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 | |
be60ffd0 | 47 | import io |
dfd7865e PG |
48 | import mmap |
49 | import operator | |
50 | import os | |
51 | import re | |
7584f5c9 ERE |
52 | import shutil |
53 | import stat | |
7584f5c9 | 54 | import struct |
dfd7865e PG |
55 | import sys |
56 | import time | |
7584f5c9 | 57 | |
c7c736b6 PG |
58 | import traceback # XXX |
59 | ||
8ab8fac5 | 60 | from . import crypto |
6e812ad9 | 61 | |
7584f5c9 ERE |
62 | try: |
63 | import grp, pwd | |
64 | except ImportError: | |
65 | grp = pwd = None | |
66 | ||
be60ffd0 ERE |
67 | # os.symlink on Windows prior to 6.0 raises NotImplementedError |
68 | symlink_exception = (AttributeError, NotImplementedError) | |
69 | try: | |
70 | # OSError (winerror=1314) will be raised if the caller does not hold the | |
71 | # SeCreateSymbolicLinkPrivilege privilege | |
72 | symlink_exception += (OSError,) | |
73 | except NameError: | |
74 | pass | |
75 | ||
7584f5c9 ERE |
76 | # from tarfile import * |
77 | __all__ = ["TarFile", "TarInfo", "is_tarfile", "TarError"] | |
78 | ||
be60ffd0 ERE |
79 | from builtins import open as _open # Since 'open' is TarFile.open |
80 | ||
7584f5c9 ERE |
81 | #--------------------------------------------------------- |
82 | # tar constants | |
83 | #--------------------------------------------------------- | |
be60ffd0 | 84 | NUL = b"\0" # the null character |
7584f5c9 ERE |
85 | BLOCKSIZE = 512 # length of processing blocks |
86 | RECORDSIZE = BLOCKSIZE * 20 # length of records | |
be60ffd0 ERE |
87 | GNU_MAGIC = b"ustar \0" # magic gnu tar string |
88 | POSIX_MAGIC = b"ustar\x0000" # magic posix tar string | |
7584f5c9 ERE |
89 | |
90 | LENGTH_NAME = 100 # maximum length of a filename | |
91 | LENGTH_LINK = 100 # maximum length of a linkname | |
92 | LENGTH_PREFIX = 155 # maximum length of the prefix field | |
93 | ||
be60ffd0 ERE |
94 | REGTYPE = b"0" # regular file |
95 | AREGTYPE = b"\0" # regular file | |
96 | LNKTYPE = b"1" # link (inside tarfile) | |
97 | SYMTYPE = b"2" # symbolic link | |
98 | CHRTYPE = b"3" # character special device | |
99 | BLKTYPE = b"4" # block special device | |
100 | DIRTYPE = b"5" # directory | |
101 | FIFOTYPE = b"6" # fifo special device | |
102 | CONTTYPE = b"7" # contiguous file | |
103 | ||
104 | GNUTYPE_LONGNAME = b"L" # GNU tar longname | |
105 | GNUTYPE_LONGLINK = b"K" # GNU tar longlink | |
106 | GNUTYPE_SPARSE = b"S" # GNU tar sparse file | |
107 | GNUTYPE_MULTIVOL = b"M" # GNU tar continuation of a file that began on | |
68ddf955 | 108 | # another volume |
7584f5c9 | 109 | |
be60ffd0 ERE |
110 | XHDTYPE = b"x" # POSIX.1-2001 extended header |
111 | XGLTYPE = b"g" # POSIX.1-2001 global header | |
112 | SOLARIS_XHDTYPE = b"X" # Solaris extended header | |
7584f5c9 ERE |
113 | |
114 | USTAR_FORMAT = 0 # POSIX.1-1988 (ustar) format | |
115 | GNU_FORMAT = 1 # GNU tar format | |
116 | PAX_FORMAT = 2 # POSIX.1-2001 (pax) format | |
117 | DEFAULT_FORMAT = GNU_FORMAT | |
118 | ||
15a81fc0 | 119 | GZ_FMT_HEADER = b"<BBBBLBB" |
203cb25e | 120 | GZ_HEADER_SIZE = 10 # not including the name |
15a81fc0 PG |
121 | GZ_MAGIC = (0x1f, 0x8b) # 0o37, 0o213 |
122 | GZ_METHOD_DEFLATE = 0x08 # 0o10 | |
dfd7865e PG |
123 | GZ_FLAG_FTEXT = 1 << 0 # ASCII payload |
124 | GZ_FLAG_FHCRC = 1 << 1 # CRC16 | |
125 | GZ_FLAG_FEXTRA = 1 << 2 # extra field | |
126 | GZ_FLAG_FNAME = 1 << 3 # set by default in gzip | |
127 | GZ_FLAG_FCOMMENT = 1 << 4 # NUL-terminated comment | |
128 | GZ_FLAG_RESERVED = 7 << 5 # unassigned | |
15a81fc0 PG |
129 | GZ_DEFLATE_FLAGS = 0x00 # 0o00, never read (deflate.c) |
130 | GZ_OS_CODE = 0x03 # 0o03, default in gzip (tailor.h) | |
d601d33b PG |
131 | GZ_MAGIC_BYTES = struct.pack ("<BB", GZ_MAGIC [0], GZ_MAGIC [1]) |
132 | GZ_MAGIC_DEFLATE = struct.pack ("<BBB", GZ_MAGIC [0], GZ_MAGIC [1], | |
133 | GZ_METHOD_DEFLATE) | |
15a81fc0 | 134 | |
04f4c7ab PG |
135 | TOLERANCE_STRICT = 0 |
136 | TOLERANCE_RECOVER = 1 # rely on offsets in index | |
137 | TOLERANCE_RESCUE = 2 # deduce metadata from archive contents | |
138 | ||
dfd7865e PG |
139 | BUFSIZE = 16 * 1024 |
140 | ||
7584f5c9 | 141 | #--------------------------------------------------------- |
d1c38f40 PG |
142 | # archive handling mode |
143 | #--------------------------------------------------------- | |
144 | ||
145 | ARCMODE_PLAIN = 0 | |
146 | ARCMODE_ENCRYPT = 1 << 0 | |
147 | ARCMODE_COMPRESS = 1 << 1 | |
148 | ARCMODE_CONCAT = 1 << 2 | |
149 | ||
150 | def arcmode_fmt (m): | |
151 | if m == ARCMODE_PLAIN: | |
152 | return "PLAIN" | |
153 | first = True | |
154 | ret = "[" | |
155 | def chkappend (b, s): | |
156 | nonlocal m | |
157 | nonlocal ret | |
158 | nonlocal first | |
159 | if m & b: | |
160 | if first is True: first = False | |
161 | else: ret += " |" | |
162 | ret += " " + s | |
163 | chkappend (ARCMODE_ENCRYPT, "ENCRYPT") | |
164 | chkappend (ARCMODE_COMPRESS, "COMPRESS") | |
165 | chkappend (ARCMODE_CONCAT, "CONCAT") | |
166 | return ret + " ]" | |
167 | ||
168 | ||
169 | def arcmode_set (concat=False, encryption=None, comptype=None, init=ARCMODE_PLAIN): | |
170 | ret = init | |
171 | if bool (concat) is True: | |
172 | ret |= ARCMODE_CONCAT | |
173 | if encryption is not None: | |
174 | ret |= ARCMODE_ENCRYPT | |
175 | if comptype == "gz": | |
176 | ret |= ARCMODE_COMPRESS | |
177 | return ret | |
178 | ||
179 | #--------------------------------------------------------- | |
7584f5c9 ERE |
180 | # tarfile constants |
181 | #--------------------------------------------------------- | |
182 | # File types that tarfile supports: | |
183 | SUPPORTED_TYPES = (REGTYPE, AREGTYPE, LNKTYPE, | |
184 | SYMTYPE, DIRTYPE, FIFOTYPE, | |
185 | CONTTYPE, CHRTYPE, BLKTYPE, | |
186 | GNUTYPE_LONGNAME, GNUTYPE_LONGLINK, | |
68ddf955 | 187 | GNUTYPE_SPARSE, GNUTYPE_MULTIVOL) |
7584f5c9 ERE |
188 | |
189 | # File types that will be treated as a regular file. | |
190 | REGULAR_TYPES = (REGTYPE, AREGTYPE, | |
191 | CONTTYPE, GNUTYPE_SPARSE) | |
192 | ||
193 | # File types that are part of the GNU tar format. | |
194 | GNU_TYPES = (GNUTYPE_LONGNAME, GNUTYPE_LONGLINK, | |
68ddf955 | 195 | GNUTYPE_SPARSE, GNUTYPE_MULTIVOL) |
7584f5c9 ERE |
196 | |
197 | # Fields from a pax header that override a TarInfo attribute. | |
198 | PAX_FIELDS = ("path", "linkpath", "size", "mtime", | |
199 | "uid", "gid", "uname", "gname") | |
200 | ||
be60ffd0 ERE |
201 | # Fields from a pax header that are affected by hdrcharset. |
202 | PAX_NAME_FIELDS = {"path", "linkpath", "uname", "gname"} | |
203 | ||
7584f5c9 ERE |
204 | # Fields in a pax header that are numbers, all other fields |
205 | # are treated as strings. | |
206 | PAX_NUMBER_FIELDS = { | |
207 | "atime": float, | |
208 | "ctime": float, | |
209 | "mtime": float, | |
210 | "uid": int, | |
211 | "gid": int, | |
212 | "size": int | |
213 | } | |
214 | ||
215 | #--------------------------------------------------------- | |
7584f5c9 ERE |
216 | # initialization |
217 | #--------------------------------------------------------- | |
be60ffd0 ERE |
218 | |
219 | if os.name in ("nt", "ce"): | |
220 | ENCODING = "utf-8" | |
221 | else: | |
222 | ENCODING = sys.getfilesystemencoding() | |
7584f5c9 ERE |
223 | |
224 | #--------------------------------------------------------- | |
225 | # Some useful functions | |
226 | #--------------------------------------------------------- | |
227 | ||
be60ffd0 ERE |
228 | def stn(s, length, encoding, errors): |
229 | """Convert a string to a null-terminated bytes object. | |
7584f5c9 | 230 | """ |
be60ffd0 | 231 | s = s.encode(encoding, errors) |
7584f5c9 ERE |
232 | return s[:length] + (length - len(s)) * NUL |
233 | ||
be60ffd0 ERE |
234 | def nts(s, encoding, errors): |
235 | """Convert a null-terminated bytes object to a string. | |
7584f5c9 | 236 | """ |
be60ffd0 ERE |
237 | p = s.find(b"\0") |
238 | if p != -1: | |
239 | s = s[:p] | |
240 | return s.decode(encoding, errors) | |
241 | ||
242 | def sbtn(s, length, encoding, errors): | |
243 | """Convert a string or a bunch of bytes to a null-terminated bytes object | |
244 | of specific size. | |
245 | """ | |
246 | if isinstance(s, str): | |
247 | s = s.encode(encoding, errors) | |
248 | return s[:length] + (length - len(s)) * NUL | |
7584f5c9 ERE |
249 | |
250 | def nti(s): | |
251 | """Convert a number field to a python number. | |
252 | """ | |
253 | # There are two possible encodings for a number field, see | |
254 | # itn() below. | |
be60ffd0 ERE |
255 | if s[0] in (0o200, 0o377): |
256 | n = 0 | |
257 | for i in range(len(s) - 1): | |
258 | n <<= 8 | |
259 | n += s[i + 1] | |
260 | if s[0] == 0o377: | |
261 | n = -(256 ** (len(s) - 1) - n) | |
262 | else: | |
7584f5c9 | 263 | try: |
be60ffd0 | 264 | n = int(nts(s, "ascii", "strict") or "0", 8) |
7584f5c9 ERE |
265 | except ValueError: |
266 | raise InvalidHeaderError("invalid header") | |
7584f5c9 ERE |
267 | return n |
268 | ||
269 | def itn(n, digits=8, format=DEFAULT_FORMAT): | |
270 | """Convert a python number to a number field. | |
271 | """ | |
272 | # POSIX 1003.1-1988 requires numbers to be encoded as a string of | |
273 | # octal digits followed by a null-byte, this allows values up to | |
274 | # (8**(digits-1))-1. GNU tar allows storing numbers greater than | |
be60ffd0 ERE |
