |
1884 | fields = None | |
1885 | sparse = [] | |
1886 | buf = tarfile.fileobj.read(BLOCKSIZE) | |
1887 | fields, buf = buf.split(b"\n", 1) | |
1888 | fields = int(fields) | |
1889 | while len(sparse) < fields * 2: | |
1890 | if b"\n" not in buf: | |
1891 | buf += tarfile.fileobj.read(BLOCKSIZE) | |
1892 | number, buf = buf.split(b"\n", 1) | |
1893 | sparse.append(int(number)) | |
1894 | next.offset_data = tarfile.fileobj.tell() | |
1895 | next.sparse = list(zip(sparse[::2], sparse[1::2])) | |
7584f5c9 | 1896 | |
be60ffd0 ERE |
1897 | def _apply_pax_info(self, pax_headers, encoding, errors): |
1898 | """Replace fields with supplemental information from a previous | |
1899 | pax extended or global header. | |
1900 | """ | |
1901 | for keyword, value in pax_headers.items(): | |
1902 | if keyword == "GNU.sparse.name": | |
1903 | setattr(self, "path", value) | |
1904 | elif keyword == "GNU.sparse.size": | |
1905 | setattr(self, "size", int(value)) | |
1906 | elif keyword == "GNU.sparse.realsize": | |
1907 | setattr(self, "size", int(value)) | |
1908 | elif keyword in PAX_FIELDS: | |
1909 | if keyword in PAX_NUMBER_FIELDS: | |
1910 | try: | |
1911 | value = PAX_NUMBER_FIELDS[keyword](value) | |
1912 | except ValueError: | |
1913 | value = 0 | |
1914 | if keyword == "path": | |
f0287fb7 | 1915 | value = value.rstrip("/") # pylint: disable=no-member |
be60ffd0 | 1916 | setattr(self, keyword, value) |
7584f5c9 ERE |
1917 | |
1918 | self.pax_headers = pax_headers.copy() | |
1919 | ||
be60ffd0 ERE |
1920 | def _decode_pax_field(self, value, encoding, fallback_encoding, fallback_errors): |
1921 | """Decode a single field from a pax record. | |
1922 | """ | |
1923 | try: | |
1924 | return value.decode(encoding, "strict") | |
1925 | except UnicodeDecodeError: | |
1926 | return value.decode(fallback_encoding, fallback_errors) | |
1927 | ||
7584f5c9 ERE |
1928 | def _block(self, count): |
1929 | """Round up a byte count by BLOCKSIZE and return it, | |
1930 | e.g. _block(834) => 1024. | |
1931 | """ | |
1932 | blocks, remainder = divmod(count, BLOCKSIZE) | |
1933 | if remainder: | |
1934 | blocks += 1 | |
1935 | return blocks * BLOCKSIZE | |
1936 | ||
1937 | def isreg(self): | |
1938 | return self.type in REGULAR_TYPES | |
1939 | def isfile(self): | |
1940 | return self.isreg() | |
1941 | def isdir(self): | |
1942 | return self.type == DIRTYPE | |
1943 | def issym(self): | |
1944 | return self.type == SYMTYPE | |
1945 | def islnk(self): | |
1946 | return self.type == LNKTYPE | |
1947 | def ischr(self): | |
1948 | return self.type == CHRTYPE | |
1949 | def isblk(self): | |
1950 | return self.type == BLKTYPE | |
1951 | def isfifo(self): | |
1952 | return self.type == FIFOTYPE | |
1953 | def issparse(self): | |
be60ffd0 | 1954 | return self.sparse is not None |
7584f5c9 ERE |
1955 | def isdev(self): |
1956 | return self.type in (CHRTYPE, BLKTYPE, FIFOTYPE) | |
68ddf955 | 1957 | def ismultivol(self): |
c04e0751 ERE |
1958 | return self.type == GNUTYPE_MULTIVOL or self.volume_offset > 0 or\ |
1959 | "GNU.volume.offset" in self.pax_headers | |
7584f5c9 ERE |
1960 | # class TarInfo |
1961 | ||
1962 | class TarFile(object): | |
1963 | """The TarFile Class provides an interface to tar archives. | |
1964 | """ | |
1965 | ||
1966 | debug = 0 # May be set from 0 (no msgs) to 3 (all msgs) | |
1967 | ||
1968 | dereference = False # If true, add content of linked file to the | |
1969 | # tar file, else the link. | |
1970 | ||
1971 | ignore_zeros = False # If true, skips empty or invalid blocks and | |
1972 | # continues processing. | |
1973 | ||
83f2d71e | 1974 | max_volume_size = None # If different from None, establishes maximum |
68ddf955 ERE |
1975 | # size of tar volumes |
1976 | ||
1977 | new_volume_handler = None # function handler to be executed before when | |
1978 | # a new volume is needed | |
1979 | ||
1980 | volume_number = 0 # current volume number, used for multi volume | |
1981 | # support | |
1982 | ||
7584f5c9 ERE |
1983 | errorlevel = 1 # If 0, fatal errors only appear in debug |
1984 | # messages (if debug >= 0). If > 0, errors | |
1985 | # are passed to the caller as exceptions. | |
1986 | ||
1987 | format = DEFAULT_FORMAT # The format to use when creating an archive. | |
1988 | ||
1989 | encoding = ENCODING # Encoding for 8-bit character strings. | |
1990 | ||
1991 | errors = None # Error handler for unicode conversion. | |
1992 | ||
1993 | tarinfo = TarInfo # The default TarInfo class to use. | |
1994 | ||
be60ffd0 | 1995 | fileobject = ExFileObject # The file-object for extractfile(). |
7584f5c9 | 1996 | |
d1c38f40 PG |
1997 | arcmode = ARCMODE_PLAIN # Object processing mode (“concat”, encryption, |
1998 | # compression) | |
5fdff89f | 1999 | |
ea625b04 ERE |
2000 | save_to_members = True # If new members are saved. This can be disabled |
2001 | # if you manage lots of files and don't want | |
2002 | # to have high memory usage | |
2003 | ||
9ef1fb87 TJ |
2004 | cache_uid2user = {} # cache to avoid getpwuid calls. It always parses /etc/passwd. |
2005 | cache_gid2group = {} # same cache for groups | |
2006 | ||
7584f5c9 ERE |
2007 | def __init__(self, name=None, mode="r", fileobj=None, format=None, |
2008 | tarinfo=None, dereference=None, ignore_zeros=None, encoding=None, | |
be60ffd0 | 2009 | errors="surrogateescape", pax_headers=None, debug=None, |
548bb8d5 | 2010 | errorlevel=None, max_volume_size=None, new_volume_handler=None, |
d1c38f40 | 2011 | concat=False, nacl=None, |
c7c736b6 | 2012 | save_to_members=True): |
7584f5c9 ERE |
2013 | """Open an (uncompressed) tar archive `name'. `mode' is either 'r' to |
2014 | read from an existing archive, 'a' to append data to an existing | |
2015 | file or 'w' to create a new file overwriting an existing one. `mode' | |
2016 | defaults to 'r'. | |
2017 | If `fileobj' is given, it is used for reading or writing data. If it | |
2018 | can be determined, `mode' is overridden by `fileobj's mode. | |
2019 | `fileobj' is not closed, when TarFile is closed. | |
2020 | """ | |
2021 | if len(mode) > 1 or mode not in "raw": | |
2022 | raise ValueError("mode must be 'r', 'a' or 'w'") | |
2023 | self.mode = mode | |
d1c38f40 | 2024 | self.arcmode = arcmode_set (concat) |
c7c736b6 | 2025 | self.nacl = nacl |
7584f5c9 ERE |
2026 | self._mode = {"r": "rb", "a": "r+b", "w": "wb"}[mode] |
2027 | ||
2028 | if not fileobj: | |
2029 | if self.mode == "a" and not os.path.exists(name): | |
2030 | # Create nonexistent files in append mode. | |
2031 | self.mode = "w" | |
2032 | self._mode = "wb" | |
2033 | fileobj = bltn_open(name, self._mode) | |
2034 | self._extfileobj = False | |
2035 | else: | |
2036 | if name is None and hasattr(fileobj, "name"): | |
2037 | name = fileobj.name | |
d5361dac | 2038 | # when fileobj is a gzip.GzipFile, fileobj.mode is an int (not valid for us) |
be60ffd0 | 2039 | if hasattr(fileobj, "mode") and isinstance(fileobj.mode, str): |
7584f5c9 ERE |
2040 | self._mode = fileobj.mode |
2041 | self._extfileobj = True | |
be60ffd0 | 2042 | self.name = os.path.abspath(name) if name else None |
2f854e77 | 2043 | self.base_name = self.name = os.path.abspath(name) if name else None |
7584f5c9 ERE |
2044 | self.fileobj = fileobj |
2045 | ||
2046 | # Init attributes. | |
2047 | if format is not None: | |
2048 | self.format = format | |
2049 | if tarinfo is not None: | |
2050 | self.tarinfo = tarinfo | |
2051 | if dereference is not None: | |
2052 | self.dereference = dereference | |
2053 | if ignore_zeros is not None: | |
2054 | self.ignore_zeros = ignore_zeros | |
2055 | if encoding is not None: | |
2056 | self.encoding = encoding | |
2057 | ||
be60ffd0 | 2058 | self.errors = errors |
7584f5c9 ERE |
2059 | |
2060 | if pax_headers is not None and self.format == PAX_FORMAT: | |
2061 | self.pax_headers = pax_headers | |
2062 | else: | |
2063 | self.pax_headers = {} | |
2064 | ||
2065 | if debug is not None: | |
2066 | self.debug = debug | |
2067 | if errorlevel is not None: | |
2068 | self.errorlevel = errorlevel | |
2069 | ||
2070 | # Init datastructures. | |
ae48acc8 | 2071 | if max_volume_size and max_volume_size < 3*BLOCKSIZE: |
0c818a18 | 2072 | raise ValueError("max_volume_size needs to be at least %d" % (3*BLOCKSIZE)) |
ae48acc8 ERE |
2073 | if max_volume_size and not callable(new_volume_handler): |
2074 | raise ValueError("new_volume_handler needs to be set and be callable for multivolume support") | |
5ab3f8f9 CH |
2075 | if max_volume_size: |
2076 | self.max_volume_size = int(max_volume_size) | |
2077 | else: | |
2078 | self.max_volume_size = None | |
ae48acc8 | 2079 | |
ea625b04 | 2080 | self.save_to_members = save_to_members |
68ddf955 | 2081 | self.new_volume_handler = new_volume_handler |
7584f5c9 ERE |
2082 | self.closed = False |
2083 | self.members = [] # list of members as TarInfo objects | |
2084 | self._loaded = False # flag if all members have been read | |
2085 | self.offset = self.fileobj.tell() | |
2086 | # current position in the archive file | |
2087 | self.inodes = {} # dictionary caching the inodes of | |
2088 | # archive members already added | |
2089 | ||
2090 | try: | |
2091 | if self.mode == "r": | |
2092 | self.firstmember = None | |
2093 | self.firstmember = self.next() | |
2094 | ||
2095 | if self.mode == "a": | |
2096 | # Move to the end of the archive, | |
2097 | # before the first empty block. | |
2098 | while True: | |
2099 | self.fileobj.seek(self.offset) | |
2100 | try: | |
2101 | tarinfo = self.tarinfo.fromtarfile(self) | |
2102 | self.members.append(tarinfo) | |
2103 | except EOFHeaderError: | |
2104 | self.fileobj.seek(self.offset) | |
2105 | break | |
be60ffd0 | 2106 | except HeaderError as e: |
7584f5c9 ERE |
2107 | raise ReadError(str(e)) |
2108 | ||
2109 | if self.mode in "aw": | |
2110 | self._loaded = True | |
2111 | ||
2112 | if self.pax_headers: | |
2113 | buf = self.tarinfo.create_pax_global_header(self.pax_headers.copy()) | |
