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