275 | # that if necessary. A leading 0o200 or 0o377 byte indicate this |
276 | # particular encoding, the following digits-1 bytes are a big-endian | |
277 | # base-256 representation. This allows values up to (256**(digits-1))-1. | |
278 | # A 0o200 byte indicates a positive number, a 0o377 byte a negative | |
279 | # number. | |
7584f5c9 | 280 | if 0 <= n < 8 ** (digits - 1): |
8112b0ed | 281 | s = bytes("%0*o" % (digits - 1, int(n)), "ascii") + NUL |
be60ffd0 ERE |
282 | elif format == GNU_FORMAT and -256 ** (digits - 1) <= n < 256 ** (digits - 1): |
283 | if n >= 0: | |
284 | s = bytearray([0o200]) | |
285 | else: | |
286 | s = bytearray([0o377]) | |
287 | n = 256 ** digits + n | |
7584f5c9 | 288 | |
be60ffd0 ERE |
289 | for i in range(digits - 1): |
290 | s.insert(1, n & 0o377) | |
7584f5c9 | 291 | n >>= 8 |
7584f5c9 | 292 | else: |
be60ffd0 ERE |
293 | raise ValueError("overflow in number field") |
294 | ||
295 | return s | |
7584f5c9 ERE |
296 | |
297 | def calc_chksums(buf): | |
298 | """Calculate the checksum for a member's header by summing up all | |
299 | characters except for the chksum field which is treated as if | |
300 | it was filled with spaces. According to the GNU tar sources, | |
301 | some tars (Sun and NeXT) calculate chksum with signed char, | |
302 | which will be different if there are chars in the buffer with | |
303 | the high bit set. So we calculate two checksums, unsigned and | |
304 | signed. | |
305 | """ | |
be60ffd0 ERE |
306 | unsigned_chksum = 256 + sum(struct.unpack_from("148B8x356B", buf)) |
307 | signed_chksum = 256 + sum(struct.unpack_from("148b8x356b", buf)) | |
7584f5c9 ERE |
308 | return unsigned_chksum, signed_chksum |
309 | ||
310 | def copyfileobj(src, dst, length=None): | |
311 | """Copy length bytes from fileobj src to fileobj dst. | |
312 | If length is None, copy the entire content. | |
313 | """ | |
314 | if length == 0: | |
315 | return | |
316 | if length is None: | |
317 | shutil.copyfileobj(src, dst) | |
318 | return | |
319 | ||
7584f5c9 | 320 | blocks, remainder = divmod(length, BUFSIZE) |
be60ffd0 | 321 | for b in range(blocks): |
7584f5c9 | 322 | buf = src.read(BUFSIZE) |
c474439c | 323 | dst.write(buf) |
7584f5c9 | 324 | if len(buf) < BUFSIZE: |
be60ffd0 | 325 | raise OSError("end of file reached") |
7584f5c9 ERE |
326 | if remainder != 0: |
327 | buf = src.read(remainder) | |
c474439c | 328 | dst.write(buf) |
7584f5c9 | 329 | if len(buf) < remainder: |
be60ffd0 | 330 | raise OSError("end of file reached") |
c7c736b6 | 331 | |
7584f5c9 | 332 | |
7584f5c9 | 333 | def filemode(mode): |
be60ffd0 ERE |
334 | """Deprecated in this location; use stat.filemode.""" |
335 | import warnings | |
336 | warnings.warn("deprecated in favor of stat.filemode", | |
337 | DeprecationWarning, 2) | |
338 | return stat.filemode(mode) | |
7584f5c9 ERE |
339 | |
340 | class TarError(Exception): | |
341 | """Base exception.""" | |
342 | pass | |
343 | class ExtractError(TarError): | |
344 | """General exception for extract errors.""" | |
345 | pass | |
346 | class ReadError(TarError): | |
be60ffd0 | 347 | """Exception for unreadable tar archives.""" |
7584f5c9 ERE |
348 | pass |
349 | class CompressionError(TarError): | |
350 | """Exception for unavailable compression methods.""" | |
351 | pass | |
352 | class StreamError(TarError): | |
353 | """Exception for unsupported operations on stream-like TarFiles.""" | |
354 | pass | |
355 | class HeaderError(TarError): | |
356 | """Base exception for header errors.""" | |
357 | pass | |
358 | class EmptyHeaderError(HeaderError): | |
359 | """Exception for empty headers.""" | |
360 | pass | |
361 | class TruncatedHeaderError(HeaderError): | |
362 | """Exception for truncated headers.""" | |
363 | pass | |
364 | class EOFHeaderError(HeaderError): | |
365 | """Exception for end of file headers.""" | |
366 | pass | |
367 | class InvalidHeaderError(HeaderError): | |
368 | """Exception for invalid headers.""" | |
369 | pass | |
370 | class SubsequentHeaderError(HeaderError): | |
371 | """Exception for missing and invalid extended headers.""" | |
372 | pass | |
8ab8fac5 PG |
373 | class InvalidEncryptionError(TarError): |
374 | """Exception for undefined crypto modes and combinations.""" | |
375 | pass | |
e4e5d0b8 PG |
376 | class DecryptionError(TarError): |
377 | """Exception for error during decryption.""" | |
378 | pass | |
c7c736b6 | 379 | class EncryptionError(TarError): |
e93f83f1 | 380 | """Exception for error during encryption.""" |
c7c736b6 | 381 | pass |
e50fa574 PG |
382 | class EndOfFile(Exception): |
383 | """Signal end of file condition when they’re not an error.""" | |
65b35c42 | 384 | pass |
7584f5c9 ERE |
385 | |
386 | #--------------------------- | |
387 | # internal stream interface | |
388 | #--------------------------- | |
389 | class _LowLevelFile: | |
390 | """Low-level file object. Supports reading and writing. | |
391 | It is used instead of a regular file object for streaming | |
392 | access. | |
393 | """ | |
394 | ||
395 | def __init__(self, name, mode): | |
ad4402e8 | 396 | _mode = { |
7584f5c9 | 397 | "r": os.O_RDONLY, |
c7c736b6 | 398 | "w": os.O_RDWR | os.O_CREAT | os.O_TRUNC, |
7584f5c9 ERE |
399 | }[mode] |
400 | if hasattr(os, "O_BINARY"): | |
f0287fb7 | 401 | _mode |= os.O_BINARY # pylint: disable=no-member |
be60ffd0 | 402 | self.fd = os.open(name, _mode, 0o666) |
ad4402e8 | 403 | self.offset = 0 |
7584f5c9 ERE |
404 | |
405 | def close(self): | |
406 | os.close(self.fd) | |
407 | ||
408 | def read(self, size): | |
ad4402e8 ERE |
409 | ret = os.read(self.fd, size) |
410 | self.offset += len(ret) | |
411 | return ret | |
7584f5c9 | 412 | |
867f75f7 PG |
413 | def write(self, s, pos=None): |
414 | if pos is not None: | |
415 | p0 = self.offset | |
416 | os.lseek (self.fd, pos, os.SEEK_SET) | |
417 | n = os.write(self.fd, s) | |
418 | if pos is None: | |
419 | self.offset += len(s) | |
420 | else: | |
421 | append = pos + n - p0 | |
422 | if append > 0: | |
423 | self.offset += append | |
424 | os.lseek (self.fd, p0, os.SEEK_SET) | |
7584f5c9 | 425 | |
ad4402e8 ERE |
426 | def tell(self): |
427 | return self.offset | |
428 | ||
c7c736b6 PG |
429 | def seek_set (self, pos): |
430 | os.lseek (self.fd, pos, os.SEEK_SET) | |
431 | self.offset = pos | |
432 | ||
8ab8fac5 | 433 | |
15a81fc0 PG |
434 | def gz_header (name=None): |
435 | timestamp = int(time.time()) | |
436 | flags = 0x0 | |
437 | ||
438 | if name is None: | |
439 | name = b"" | |
440 | else: | |
dfd7865e | 441 | flags |= GZ_FLAG_FNAME |
15a81fc0 PG |
442 | if type(name) is str: |
443 | name = name.encode("iso-8859-1", "replace") | |
6e99d23a PG |
444 | if name.endswith(b".pdtcrypt"): |
445 | name = name[:-9] | |
15a81fc0 PG |
446 | if name.endswith(b".gz"): |
447 | name = name[:-3] | |
448 | # RFC1952 says we must use ISO-8859-1 for the FNAME field. | |
449 | name += NUL | |
450 | ||
451 | hdr = struct.pack (GZ_FMT_HEADER, | |
452 | GZ_MAGIC [0], GZ_MAGIC [1], | |
453 | GZ_METHOD_DEFLATE, flags, | |
454 | timestamp, | |
455 | GZ_DEFLATE_FLAGS, GZ_OS_CODE) | |
456 | ||
457 | return hdr + name | |
458 | ||
d601d33b | 459 | |
7584f5c9 ERE |
460 | class _Stream: |
461 | """Class that serves as an adapter between TarFile and | |
462 | a stream-like object. The stream-like object only | |
463 | needs to have a read() or write() method and is accessed | |
464 | blockwise. Use of gzip or bzip2 compression is possible. | |
465 | A stream-like object could be for example: sys.stdin, | |
466 | sys.stdout, a socket, a tape device etc. | |
467 | ||
3031b7ae PG |
468 | _Stream is intended to be used only internally but is |
469 | nevertherless used externally by Deltatar. | |
470 | ||
471 | When encrypting, the ``enccounter`` will be used for | |
472 | initializing the first cryptographic context. When | |
473 | decrypting, its value will be compared to the decrypted | |
474 | object. Decryption fails if the value does not match. | |
475 | In effect, this means that a ``_Stream`` whose ctor was | |
476 | passed ``enccounter`` can only be used to encrypt or | |
477 | decrypt a single object. | |
7584f5c9 ERE |
478 | """ |
479 | ||
c7c736b6 | 480 | remainder = -1 # track size in encrypted entries |
04f4c7ab | 481 | tolerance = TOLERANCE_STRICT |
c7c736b6 | 482 | |
6e812ad9 | 483 | def __init__(self, name, mode, comptype, fileobj, bufsize, |
d1c38f40 | 484 | concat=False, encryption=None, enccounter=None, |
04f4c7ab | 485 | compresslevel=9, tolerance=TOLERANCE_STRICT): |
7584f5c9 ERE |
486 | """Construct a _Stream object. |
487 | """ | |
d1c38f40 | 488 | self.arcmode = arcmode_set (concat, encryption, comptype) |
04f4c7ab | 489 | self.tolerance = tolerance |
d1c38f40 | 490 | |
7584f5c9 ERE |
491 | self._extfileobj = True |
492 | if fileobj is None: | |
493 | fileobj = _LowLevelFile(name, mode) | |
494 | self._extfileobj = False | |
495 | ||
496 | if comptype == '*': | |
497 | # Enable transparent compression detection for the | |
498 | # stream interface | |
499 | fileobj = _StreamProxy(fileobj) | |
500 | comptype = fileobj.getcomptype() | |
d1c38f40 PG |
501 | if comptype == '': |
502 | comptype = "tar" | |
7584f5c9 | 503 | |
3031b7ae PG |
504 | self.enccounter = None |
505 | if self.arcmode & ARCMODE_ENCRYPT: | |