2114 | self.fileobj.write(buf) | |
2115 | self.offset += len(buf) | |
2116 | except: | |
2117 | if not self._extfileobj: | |
2118 | self.fileobj.close() | |
2119 | self.closed = True | |
2120 | raise | |
2121 | ||
7584f5c9 ERE |
2122 | #-------------------------------------------------------------------------- |
2123 | # Below are the classmethods which act as alternate constructors to the | |
2124 | # TarFile class. The open() method is the only one that is needed for | |
2125 | # public use; it is the "super"-constructor and is able to select an | |
2126 | # adequate "sub"-constructor for a particular compression using the mapping | |
2127 | # from OPEN_METH. | |
2128 | # | |
2129 | # This concept allows one to subclass TarFile without losing the comfort of | |
2130 | # the super-constructor. A sub-constructor is registered and made available | |
2131 | # by adding it to the mapping in OPEN_METH. | |
2132 | ||
2133 | @classmethod | |
2b82f50c | 2134 | def open(cls, name=None, mode="r", fileobj=None, bufsize=RECORDSIZE, |
04f4c7ab PG |
2135 | encryption=None, compresslevel=9, tolerance=TOLERANCE_STRICT, |
2136 | **kwargs): | |
7584f5c9 ERE |
2137 | """Open a tar archive for reading, writing or appending. Return |
2138 | an appropriate TarFile class. | |
2139 | ||
2140 | mode: | |
2141 | 'r' or 'r:*' open for reading with transparent compression | |
2142 | 'r:' open for reading exclusively uncompressed | |
2143 | 'r:gz' open for reading with gzip compression | |
2144 | 'r:bz2' open for reading with bzip2 compression | |
be60ffd0 | 2145 | 'r:xz' open for reading with lzma compression |
7584f5c9 ERE |
2146 | 'a' or 'a:' open for appending, creating the file if necessary |
2147 | 'w' or 'w:' open for writing without compression | |
2148 | 'w:gz' open for writing with gzip compression | |
2149 | 'w:bz2' open for writing with bzip2 compression | |
be60ffd0 | 2150 | 'w:xz' open for writing with lzma compression |
7584f5c9 ERE |
2151 | |
2152 | 'r|*' open a stream of tar blocks with transparent compression | |
2153 | 'r|' open an uncompressed stream of tar blocks for reading | |
2154 | 'r|gz' open a gzip compressed stream of tar blocks | |
2155 | 'r|bz2' open a bzip2 compressed stream of tar blocks | |
be60ffd0 | 2156 | 'r|xz' open an lzma compressed stream of tar blocks |
7584f5c9 ERE |
2157 | 'w|' open an uncompressed stream for writing |
2158 | 'w|gz' open a gzip compressed stream for writing | |
2159 | 'w|bz2' open a bzip2 compressed stream for writing | |
be60ffd0 | 2160 | 'w|xz' open an lzma compressed stream for writing |
85737f48 ERE |
2161 | |
2162 | 'r#gz' open a stream of gzip compressed tar blocks for reading | |
2163 | 'w#gz' open a stream of gzip compressed tar blocks for writing | |
7584f5c9 | 2164 | """ |
7584f5c9 ERE |
2165 | if not name and not fileobj: |
2166 | raise ValueError("nothing to open") | |
2167 | ||
2168 | if mode in ("r", "r:*"): | |
2169 | # Find out which *open() is appropriate for opening the file. | |
2170 | for comptype in cls.OPEN_METH: | |
2171 | func = getattr(cls, cls.OPEN_METH[comptype]) | |
2172 | if fileobj is not None: | |
2173 | saved_pos = fileobj.tell() | |
2174 | try: | |
2175 | return func(name, "r", fileobj, **kwargs) | |
be60ffd0 | 2176 | except (ReadError, CompressionError) as e: |
c7c736b6 | 2177 | # usually nothing exceptional but sometimes is |
7584f5c9 ERE |
2178 | if fileobj is not None: |
2179 | fileobj.seek(saved_pos) | |
2180 | continue | |
2181 | raise ReadError("file could not be opened successfully") | |
2182 | ||
2183 | elif ":" in mode: | |
2184 | filemode, comptype = mode.split(":", 1) | |
2185 | filemode = filemode or "r" | |
2186 | comptype = comptype or "tar" | |
2187 | ||
2188 | # Select the *open() function according to | |
2189 | # given compression. | |
2190 | if comptype in cls.OPEN_METH: | |
2191 | func = getattr(cls, cls.OPEN_METH[comptype]) | |
2192 | else: | |
2193 | raise CompressionError("unknown compression type %r" % comptype) | |
e05f0440 TJ |
2194 | |
2195 | # Pass on compression level for gzip / bzip2. | |
2196 | if comptype == 'gz' or comptype == 'bz2': | |
2197 | kwargs['compresslevel'] = compresslevel | |
2198 | ||
7a2b9329 CH |
2199 | if 'max_volume_size' in kwargs: |
2200 | if comptype != 'tar' and filemode in 'wa' \ | |
2201 | and kwargs['max_volume_size']: | |
2202 | import warnings | |
2203 | warnings.warn('Only the first volume will be compressed ' | |
2204 | 'for modes with "w:"!') | |
2205 | ||
e05f0440 | 2206 | return func(name, filemode, fileobj, **kwargs) |
7584f5c9 ERE |
2207 | |
2208 | elif "|" in mode: | |
2209 | filemode, comptype = mode.split("|", 1) | |
2210 | filemode = filemode or "r" | |
2211 | comptype = comptype or "tar" | |
2212 | ||
2213 | if filemode not in "rw": | |
2214 | raise ValueError("mode must be 'r' or 'w'") | |
2215 | ||
2216 | t = cls(name, filemode, | |
2b82f50c ERE |
2217 | _Stream(name, filemode, comptype, fileobj, bufsize, |
2218 | compresslevel=compresslevel), | |
7584f5c9 ERE |
2219 | **kwargs) |
2220 | t._extfileobj = False | |
2221 | return t | |
2222 | ||
5fdff89f ERE |
2223 | elif "#" in mode: |
2224 | filemode, comptype = mode.split("#", 1) | |
2225 | filemode = filemode or "r" | |
5fdff89f ERE |
2226 | |
2227 | if filemode not in "rw": | |
5faea0e1 PG |
2228 | raise ValueError ("mode %s not compatible with concat " |
2229 | "archive; must be 'r' or 'w'" % mode) | |
5fdff89f | 2230 | |
be60ffd0 | 2231 | stream = _Stream(name, filemode, comptype, fileobj, bufsize, |
d1c38f40 | 2232 | concat=True, encryption=encryption, |
04f4c7ab | 2233 | compresslevel=compresslevel, tolerance=tolerance) |
d1c38f40 | 2234 | kwargs ["concat"] = True |
be60ffd0 ERE |
2235 | try: |
2236 | t = cls(name, filemode, stream, **kwargs) | |
c7c736b6 | 2237 | except: # XXX except what? |
be60ffd0 | 2238 | stream.close() |
c7c736b6 | 2239 | raise # XXX raise what? |
5fdff89f ERE |
2240 | t._extfileobj = False |
2241 | return t | |
2242 | ||
7584f5c9 ERE |
2243 | elif mode in "aw": |
2244 | return cls.taropen(name, mode, fileobj, **kwargs) | |
2245 | ||
133d30da | 2246 | raise ValueError("undiscernible mode %r" % mode) |
7584f5c9 | 2247 | |
d39d4cbf PG |
2248 | |
2249 | @classmethod | |
2250 | def open_at_offset(cls, offset, *a, **kwa): | |
2251 | """ | |
2252 | Same as ``.open()``, but start reading at the given offset. Assumes a | |
5bd2d4b5 PG |
2253 | seekable file object. Returns *None* if opening failed due to a read |
2254 | problem. | |
d39d4cbf PG |
2255 | """ |
2256 | fileobj = kwa.get ("fileobj") | |
2257 | if fileobj is not None: | |
2258 | fileobj.seek (offset) | |
5bd2d4b5 | 2259 | |
d39d4cbf PG |
2260 | return cls.open (*a, **kwa) |
2261 | ||
2262 | ||
7584f5c9 ERE |
2263 | @classmethod |
2264 | def taropen(cls, name, mode="r", fileobj=None, **kwargs): | |
2265 | """Open uncompressed tar archive name for reading or writing. | |
2266 | """ | |
2267 | if len(mode) > 1 or mode not in "raw": | |
2268 | raise ValueError("mode must be 'r', 'a' or 'w'") | |
2269 | return cls(name, mode, fileobj, **kwargs) | |
2270 | ||
2271 | @classmethod | |
2272 | def gzopen(cls, name, mode="r", fileobj=None, compresslevel=9, **kwargs): | |
2273 | """Open gzip compressed tar archive name for reading or writing. | |
2274 | Appending is not allowed. | |
2275 | """ | |
2276 | if len(mode) > 1 or mode not in "rw": | |
2277 | raise ValueError("mode must be 'r' or 'w'") | |
2278 | ||
2279 | try: | |
2280 | import gzip | |
2281 | gzip.GzipFile | |
2282 | except (ImportError, AttributeError): | |
2283 | raise CompressionError("gzip module is not available") | |
2284 | ||
be60ffd0 | 2285 | extfileobj = fileobj is not None |
7584f5c9 | 2286 | try: |
be60ffd0 ERE |
2287 | fileobj = gzip.GzipFile(name, mode + "b", compresslevel, fileobj) |
2288 | t = cls.taropen(name, mode, fileobj, **kwargs) | |
2289 | except OSError: | |
2290 | if not extfileobj and fileobj is not None: | |
2291 | fileobj.close() | |
2292 | if fileobj is None: | |
2293 | raise | |
7584f5c9 | 2294 | raise ReadError("not a gzip file") |
be60ffd0 ERE |
2295 | except: |
2296 | if not extfileobj and fileobj is not None: | |
2297 | fileobj.close() | |
2298 | raise | |
2299 | t._extfileobj = extfileobj | |
7584f5c9 ERE |
2300 | return t |
2301 | ||
2302 | @classmethod | |
2303 | def bz2open(cls, name, mode="r", fileobj=None, compresslevel=9, **kwargs): | |
2304 | """Open bzip2 compressed tar archive name for reading or writing. | |
2305 | Appending is not allowed. | |
2306 | """ | |
2307 | if len(mode) > 1 or mode not in "rw": | |
2308 | raise ValueError("mode must be 'r' or 'w'.") | |
2309 | ||
2310 | try: | |
2311 | import bz2 | |
2312 | except ImportError: | |
2313 | raise CompressionError("bz2 module is not available") | |
2314 | ||
be60ffd0 ERE |
2315 | fileobj = bz2.BZ2File(fileobj or name, mode, |
2316 | compresslevel=compresslevel) | |
7584f5c9 ERE |
2317 | |
2318 | try: | |
2319 | t = cls.taropen(name, mode, fileobj, **kwargs) | |
be60ffd0 ERE |
2320 | except (OSError, EOFError): |
2321 | fileobj.close() | |
7584f5c9 ERE |
2322 | raise ReadError("not a bzip2 file") |
2323 | t._extfileobj = False | |
2324 | return t | |
2325 | ||
be60ffd0 ERE |
2326 | @classmethod |
2327 | def xzopen(cls, name, mode="r", fileobj=None, preset=None, **kwargs): | |
2328 | """Open lzma compressed tar archive name for reading or writing. | |
2329 | Appending is not allowed. | |
2330 | """ | |
2331 | if mode not in ("r", "w"): | |
2332 | raise ValueError("mode must be 'r' or 'w'") | |
2333 | ||
2334 | try: | |
2335 | import lzma | |
2336 | except ImportError: | |
2337 | raise CompressionError("lzma module is not available") | |
2338 | ||
2339 | fileobj = lzma.LZMAFile(fileobj or name, mode, preset=preset) | |
2340 | ||
2341 | try: | |
2342 | t = cls.taropen(name, mode, fileobj, **kwargs) | |
2343 | except (lzma.LZMAError, EOFError): | |
2344 | fileobj.close() | |
2345 | raise ReadError("not an lzma file") | |
2346 | t._extfileobj = False | |
2347 | return t | |
2348 | ||
7584f5c9 ERE |
2349 | # All *open() methods are registered here. |
2350 | OPEN_METH = { | |
2351 | "tar": "taropen", # uncompressed tar | |
2352 | "gz": "gzopen", # gzip compressed tar | |
be60ffd0 ERE |
2353 | "bz2": "bz2open", # bzip2 compressed tar |
2354 | "xz": "xzopen" # lzma compressed tar | |
7584f5c9 ERE |
2355 | } |
2356 | ||
2357 | #-------------------------------------------------------------------------- | |
2358 | # The public methods which TarFile provides: | |
2359 | ||
2360 | def close(self): | |
2361 | """Close the TarFile. In write-mode, two finishing zero blocks are | |
fd2f01f2 PG |
2362 | appended to the archive. A special case are empty archives which are |
2363 | initialized accordingly so the two mandatory blocks of zeros are | |
2364 | written abiding by the requested encryption and compression settings. | |
7584f5c9 ERE |
2365 | """ |
2366 | if self.closed: | |
2367 | return | |
2368 | ||
2369 | if self.mode in "aw": | |
fd2f01f2 PG |
2370 | if self.arcmode & ARCMODE_CONCAT and self.fileobj.tell () == 0: |
2371 | self.fileobj.next ("") | |
7584f5c9 ERE |
2372 | self.fileobj.write(NUL * (BLOCKSIZE * 2)) |
2373 | self.offset += (BLOCKSIZE * 2) | |
2374 | # fill up the end with zero-blocks | |
2375 | # (like option -b20 for tar does) | |
2376 | blocks, remainder = divmod(self.offset, RECORDSIZE) | |
2377 | if remainder > 0: | |
2378 | self.fileobj.write(NUL * (RECORDSIZE - remainder)) | |
7584f5c9 ERE |
2379 | if not self._extfileobj: |
2380 | self.fileobj.close() | |
2381 | self.closed = True | |
2382 | ||
2383 | def getmember(self, name): | |
2384 | """Return a TarInfo object for member `name'. If `name' can not be | |
2385 | found in the archive, KeyError is raised. If a member occurs more | |
2386 | than once in the archive, its last occurrence is assumed to be the | |
2387 | most up-to-date version. | |
2388 | """ | |
2389 | tarinfo = self._getmember(name) | |
2390 | if tarinfo is None: | |
2391 | raise KeyError("filename %r not found" % name) | |
2392 | return tarinfo | |
2393 | ||
2394 | def getmembers(self): | |
2395 | """Return the members of the archive as a list of TarInfo objects. The | |
2396 | list has the same order as the members in the archive. | |
2397 | """ | |
2398 | self._check() | |
2399 | if not self._loaded: # if we want to obtain a list of | |
2400 | self._load() # all members, we first have to | |
2401 | # scan the whole archive. | |
2402 | return self.members | |
2403 | ||
ad4402e8 ERE |
2404 | def get_last_member_offset(self): |
2405 | """Return the last member offset. Usually this is self.fileobj.tell(), | |
2406 | but when there's encryption or concat compression going on it's more | |
2407 | complicated than that. | |
2408 | """ | |
b8fc2f5d | 2409 | return self.last_block_offset |
ad4402e8 | 2410 | |
7584f5c9 ERE |
2411 | def getnames(self): |
2412 | """Return the members of the archive as a list of their names. It has | |
2413 | the same order as the list returned by getmembers(). | |
2414 | """ | |
2415 | return [tarinfo.name for tarinfo in self.getmembers()] | |
2416 | ||
2417 | def gettarinfo(self, name=None, arcname=None, fileobj=None): | |
2418 | """Create a TarInfo object for either the file `name' or the file | |
2419 | object `fileobj' (using os.fstat on its file descriptor). You can | |
2420 | modify some of the TarInfo's attributes before you add it using | |
2421 | addfile(). If given, `arcname' specifies an alternative name for the | |
2422 | file in the archive. | |
2423 | """ | |
2424 | self._check("aw") | |
2425 | ||
2426 | # When fileobj is given, replace name by | |
2427 | # fileobj's real name. | |
2428 | if fileobj is not None: | |
2429 | name = fileobj.name | |
2430 | ||
2431 | # Building the name of the member in the archive. | |
2432 | # Backward slashes are converted to forward slashes, | |
2433 | # Absolute paths are turned to relative paths. | |
2434 | if arcname is None: | |
2435 | arcname = name | |
2436 | drv, arcname = os.path.splitdrive(arcname) | |
be60ffd0 | 2437 | arcname = arcname.replace(os.sep, "/") |
7584f5c9 ERE |
2438 | arcname = arcname.lstrip("/") |
2439 | ||
2440 | # Now, fill the TarInfo object with | |
2441 | # information specific for the file. | |
2442 | tarinfo = self.tarinfo() | |
2443 | tarinfo.tarfile = self | |
2444 | ||
2445 | # Use os.stat or os.lstat, depending on platform | |
2446 | # and if symlinks shall be resolved. | |
2447 | if fileobj is None: | |
2448 | if hasattr(os, "lstat") and not self.dereference: | |
2449 | statres = os.lstat(name) | |
2450 | else: | |
2451 | statres = os.stat(name) | |
2452 | else: | |
2453 | statres = os.fstat(fileobj.fileno()) | |
2454 | linkname = "" | |
2455 | ||
2456 | stmd = statres.st_mode | |
2457 | if stat.S_ISREG(stmd): | |
2458 | inode = (statres.st_ino, statres.st_dev) | |
2459 | if not self.dereference and statres.st_nlink > 1 and \ | |
2460 | inode in self.inodes and arcname != self.inodes[inode]: | |
2461 | # Is it a hardlink to an already | |
2462 | # archived file? | |
2463 | type = LNKTYPE | |
2464 | linkname = self.inodes[inode] | |
2465 | else: | |
2466 | # The inode is added only if its valid. | |
2467 | # For win32 it is always 0. | |
2468 | type = REGTYPE | |
6f422b65 | 2469 | if inode[0] and self.save_to_members: |
7584f5c9 ERE |
2470 | self.inodes[inode] = arcname |
2471 | elif stat.S_ISDIR(stmd): | |
2472 | type = DIRTYPE | |
2473 | elif stat.S_ISFIFO(stmd): | |
2474 | type = FIFOTYPE | |
2475 | elif stat.S_ISLNK(stmd): | |
2476 | type = SYMTYPE | |
2477 | linkname = os.readlink(name) | |
2478 | elif stat.S_ISCHR(stmd): | |
2479 | type = CHRTYPE | |
2480 | elif stat.S_ISBLK(stmd): | |
2481 | type = BLKTYPE | |
2482 | else: | |
2483 | return None | |
2484 | ||
2485 | # Fill the TarInfo object with all | |
2486 | # information we can get. | |
2487 | tarinfo.name = arcname | |
2488 | tarinfo.mode = stmd | |
2489 | tarinfo.uid = statres.st_uid | |
2490 | tarinfo.gid = statres.st_gid | |
2491 | if type == REGTYPE: | |
2492 | tarinfo.size = statres.st_size | |
2493 | else: | |
be60ffd0 | 2494 | tarinfo.size = 0 |
7584f5c9 ERE |
2495 | tarinfo.mtime = statres.st_mtime |
2496 | tarinfo.type = type | |
2497 | tarinfo.linkname = linkname | |
2498 | if pwd: | |
9ef1fb87 TJ |
2499 | if tarinfo.uid in self.cache_uid2user: |
2500 | tarinfo.uname = self.cache_uid2user[tarinfo.uid] | |
2501 | else: | |
2502 | try: | |
2503 | tarinfo.uname = pwd.getpwuid(tarinfo.uid)[0] | |
2504 | self.cache_uid2user[tarinfo.uid] = tarinfo.uname | |
2505 | except KeyError: | |
2506 | # remember user does not exist: | |
2507 | # same default value as in tarinfo class | |
2508 | self.cache_uid2user[tarinfo.uid] = "" | |
7584f5c9 | 2509 | if grp: |
9ef1fb87 TJ |
2510 | if tarinfo.gid in self.cache_gid2group: |
2511 | tarinfo.gname = self.cache_gid2group[tarinfo.gid] | |
2512 | else: | |
2513 | try: | |
2514 | tarinfo.gname = grp.getgrgid(tarinfo.gid)[0] | |
2515 | self.cache_gid2group[tarinfo.gid] = tarinfo.gname | |
2516 | except KeyError: | |
2517 | # remember group does not exist: | |
2518 | # same default value as in tarinfo class | |
2519 | self.cache_gid2group[tarinfo.gid] = "" | |
7584f5c9 ERE |
2520 | |
2521 | if type in (CHRTYPE, BLKTYPE): | |
2522 | if hasattr(os, "major") and hasattr(os, "minor"): | |
2523 | tarinfo.devmajor = os.major(statres.st_rdev) | |
2524 | tarinfo.devminor = os.minor(statres.st_rdev) | |
2525 | return tarinfo | |
2526 | ||
2527 | def list(self, verbose=True): | |
2528 | """Print a table of contents to sys.stdout. If `verbose' is False, only | |
2529 | the names of the members are printed. If it is True, an `ls -l'-like | |
2530 | output is produced. | |
2531 | """ | |
2532 | self._check() | |
2533 | ||
2534 | for tarinfo in self: | |
2535 | if verbose: | |
be60ffd0 ERE |
2536 | print(stat.filemode(tarinfo.mode), end=' ') |
2537 | print("%s/%s" % (tarinfo.uname or tarinfo.uid, | |
2538 | tarinfo.gname or tarinfo.gid), end=' ') | |
7584f5c9 | 2539 | if tarinfo.ischr() or tarinfo.isblk(): |
be60ffd0 ERE |
2540 | print("%10s" % ("%d,%d" \ |
2541 | % (tarinfo.devmajor, tarinfo.devminor)), end=' ') | |
7584f5c9 | 2542 | else: |
be60ffd0 ERE |
2543 | print("%10d" % tarinfo.size, end=' ') |
2544 | print("%d-%02d-%02d %02d:%02d:%02d" \ | |
2545 | % time.localtime(tarinfo.mtime)[:6], end=' ') | |
7584f5c9 | 2546 | |
be60ffd0 | 2547 | print(tarinfo.name + ("/" if tarinfo.isdir() else ""), end=' ') |
7584f5c9 ERE |
2548 | |
2549 | if verbose: | |
2550 | if tarinfo.issym(): | |
be60ffd0 | 2551 | print("->", tarinfo.linkname, end=' ') |
7584f5c9 | 2552 | if tarinfo.islnk(): |
be60ffd0 ERE |
2553 | print("link to", tarinfo.linkname, end=' ') |
2554 | print() | |
7584f5c9 | 2555 | |
be60ffd0 | 2556 | def add(self, name, arcname=None, recursive=True, exclude=None, *, filter=None): |
7584f5c9 ERE |
2557 | """Add the file `name' to the archive. `name' may be any type of file |
2558 | (directory, fifo, symbolic link, etc.). If given, `arcname' | |
2559 | specifies an alternative name for the file in the archive. | |
2560 | Directories are added recursively by default. This can be avoided by | |
2561 | setting `recursive' to False. `exclude' is a function that should | |
2562 | return True for each filename to be excluded. `filter' is a function | |
2563 | that expects a TarInfo object argument and returns the changed | |
2564 | TarInfo object, if it returns None the TarInfo object will be | |
2565 | excluded from the archive. | |
2566 | """ | |
2567 | self._check("aw") | |
2568 | ||
2569 | if arcname is None: | |
2570 | arcname = name | |
2571 | ||
2572 | # Exclude pathnames. | |
2573 | if exclude is not None: | |
2574 | import warnings | |
2575 | warnings.warn("use the filter argument instead", | |