506 | self.enccounter = enccounter | |
507 | ||
7584f5c9 ERE |
508 | self.name = name or "" |
509 | self.mode = mode | |
510 | self.comptype = comptype | |
53732900 | 511 | self.cmp = None |
7584f5c9 ERE |
512 | self.fileobj = fileobj |
513 | self.bufsize = bufsize | |
be60ffd0 ERE |
514 | self.buf = b"" |
515 | self.pos = 0 | |
516 | self.concat_pos = 0 | |
7584f5c9 | 517 | self.closed = False |
be60ffd0 | 518 | self.flags = 0 |
be60ffd0 | 519 | self.last_block_offset = 0 |
e4e5d0b8 | 520 | self.dbuf = b"" # ??? |
46c03c02 | 521 | self.exception = None # communicate decompression failure |
2b82f50c | 522 | self.compresslevel = compresslevel |
784175ba | 523 | self.bytes_written = 0 |
c7c736b6 | 524 | # crypto parameters |
2ae46844 | 525 | self.encryption = encryption |
c7c736b6 | 526 | self.lasthdr = None |
7584f5c9 | 527 | |
be60ffd0 ERE |
528 | try: |
529 | if comptype == "gz": | |
530 | try: | |
531 | import zlib | |
532 | except ImportError: | |
533 | raise CompressionError("zlib module is not available") | |
534 | self.zlib = zlib | |
bec34b42 PG |
535 | if mode == "r": |
536 | self.exception = zlib.error | |
8ae983c4 | 537 | self._init_read_gz() |
bec34b42 | 538 | elif mode == "w": |
d1c38f40 PG |
539 | if not (self.arcmode & ARCMODE_CONCAT): |
540 | if self.arcmode & ARCMODE_ENCRYPT: | |
3031b7ae | 541 | self._init_write_encrypt (name) |
a0873dcc | 542 | self._init_write_gz () |
c2ffe2ec | 543 | self.crc = zlib.crc32(b"") & 0xFFFFffff |
7584f5c9 | 544 | |
be60ffd0 | 545 | elif comptype == "bz2": |
d1c38f40 | 546 | if self.arcmode & ARCMODE_ENCRYPT: |
2ae46844 | 547 | raise InvalidEncryptionError("encryption not available for " |
d1c38f40 | 548 | "compression “%s”" % comptype) |
be60ffd0 ERE |
549 | try: |
550 | import bz2 | |
551 | except ImportError: | |
552 | raise CompressionError("bz2 module is not available") | |
553 | if mode == "r": | |
554 | self.dbuf = b"" | |
555 | self.cmp = bz2.BZ2Decompressor() | |
556 | self.exception = OSError | |
557 | else: | |
558 | self.cmp = bz2.BZ2Compressor() | |
7584f5c9 | 559 | |
be60ffd0 | 560 | elif comptype == 'xz': |
d1c38f40 | 561 | if self.arcmode & ARCMODE_ENCRYPT: |
2ae46844 | 562 | raise InvalidEncryptionError("encryption not available for " |
d1c38f40 | 563 | "compression “%s”" % comptype) |
c7c736b6 PG |
564 | try: |
565 | import lzma | |
566 | except ImportError: | |
567 | raise CompressionError("lzma module is not available") | |
568 | if mode == "r": | |
569 | self.dbuf = b"" | |
570 | self.cmp = lzma.LZMADecompressor() | |
571 | self.exception = lzma.LZMAError | |
572 | else: | |
573 | self.cmp = lzma.LZMACompressor() | |
574 | ||
6de9444a | 575 | elif comptype == "tar": |
d1c38f40 | 576 | if not (self.arcmode & ARCMODE_CONCAT) \ |
6de9444a | 577 | and mode == "w" \ |
d1c38f40 | 578 | and self.arcmode & ARCMODE_ENCRYPT: |
3031b7ae | 579 | self._init_write_encrypt (name) |
6de9444a PG |
580 | |
581 | else: | |
d1c38f40 | 582 | if self.arcmode & ARCMODE_ENCRYPT: |
2ae46844 | 583 | raise InvalidEncryptionError("encryption not available for " |
d1c38f40 | 584 | "compression “%s”" % comptype) |
c7c736b6 | 585 | raise CompressionError("unknown compression type %r" % comptype) |
be60ffd0 | 586 | |
200d4866 | 587 | except: |
be60ffd0 ERE |
588 | if not self._extfileobj: |
589 | self.fileobj.close() | |
590 | self.closed = True | |
591 | raise | |
ac5e4184 | 592 | |
7584f5c9 ERE |
593 | def __del__(self): |
594 | if hasattr(self, "closed") and not self.closed: | |
fac2cfe1 PG |
595 | try: |
596 | self.close() | |
597 | except crypto.InternalError: | |
598 | # context already finalized due to abort but close() tried | |
599 | # to use it | |
600 | pass | |
7584f5c9 | 601 | |
c7c736b6 | 602 | |
d1c38f40 PG |
603 | def next (self, name): |
604 | if self.arcmode & ARCMODE_COMPRESS: | |
605 | if getattr (self, "cmp", None) is not None: | |
606 | self._finalize_write_gz () | |
0349168a PG |
607 | self.__sync() |
608 | if self.arcmode & ~(ARCMODE_ENCRYPT | ARCMODE_COMPRESS): | |
609 | self.last_block_offset = self.fileobj.tell() | |
d1c38f40 PG |
610 | if self.arcmode & ARCMODE_ENCRYPT: |
611 | self._finalize_write_encrypt () | |
612 | self._init_write_encrypt (name, set_last_block_offset=True) | |
613 | if self.arcmode & ARCMODE_COMPRESS: | |
614 | self._init_write_gz (set_last_block_offset = | |
0349168a | 615 | not (self.arcmode & ARCMODE_ENCRYPT)) |
d1c38f40 PG |
616 | return self.last_block_offset |
617 | ||
618 | ||
619 | def next_volume (self, name): | |
620 | # with non-concat modes, this is taken care by the _Stream | |
621 | # ctor as invoked by the newvol handler | |
622 | if self.arcmode & ARCMODE_COMPRESS: | |
623 | if getattr (self, "cmp", None) is not None: | |
624 | # e. g. compressed PAX header written | |
625 | self._finalize_write_gz () | |
626 | if self.arcmode & ARCMODE_ENCRYPT: | |
627 | self._init_write_encrypt (name) | |
628 | if self.arcmode & ARCMODE_COMPRESS: | |
629 | self._init_write_gz () | |
630 | ||
c7c736b6 | 631 | |
d1c38f40 PG |
632 | def _init_write_encrypt (self, entry=None, set_last_block_offset=False): |
633 | """ | |
634 | Save position for delayed write of header; fill the header location | |
635 | with dummy bytes. | |
636 | """ | |
637 | # first thing, proclaim new object to the encryption context | |
638 | # secondly, assemble the header with the updated parameters | |
639 | # and commit it directly to the underlying stream, bypassing the | |
640 | # encryption layer in .__write(). | |
641 | dummyhdr = self.encryption.next (entry, counter=self.enccounter) | |
642 | if dummyhdr is None: | |
643 | raise EncryptionError ("Crypto.next(): bad dummy header") # XXX | |
644 | self.lasthdr = self.fileobj.tell() | |
645 | self.__write_to_file(dummyhdr) | |
646 | if set_last_block_offset is True: | |
647 | self.last_block_offset = self.lasthdr | |
c7c736b6 PG |
648 | |
649 | ||
650 | def _finalize_write_encrypt (self): | |
651 | """ | |
652 | Seek back to header position, read dummy bytes, finalize crypto | |
653 | obtaining the actual header, write header, seek back to current | |
654 | position. | |
963d0db4 PG |
655 | |
656 | Returns the list of IV fixed parts as used during encryption. | |
c7c736b6 | 657 | """ |
d1c38f40 | 658 | if self.lasthdr is not None: |
c7c736b6 PG |
659 | pos0 = self.fileobj.tell () |
660 | self.fileobj.seek_set (self.lasthdr) | |
dd47d6a2 | 661 | dummy = self.fileobj.read (crypto.PDTCRYPT_HDR_SIZE) |
c7c736b6 PG |
662 | pos1 = self.fileobj.tell () |
663 | dpos = pos1 - self.lasthdr | |
dd47d6a2 | 664 | assert dpos == crypto.PDTCRYPT_HDR_SIZE |
c7c736b6 | 665 | self.fileobj.seek_set (pos0) |
c8c72fe1 | 666 | data, hdr, _ = self.encryption.done (dummy) |
5f38bff6 | 667 | self.__write_to_file(hdr, pos=self.lasthdr) |
c7c736b6 PG |
668 | self.__write_to_file(data) # append remainder of data |
669 | self.lasthdr = -1 | |
670 | ||
671 | ||
57db1546 PG |
672 | def _finalize_write_gz (self): |
673 | if self.cmp is not None: | |
674 | chunk = self.buf + self.cmp.flush() | |
675 | if chunk: | |
676 | if self.comptype == "gz": | |
677 | # The native zlib crc is an unsigned 32-bit integer, but | |
678 | # the Python wrapper implicitly casts that to a signed C | |
679 | # long. So, on a 32-bit box self.crc may "look negative", | |
680 | # while the same crc on a 64-bit box may "look positive". | |
681 | # To avoid irksome warnings from the `struct` module, force | |
682 | # it to look positive on all boxes. | |
683 | chunk += struct.pack("<L", self.crc & 0xffffffff) | |
684 | chunk += struct.pack("<L", self.concat_pos & 0xffffFFFF) | |
685 | self.__enc_write (chunk) | |
15a81fc0 | 686 | self.buf = b"" |
57db1546 PG |
687 | |
688 | ||
a0873dcc | 689 | def _init_write_gz (self, set_last_block_offset=False): |
5fdff89f ERE |
690 | ''' |
691 | Add a new gzip block, closing last one | |
692 | ''' | |
be60ffd0 | 693 | self.concat_pos = 0 |
c2ffe2ec | 694 | self.crc = self.zlib.crc32(b"") & 0xFFFFffff |
6de9444a | 695 | first = self.cmp is None |
2b82f50c ERE |
696 | self.cmp = self.zlib.compressobj(self.compresslevel, |
697 | self.zlib.DEFLATED, | |
698 | -self.zlib.MAX_WBITS, | |
699 | self.zlib.DEF_MEM_LEVEL, | |
700 | 0) | |
6e812ad9 DGM |
701 | |
702 | # if aes, we encrypt after compression | |
6de9444a | 703 | if set_last_block_offset is True: |
ad4402e8 | 704 | self.last_block_offset = self.fileobj.tell() |
6e812ad9 | 705 | |
15a81fc0 | 706 | self.__write(gz_header (self.name if first is True else None)) |
5fdff89f | 707 | |
ac5e4184 | 708 | |
7584f5c9 ERE |
709 | def write(self, s): |
710 | """Write string s to the stream. | |
711 | """ | |
712 | if self.comptype == "gz": | |
c2ffe2ec | 713 | self.crc = self.zlib.crc32(s, self.crc) & 0xFFFFffff |
7584f5c9 | 714 | self.pos += len(s) |
5fdff89f | 715 | self.concat_pos += len(s) |
53732900 | 716 | if self.cmp is not None: |
7584f5c9 ERE |
717 | s = self.cmp.compress(s) |
718 | self.__write(s) | |
719 | ||
c7c736b6 | 720 | def __sync(self): |
cb7a3911 | 721 | """Write what’s left in the buffer to the stream.""" |
c7c736b6 PG |
722 | self.__write (b"") # → len (buf) <= bufsiz |
723 | self.__enc_write (self.buf) | |
724 | self.buf = b"" | |
725 | ||
7584f5c9 | 726 | def __write(self, s): |
548bb8d5 CH |
727 | """Writes (and encodes) string s to the stream blockwise |