2576 | DeprecationWarning, 2) | |
2577 | if exclude(name): | |
2578 | self._dbg(2, "tarfile: Excluded %r" % name) | |
2579 | return | |
2580 | ||
2581 | # Skip if somebody tries to archive the archive... | |
2582 | if self.name is not None and os.path.abspath(name) == self.name: | |
2583 | self._dbg(2, "tarfile: Skipped %r" % name) | |
2584 | return | |
2585 | ||
2586 | self._dbg(1, name) | |
2587 | ||
2588 | # Create a TarInfo object from the file. | |
2589 | tarinfo = self.gettarinfo(name, arcname) | |
2590 | ||
2591 | if tarinfo is None: | |
2592 | self._dbg(1, "tarfile: Unsupported type %r" % name) | |
2593 | return | |
2594 | ||
2595 | # Change or exclude the TarInfo object. | |
2596 | if filter is not None: | |
2597 | tarinfo = filter(tarinfo) | |
2598 | if tarinfo is None: | |
2599 | self._dbg(2, "tarfile: Excluded %r" % name) | |
2600 | return | |
2601 | ||
2602 | # Append the tar header and data to the archive. | |
2603 | if tarinfo.isreg(): | |
2604 | with bltn_open(name, "rb") as f: | |
2605 | self.addfile(tarinfo, f) | |
2606 | ||
2607 | elif tarinfo.isdir(): | |
2608 | self.addfile(tarinfo) | |
2609 | if recursive: | |
2610 | for f in os.listdir(name): | |
2611 | self.add(os.path.join(name, f), os.path.join(arcname, f), | |
be60ffd0 | 2612 | recursive, exclude, filter=filter) |
7584f5c9 ERE |
2613 | |
2614 | else: | |
2615 | self.addfile(tarinfo) | |
2616 | ||
defc9a22 | 2617 | def _size_left_file(self): |
be60ffd0 | 2618 | """Calculates size left in a volume with a maximum volume size. |
ba5a449e | 2619 | |
be60ffd0 | 2620 | Assumes self.max_volume_size is set. |
ba5a449e | 2621 | If using compression through a _Stream, use _size_left_stream instead |
be60ffd0 | 2622 | """ |
ba5a449e | 2623 | # left-over size = max_size - offset - 2 zero-blocks written in close |
ae48acc8 ERE |
2624 | size_left = self.max_volume_size - 2*BLOCKSIZE - self.offset |
2625 | # limit size left to a discrete number of blocks, because we won't | |
be60ffd0 | 2626 | # write only half a block when writting the end of a volume |
ae48acc8 | 2627 | # and filling with zeros |
defc9a22 CH |
2628 | return BLOCKSIZE * (size_left // BLOCKSIZE) |
2629 | ||
2630 | def _size_left_stream(self): | |
ba5a449e CH |
2631 | """ Calculates size left in a volume if using comression/encryption |
2632 | ||
2633 | Assumes self.max_volume_size is set and self.fileobj is a _Stream | |
2634 | (otherwise use _size_left_file) | |
2635 | """ | |
2636 | # left-over size = max_size - bytes written - 2 zero-blocks (close) | |
defc9a22 CH |
2637 | size_left = self.max_volume_size - self.fileobj.estim_file_size() \ |
2638 | - 2*BLOCKSIZE | |
2639 | return BLOCKSIZE * (size_left // BLOCKSIZE) | |
ae48acc8 | 2640 | |
7584f5c9 ERE |
2641 | def addfile(self, tarinfo, fileobj=None): |
2642 | """Add the TarInfo object `tarinfo' to the archive. If `fileobj' is | |
2643 | given, tarinfo.size bytes are read from it and added to the archive. | |
2644 | You can create TarInfo objects using gettarinfo(). | |
2645 | On Windows platforms, `fileobj' should always be opened with mode | |
2646 | 'rb' to avoid irritation about the file size. | |
2647 | """ | |
2648 | self._check("aw") | |
2649 | ||
2650 | tarinfo = copy.copy(tarinfo) | |
cbf55ffb | 2651 | |
d1c38f40 PG |
2652 | if self.arcmode & ARCMODE_CONCAT: |
2653 | self.last_block_offset = self.fileobj.next (tarinfo.name) | |
11684b1d ERE |
2654 | else: |
2655 | self.last_block_offset = self.fileobj.tell() | |
7584f5c9 ERE |
2656 | |
2657 | buf = tarinfo.tobuf(self.format, self.encoding, self.errors) | |
2658 | self.fileobj.write(buf) | |
2659 | self.offset += len(buf) | |
2660 | ||
ae9c8de2 CH |
2661 | if self.max_volume_size: |
2662 | if isinstance(self.fileobj, _Stream): | |
2663 | _size_left = self._size_left_stream | |
2664 | else: | |
2665 | _size_left = self._size_left_file | |
2666 | else: | |
2667 | _size_left = lambda: tarinfo.size | |
68ddf955 | 2668 | |
29c354ac PG |
2669 | # If there's no data to follow, finish |
2670 | if not fileobj: | |
29c354ac PG |
2671 | if self.save_to_members: |
2672 | self.members.append(tarinfo) | |
2673 | return | |
2674 | ||
2675 | target_size_left = _size_left() | |
2676 | source_size_left = tarinfo.size | |
2677 | assert tarinfo.volume_offset == 0 | |
2678 | ||
2679 | # we only split volumes in the middle of a file, that means we have | |
2680 | # to write at least one block | |
2681 | if target_size_left < BLOCKSIZE: | |
2682 | target_size_left = BLOCKSIZE | |
2683 | ||
ae9c8de2 CH |
2684 | # loop over multiple volumes |
2685 | while source_size_left > 0: | |
ae48acc8 | 2686 | |
ae9c8de2 CH |
2687 | # Write as much data as possble from source into target. |
2688 | # When compressing data, we cannot easily predict how much data we | |
2689 | # can write until target_size_left == 0 --> need to iterate | |
2690 | size_can_write = min(target_size_left, source_size_left) | |
c04e0751 | 2691 | |
ae9c8de2 CH |
2692 | while size_can_write > 0: |
2693 | copyfileobj(fileobj, self.fileobj, size_can_write) | |
2694 | self.offset += size_can_write | |
2695 | source_size_left -= size_can_write | |
2696 | target_size_left = _size_left() | |
2697 | size_can_write = min(target_size_left, source_size_left) | |
68ddf955 | 2698 | |
ae9c8de2 CH |
2699 | # now target_size_left == 0 or source_size_left == 0 |
2700 | ||
2701 | # if there is data left to write, we need to create a new volume | |
2702 | if source_size_left > 0: | |
5f38bff6 PG |
2703 | # Only finalize the crypto entry here if we’re continuing with |
2704 | # another one; otherwise, the encryption must include the block | |
2705 | # padding below. | |
2f854e77 | 2706 | tarinfo.type = GNUTYPE_MULTIVOL |
68ddf955 ERE |
2707 | |
2708 | if not self.new_volume_handler or\ | |
2709 | not callable(self.new_volume_handler): | |
c04e0751 | 2710 | raise Exception("We need to create a new volume and you " |
ae9c8de2 | 2711 | "didn't supply a new_volume_handler") |
68ddf955 | 2712 | |
54128a00 | 2713 | |
68ddf955 ERE |
2714 | # the new volume handler should do everything needed to |
2715 | # start working in a new volume. usually, the handler calls | |
2716 | # to self.open_volume | |
2f854e77 | 2717 | self.volume_number += 1 |
0eb5048f | 2718 | |
ae9c8de2 | 2719 | # set to be used by open_volume, because in the case of a PAX |
0eb5048f ERE |
2720 | # tar it needs to write information about the volume and offset |
2721 | # in the global header | |
ae9c8de2 | 2722 | tarinfo.volume_offset = tarinfo.size - source_size_left |
0eb5048f | 2723 | self.volume_tarinfo = tarinfo |
ae9c8de2 | 2724 | |
a0873dcc PG |
2725 | # the “new_volume_handler” is supposed to call .close() on the |
2726 | # “fileobj” _Stream | |
2f854e77 ERE |
2727 | self.new_volume_handler(self, self.base_name, self.volume_number) |
2728 | ||
0eb5048f ERE |
2729 | self.volume_tarinfo = None |
2730 | ||
d1c38f40 PG |
2731 | if self.arcmode & ARCMODE_CONCAT: |
2732 | self.fileobj.next_volume (tarinfo.name) | |
5f38bff6 | 2733 | |
2f854e77 ERE |
2734 | # write new volume header |
2735 | buf = tarinfo.tobuf(self.format, self.encoding, self.errors) | |
2f854e77 | 2736 | self.fileobj.write(buf) |
ae9c8de2 CH |
2737 | self.offset += len(buf) |
2738 | ||
2739 | # adjust variables; open_volume should have reset self.offset | |
2740 | # --> _size_left should be big again | |
2741 | target_size_left = _size_left() | |
2742 | size_can_write = min(target_size_left, source_size_left) | |
e0da4709 | 2743 | self._dbg(3, 'new volume') |
ae9c8de2 CH |
2744 | |
2745 | # now, all data has been written. We may have to fill up the rest of | |
2746 | # the block in target with 0s | |
2747 | remainder = (tarinfo.size - tarinfo.volume_offset) % BLOCKSIZE | |
2748 | if remainder > 0: | |
2749 | self.fileobj.write(NUL * (BLOCKSIZE - remainder)) | |
2750 | self.offset += BLOCKSIZE - remainder | |
7584f5c9 | 2751 | |
ea625b04 ERE |
2752 | if self.save_to_members: |
2753 | self.members.append(tarinfo) | |
7584f5c9 | 2754 | |
170c6c52 | 2755 | def open_volume(self, name="", fileobj=None, encryption=None): |
68ddf955 | 2756 | ''' |
0eb5048f | 2757 | Called by the user to change this tar file to point to a new volume. |
68ddf955 | 2758 | ''' |
27ee4dd4 | 2759 | |
68ddf955 ERE |
2760 | # open the file using either fileobj or name |
2761 | if not fileobj: | |
2762 | if self.mode == "a" and not os.path.exists(name): | |
2763 | # Create nonexistent files in append mode. | |
2764 | self.mode = "w" | |
2765 | self._mode = "wb" | |
68ddf955 | 2766 | self._extfileobj = False |
26fa5ad5 ERE |
2767 | |
2768 | if isinstance(self.fileobj, _Stream): | |
e0da4709 | 2769 | self._dbg(3, 'open_volume: create a _Stream') |
26fa5ad5 ERE |
2770 | fileobj = _Stream(name=name, |
2771 | mode=self.fileobj.mode, | |
2772 | comptype=self.fileobj.comptype, | |
2773 | fileobj=None, | |
2774 | bufsize=self.fileobj.bufsize, | |
cea130ec | 2775 | encryption=encryption or self.fileobj.encryption, |
27ee4dd4 PG |
2776 | concat=self.fileobj.arcmode & ARCMODE_CONCAT, |
2777 | tolerance=self.fileobj.tolerance) | |
26fa5ad5 | 2778 | else: |
7a2b9329 | 2779 | # here, we lose information about compression/encryption! |
e0da4709 | 2780 | self._dbg(3, 'open_volume: builtin open') |
26fa5ad5 | 2781 | fileobj = bltn_open(name, self._mode) |
68ddf955 ERE |
2782 | else: |
2783 | if name is None and hasattr(fileobj, "name"): | |
2784 | name = fileobj.name | |
2785 | if hasattr(fileobj, "mode"): | |
2786 | self._mode = fileobj.mode | |
2787 | self._extfileobj = True | |
1027433a | 2788 | self._dbg(3, 'open_volume: using external fileobj {}', fileobj) |
68ddf955 ERE |
2789 | self.name = os.path.abspath(name) if name else None |
2790 | self.fileobj = fileobj | |
2791 | ||
2792 | # init data structures | |
2793 | self.closed = False | |
2794 | self.members = [] # list of members as TarInfo objects | |