728 | ||
729 | will wait with encoding/writing until block is complete | |
7584f5c9 ERE |
730 | """ |
731 | self.buf += s | |
732 | while len(self.buf) > self.bufsize: | |
6e812ad9 | 733 | self.__enc_write(self.buf[:self.bufsize]) |
7584f5c9 ERE |
734 | self.buf = self.buf[self.bufsize:] |
735 | ||
867f75f7 | 736 | |
5f38bff6 | 737 | def __write_to_file(self, s, pos=None): |
6e812ad9 | 738 | ''' |
5f38bff6 | 739 | Writes directly to the fileobj; updates self.bytes_written. If “pos” is |
cb7a3911 | 740 | given, the stream will seek to that position first and back afterwards, |
5f38bff6 | 741 | and the total of bytes written is not updated. |
6e812ad9 | 742 | ''' |
867f75f7 | 743 | self.fileobj.write(s, pos) |
5f38bff6 PG |
744 | if pos is None: |
745 | self.bytes_written += len(s) | |
867f75f7 | 746 | |
6e812ad9 DGM |
747 | |
748 | def __enc_write(self, s): | |
cb7a3911 PG |
749 | """ |
750 | If encryption is active, the string s is encrypted before being written | |
751 | to the file. | |
752 | """ | |
753 | if len (s) == 0: | |
754 | return | |
d1c38f40 | 755 | if self.arcmode & ARCMODE_ENCRYPT: |
cb7a3911 PG |
756 | buf = s |
757 | while len (buf) > 0: | |
758 | n, ct = self.encryption.process(buf) | |
759 | self.__write_to_file(ct) | |
760 | buf = buf [n:] | |
761 | if len (buf) > 0: | |
762 | # The entire plaintext was not consumed: The size limit | |
763 | # for encrypted objects was reached. Transparently create | |
764 | # a new encrypted object and continue processing the input. | |
765 | self._finalize_write_encrypt () | |
766 | self._init_write_encrypt () | |
767 | else: | |
768 | self.__write_to_file(s) | |
769 | ||
6e812ad9 | 770 | |
784175ba CH |
771 | def estim_file_size(self): |
772 | """ estimates size of file if closing it now | |
773 | ||
774 | The result may differ greatly from the amount of data sent to write() | |
775 | due to compression, encryption and buffering. | |
776 | ||
777 | In tests the result (before calling close()) was up to 12k smaller than | |
778 | the final file size if compression is being used because zlib/bz2 | |
779 | compressors do not allow inspection of their buffered data :-( | |
780 | ||
ba5a449e CH |
781 | Still, we add what close() would add: 8 bytes for gz checksum, one |
782 | encryption block size if encryption is used and the size of our own | |
783 | buffer | |
784175ba CH |
784 | """ |
785 | if self.closed: | |
786 | return self.bytes_written | |
787 | ||
788 | result = self.bytes_written | |
789 | if self.buf: | |
790 | result += len(self.buf) | |
791 | if self.comptype == 'gz': | |
ba5a449e | 792 | result += 8 # 2 longs = 8 byte (no extra info written for bzip2) |
784175ba CH |
793 | return result |
794 | ||
5fdff89f | 795 | def close(self, close_fileobj=True): |
7584f5c9 ERE |
796 | """Close the _Stream object. No operation should be |
797 | done on it afterwards. | |
798 | """ | |
963d0db4 | 799 | |
7584f5c9 ERE |
800 | if self.closed: |
801 | return | |
802 | ||
963d0db4 | 803 | if close_fileobj is True: |
a0873dcc | 804 | |
ae3d0f2a | 805 | if self.mode == "w": |
d1c38f40 | 806 | if self.arcmode & ARCMODE_COMPRESS: |
a0873dcc | 807 | self._finalize_write_gz () |
ae3d0f2a | 808 | # end of Tar archive marker (two empty blocks) was written |
267bc643 PG |
809 | # finalize encryption last; no writes may be performed after |
810 | # this point | |
cb7a3911 | 811 | self.__sync () |
d1c38f40 PG |
812 | if self.arcmode & ARCMODE_ENCRYPT: |
813 | self._finalize_write_encrypt () | |
267bc643 | 814 | |
963d0db4 PG |
815 | if not self._extfileobj: |
816 | self.fileobj.close() | |
817 | else: | |
818 | # read the zlib crc and length and check them | |
819 | if self.mode == "r" and self.comptype == "gz": | |
820 | read_crc = self.__read(4) | |
821 | read_length = self.__read(4) | |
822 | calculated_crc = self.crc | |
823 | if struct.unpack("<L", read_crc)[0] != calculated_crc: | |
824 | raise CompressionError("bad gzip crc") | |
7584f5c9 ERE |
825 | self.closed = True |
826 | ||
54128a00 | 827 | |
7584f5c9 ERE |
828 | def _init_read_gz(self): |
829 | """Initialize for reading a gzip compressed fileobj. | |
830 | """ | |
831 | self.cmp = self.zlib.decompressobj(-self.zlib.MAX_WBITS) | |
7584f5c9 | 832 | |
85737f48 | 833 | read2 = self.__read(2) |
e50fa574 PG |
834 | if read2 == b"": |
835 | raise EndOfFile ("_init_read_gz(): read returned zero bytes at pos " | |
836 | "%d" % self.fileobj.tell()) | |
837 | # taken from gzip.GzipFile with some alterations | |
d601d33b | 838 | if read2 != GZ_MAGIC_BYTES: |
7584f5c9 | 839 | raise ReadError("not a gzip file") |
85737f48 | 840 | |
dfd7865e PG |
841 | read1 = ord (self.__read(1)) |
842 | if read1 != GZ_METHOD_DEFLATE: | |
7584f5c9 ERE |
843 | raise CompressionError("unsupported compression method") |
844 | ||
85737f48 | 845 | self.flags = flag = ord(self.__read(1)) |
dfd7865e | 846 | self.__read(6) # discard timestamp[4], deflate flags, os code |
7584f5c9 | 847 | |
dfd7865e | 848 | if flag & GZ_FLAG_FEXTRA: |
7584f5c9 ERE |
849 | xlen = ord(self.__read(1)) + 256 * ord(self.__read(1)) |
850 | self.read(xlen) | |
dfd7865e | 851 | if flag & GZ_FLAG_FNAME: |
7584f5c9 ERE |
852 | while True: |
853 | s = self.__read(1) | |
854 | if not s or s == NUL: | |
855 | break | |
dfd7865e | 856 | if flag & GZ_FLAG_FCOMMENT: |
7584f5c9 ERE |
857 | while True: |
858 | s = self.__read(1) | |
859 | if not s or s == NUL: | |
860 | break | |
dfd7865e | 861 | if flag & GZ_FLAG_FHCRC: |
7584f5c9 ERE |
862 | self.__read(2) |
863 | ||
c7c736b6 PG |
864 | def _init_read_encrypt (self): |
865 | """Initialize encryption for next entry in archive. Read a header and | |
866 | notify the crypto context.""" | |
d1c38f40 | 867 | if self.arcmode & ARCMODE_ENCRYPT: |
6e99d23a | 868 | lasthdr = self.fileobj.tell () |
15d3eefd PG |
869 | try: |
870 | hdr = crypto.hdr_read_stream (self.fileobj) | |
8a8ac469 PG |
871 | except crypto.EndOfFile: |
872 | return False | |
6e99d23a | 873 | except crypto.InvalidHeader as exn: |
c7c736b6 | 874 | raise DecryptionError ("Crypto.hdr_read_stream(): error “%s” " |
6e99d23a PG |
875 | "processing %r at pos %d" |
876 | % (exn, self.fileobj, lasthdr)) \ | |
ae3d0f2a | 877 | from exn |
3031b7ae PG |
878 | if self.enccounter is not None: |
879 | # enforce that the iv counter in the header matches an | |
880 | # explicitly requested one | |
881 | iv = crypto.hdr_iv_counter (hdr) | |
882 | if iv != self.enccounter: | |
883 | raise DecryptionError ("expected IV counter %d, got %d" | |
884 | % (self.enccounter, iv)) | |
6e99d23a | 885 | self.lasthdr = lasthdr |
c7c736b6 | 886 | self.remainder = hdr ["ctsize"] # distance to next header |
1ed44e7b PG |
887 | try: |
888 | self.encryption.next (hdr) | |
889 | except crypto.InvalidParameter as exn: | |
890 | raise DecryptionError ("Crypto.next(): error “%s” " | |
891 | "processing %r at pos %d" | |
892 | % (exn, self.fileobj, lasthdr)) \ | |
893 | from exn | |
8a8ac469 PG |
894 | |
895 | return True | |
c7c736b6 PG |
896 | |
897 | ||
8de91f4f PG |
898 | def _read_encrypt (self, buf): |
899 | """ | |
900 | Demote a program error to a decryption error in tolerant mode. This | |
901 | allows recovery from corrupted headers and invalid data. | |
902 | """ | |
903 | try: | |
904 | return self.encryption.process (buf) | |
905 | except RuntimeError as exn: | |
04f4c7ab | 906 | if self.tolerance != TOLERANCE_STRICT: |
8de91f4f PG |
907 | raise DecryptionError (exn) |
908 | raise | |
909 | ||
910 | ||
c7c736b6 PG |
911 | def _finalize_read_encrypt (self): |
912 | """ | |
913 | Finalize decryption. | |
914 | """ | |
d1c38f40 PG |
915 | if self.arcmode & ARCMODE_ENCRYPT \ |
916 | and self.lasthdr is not None : | |
c7c736b6 PG |
917 | assert self.remainder >= 0 |
918 | if self.remainder > 0: | |
c7c736b6 | 919 | self.remainder = 0 |
b0078f26 PG |
920 | try: |
921 | data = self.encryption.done () | |
922 | except crypto.InvalidGCMTag as exn: | |
923 | raise DecryptionError ("decryption failed: %s" % exn) | |
c7c736b6 PG |
924 | return data |
925 | ||
926 | ||
7584f5c9 ERE |
927 | def tell(self): |
928 | """Return the stream's file pointer position. | |
929 | """ | |
930 | return self.pos | |
931 | ||
932 | def seek(self, pos=0): | |
933 | """Set the stream's file pointer to pos. Negative seeking | |
934 | is forbidden. | |
935 | """ | |
936 | if pos - self.pos >= 0: | |
937 | blocks, remainder = divmod(pos - self.pos, self.bufsize) | |
be60ffd0 | 938 | for i in range(blocks): |
7584f5c9 ERE |
939 | self.read(self.bufsize) |
940 | self.read(remainder) | |
941 | else: | |
942 | raise StreamError("seeking backwards is not allowed") | |
943 | return self.pos | |
944 | ||
945 | def read(self, size=None): | |
946 | """Return the next size number of bytes from the stream. | |
947 | If size is not defined, return all bytes of the stream | |
948 | up to EOF. | |
949 | """ | |
950 | if size is None: | |
951 | t = [] | |
952 | while True: | |
953 | buf = self._read(self.bufsize) | |
954 | if not buf: | |
955 | break | |
956 | t.append(buf) | |
9dc7ac5c | 957 | buf = b"".join(t) |
7584f5c9 ERE |
958 | else: |
959 | buf = self._read(size) | |
960 | self.pos += len(buf) | |
961 | return buf | |
962 | ||
3a7e1a50 ERE |
963 | def readline(self): |
964 | """Reads just one line, new line character included | |
965 | """ | |
f0fd5e3a | 966 | # if \n in dbuf, no read neads to be done |
be60ffd0 ERE |
967 | if b'\n' in self.dbuf: |
968 | pos = self.dbuf.index(b'\n') + 1 | |
f0fd5e3a ERE |
969 | ret = self.dbuf[:pos] |
970 | self.dbuf = self.dbuf[pos:] | |
971 | return ret | |
972 | ||
1215b602 | 973 | buf = [] |
3a7e1a50 ERE |
974 | while True: |
975 | chunk = self._read(self.bufsize) | |
976 | ||
f0fd5e3a | 977 | # nothing more to read, so return the buffer |
3a7e1a50 | 978 | if not chunk: |
be60ffd0 | 979 | return b''.join(buf) |
3a7e1a50 ERE |
980 | |
981 | buf.append(chunk) | |
f0fd5e3a ERE |
982 | |
983 | # if \n found, return the new line | |
be60ffd0 ERE |
984 | if b'\n' in chunk: |
985 | dbuf = b''.join(buf) | |
986 | pos = dbuf.index(b'\n') + 1 | |
1215b602 | 987 | self.dbuf = dbuf[pos:] + self.dbuf |
3a7e1a50 ERE |
988 | return dbuf[:pos] |
989 | ||
7584f5c9 ERE |
990 | def _read(self, size): |
991 | """Return size bytes from the stream. | |
992 | """ | |
7584f5c9 ERE |
993 | c = len(self.dbuf) |
994 | t = [self.dbuf] | |
e4e5d0b8 | 995 | |
7584f5c9 | 996 | while c < size: |
867f75f7 | 997 | buf = self.__read(self.bufsize) |
7584f5c9 ERE |
998 | if not buf: |
999 | break | |
3a7e1a50 | 1000 | |
53732900 | 1001 | if self.cmp is not None: |
85737f48 | 1002 | try: |
3a7e1a50 | 1003 | buf = self.cmp.decompress(buf) |
54128a00 PG |
1004 | except self.exception as exn: |
1005 | raise ReadError("invalid compressed data (%r)" % exn) | |
be60ffd0 | 1006 | except Exception as e: |
04fb06f4 DGM |
1007 | # happens at the end of the file |
1008 | # _init_read_gz failed in the previous iteration so | |
e4e5d0b8 | 1009 | # self.cmp.decompress fails here |
d1c38f40 | 1010 | if self.arcmode & ARCMODE_CONCAT: |
be60ffd0 ERE |
1011 | pass |
1012 | else: | |
1013 | raise ReadError("invalid compressed data") | |
d1c38f40 | 1014 | if self.arcmode & ARCMODE_COMPRESS and hasattr(self, "crc"): |
c2ffe2ec | 1015 | self.crc = self.zlib.crc32(buf, self.crc) & 0xFFFFffff |
d1c38f40 PG |
1016 | if self.arcmode & ARCMODE_CONCAT \ |
1017 | and len(self.cmp.unused_data) != 0: | |
3a7e1a50 ERE |
1018 | self.buf = self.cmp.unused_data + self.buf |
1019 | self.close(close_fileobj=False) | |
1020 | try: | |
1021 | self._init_read_gz() | |
8de91f4f | 1022 | except DecryptionError: |
04f4c7ab | 1023 | if self.tolerance != TOLERANCE_STRICT: |
8de91f4f PG |
1024 | # return whatever data was processed successfully |
1025 | if len (buf) > 0: | |
1026 | t.append (buf) | |
1027 | if len (t) > 0: | |
1028 | break | |
24afaf18 PG |
1029 | raise |
1030 | except ReadError: # gzip troubles | |
1031 | if self.tolerance == TOLERANCE_RESCUE: | |
1032 | if len (buf) > 0: | |
1033 | t.append (buf) | |
1034 | if len (t) > 0: | |
1035 | break | |
1036 | raise | |
e50fa574 | 1037 | except EndOfFile: |
3a7e1a50 ERE |
1038 | # happens at the end of the file |
1039 | pass | |
c2ffe2ec | 1040 | self.crc = self.zlib.crc32(b"") & 0xFFFFffff |
3a7e1a50 | 1041 | self.closed = False |
7584f5c9 ERE |
1042 | t.append(buf) |
1043 | c += len(buf) | |
be60ffd0 | 1044 | t = b"".join(t) |
7584f5c9 ERE |
1045 | self.dbuf = t[size:] |
1046 | return t[:size] | |
1047 | ||
e4e5d0b8 | 1048 | |
7584f5c9 | 1049 | def __read(self, size): |
ef3b4499 PG |
1050 | """ |
1051 | Return size bytes from stream. If internal buffer is empty, read | |
1052 | another block from the stream. | |
1053 | ||
1054 | The function returns up to size bytes of data. When an error occurs | |
1055 | during decryption, everything until the end of the last successfully | |
1056 | finalized object is returned. | |
7584f5c9 ERE |
1057 | """ |
1058 | c = len(self.buf) | |
8de91f4f | 1059 | t = [self.buf] if c > 0 else [] |
1ed44e7b | 1060 | good_crypto = len (t) |
8de91f4f | 1061 | |
7584f5c9 | 1062 | while c < size: |
c7c736b6 | 1063 | todo = size |
8de91f4f PG |
1064 | try: |
1065 | if self.arcmode & ARCMODE_ENCRYPT: | |
1066 | if self.remainder <= 0: | |
1067 | # prepare next object | |
044585c6 PG |
1068 | if self._init_read_encrypt () is False: # EOF |
1069 | buf = None | |
1070 | break # while | |
8de91f4f PG |
1071 | |
1072 | # only read up to the end of the encrypted object | |
1073 | todo = min (size, self.remainder) | |
1074 | buf = self.fileobj.read(todo) | |
1075 | if self.arcmode & ARCMODE_ENCRYPT: | |
1076 | # decrypt the thing | |
1077 | buf = self._read_encrypt (buf) | |
1078 | if todo == self.remainder: | |
1079 | # at the end of a crypto object; finalization will fail if | |
1080 | # the GCM tag does not match | |
ef3b4499 | 1081 | trailing = self._finalize_read_encrypt () |
8de91f4f PG |
1082 | good_crypto = len (t) + 1 |
1083 | if len (trailing) > 0: | |
1084 | buf += trailing | |
1085 | self.remainder = 0 | |
1086 | else: | |
1087 | self.remainder -= todo | |
1088 | except DecryptionError: | |
04f4c7ab | 1089 | if self.tolerance == TOLERANCE_STRICT: |
8de91f4f PG |
1090 | raise |
1091 | self.encryption.drop () | |
24afaf18 PG |
1092 | if self.tolerance == TOLERANCE_RECOVER: |
1093 | if good_crypto == 0: | |
1094 | raise | |
1095 | # this may occur at any of the three crypto operations above. | |
1096 | # some objects did validate; discard all data after it; next | |
1097 | # call will start with the bad object and error out immediately | |
1098 | self.buf = b"".join (t [good_crypto:]) | |
1099 | return b"".join (t [:good_crypto]) | |
1100 | elif self.tolerance == TOLERANCE_RESCUE: | |
1101 | # keep what we have so far despite the finalization issue | |
1102 | t.append (buf) | |
1103 | c += len (buf) | |
1104 | break | |
1105 | else: | |
1106 | raise RuntimeError("internal error: bad tolerance level") | |
c7c736b6 PG |
1107 | |
1108 | if not buf: ## XXX stream terminated prematurely; this should be an error | |
7584f5c9 | 1109 | break |
c7c736b6 | 1110 | |
7584f5c9 ERE |
1111 | t.append(buf) |
1112 | c += len(buf) | |
be60ffd0 | 1113 | t = b"".join(t) |
7584f5c9 | 1114 | self.buf = t[size:] |
fb27c6e8 | 1115 | |
7584f5c9 | 1116 | return t[:size] |
7d372216 | 1117 | |
7584f5c9 ERE |
1118 | |
1119 | class _StreamProxy(object): | |
1120 | """Small proxy class that enables transparent compression | |
1121 | detection for the Stream interface (mode 'r|*'). | |
1122 | """ | |
1123 | ||
1124 | def __init__(self, fileobj): | |
1125 | self.fileobj = fileobj | |
1126 | self.buf = self.fileobj.read(BLOCKSIZE) | |
1127 | ||
f0287fb7 | 1128 | def read(self, size): # pylint: disable=method-hidden |
7584f5c9 ERE |
1129 | self.read = self.fileobj.read |
1130 | return self.buf | |
1131 | ||
1132 | def getcomptype(self): | |
d601d33b | 1133 | if self.buf.startswith(GZ_MAGIC_DEFLATE): |
7584f5c9 | 1134 | return "gz" |
be60ffd0 | 1135 | elif self.buf[0:3] == b"BZh" and self.buf[4:10] == b"1AY&SY": |
7584f5c9 | 1136 | return "bz2" |
be60ffd0 ERE |
1137 | elif self.buf.startswith((b"\x5d\x00\x00\x80", b"\xfd7zXZ")): |
1138 | return "xz" | |
1139 | else: | |
1140 | return "tar" | |
7584f5c9 ERE |
1141 | |
1142 | def close(self): | |
1143 | self.fileobj.close() | |
1144 | # class StreamProxy | |
1145 | ||
7584f5c9 ERE |
1146 | #------------------------ |
1147 | # Extraction file object | |
1148 | #------------------------ | |
1149 | class _FileInFile(object): | |
1150 | """A thin wrapper around an existing file object that | |
1151 | provides a part of its data as an individual file | |
1152 | object. | |
1153 | """ | |
1154 | ||
be60ffd0 | 1155 | def __init__(self, fileobj, offset, size, blockinfo=None): |
7584f5c9 ERE |
1156 | self.fileobj = fileobj |
1157 | self.offset = offset | |
1158 | self.size = size | |
7584f5c9 | 1159 | self.position = 0 |
be60ffd0 ERE |
1160 | self.name = getattr(fileobj, "name", None) |
1161 | self.closed = False | |
1162 | ||
1163 | if blockinfo is None: | |
1164 | blockinfo = [(0, size)] | |
1165 | ||
1166 | # Construct a map with data and zero blocks. | |
1167 | self.map_index = 0 | |
1168 | self.map = [] | |
1169 | lastpos = 0 | |
1170 | realpos = self.offset | |
1171 | for offset, size in blockinfo: | |
1172 | if offset > lastpos: | |
1173 | self.map.append((False, lastpos, offset, None)) | |
1174 | self.map.append((True, offset, offset + size, realpos)) | |
1175 | realpos += size | |
1176 | lastpos = offset + size | |
1177 | if lastpos < self.size: | |
1178 | self.map.append((False, lastpos, self.size, None)) | |
1179 | ||
1180 | def flush(self): | |
1181 | pass | |
1182 | ||
1183 | def readable(self): | |
1184 | return True | |
1185 | ||
1186 | def writable(self): | |
1187 | return False | |
1188 | ||
1189 | def seekable(self): | |
1190 | return self.fileobj.seekable() | |
7584f5c9 ERE |
1191 | |
1192 | def tell(self): | |
1193 | """Return the current file position. | |
1194 | """ | |
1195 | return self.position | |
1196 | ||
be60ffd0 | 1197 | def seek(self, position, whence=io.SEEK_SET): |
7584f5c9 ERE |
1198 | """Seek to a position in the file. |
1199 | """ | |
be60ffd0 ERE |
1200 | if whence == io.SEEK_SET: |
1201 | self.position = min(max(position, 0), self.size) | |
1202 | elif whence == io.SEEK_CUR: | |
1203 | if position < 0: | |
1204 | self.position = max(self.position + position, 0) | |
1205 | else: | |
1206 | self.position = min(self.position + position, self.size) | |