2795 | self._loaded = False # flag if all members have been read | |
2796 | self.offset = self.fileobj.tell() | |
2797 | # current position in the archive file | |
2798 | self.inodes = {} # dictionary caching the inodes of | |
2799 | # archive members already added | |
2800 | ||
2801 | try: | |
2802 | if self.mode == "r": | |
2803 | self.firstmember = None | |
2804 | self.firstmember = self.next() | |
2805 | ||
2806 | if self.mode == "a": | |
2807 | # Move to the end of the archive, | |
2808 | # before the first empty block. | |
2809 | while True: | |
2810 | self.fileobj.seek(self.offset) | |
2811 | try: | |
2812 | tarinfo = self.tarinfo.fromtarfile(self) | |
2813 | self.members.append(tarinfo) | |
2814 | except EOFHeaderError: | |
2815 | self.fileobj.seek(self.offset) | |
2816 | break | |
be60ffd0 | 2817 | except HeaderError as e: |
68ddf955 ERE |
2818 | raise ReadError(str(e)) |
2819 | ||
2820 | if self.mode in "aw": | |
2821 | self._loaded = True | |
2822 | ||
c04e0751 ERE |
2823 | if self.format == PAX_FORMAT: |
2824 | volume_info = { | |
be60ffd0 ERE |
2825 | "GNU.volume.filename": str(self.volume_tarinfo.name), |
2826 | "GNU.volume.size": str(self.volume_tarinfo.size - self.volume_tarinfo.volume_offset), | |
2827 | "GNU.volume.offset": str(self.volume_tarinfo.volume_offset), | |
c04e0751 | 2828 | } |
0eb5048f | 2829 | |
c04e0751 ERE |
2830 | self.pax_headers.update(volume_info) |
2831 | ||
a0873dcc PG |
2832 | if isinstance(self.fileobj, _Stream): |
2833 | self.fileobj._init_write_gz () | |
c04e0751 ERE |
2834 | buf = self.tarinfo.create_pax_global_header(volume_info.copy()) |
2835 | self.fileobj.write(buf) | |
2836 | self.offset += len(buf) | |
54128a00 | 2837 | except Exception as exn: |
68ddf955 ERE |
2838 | if not self._extfileobj: |
2839 | self.fileobj.close() | |
2840 | self.closed = True | |
2841 | raise | |
2842 | ||
e5f5681b | 2843 | def extractall(self, path=".", members=None, filter=None): |
7584f5c9 ERE |
2844 | """Extract all members from the archive to the current working |
2845 | directory and set owner, modification time and permissions on | |
2846 | directories afterwards. `path' specifies a different directory | |
2847 | to extract to. `members' is optional and must be a subset of the | |
2848 | list returned by getmembers(). | |
2849 | """ | |
2850 | directories = [] | |
2851 | ||
2852 | if members is None: | |
2853 | members = self | |
2854 | ||
2855 | for tarinfo in members: | |
c474439c ERE |
2856 | if self.volume_number > 0 and tarinfo.ismultivol(): |
2857 | continue | |
2858 | ||
974408b5 | 2859 | if filter and not filter(tarinfo): |
e5f5681b ERE |
2860 | continue |
2861 | ||
7584f5c9 ERE |
2862 | if tarinfo.isdir(): |
2863 | # Extract directories with a safe mode. | |
2864 | directories.append(tarinfo) | |
2865 | tarinfo = copy.copy(tarinfo) | |
be60ffd0 ERE |
2866 | tarinfo.mode = 0o0700 |
2867 | # Do not set_attrs directories, as we will do that further down | |
2868 | self.extract(tarinfo, path, set_attrs=not tarinfo.isdir()) | |
7584f5c9 ERE |
2869 | |
2870 | # Reverse sort directories. | |
be60ffd0 | 2871 | directories.sort(key=lambda a: a.name) |
7584f5c9 ERE |
2872 | directories.reverse() |
2873 | ||
2874 | # Set correct owner, mtime and filemode on directories. | |
2875 | for tarinfo in directories: | |
2876 | dirpath = os.path.join(path, tarinfo.name) | |
2877 | try: | |
2878 | self.chown(tarinfo, dirpath) | |
2879 | self.utime(tarinfo, dirpath) | |
2880 | self.chmod(tarinfo, dirpath) | |
be60ffd0 | 2881 | except ExtractError as e: |
7584f5c9 ERE |
2882 | if self.errorlevel > 1: |
2883 | raise | |
2884 | else: | |
2885 | self._dbg(1, "tarfile: %s" % e) | |
2886 | ||
786addd6 | 2887 | def extract(self, member, path="", set_attrs=True, symlink_cb=None): |
7584f5c9 ERE |
2888 | """Extract a member from the archive to the current working directory, |
2889 | using its full name. Its file information is extracted as accurately | |
2890 | as possible. `member' may be a filename or a TarInfo object. You can | |
be60ffd0 ERE |
2891 | specify a different directory using `path'. File attributes (owner, |
2892 | mtime, mode) are set unless `set_attrs' is False. | |
786addd6 PG |
2893 | ``symlink_cb`` is a hook accepting a function that is passed the |
2894 | ``member``, ``path``, and ``set_attrs`` arguments if the tarinfo for | |
2895 | ``member`` indicates a symlink in which case only the callback | |
9b13f5c4 PG |
2896 | passed will be applied, skipping the actual extraction. In case the |
2897 | callback is invoked, its return value is passed on to the caller. | |
7584f5c9 ERE |
2898 | """ |
2899 | self._check("r") | |
2900 | ||
be60ffd0 | 2901 | if isinstance(member, str): |
7584f5c9 ERE |
2902 | tarinfo = self.getmember(member) |
2903 | else: | |
2904 | tarinfo = member | |
2905 | ||
2906 | # Prepare the link target for makelink(). | |
2907 | if tarinfo.islnk(): | |
2908 | tarinfo._link_target = os.path.join(path, tarinfo.linkname) | |
2909 | ||
9b13f5c4 | 2910 | if symlink_cb is not None and tarinfo.issym(): |
83f5fd71 | 2911 | return symlink_cb(member, path, set_attrs) |
786addd6 | 2912 | |
7584f5c9 | 2913 | try: |
be60ffd0 ERE |
2914 | self._extract_member(tarinfo, os.path.join(path, tarinfo.name), |
2915 | set_attrs=set_attrs) | |
2916 | except EnvironmentError as e: | |
7584f5c9 ERE |
2917 | if self.errorlevel > 0: |
2918 | raise | |
2919 | else: | |
2920 | if e.filename is None: | |
2921 | self._dbg(1, "tarfile: %s" % e.strerror) | |
2922 | else: | |
2923 | self._dbg(1, "tarfile: %s %r" % (e.strerror, e.filename)) | |
be60ffd0 | 2924 | except ExtractError as e: |
7584f5c9 ERE |
2925 | if self.errorlevel > 1: |
2926 | raise | |
2927 | else: | |
2928 | self._dbg(1, "tarfile: %s" % e) | |
2929 | ||
2930 | def extractfile(self, member): | |
2931 | """Extract a member from the archive as a file object. `member' may be | |
be60ffd0 ERE |
2932 | a filename or a TarInfo object. If `member' is a regular file or a |
2933 | link, an io.BufferedReader object is returned. Otherwise, None is | |
2934 | returned. | |
7584f5c9 ERE |
2935 | """ |
2936 | self._check("r") | |
2937 | ||
be60ffd0 | 2938 | if isinstance(member, str): |
7584f5c9 ERE |
2939 | tarinfo = self.getmember(member) |
2940 | else: | |
2941 | tarinfo = member | |
2942 | ||
be60ffd0 ERE |
2943 | if tarinfo.isreg() or tarinfo.ismultivol() or\ |
2944 | tarinfo.type not in SUPPORTED_TYPES: | |
7584f5c9 ERE |
2945 | # If a member's type is unknown, it is treated as a |
2946 | # regular file. | |
2947 | return self.fileobject(self, tarinfo) | |
2948 | ||
2949 | elif tarinfo.islnk() or tarinfo.issym(): | |
2950 | if isinstance(self.fileobj, _Stream): | |
2951 | # A small but ugly workaround for the case that someone tries | |
2952 | # to extract a (sym)link as a file-object from a non-seekable | |
2953 | # stream of tar blocks. | |
2954 | raise StreamError("cannot extract (sym)link as file object") | |
2955 | else: | |
2956 | # A (sym)link's file object is its target's file object. | |
2957 | return self.extractfile(self._find_link_target(tarinfo)) | |
2958 | else: | |
2959 | # If there's no data associated with the member (directory, chrdev, | |
2960 | # blkdev, etc.), return None instead of a file object. | |
2961 | return None | |
2962 | ||
be60ffd0 | 2963 | def _extract_member(self, tarinfo, targetpath, set_attrs=True): |
7584f5c9 ERE |
2964 | """Extract the TarInfo object tarinfo to a physical |
2965 | file called targetpath. | |
2966 | """ | |
2967 | # Fetch the TarInfo object for the given name | |
2968 | # and build the destination pathname, replacing | |
2969 | # forward slashes to platform specific separators. | |
2970 | targetpath = targetpath.rstrip("/") | |
2971 | targetpath = targetpath.replace("/", os.sep) | |
2972 | ||
2973 | # Create all upper directories. | |
2974 | upperdirs = os.path.dirname(targetpath) | |
2975 | if upperdirs and not os.path.exists(upperdirs): | |
2976 | # Create directories that are not part of the archive with | |
2977 | # default permissions. | |
2978 | os.makedirs(upperdirs) | |
2979 | ||
2980 | if tarinfo.islnk() or tarinfo.issym(): | |
2981 | self._dbg(1, "%s -> %s" % (tarinfo.name, tarinfo.linkname)) | |
2982 | else: | |
2983 | self._dbg(1, tarinfo.name) | |
2984 | ||
2985 | if tarinfo.isreg(): | |
2986 | self.makefile(tarinfo, targetpath) | |
2987 | elif tarinfo.isdir(): | |
2988 | self.makedir(tarinfo, targetpath) | |
2989 | elif tarinfo.isfifo(): | |
2990 | self.makefifo(tarinfo, targetpath) | |
2991 | elif tarinfo.ischr() or tarinfo.isblk(): | |
2992 | self.makedev(tarinfo, targetpath) | |
2993 | elif tarinfo.islnk() or tarinfo.issym(): | |
2994 | self.makelink(tarinfo, targetpath) | |
2995 | elif tarinfo.type not in SUPPORTED_TYPES: | |
2996 | self.makeunknown(tarinfo, targetpath) | |
2997 | else: | |
2998 | self.makefile(tarinfo, targetpath) | |
2999 | ||
be60ffd0 ERE |
3000 | if set_attrs: |
3001 | self.chown(tarinfo, targetpath) | |
3002 | if not tarinfo.issym(): | |
3003 | self.chmod(tarinfo, targetpath) | |
3004 | self.utime(tarinfo, targetpath) | |
7584f5c9 ERE |
3005 | |
3006 | #-------------------------------------------------------------------------- | |
3007 | # Below are the different file methods. They are called via | |
3008 | # _extract_member() when extract() is called. They can be replaced in a | |
3009 | # subclass to implement other functionality. | |
3010 | ||
3011 | def makedir(self, tarinfo, targetpath): | |
3012 | """Make a directory called targetpath. | |
3013 | """ | |
3014 | try: | |
3015 | # Use a safe mode for the directory, the real mode is set | |
3016 | # later in _extract_member(). | |
be60ffd0 ERE |
3017 | os.mkdir(targetpath, 0o0700) |
3018 | except FileExistsError: | |
3019 | pass | |
7584f5c9 ERE |
3020 | |
3021 | def makefile(self, tarinfo, targetpath): | |
3022 | """Make a file called targetpath. | |
3023 | """ | |