1207 | elif whence == io.SEEK_END: | |
1208 | self.position = max(min(self.size + position, self.size), 0) | |
1209 | else: | |
1210 | raise ValueError("Invalid argument") | |
1211 | return self.position | |
7584f5c9 ERE |
1212 | |
1213 | def read(self, size=None): | |
1214 | """Read data from the file. | |
1215 | """ | |
1216 | if size is None: | |
1217 | size = self.size - self.position | |
1218 | else: | |
1219 | size = min(size, self.size - self.position) | |
1220 | ||
be60ffd0 | 1221 | buf = b"" |
7584f5c9 | 1222 | while size > 0: |
7584f5c9 | 1223 | while True: |
be60ffd0 ERE |
1224 | data, start, stop, offset = self.map[self.map_index] |
1225 | if start <= self.position < stop: | |
7584f5c9 | 1226 | break |
be60ffd0 ERE |
1227 | else: |
1228 | self.map_index += 1 | |
1229 | if self.map_index == len(self.map): | |
1230 | self.map_index = 0 | |
1231 | length = min(size, stop - self.position) | |
1232 | if data: | |
1233 | self.fileobj.seek(offset + (self.position - start)) | |
1234 | buf += self.fileobj.read(length) | |
7584f5c9 | 1235 | else: |
be60ffd0 ERE |
1236 | buf += NUL * length |
1237 | size -= length | |
1238 | self.position += length | |
1239 | return buf | |
7584f5c9 | 1240 | |
be60ffd0 ERE |
1241 | def readinto(self, b): |
1242 | buf = self.read(len(b)) | |
1243 | b[:len(buf)] = buf | |
1244 | return len(buf) | |
7584f5c9 ERE |
1245 | |
1246 | def close(self): | |
7584f5c9 | 1247 | self.closed = True |
be60ffd0 | 1248 | #class _FileInFile |
7584f5c9 | 1249 | |
be60ffd0 ERE |
1250 | |
1251 | class ExFileObject(io.BufferedReader): | |
1252 | ||
1253 | def __init__(self, tarfile, tarinfo): | |
1254 | fileobj = _FileInFile(tarfile.fileobj, tarinfo.offset_data, | |
1255 | tarinfo.size, tarinfo.sparse) | |
1256 | super().__init__(fileobj) | |
7584f5c9 ERE |
1257 | #class ExFileObject |
1258 | ||
1259 | #------------------ | |
1260 | # Exported Classes | |
1261 | #------------------ | |
1262 | class TarInfo(object): | |
1263 | """Informational class which holds the details about an | |
1264 | archive member given by a tar header block. | |
1265 | TarInfo objects are returned by TarFile.getmember(), | |
1266 | TarFile.getmembers() and TarFile.gettarinfo() and are | |
1267 | usually created internally. | |
1268 | """ | |
1269 | ||
be60ffd0 ERE |
1270 | __slots__ = ("name", "mode", "uid", "gid", "size", "mtime", |
1271 | "chksum", "type", "linkname", "uname", "gname", | |
1272 | "devmajor", "devminor", "volume_offset", | |
1273 | "offset", "offset_data", "pax_headers", "sparse", | |
1274 | "tarfile", "_sparse_structs", "_link_target") | |
1275 | ||
7584f5c9 ERE |
1276 | def __init__(self, name=""): |
1277 | """Construct a TarInfo object. name is the optional name | |
1278 | of the member. | |
1279 | """ | |
1280 | self.name = name # member name | |
be60ffd0 | 1281 | self.mode = 0o644 # file permissions |
7584f5c9 ERE |
1282 | self.uid = 0 # user id |
1283 | self.gid = 0 # group id | |
1284 | self.size = 0 # file size | |
1285 | self.mtime = 0 # modification time | |
1286 | self.chksum = 0 # header checksum | |
1287 | self.type = REGTYPE # member type | |
1288 | self.linkname = "" # link name | |
1289 | self.uname = "" # user name | |
1290 | self.gname = "" # group name | |
1291 | self.devmajor = 0 # device major number | |
1292 | self.devminor = 0 # device minor number | |
1293 | ||
1294 | self.offset = 0 # the tar header starts here | |
1295 | self.offset_data = 0 # the file's data starts here | |
0eb5048f ERE |
1296 | self.volume_offset = 0 # the file's data corresponds with the data |
1297 | # starting at this position | |
7584f5c9 | 1298 | |
be60ffd0 | 1299 | self.sparse = None # sparse member information |
7584f5c9 ERE |
1300 | self.pax_headers = {} # pax header information |
1301 | ||
1302 | # In pax headers the "name" and "linkname" field are called | |
1303 | # "path" and "linkpath". | |
1304 | def _getpath(self): | |
1305 | return self.name | |
1306 | def _setpath(self, name): | |
1307 | self.name = name | |
1308 | path = property(_getpath, _setpath) | |
1309 | ||
1310 | def _getlinkpath(self): | |
1311 | return self.linkname | |
1312 | def _setlinkpath(self, linkname): | |
1313 | self.linkname = linkname | |
1314 | linkpath = property(_getlinkpath, _setlinkpath) | |
1315 | ||
1316 | def __repr__(self): | |
1317 | return "<%s %r at %#x>" % (self.__class__.__name__,self.name,id(self)) | |
1318 | ||
be60ffd0 | 1319 | def get_info(self, encoding=None, errors=None): |
7584f5c9 ERE |
1320 | """Return the TarInfo's attributes as a dictionary. |
1321 | """ | |
1322 | info = { | |
1323 | "name": self.name, | |
be60ffd0 | 1324 | "mode": self.mode & 0o7777, |
7584f5c9 ERE |
1325 | "uid": self.uid, |
1326 | "gid": self.gid, | |
1327 | "size": self.size, | |
1328 | "mtime": self.mtime, | |
1329 | "chksum": self.chksum, | |
1330 | "type": self.type, | |
1331 | "linkname": self.linkname, | |
1332 | "uname": self.uname, | |
1333 | "gname": self.gname, | |
1334 | "devmajor": self.devmajor, | |
36a315a0 | 1335 | "devminor": self.devminor, |
0eb5048f ERE |
1336 | "offset_data": self.offset_data, |
1337 | "volume_offset": self.volume_offset | |
7584f5c9 ERE |
1338 | } |
1339 | ||
1340 | if info["type"] == DIRTYPE and not info["name"].endswith("/"): | |
1341 | info["name"] += "/" | |
1342 | ||
7584f5c9 ERE |
1343 | return info |
1344 | ||
be60ffd0 ERE |
1345 | def tobuf(self, format=DEFAULT_FORMAT, encoding=ENCODING, |
1346 | errors="surrogateescape"): | |
7584f5c9 ERE |
1347 | """Return a tar header as a string of 512 byte blocks. |
1348 | """ | |
1349 | info = self.get_info(encoding, errors) | |
1350 | ||
1351 | if format == USTAR_FORMAT: | |
be60ffd0 | 1352 | return self.create_ustar_header(info, encoding, errors) |
7584f5c9 | 1353 | elif format == GNU_FORMAT: |
be60ffd0 | 1354 | return self.create_gnu_header(info, encoding, errors) |
7584f5c9 ERE |
1355 | elif format == PAX_FORMAT: |
1356 | return self.create_pax_header(info, encoding, errors) | |
1357 | else: | |
1358 | raise ValueError("invalid format") | |
1359 | ||
be60ffd0 | 1360 | def create_ustar_header(self, info, encoding, errors): |
7584f5c9 ERE |
1361 | """Return the object as a ustar header block. |
1362 | """ | |
1363 | info["magic"] = POSIX_MAGIC | |
1364 | ||
1365 | if len(info["linkname"]) > LENGTH_LINK: | |
1366 | raise ValueError("linkname is too long") | |
1367 | ||
1368 | if len(info["name"]) > LENGTH_NAME: | |
1369 | info["prefix"], info["name"] = self._posix_split_name(info["name"]) | |
1370 | ||
be60ffd0 | 1371 | return self._create_header(info, USTAR_FORMAT, encoding, errors) |
7584f5c9 | 1372 | |
be60ffd0 | 1373 | def create_gnu_header(self, info, encoding, errors): |
7584f5c9 ERE |
1374 | """Return the object as a GNU header block sequence. |
1375 | """ | |
1376 | info["magic"] = GNU_MAGIC | |
1377 | ||
2f854e77 ERE |
1378 | if self.ismultivol(): |
1379 | prefix = [ | |
1380 | itn(info.get("atime", 0), 12, GNU_FORMAT), | |
1381 | itn(info.get("ctime", 0), 12, GNU_FORMAT), | |
0eb5048f | 1382 | itn(self.volume_offset, 12, GNU_FORMAT), |
2f854e77 ERE |
1383 | itn(0, 119, GNU_FORMAT), # stuff unused in this tar implementation, set to zero |
1384 | ] | |
be60ffd0 | 1385 | info['prefix'] = b"".join(prefix) |
0eb5048f | 1386 | info['size'] = info['size'] - self.volume_offset |
2f854e77 | 1387 | |
be60ffd0 | 1388 | buf = b"" |
7584f5c9 | 1389 | if len(info["linkname"]) > LENGTH_LINK: |
be60ffd0 ERE |
1390 | buf += self._create_gnu_long_header(info["linkname"], |
1391 | GNUTYPE_LONGLINK, encoding, errors) | |
7584f5c9 ERE |
1392 | |
1393 | if len(info["name"]) > LENGTH_NAME: | |
be60ffd0 ERE |
1394 | buf += self._create_gnu_long_header(info["name"], GNUTYPE_LONGNAME, |
1395 | encoding, errors) | |
7584f5c9 | 1396 | |
be60ffd0 | 1397 | return buf + self._create_header(info, GNU_FORMAT, encoding, errors) |
7584f5c9 ERE |
1398 | |
1399 | def create_pax_header(self, info, encoding, errors): | |
1400 | """Return the object as a ustar header block. If it cannot be | |
1401 | represented this way, prepend a pax extended header sequence | |
1402 | with supplement information. | |
1403 | """ | |
1404 | info["magic"] = POSIX_MAGIC | |
1405 | pax_headers = self.pax_headers.copy() | |
c04e0751 ERE |
1406 | if self.ismultivol(): |
1407 | info['size'] = info['size'] - self.volume_offset | |
7584f5c9 ERE |
1408 | |
1409 | # Test string fields for values that exceed the field length or cannot | |
1410 | # be represented in ASCII encoding. | |
1411 | for name, hname, length in ( | |
36a315a0 ERE |
1412 | ("name", "path", LENGTH_NAME), |
1413 | ("linkname", "linkpath", LENGTH_LINK), | |
1414 | ("uname", "uname", 32), | |
1415 | ("gname", "gname", 32)): | |
7584f5c9 ERE |
1416 | |
1417 | if hname in pax_headers: | |
1418 | # The pax header has priority. | |
1419 | continue | |
1420 | ||
7584f5c9 ERE |
1421 | # Try to encode the string as ASCII. |
1422 | try: | |
be60ffd0 | 1423 | info[name].encode("ascii", "strict") |
7584f5c9 | 1424 | except UnicodeEncodeError: |
be60ffd0 | 1425 | pax_headers[hname] = info[name] |
7584f5c9 ERE |
1426 | continue |
1427 | ||
1428 | if len(info[name]) > length: | |
be60ffd0 | 1429 | pax_headers[hname] = info[name] |
7584f5c9 ERE |
1430 | |
1431 | # Test number fields for values that exceed the field limit or values | |
1432 | # that like to be stored as float. | |
1433 | for name, digits in (("uid", 8), ("gid", 8), ("size", 12), ("mtime", 12)): | |
1434 | if name in pax_headers: | |
1435 | # The pax header has priority. Avoid overflow. | |
1436 | info[name] = 0 | |
1437 | continue | |
1438 | ||
1439 | val = info[name] | |
1440 | if not 0 <= val < 8 ** (digits - 1) or isinstance(val, float): | |
be60ffd0 | 1441 | pax_headers[name] = str(val) |
7584f5c9 ERE |
1442 | info[name] = 0 |
1443 | ||
1444 | # Create a pax extended header if necessary. | |
1445 | if pax_headers: | |
be60ffd0 | 1446 | buf = self._create_pax_generic_header(pax_headers, XHDTYPE, encoding) |
7584f5c9 | 1447 | else: |
be60ffd0 | 1448 | buf = b"" |
7584f5c9 | 1449 | |
be60ffd0 | 1450 | return buf + self._create_header(info, USTAR_FORMAT, "ascii", "replace") |
7584f5c9 ERE |
1451 | |
1452 | @classmethod | |
1453 | def create_pax_global_header(cls, pax_headers): | |
1454 | """Return the object as a pax global header block sequence. | |
1455 | """ | |
be60ffd0 | 1456 | return cls._create_pax_generic_header(pax_headers, XGLTYPE, "utf-8") |
7584f5c9 ERE |
1457 | |
1458 | def _posix_split_name(self, name): | |
1459 | """Split a name longer than 100 chars into a prefix | |
1460 | and a name part. | |
1461 | """ | |
1462 | prefix = name[:LENGTH_PREFIX + 1] | |
1463 | while prefix and prefix[-1] != "/": | |
1464 | prefix = prefix[:-1] | |
1465 | ||
1466 | name = name[len(prefix):] | |
1467 | prefix = prefix[:-1] | |
1468 | ||
1469 | if not prefix or len(name) > LENGTH_NAME: | |
1470 | raise ValueError("name is too long") | |
1471 | return prefix, name | |
1472 | ||
1473 | @staticmethod | |
be60ffd0 | 1474 | def _create_header(info, format, encoding, errors): |
7584f5c9 ERE |
1475 | """Return a header block. info is a dictionary with file |
1476 | information, format must be one of the *_FORMAT constants. | |
1477 | """ | |
1478 | parts = [ | |
be60ffd0 ERE |
1479 | stn(info.get("name", ""), 100, encoding, errors), |
1480 | itn(info.get("mode", 0) & 0o7777, 8, format), | |
7584f5c9 ERE |
1481 | itn(info.get("uid", 0), 8, format), |
1482 | itn(info.get("gid", 0), 8, format), | |
1483 | itn(info.get("size", 0), 12, format), | |
1484 | itn(info.get("mtime", 0), 12, format), | |
be60ffd0 | 1485 | b" ", # checksum field |
2f854e77 | 1486 | info.get("type", REGTYPE), |
be60ffd0 ERE |
1487 | stn(info.get("linkname", ""), 100, encoding, errors), |
1488 | info.get("magic", POSIX_MAGIC), | |
1489 | stn(info.get("uname", ""), 32, encoding, errors), | |
1490 | stn(info.get("gname", ""), 32, encoding, errors), | |
7584f5c9 ERE |
1491 | itn(info.get("devmajor", 0), 8, format), |
1492 | itn(info.get("devminor", 0), 8, format), | |
be60ffd0 | 1493 | sbtn(info.get("prefix", ""), 155, encoding, errors) |
7584f5c9 ERE |
1494 | ] |
1495 | ||
be60ffd0 | 1496 | buf = struct.pack("%ds" % BLOCKSIZE, b"".join(parts)) |
7584f5c9 | 1497 | chksum = calc_chksums(buf[-BLOCKSIZE:])[0] |
be60ffd0 | 1498 | buf = buf[:-364] + bytes("%06o\0" % chksum, "ascii") + buf[-357:] |
7584f5c9 ERE |
1499 | return buf |
1500 | ||
1501 | @staticmethod | |
1502 | def _create_payload(payload): | |
1503 | """Return the string payload filled with zero bytes | |
1504 | up to the next 512 byte border. | |
1505 | """ | |
1506 | blocks, remainder = divmod(len(payload), BLOCKSIZE) | |
1507 | if remainder > 0: | |
1508 | payload += (BLOCKSIZE - remainder) * NUL | |
1509 | return payload | |
1510 | ||
1511 | @classmethod | |
be60ffd0 | 1512 | def _create_gnu_long_header(cls, name, type, encoding, errors): |
7584f5c9 ERE |
1513 | """Return a GNUTYPE_LONGNAME or GNUTYPE_LONGLINK sequence |
1514 | for name. | |
1515 | """ | |
be60ffd0 | 1516 | name = name.encode(encoding, errors) + NUL |
7584f5c9 ERE |
1517 | |
1518 | info = {} | |
1519 | info["name"] = "././@LongLink" | |
1520 | info["type"] = type | |
1521 | info["size"] = len(name) | |
1522 | info["magic"] = GNU_MAGIC | |
1523 | ||
1524 | # create extended header + name blocks. | |
be60ffd0 | 1525 | return cls._create_header(info, USTAR_FORMAT, encoding, errors) + \ |
7584f5c9 ERE |
1526 | cls._create_payload(name) |
1527 | ||
1528 | @classmethod | |
be60ffd0 ERE |
1529 | def _create_pax_generic_header(cls, pax_headers, type, encoding): |
1530 | """Return a POSIX.1-2008 extended or global header sequence | |
7584f5c9 | 1531 | that contains a list of keyword, value pairs. The values |
be60ffd0 | 1532 | must be strings. |
7584f5c9 | 1533 | """ |
be60ffd0 ERE |
1534 | # Check if one of the fields contains surrogate characters and thereby |
1535 | # forces hdrcharset=BINARY, see _proc_pax() for more information. | |
1536 | binary = False | |
1537 | for keyword, value in pax_headers.items(): | |
1538 | try: | |
1539 | value.encode("utf-8", "strict") | |
1540 | except UnicodeEncodeError: | |
1541 | binary = True | |
1542 | break | |
1543 | ||
1544 | records = b"" | |
1545 | if binary: | |
1546 | # Put the hdrcharset field at the beginning of the header. | |
1547 | records += b"21 hdrcharset=BINARY\n" | |
1548 | ||
1549 | for keyword, value in pax_headers.items(): | |
1550 | keyword = keyword.encode("utf-8") | |
1551 | if binary: | |
1552 | # Try to restore the original byte representation of `value'. | |
1553 | # Needless to say, that the encoding must match the string. | |
1554 | value = value.encode(encoding, "surrogateescape") | |
1555 | else: | |
1556 | value = value.encode("utf-8") | |
1557 | ||
7584f5c9 ERE |
1558 | l = len(keyword) + len(value) + 3 # ' ' + '=' + '\n' |
1559 | n = p = 0 | |
1560 | while True: | |
1561 | n = l + len(str(p)) | |
1562 | if n == p: | |
1563 | break | |
1564 | p = n | |
be60ffd0 | 1565 | records += bytes(str(p), "ascii") + b" " + keyword + b"=" + value + b"\n" |
7584f5c9 ERE |
1566 | |
1567 | # We use a hardcoded "././@PaxHeader" name like star does | |
1568 | # instead of the one that POSIX recommends. | |
1569 | info = {} | |
1570 | info["name"] = "././@PaxHeader" | |
1571 | info["type"] = type | |
1572 | info["size"] = len(records) | |
1573 | info["magic"] = POSIX_MAGIC | |
1574 | ||
1575 | # Create pax header + record blocks. | |
be60ffd0 | 1576 | return cls._create_header(info, USTAR_FORMAT, "ascii", "replace") + \ |
7584f5c9 ERE |
1577 | cls._create_payload(records) |
1578 | ||
1579 | @classmethod | |
be60ffd0 ERE |
1580 | def frombuf(cls, buf, encoding, errors): |
1581 | """Construct a TarInfo object from a 512 byte bytes object. | |
7584f5c9 ERE |
1582 | """ |
1583 | if len(buf) == 0: | |
1584 | raise EmptyHeaderError("empty header") | |
1585 | if len(buf) != BLOCKSIZE: | |
1586 | raise TruncatedHeaderError("truncated header") | |
1587 | if buf.count(NUL) == BLOCKSIZE: | |
1588 | raise EOFHeaderError("end of file header") | |
1589 | ||
1590 | chksum = nti(buf[148:156]) | |
1591 | if chksum not in calc_chksums(buf): | |
1592 | raise InvalidHeaderError("bad checksum") | |
1593 | ||
1594 | obj = cls() | |
be60ffd0 | 1595 | obj.name = nts(buf[0:100], encoding, errors) |
7584f5c9 ERE |
1596 | obj.mode = nti(buf[100:108]) |
1597 | obj.uid = nti(buf[108:116]) | |
1598 | obj.gid = nti(buf[116:124]) | |
1599 | obj.size = nti(buf[124:136]) | |
1600 | obj.mtime = nti(buf[136:148]) | |
1601 | obj.chksum = chksum | |
1602 | obj.type = buf[156:157] | |
be60ffd0 ERE |
1603 | obj.linkname = nts(buf[157:257], encoding, errors) |
1604 | obj.uname = nts(buf[265:297], encoding, errors) | |
1605 | obj.gname = nts(buf[297:329], encoding, errors) | |
7584f5c9 ERE |
1606 | obj.devmajor = nti(buf[329:337]) |
1607 | obj.devminor = nti(buf[337:345]) | |
be60ffd0 ERE |
1608 | prefix = nts(buf[345:500], encoding, errors) |
1609 | ||
1610 | # The old GNU sparse format occupies some of the unused | |
1611 | # space in the buffer for up to 4 sparse structures. | |
1612 | # Save the them for later processing in _proc_sparse(). | |
1613 | if obj.type == GNUTYPE_SPARSE: | |
1614 | pos = 386 | |
1615 | structs = [] | |
1616 | for i in range(4): | |
1617 | try: | |
1618 | offset = nti(buf[pos:pos + 12]) | |
1619 | numbytes = nti(buf[pos + 12:pos + 24]) | |
1620 | except ValueError: | |
1621 | break | |
1622 | structs.append((offset, numbytes)) | |
1623 | pos += 24 | |
1624 | isextended = bool(buf[482]) | |
1625 | origsize = nti(buf[483:495]) | |
1626 | obj._sparse_structs = (structs, isextended, origsize) | |
7584f5c9 ERE |
1627 | |
1628 | # Old V7 tar format represents a directory as a regular | |
1629 | # file with a trailing slash. | |
1630 | if obj.type == AREGTYPE and obj.name.endswith("/"): | |
1631 | obj.type = DIRTYPE | |
1632 | ||
1633 | # Remove redundant slashes from directories. | |
1634 | if obj.isdir(): | |
1635 | obj.name = obj.name.rstrip("/") | |
1636 | ||
1637 | # Reconstruct a ustar longname. | |
1638 | if prefix and obj.type not in GNU_TYPES: | |
1639 | obj.name = prefix + "/" + obj.name | |
c474439c ERE |
1640 | else: |
1641 | obj.offset_data = nti(buf[369:381]) | |
7584f5c9 ERE |
1642 | return obj |
1643 | ||
1644 | @classmethod | |
1645 | def fromtarfile(cls, tarfile): | |
1646 | """Return the next TarInfo object from TarFile object | |
1647 | tarfile. | |
1648 | """ | |
1649 | buf = tarfile.fileobj.read(BLOCKSIZE) | |
be60ffd0 | 1650 | obj = cls.frombuf(buf, tarfile.encoding, tarfile.errors) |