be60ffd0 ERE |
3024 | source = self.fileobj |
3025 | source.seek(tarinfo.offset_data) | |
c7c736b6 | 3026 | decrypt = False |
c474439c ERE |
3027 | iterate = True |
3028 | target = bltn_open(targetpath, "wb") | |
3029 | ||
be60ffd0 ERE |
3030 | if tarinfo.sparse is not None: |
3031 | try: | |
3032 | for offset, size in tarinfo.sparse: | |
3033 | target.seek(offset) | |
3034 | copyfileobj(source, target, size) | |
3035 | target.seek(tarinfo.size) | |
3036 | target.truncate() | |
3037 | finally: | |
3038 | target.close() | |
3039 | return | |
3040 | ||
c474439c ERE |
3041 | while iterate: |
3042 | iterate = False | |
3043 | try: | |
3044 | copyfileobj(source, target, tarinfo.size) | |
aa828cd1 | 3045 | except OSError: |
c474439c ERE |
3046 | source.close() |
3047 | # only if we are extracting a multivolume this can be treated | |
3048 | if not self.new_volume_handler: | |
3049 | target.close() | |
3050 | raise Exception("We need to read a new volume and you" | |
3051 | " didn't supply a new_volume_handler") | |
3052 | ||
3053 | # the new volume handler should do everything needed to | |
3054 | # start working in a new volume. usually, the handler calls | |
3055 | # to self.open_volume | |
3056 | self.volume_number += 1 | |
3057 | self.new_volume_handler(self, self.base_name, self.volume_number) | |
be60ffd0 ERE |
3058 | tarinfo = self.firstmember |
3059 | source = self.fileobj | |
c474439c | 3060 | iterate = True |
c474439c ERE |
3061 | target.close() |
3062 | ||
7584f5c9 ERE |
3063 | |
3064 | def makeunknown(self, tarinfo, targetpath): | |
3065 | """Make a file from a TarInfo object with an unknown type | |
3066 | at targetpath. | |
3067 | """ | |
3068 | self.makefile(tarinfo, targetpath) | |
3069 | self._dbg(1, "tarfile: Unknown file type %r, " \ | |
3070 | "extracted as regular file." % tarinfo.type) | |
3071 | ||
3072 | def makefifo(self, tarinfo, targetpath): | |
3073 | """Make a fifo called targetpath. | |
3074 | """ | |
3075 | if hasattr(os, "mkfifo"): | |
3076 | os.mkfifo(targetpath) | |
3077 | else: | |
3078 | raise ExtractError("fifo not supported by system") | |
3079 | ||
3080 | def makedev(self, tarinfo, targetpath): | |
3081 | """Make a character or block device called targetpath. | |
3082 | """ | |
3083 | if not hasattr(os, "mknod") or not hasattr(os, "makedev"): | |
3084 | raise ExtractError("special devices not supported by system") | |
3085 | ||
3086 | mode = tarinfo.mode | |
3087 | if tarinfo.isblk(): | |
3088 | mode |= stat.S_IFBLK | |
3089 | else: | |
3090 | mode |= stat.S_IFCHR | |
3091 | ||
3092 | os.mknod(targetpath, mode, | |
3093 | os.makedev(tarinfo.devmajor, tarinfo.devminor)) | |
3094 | ||
3095 | def makelink(self, tarinfo, targetpath): | |
3096 | """Make a (symbolic) link called targetpath. If it cannot be created | |
3097 | (platform limitation), we try to make a copy of the referenced file | |
3098 | instead of a link. | |
3099 | """ | |
be60ffd0 | 3100 | try: |
7584f5c9 ERE |
3101 | # For systems that support symbolic and hard links. |
3102 | if tarinfo.issym(): | |
7584f5c9 ERE |
3103 | os.symlink(tarinfo.linkname, targetpath) |
3104 | else: | |
3105 | # See extract(). | |
3106 | if os.path.exists(tarinfo._link_target): | |
7584f5c9 ERE |
3107 | os.link(tarinfo._link_target, targetpath) |
3108 | else: | |
be60ffd0 ERE |
3109 | self._extract_member(self._find_link_target(tarinfo), |
3110 | targetpath) | |
3111 | except symlink_exception: | |
7584f5c9 | 3112 | try: |
be60ffd0 ERE |
3113 | self._extract_member(self._find_link_target(tarinfo), |
3114 | targetpath) | |
7584f5c9 ERE |
3115 | except KeyError: |
3116 | raise ExtractError("unable to resolve link inside archive") | |
3117 | ||
3118 | def chown(self, tarinfo, targetpath): | |
3119 | """Set owner of targetpath according to tarinfo. | |
3120 | """ | |
3121 | if pwd and hasattr(os, "geteuid") and os.geteuid() == 0: | |
3122 | # We have to be root to do so. | |
3123 | try: | |
3124 | g = grp.getgrnam(tarinfo.gname)[2] | |
3125 | except KeyError: | |
3126 | g = tarinfo.gid | |
3127 | try: | |
3128 | u = pwd.getpwnam(tarinfo.uname)[2] | |
3129 | except KeyError: | |
3130 | u = tarinfo.uid | |
3131 | try: | |
3132 | if tarinfo.issym() and hasattr(os, "lchown"): | |
3133 | os.lchown(targetpath, u, g) | |
3134 | else: | |
be60ffd0 ERE |
3135 | os.chown(targetpath, u, g) |
3136 | except OSError as e: | |
7584f5c9 ERE |
3137 | raise ExtractError("could not change owner") |
3138 | ||
3139 | def chmod(self, tarinfo, targetpath): | |
3140 | """Set file permissions of targetpath according to tarinfo. | |
3141 | """ | |
3142 | if hasattr(os, 'chmod'): | |
3143 | try: | |
3144 | os.chmod(targetpath, tarinfo.mode) | |
be60ffd0 | 3145 | except OSError as e: |
7584f5c9 ERE |
3146 | raise ExtractError("could not change mode") |
3147 | ||
3148 | def utime(self, tarinfo, targetpath): | |
3149 | """Set modification time of targetpath according to tarinfo. | |
3150 | """ | |
3151 | if not hasattr(os, 'utime'): | |
3152 | return | |
3153 | try: | |
3154 | os.utime(targetpath, (tarinfo.mtime, tarinfo.mtime)) | |
be60ffd0 | 3155 | except OSError as e: |
7584f5c9 ERE |
3156 | raise ExtractError("could not change modification time") |
3157 | ||
3158 | #-------------------------------------------------------------------------- | |
3159 | def next(self): | |
3160 | """Return the next member of the archive as a TarInfo object, when | |
3161 | TarFile is opened for reading. Return None if there is no more | |
3162 | available. | |
3163 | """ | |
3164 | self._check("ra") | |
3165 | if self.firstmember is not None: | |
3166 | m = self.firstmember | |
3167 | self.firstmember = None | |
3168 | return m | |
3169 | ||
be60ffd0 ERE |
3170 | # Read the next block. |
3171 | self.fileobj.seek(self.offset) | |
7584f5c9 ERE |
3172 | tarinfo = None |
3173 | while True: | |
3174 | try: | |
3175 | tarinfo = self.tarinfo.fromtarfile(self) | |
be60ffd0 | 3176 | except EOFHeaderError as e: |
7584f5c9 ERE |
3177 | if self.ignore_zeros: |
3178 | self._dbg(2, "0x%X: %s" % (self.offset, e)) | |
3179 | self.offset += BLOCKSIZE | |
3180 | continue | |
be60ffd0 | 3181 | except InvalidHeaderError as e: |
7584f5c9 ERE |
3182 | if self.ignore_zeros: |
3183 | self._dbg(2, "0x%X: %s" % (self.offset, e)) | |
3184 | self.offset += BLOCKSIZE | |
3185 | continue | |
3186 | elif self.offset == 0: | |
3187 | raise ReadError(str(e)) | |
3188 | except EmptyHeaderError: | |
3189 | if self.offset == 0: | |
3190 | raise ReadError("empty file") | |
be60ffd0 | 3191 | except TruncatedHeaderError as e: |
7584f5c9 ERE |
3192 | if self.offset == 0: |
3193 | raise ReadError(str(e)) | |
be60ffd0 | 3194 | except SubsequentHeaderError as e: |
7584f5c9 ERE |
3195 | raise ReadError(str(e)) |
3196 | break | |
3197 | ||
3198 | if tarinfo is not None: | |
ea625b04 ERE |
3199 | if self.save_to_members: |
3200 | self.members.append(tarinfo) | |
7584f5c9 ERE |
3201 | else: |
3202 | self._loaded = True | |
3203 | ||
3204 | return tarinfo | |
3205 | ||
3206 | #-------------------------------------------------------------------------- | |
3207 | # Little helper methods: | |
3208 | ||
3209 | def _getmember(self, name, tarinfo=None, normalize=False): | |
3210 | """Find an archive member by name from bottom to top. | |
3211 | If tarinfo is given, it is used as the starting point. | |
3212 | """ | |
3213 | # Ensure that all members have been loaded. | |
3214 | members = self.getmembers() | |
3215 | ||
3216 | # Limit the member search list up to tarinfo. | |
3217 | if tarinfo is not None: | |
3218 | members = members[:members.index(tarinfo)] | |
3219 | ||
3220 | if normalize: | |
3221 | name = os.path.normpath(name) | |
3222 | ||
3223 | for member in reversed(members): | |
3224 | if normalize: | |
3225 | member_name = os.path.normpath(member.name) | |
3226 | else: | |
3227 | member_name = member.name | |
3228 | ||
3229 | if name == member_name: | |
3230 | return member | |
3231 | ||
3232 | def _load(self): | |
3233 | """Read through the entire archive file and look for readable | |
3234 | members. | |
3235 | """ | |
3236 | while True: | |
3237 | tarinfo = self.next() | |
3238 | if tarinfo is None: | |
3239 | break | |
3240 | self._loaded = True | |
3241 | ||
3242 | def _check(self, mode=None): | |
3243 | """Check if TarFile is still open, and if the operation's mode | |
3244 | corresponds to TarFile's mode. | |
3245 | """ | |
3246 | if self.closed: | |
be60ffd0 | 3247 | raise OSError("%s is closed" % self.__class__.__name__) |
7584f5c9 | 3248 | if mode is not None and self.mode not in mode: |
be60ffd0 | 3249 | raise OSError("bad operation for mode %r" % self.mode) |
7584f5c9 ERE |
3250 | |
3251 | def _find_link_target(self, tarinfo): | |
3252 | """Find the target member of a symlink or hardlink member in the | |
3253 | archive. | |
3254 | """ | |
3255 | if tarinfo.issym(): | |
3256 | # Always search the entire archive. | |
3257 | linkname = "/".join(filter(None, (os.path.dirname(tarinfo.name), tarinfo.linkname))) | |
3258 | limit = None | |
3259 | else: | |
3260 | # Search the archive before the link, because a hard link is | |
3261 | # just a reference to an already archived file. | |
3262 | linkname = tarinfo.linkname | |
3263 | limit = tarinfo | |
3264 | ||
3265 | member = self._getmember(linkname, tarinfo=limit, normalize=True) | |
3266 | if member is None: | |
3267 | raise KeyError("linkname %r not found" % linkname) | |
3268 | return member | |
3269 | ||
3270 | def __iter__(self): | |
3271 | """Provide an iterator object. | |
3272 | """ | |
3273 | if self._loaded: | |
3274 | return iter(self.members) | |
3275 | else: | |
3276 | return TarIter(self) | |
3277 | ||
1027433a | 3278 | def _dbg(self, level, msg, *args): |
7584f5c9 ERE |
3279 | """Write debugging output to sys.stderr. |
3280 | """ | |
3281 | if level <= self.debug: | |
1027433a | 3282 | print(msg.format(*args), file=sys.stderr) |
7584f5c9 ERE |
3283 | |
3284 | def __enter__(self): | |
3285 | self._check() | |
3286 | return self | |
3287 | ||
3288 | def __exit__(self, type, value, traceback): | |
3289 | if type is None: | |
3290 | self.close() | |
3291 | else: | |
3292 | # An exception occurred. We must not call close() because | |
3293 | # it would try to write end-of-archive blocks and padding. | |
3294 | if not self._extfileobj: | |
3295 | self.fileobj.close() | |
3296 | self.closed = True | |
3297 | # class TarFile | |
3298 | ||
3299 | class TarIter: | |
3300 | """Iterator Class. | |
3301 | ||
3302 | for tarinfo in TarFile(...): | |
3303 | suite... | |
3304 | """ | |
3305 | ||
3306 | def __init__(self, tarfile): | |
3307 | """Construct a TarIter object. | |
3308 | """ | |
3309 | self.tarfile = tarfile | |
3310 | self.index = 0 | |
3311 | def __iter__(self): | |
3312 | """Return iterator object. | |
3313 | """ | |
3314 | return self | |
be60ffd0 | 3315 | def __next__(self): |
7584f5c9 ERE |
3316 | """Return the next item using TarFile's next() method. |
3317 | When all members have been read, set TarFile as _loaded. | |
3318 | """ | |
3319 | # Fix for SF #1100429: Under rare circumstances it can | |
3320 | # happen that getmembers() is called during iteration, | |
3321 | # which will cause TarIter to stop prematurely. | |
3322 | ||
3323 | if self.index == 0 and self.tarfile.firstmember is not None: | |
3324 | tarinfo = self.tarfile.next() | |
3325 | elif self.index < len(self.tarfile.members): | |
3326 | tarinfo = self.tarfile.members[self.index] | |
3327 | elif not self.tarfile._loaded: | |
3328 | tarinfo = self.tarfile.next() | |
3329 | if not tarinfo: | |
3330 | self.tarfile._loaded = True | |
3331 | raise StopIteration | |
3332 | else: | |
3333 | raise StopIteration | |
3334 | self.index += 1 | |
fb27c6e8 | 3335 | |
7584f5c9 ERE |
3336 | return tarinfo |
3337 | ||
6690f5e0 PG |
3338 | #--------------------------------------------------------- |
3339 | # support functionality for rescue mode | |
3340 | #--------------------------------------------------------- | |
3341 | ||
8fc6040c PG |
3342 | TAR_FMT_HDR = (# See tar(5): |
3343 | "<" | |
3344 | "100s" # ← char name[100]; /* 100 */ | |
3345 | "8s" # ← char mode[8]; /* 108 */ | |
3346 | "8s" # ← char uid[8]; /* 116 */ | |
3347 | "8s" # ← char gid[8]; /* 124 */ | |
3348 | "12s" # ← char size[12]; /* 136 */ | |
3349 | "12s" # ← char mtime[12]; /* 148 */ | |
3350 | "8s" # ← char checksum[8]; /* 156 */ | |
3351 | "B" # ← char typeflag[1]; /* 157 */ | |
3352 | "100s" # ← char linkname[100]; /* 257 */ | |
3353 | "6s" # ← char magic[6]; /* 263 */ | |
3354 | "2s" # ← char version[2]; /* 265 */ | |
3355 | "32s" # ← char uname[32]; /* 297 */ | |
3356 | "32s" # ← char gname[32]; /* 329 */ | |
3357 | "8s" # ← char devmajor[8]; /* 337 */ | |
3358 | "8s" # ← char devminor[8]; /* 345 */ | |
3359 | "12s" # ← char atime[12]; /* 357 */ | |
3360 | "12s" # ← char ctime[12]; /* 369 */ | |
3361 | "12s" # ← char offset[12]; /* 381 */ | |
3362 | "4s" # ← char longnames[4]; /* 385 */ | |
3363 | "B" # ← char unused[1]; /* 386 */ | |
3364 | "" # struct { | |
3365 | "12s" # ← char offset[12]; | |
3366 | "12s" # ← char numbytes[12]; | |
3367 | "12s" # ← char offset[12]; | |
3368 | "12s" # ← char numbytes[12]; | |
3369 | "12s" # ← char offset[12]; | |
3370 | "12s" # ← char numbytes[12]; | |
3371 | "12s" # ← char offset[12]; | |
3372 | "12s" # ← char numbytes[12]; | |
3373 | "" # } sparse[4]; /* 482 */ | |
3374 | "B" # ← char isextended[1]; /* 483 */ | |
3375 | "12s" # ← char realsize[12]; /* 495 */ | |
3376 | "17s" # ← char pad[17]; /* 512 */ | |
3377 | ) | |
3378 | ||
3379 | # The “magic” and “version” fields are special: | |
3380 | # | |
3381 | # tar(5) | |
3382 | # magic The magic field holds the five characters “ustar” followed by a | |
3383 | # space. Note that POSIX ustar archives have a trailing null. | |
3384 | # | |
3385 | # however, “tar.h”: | |
3386 | # | |
3387 | # /* OLDGNU_MAGIC uses both magic and version fields, which are contiguous. | |
3388 | # Found in an archive, it indicates an old GNU header format, which will be | |
3389 | # hopefully become obsolescent. With OLDGNU_MAGIC, uname and gname are | |
3390 | # valid, though the header is not truly POSIX conforming. */ | |
3391 | # | |
3392 | # | |
a793ee30 | 3393 | TAR_HDR_OFF_MAGIC = 257 |
8fc6040c PG |
3394 | TAR_FMT_OLDGNU_MAGIC = b"ustar " |
3395 | ||
3396 | def read_gnu_tar_hdr (data): | |
3397 | if len (data) != BLOCKSIZE: # header requires one complete block | |
3398 | return None | |
65b35c42 | 3399 | |
8fc6040c PG |
3400 | try: |
3401 | name, mode, \ | |
3402 | uid, gid, \ | |
3403 | size, mtime, \ | |
3404 | checksum, \ | |
3405 | typeflag, \ | |
3406 | linkname, \ | |
3407 | magic, \ | |
3408 | version, \ | |
3409 | uname, \ | |
3410 | gname, \ | |
3411 | devmajor, \ | |
3412 | devminor, \ | |
3413 | atime, \ | |
3414 | ctime, \ | |
3415 | offset, \ | |
3416 | longnames, \ | |
3417 | unused, \ | |
3418 | offset1, numbytes1, \ | |
3419 | offset2, numbytes2, \ | |
3420 | offset3, numbytes3, \ | |
3421 | offset4, numbytes4, \ | |
3422 | isextended, \ | |
3423 | realsize, \ | |
3424 | pad = struct.unpack (TAR_FMT_HDR, data) | |
3425 | except struct.error: | |
3426 | return None | |
3427 | ||
3428 | if magic != TAR_FMT_OLDGNU_MAGIC: | |
3429 | return None | |
3430 | ||
3431 | # return all except “unused” and “pad” | |
3432 | return \ | |
3433 | { "name" : name, "mode" : mode | |
3434 | , "uid" : uid , "gid" : gid | |
3435 | , "size" : size, "mtime" : mtime | |
3436 | , "checksum" : checksum | |
3437 | , "typeflag" : typeflag | |
3438 | , "linkname" : linkname | |
3439 | , "magic" : magic | |
3440 | , "version" : version | |
3441 | , "uname" : uname, "gname" : gname | |
3442 | , "devmajor" : devmajor, "devminor" : devminor | |
3443 | , "atime" : atime, "ctime" : ctime | |
3444 | , "offset" : offset | |
3445 | , "longnames" : longnames | |
3446 | , "offset1" : offset1, "numbytes1" : numbytes1 | |
3447 | , "offset2" : offset2, "numbytes2" : numbytes2 | |
3448 | , "offset3" : offset3, "numbytes3" : numbytes3 | |
3449 | , "offset4" : offset4, "numbytes4" : numbytes4 | |
3450 | , "isextended" : isextended | |
3451 | , "realsize" : realsize | |
3452 | } | |
3453 | ||
3454 | ||
a793ee30 PG |
3455 | def tar_hdr_check_chksum (data): |
3456 | hdr = read_gnu_tar_hdr (data) | |
3457 | if hdr is None: | |
3458 | return False | |
3459 | s = calc_chksums (data) | |
3460 | return nti (hdr ["checksum"]) in s | |
3461 | ||
3462 | ||
8fc6040c PG |
3463 | def readable_tar_objects_offsets (ifd): |
3464 | """ | |
3465 | Traverse blocks in file, trying to extract tar headers. | |
3466 | """ | |
3467 | pos = 0 | |
3468 | offsets = [] | |
3469 | ||
a793ee30 PG |
3470 | mm = mmap.mmap(ifd, 0, mmap.MAP_SHARED, mmap.PROT_READ) |
3471 | pos = TAR_HDR_OFF_MAGIC | |
3472 | ||
8fc6040c | 3473 | while True: |
a793ee30 PG |
3474 | pos = mm.find (TAR_FMT_OLDGNU_MAGIC, pos) |
3475 | if pos == -1: | |
8fc6040c | 3476 | break |
a793ee30 PG |
3477 | off = pos - TAR_HDR_OFF_MAGIC |
3478 | mm.seek (off) | |
3479 | blk = mm.read (BLOCKSIZE) | |
3480 | if tar_hdr_check_chksum (blk) is True: | |
3481 | offsets.append (off) | |
3482 | pos += 1 | |
65b35c42 | 3483 | |
8fc6040c | 3484 | return offsets |
65b35c42 PG |
3485 | |
3486 | ||
dfd7865e PG |
3487 | def locate_gz_hdr_candidates (fd): |
3488 | """ | |
3489 | Walk over instances of the GZ magic in the payload, collecting their | |
3490 | positions. If the offset of the first found instance is not zero, the file | |
3491 | begins with leading garbage. | |
3492 | ||
3493 | Note that since the GZ magic consists of only two bytes, we expect a lot of | |
3494 | false positives inside binary data. | |
3495 | ||
3496 | :return: The list of offsets in the file. | |
3497 | """ | |
3498 | pos = 0 | |
3499 | cands = [] | |
3500 | mm = mmap.mmap(fd, 0, mmap.MAP_SHARED, mmap.PROT_READ) | |
3501 | ||
3502 | while True: | |
3503 | pos = mm.find (GZ_MAGIC_BYTES, pos) | |
3504 | if pos == -1: | |
3505 | break | |
3506 | cands.append (pos) | |
3507 | pos += len (GZ_MAGIC_BYTES) | |
3508 | ||
3509 | return cands | |
3510 | ||
3511 | ||
3512 | HDR_CAND_GOOD = 0 # header marks begin of valid object | |
3513 | HDR_CAND_FISHY = 1 # inconclusive | |
3514 | HDR_CAND_JUNK = 2 # not a header / object unreadable | |
3515 | ||
3516 | ||
3517 | def read_cstring (fd, max=-1, encoding=None): | |
3518 | """ | |
3519 | Read one NUL-terminated string from *fd* into a Python string. If *max* is | |
3520 | non-negative, reading will terminate after the specified number of bytes. | |
3521 | ||
3522 | Optionally, an *encoding* may be specified to interpret the data as. | |
3523 | ||
3524 | :returns: *None* if parsing failed or the maximum number of bytes has been | |
3525 | exceeded; a Python string with the data otherwise. | |
3526 | """ | |
3527 | buf = b"" | |
3528 | l = 0 | |
3529 | ||
3530 | while True: | |
3531 | c = os.read (fd, 1) | |
3532 | if c == NUL: | |
3533 | break | |
3534 | if max >= 0 and l > max: | |
3535 | return None | |
3536 | buf += c | |
3537 | l += 1 | |
3538 | if encoding is not None: | |
3539 | buf = buf.decode (encoding) | |
3540 | ||
3541 | return buf | |
3542 | ||
3543 | ||
3544 | def inspect_gz_hdr (fd, off): | |
3545 | """ | |
3546 | Attempt to parse a Gzip header in *fd* at position *off*. The format is | |
3547 | documented as RFC1952. | |
3548 | ||
3549 | Returns a verdict about the quality of that header plus the parsed header | |
3550 | when readable. Problematic sizes such as fields running past the EOF are | |
3551 | treated as garbage. Properties in which the header merely doesn’t conform | |
3552 | to the spec (garbage flag bits, bogus timestamp) are considered “fishy”. No | |
3553 | validation is possible on embedded strings because they are single-byte | |
3554 | encoded. | |
3555 | """ | |
3556 | fname = None | |
3557 | flags = 0x00 | |
3558 | dflags = 0x00 | |
3559 | mtime = 0x00000000 | |
3560 | oscode = 0x00 | |
3561 | verdict = HDR_CAND_GOOD | |
3562 | ||
3563 | os.lseek (fd, off, os.SEEK_SET) | |
3564 | if os.lseek (fd, 0, os.SEEK_CUR) != off: | |
3565 | return HDR_CAND_JUNK, None | |
3566 | ||
3567 | raw = os.read (fd, GZ_HEADER_SIZE) | |
3568 | if len (raw) != GZ_HEADER_SIZE: | |
3569 | return HDR_CAND_JUNK, None | |
3570 | ||
3571 | flags = 0x0 | |
3572 | try: | |
3573 | _m1, _m2, meth, flags, mtime, dflags, oscode = \ | |
3574 | struct.unpack (GZ_FMT_HEADER, raw) | |
3575 | if meth != GZ_METHOD_DEFLATE: # only deflate is supported | |
3576 | return HDR_CAND_JUNK, None | |
3577 | except struct.error as exn: | |
3578 | return HDR_CAND_JUNK, None | |
3579 | ||
3580 | if mtime > int (time.time ()): | |
3581 | verdict = HDR_CAND_FISHY | |
3582 | ||
3583 | if dflags != GZ_DEFLATE_FLAGS: | |
3584 | verdict = HDR_CAND_FISHY | |
3585 | ||
3586 | if oscode != GZ_OS_CODE: | |
3587 | verdict = HDR_CAND_FISHY | |
3588 | ||
3589 | if flags & GZ_FLAG_FTEXT: # created by some contrarian | |
3590 | verdict = HDR_CAND_FISHY | |
3591 | if flags & GZ_FLAG_FEXTRA: | |
04263795 | 3592 | xlen = struct.unpack ("<H", os.read (fd, 2))[0] |
dfd7865e PG |
3593 | xtra = os.read (fd, xlen) |
3594 | if len (xtra) != xlen: # eof inside header | |
3595 | return HDR_CAND_JUNK, None | |
3596 | if flags & GZ_FLAG_FNAME: | |
3597 | # read up to the next NUL byte, not exceeding the maximum path length | |
3598 | # allowed by tar(5) | |
3599 | fname = read_cstring (fd, max=(LENGTH_NAME + LENGTH_PREFIX), | |
3600 | encoding="iso-8859-1") | |
3601 | if fname is None: | |
3602 | return HDR_CAND_JUNK, None | |
3603 | if flags & GZ_FLAG_FCOMMENT: | |
3604 | fname = read_cstring (fd, max=(LENGTH_NAME + LENGTH_PREFIX), | |
3605 | encoding="iso-8859-1") | |
3606 | if fname is None: | |
3607 | return HDR_CAND_JUNK, None | |
3608 | if flags & GZ_FLAG_FHCRC: # half a CRC32; discarded | |
3609 | crc16 = os.read (fd, 2) | |
3610 | if len (crc16) != 2: # eof inside header | |
3611 | return HDR_CAND_JUNK, None | |
3612 | if flags & GZ_FLAG_RESERVED: | |
3613 | # according to the RFC, these must not be set | |
3614 | verdict = HDR_CAND_FISHY | |
3615 | ||
3616 | hlen = os.lseek (fd, 0, os.SEEK_CUR) - off | |
3617 | ||
3618 | return verdict, \ | |
3619 | { "fname" : fname | |
3620 | , "flags" : flags | |
3621 | , "dflags" : dflags | |
3622 | , "mtime" : mtime | |
3623 | , "oscode" : oscode | |
3624 | , "hlen" : hlen | |
3625 | } | |
3626 | ||
3627 | ||
3628 | def try_decompress (ifd, off, hdr): | |
3629 | """ | |
3630 | Attempt to process the object starting at *off* with gzip. | |
3631 | ||
3632 | :returns: A pair containing the values of the decompressed data and | |
3633 | the length of the input consumed. Note that the latter value | |
3634 | may exceed the length of the compressed data because the | |
3635 | *zlib* module does not provide a means to query how much | |
3636 | of the input it processed before the end of an object. | |
3637 | """ | |
3638 | import zlib | |
3639 | decmp = zlib.decompressobj (-zlib.MAX_WBITS) | |
3640 | pos = off | |
3641 | dlen = 0 # size of decompressed data | |
3642 | ||
3643 | os.lseek (ifd, pos, os.SEEK_SET) | |
3644 | while True: | |
3645 | cnk = os.read (ifd, BUFSIZE) | |
3646 | pos += len (cnk) | |
3647 | try: | |
3648 | data = decmp.decompress (cnk) | |
3649 | except zlib.error as exn: # probably CRC32 mismatch; terminate softly | |
3650 | break # fishy | |
3651 | dlen += len (data) | |
3652 | if decmp.eof is True: | |
3653 | break | |
3654 | if len (cnk) != BUFSIZE: # eof, but not end of decompressed object! | |
3655 | break # fishy | |
3656 | ||
3657 | return dlen, pos - off | |
3658 | ||
3659 | def readable_gz_objects_offsets (ifd, cands): | |
3660 | """ | |
3661 | Inspect header candidates for parseable *ifd* gzipped objects. | |
3662 | """ | |
3663 | good = [] | |
3664 | nobj = 0 | |
3665 | ||
3666 | for cand in cands: | |
3667 | nobj += 1 | |
3668 | vdt, hdr = inspect_gz_hdr (ifd, cand) | |
3669 | if vdt == HDR_CAND_JUNK: | |
3670 | pass # ignore unreadable ones | |
3671 | elif vdt in [HDR_CAND_GOOD, HDR_CAND_FISHY]: | |
3672 | off0 = cand + hdr ["hlen"] | |
3673 | dlen, clen = try_decompress (ifd, off0, hdr) | |
3674 | if dlen > 0 and clen > 0: | |
3675 | good.append (cand) | |
3676 | ||
3677 | return good | |
3678 | ||
3679 | ||
3680 | def reconstruct_offsets_gz (fname): | |
3681 | """ | |
3682 | From the given file, retrieve all GZ header-like offsets (“candidates”). | |
3683 | Then check each of those locations whether they can be processed as | |
3684 | compressed data. | |
3685 | """ | |
3686 | ifd = os.open (fname, os.O_RDONLY) | |
3687 | ||
3688 | try: | |
3689 | cands = locate_gz_hdr_candidates (ifd) | |
3690 | return readable_gz_objects_offsets (ifd, cands) | |
3691 | finally: | |
3692 | os.close (ifd) | |
3693 | ||
3694 | ||
65b35c42 PG |
3695 | def reconstruct_offsets_tar (fname): |
3696 | """ | |
3697 | From the given file, retrieve all tar header-like offsets (“candidates”). | |
3698 | Then check each of those locations whether they can be processed as tar | |
3699 | data. | |
3700 | """ | |
3701 | ifd = os.open (fname, os.O_RDONLY) | |
3702 | ||
3703 | try: | |
8fc6040c | 3704 | return readable_tar_objects_offsets (ifd) |
65b35c42 PG |
3705 | finally: |
3706 | os.close (ifd) | |
3707 | ||
3708 | ||
d39d4cbf PG |
3709 | def read_tarobj_at_offset (fileobj, offset, mode, secret=None): |
3710 | decr = None | |
d39d4cbf | 3711 | |
dfd7865e PG |
3712 | if secret is not None: |
3713 | ks = secret [0] | |
3714 | ||
3715 | if ks == crypto.PDTCRYPT_SECRET_PW: | |
3716 | decr = crypto.Decrypt (password=secret [1]) | |
3717 | elif ks == crypto.PDTCRYPT_SECRET_KEY: | |
3718 | key = binascii.unhexlify (secret [1]) | |
3719 | decr = crypto.Decrypt (key=key) | |
3720 | else: | |
3721 | raise RuntimeError | |
d39d4cbf | 3722 | |
5bd2d4b5 PG |
3723 | try: |
3724 | tarobj = \ | |
3725 | TarFile.open_at_offset (offset, | |
3726 | mode=mode, | |
3727 | fileobj=fileobj, | |
3728 | format=GNU_FORMAT, | |
3729 | concat='#' in mode, | |
3730 | encryption=decr, | |
3731 | save_to_members=False, | |
3732 | tolerance=TOLERANCE_RESCUE) | |
3733 | except (ReadError, EndOfFile): | |
3734 | return None | |
d39d4cbf PG |
3735 | |
3736 | return tarobj.next () | |
3737 | ||
3738 | ||
2d50b7f7 PG |
3739 | def idxent_of_tarinfo (tarinfo): |
3740 | """ | |
3741 | Scrape the information relevant for the index from a *TarInfo* object. | |
3742 | Keys like the inode number that lack a corresponding field in a TarInfo | |
3743 | will be set to some neutral value. | |
3744 | Example output: | |
3745 | ||
3746 | { "inode" : 0 | |
3747 | , "uid" : 0 | |
3748 | , "path" : "snapshot://annotations.db" | |
3749 | , "offset" : 0 | |
3750 | , "volume" : 0 | |
3751 | , "mode" : 33152 | |
3752 | , "ctime" : 1502798115 | |
3753 | , "mtime" : 1502196423 | |
3754 | , "size" : 144 | |
3755 | , "type" : "file" | |
3756 | , "gid" : 0 | |
3757 | } | |
3758 | ||
3759 | """ | |
3760 | ||
3761 | return \ | |
3762 | { "inode" : 0 # ignored when reading the index | |
3763 | , "uid" : tarinfo.uid | |
3764 | , "gid" : tarinfo.gid | |
3765 | , "path" : tarinfo.name # keeping URI scheme | |
3766 | , "offset" : 0 # to be added by the caller | |
3767 | , "volume" : tarinfo.volume_offset | |
3768 | , "mode" : tarinfo.mode | |
3769 | , "ctime" : tarinfo.mtime | |
3770 | , "mtime" : tarinfo.mtime | |
3771 | , "size" : tarinfo.size | |
3772 | , "type" : tarinfo.type | |
3773 | } | |
3774 | ||
3775 | ||
27ee4dd4 PG |
3776 | def gen_rescue_index (gen_volume_name, mode, maxvol=None, password=None, key=None): |
3777 | infos = [] | |
6690f5e0 PG |
3778 | psidx = [] # pseudo index, return value |
3779 | offsets = None | |
addcec42 | 3780 | secret = crypto.make_secret (password=password, key=key) |
6690f5e0 | 3781 | |
27ee4dd4 | 3782 | nvol = 0 |
dfd7865e | 3783 | |
27ee4dd4 PG |
3784 | while True: |
3785 | vpath = gen_volume_name (nvol) | |
3786 | try: | |
3787 | if secret is not None: | |
3788 | offsets = crypto.reconstruct_offsets (vpath, secret) | |
3789 | elif mode == "#gz": | |
3790 | offsets = reconstruct_offsets_gz (vpath) | |
3791 | elif mode == "#": | |
3792 | offsets = reconstruct_offsets_tar (vpath) | |
3793 | else: | |
3794 | raise TarError ("no rescue handling for mode “%s”" % mode) | |
3795 | except FileNotFoundError as exn: | |
3796 | # volume does not exist | |
611c5d03 | 3797 | if maxvol is not None and nvol < maxvol: |
27ee4dd4 PG |
3798 | continue # explicit volume number specified, ignore missing ones |
3799 | else: | |
3800 | break | |
3801 | ||
3802 | fileobj = bltn_open (vpath, "rb") | |
5bd2d4b5 PG |
3803 | |
3804 | def aux (acc, off): | |
3805 | obj = read_tarobj_at_offset (fileobj, off, mode, secret=secret) | |
3806 | if obj is not None: | |
3807 | acc.append ((off, nvol, obj)) | |
3808 | return acc | |
3809 | infos += functools.reduce (aux, offsets, []) | |
3810 | ||
27ee4dd4 PG |
3811 | nvol += 1 |
3812 | ||
5bd2d4b5 PG |
3813 | def aux (o, nvol, ti): |
3814 | ie = idxent_of_tarinfo (ti) | |
3815 | ie ["offset"] = o | |
3816 | ie ["volume"] = nvol | |
3817 | return ie | |
3818 | ||
27ee4dd4 | 3819 | psidx = [ aux (o, nvol, ti) for o, nvol, ti in infos ] |
6690f5e0 PG |
3820 | |
3821 | return psidx | |
7584f5c9 ERE |
3822 | |
3823 | #-------------------- | |
3824 | # exported functions | |
3825 | #-------------------- | |
3826 | def is_tarfile(name): | |
3827 | """Return True if name points to a tar archive that we | |
3828 | are able to handle, else return False. | |
3829 | """ | |
3830 | try: | |
3831 | t = open(name) | |
3832 | t.close() | |
3833 | return True | |
3834 | except TarError: | |
3835 | return False | |
3836 | ||
3837 | bltn_open = open | |
3838 | open = TarFile.open |