7584f5c9 ERE |
1651 | obj.offset = tarfile.fileobj.tell() - BLOCKSIZE |
1652 | return obj._proc_member(tarfile) | |
1653 | ||
1654 | #-------------------------------------------------------------------------- | |
1655 | # The following are methods that are called depending on the type of a | |
1656 | # member. The entry point is _proc_member() which can be overridden in a | |
1657 | # subclass to add custom _proc_*() methods. A _proc_*() method MUST | |
1658 | # implement the following | |
1659 | # operations: | |
1660 | # 1. Set self.offset_data to the position where the data blocks begin, | |
1661 | # if there is data that follows. | |
1662 | # 2. Set tarfile.offset to the position where the next member's header will | |
1663 | # begin. | |
1664 | # 3. Return self or another valid TarInfo object. | |
1665 | def _proc_member(self, tarfile): | |
1666 | """Choose the right processing method depending on | |
1667 | the type and call it. | |
1668 | """ | |
1669 | if self.type in (GNUTYPE_LONGNAME, GNUTYPE_LONGLINK): | |
1670 | return self._proc_gnulong(tarfile) | |
1671 | elif self.type == GNUTYPE_SPARSE: | |
1672 | return self._proc_sparse(tarfile) | |
1673 | elif self.type in (XHDTYPE, XGLTYPE, SOLARIS_XHDTYPE): | |
1674 | return self._proc_pax(tarfile) | |
1675 | else: | |
1676 | return self._proc_builtin(tarfile) | |
1677 | ||
1678 | def _proc_builtin(self, tarfile): | |
1679 | """Process a builtin type or an unknown type which | |
1680 | will be treated as a regular file. | |
1681 | """ | |
1682 | self.offset_data = tarfile.fileobj.tell() | |
1683 | offset = self.offset_data | |
00c34a12 | 1684 | if self.isreg() or self.ismultivol() or self.type not in SUPPORTED_TYPES: |
7584f5c9 ERE |
1685 | # Skip the following data blocks. |
1686 | offset += self._block(self.size) | |
1687 | tarfile.offset = offset | |
1688 | ||
1689 | # Patch the TarInfo object with saved global | |
1690 | # header information. | |
1691 | self._apply_pax_info(tarfile.pax_headers, tarfile.encoding, tarfile.errors) | |
1692 | ||
1693 | return self | |
1694 | ||
1695 | def _proc_gnulong(self, tarfile): | |
1696 | """Process the blocks that hold a GNU longname | |
1697 | or longlink member. | |
1698 | """ | |
1699 | buf = tarfile.fileobj.read(self._block(self.size)) | |
1700 | ||
1701 | # Fetch the next header and process it. | |
1702 | try: | |
1703 | next = self.fromtarfile(tarfile) | |
1704 | except HeaderError: | |
1705 | raise SubsequentHeaderError("missing or bad subsequent header") | |
1706 | ||
1707 | # Patch the TarInfo object from the next header with | |
1708 | # the longname information. | |
1709 | next.offset = self.offset | |
1710 | if self.type == GNUTYPE_LONGNAME: | |
be60ffd0 | 1711 | next.name = nts(buf, tarfile.encoding, tarfile.errors) |
7584f5c9 | 1712 | elif self.type == GNUTYPE_LONGLINK: |
be60ffd0 | 1713 | next.linkname = nts(buf, tarfile.encoding, tarfile.errors) |
7584f5c9 ERE |
1714 | |
1715 | return next | |
1716 | ||
1717 | def _proc_sparse(self, tarfile): | |
1718 | """Process a GNU sparse header plus extra headers. | |
1719 | """ | |
be60ffd0 ERE |
1720 | # We already collected some sparse structures in frombuf(). |
1721 | structs, isextended, origsize = self._sparse_structs | |
1722 | del self._sparse_structs | |
1723 | ||
1724 | # Collect sparse structures from extended header blocks. | |
1725 | while isextended: | |
7584f5c9 ERE |
1726 | buf = tarfile.fileobj.read(BLOCKSIZE) |
1727 | pos = 0 | |
be60ffd0 | 1728 | for i in range(21): |
7584f5c9 ERE |
1729 | try: |
1730 | offset = nti(buf[pos:pos + 12]) | |
1731 | numbytes = nti(buf[pos + 12:pos + 24]) | |
1732 | except ValueError: | |
1733 | break | |
be60ffd0 ERE |
1734 | if offset and numbytes: |
1735 | structs.append((offset, numbytes)) | |
7584f5c9 | 1736 | pos += 24 |
be60ffd0 ERE |
1737 | isextended = bool(buf[504]) |
1738 | self.sparse = structs | |
7584f5c9 ERE |
1739 | |
1740 | self.offset_data = tarfile.fileobj.tell() | |
1741 | tarfile.offset = self.offset_data + self._block(self.size) | |
1742 | self.size = origsize | |
7584f5c9 ERE |
1743 | return self |
1744 | ||
1745 | def _proc_pax(self, tarfile): | |
1746 | """Process an extended or global header as described in | |
be60ffd0 | 1747 | POSIX.1-2008. |
7584f5c9 ERE |
1748 | """ |
1749 | # Read the header information. | |
1750 | buf = tarfile.fileobj.read(self._block(self.size)) | |
1751 | ||
1752 | # A pax header stores supplemental information for either | |
1753 | # the following file (extended) or all following files | |
1754 | # (global). | |
1755 | if self.type == XGLTYPE: | |
1756 | pax_headers = tarfile.pax_headers | |
1757 | else: | |
1758 | pax_headers = tarfile.pax_headers.copy() | |
1759 | ||
be60ffd0 ERE |
1760 | # Check if the pax header contains a hdrcharset field. This tells us |
1761 | # the encoding of the path, linkpath, uname and gname fields. Normally, | |
1762 | # these fields are UTF-8 encoded but since POSIX.1-2008 tar | |
1763 | # implementations are allowed to store them as raw binary strings if | |
1764 | # the translation to UTF-8 fails. | |
1765 | match = re.search(br"\d+ hdrcharset=([^\n]+)\n", buf) | |
1766 | if match is not None: | |
1767 | pax_headers["hdrcharset"] = match.group(1).decode("utf-8") | |
1768 | ||
1769 | # For the time being, we don't care about anything other than "BINARY". | |
1770 | # The only other value that is currently allowed by the standard is | |
1771 | # "ISO-IR 10646 2000 UTF-8" in other words UTF-8. | |
1772 | hdrcharset = pax_headers.get("hdrcharset") | |
1773 | if hdrcharset == "BINARY": | |
1774 | encoding = tarfile.encoding | |
1775 | else: | |
1776 | encoding = "utf-8" | |
1777 | ||
7584f5c9 ERE |
1778 | # Parse pax header information. A record looks like that: |
1779 | # "%d %s=%s\n" % (length, keyword, value). length is the size | |
1780 | # of the complete record including the length field itself and | |
1781 | # the newline. keyword and value are both UTF-8 encoded strings. | |
be60ffd0 | 1782 | regex = re.compile(br"(\d+) ([^=]+)=") |
7584f5c9 ERE |
1783 | pos = 0 |
1784 | while True: | |
1785 | match = regex.match(buf, pos) | |
1786 | if not match: | |
1787 | break | |
1788 | ||
1789 | length, keyword = match.groups() | |
1790 | length = int(length) | |
1791 | value = buf[match.end(2) + 1:match.start(1) + length - 1] | |
1792 | ||
be60ffd0 ERE |
1793 | # Normally, we could just use "utf-8" as the encoding and "strict" |
1794 | # as the error handler, but we better not take the risk. For | |
1795 | # example, GNU tar <= 1.23 is known to store filenames it cannot | |
1796 | # translate to UTF-8 as raw strings (unfortunately without a | |
1797 | # hdrcharset=BINARY header). | |
1798 | # We first try the strict standard encoding, and if that fails we | |
1799 | # fall back on the user's encoding and error handler. | |
1800 | keyword = self._decode_pax_field(keyword, "utf-8", "utf-8", | |
1801 | tarfile.errors) | |
1802 | if keyword in PAX_NAME_FIELDS: | |
1803 | value = self._decode_pax_field(value, encoding, tarfile.encoding, | |
1804 | tarfile.errors) | |
1805 | else: | |
1806 | value = self._decode_pax_field(value, "utf-8", "utf-8", | |
1807 | tarfile.errors) | |
7584f5c9 ERE |
1808 | |
1809 | pax_headers[keyword] = value | |
1810 | pos += length | |
1811 | ||
36a315a0 | 1812 | |
7584f5c9 ERE |
1813 | # Fetch the next header. |
1814 | try: | |
1815 | next = self.fromtarfile(tarfile) | |
1816 | except HeaderError: | |
1817 | raise SubsequentHeaderError("missing or bad subsequent header") | |
1818 | ||
be60ffd0 ERE |
1819 | # Process GNU sparse information. |
1820 | if "GNU.sparse.map" in pax_headers: | |
1821 | # GNU extended sparse format version 0.1. | |
1822 | self._proc_gnusparse_01(next, pax_headers) | |
1823 | ||
1824 | elif "GNU.sparse.size" in pax_headers: | |
1825 | # GNU extended sparse format version 0.0. | |
1826 | self._proc_gnusparse_00(next, pax_headers, buf) | |
1827 | ||
1828 | elif pax_headers.get("GNU.sparse.major") == "1" and pax_headers.get("GNU.sparse.minor") == "0": | |
1829 | # GNU extended sparse format version 1.0. | |
1830 | self._proc_gnusparse_10(next, pax_headers, tarfile) | |
1831 | ||
7584f5c9 ERE |
1832 | if self.type in (XHDTYPE, SOLARIS_XHDTYPE): |
1833 | # Patch the TarInfo object with the extended header info. | |
1834 | next._apply_pax_info(pax_headers, tarfile.encoding, tarfile.errors) | |
1835 | next.offset = self.offset | |
1836 | ||
1837 | if "size" in pax_headers: | |
1838 | # If the extended header replaces the size field, | |
1839 | # we need to recalculate the offset where the next | |
1840 | # header starts. | |
1841 | offset = next.offset_data | |
1842 | if next.isreg() or next.type not in SUPPORTED_TYPES: | |
1843 | offset += next._block(next.size) | |
1844 | tarfile.offset = offset | |
1845 | ||
c04e0751 ERE |
1846 | if next is not None: |
1847 | if "GNU.volume.filename" in pax_headers: | |
1848 | if pax_headers["GNU.volume.filename"] == next.name: | |
1849 | if "GNU.volume.size" in pax_headers: | |
1850 | next.size = int(pax_headers["GNU.volume.size"]) | |
1851 | if "GNU.volume.offset" in pax_headers: | |
1852 | next.volume_offset = int(pax_headers["GNU.volume.offset"]) | |
1853 | ||
1854 | for key in pax_headers.keys(): | |
1855 | if key.startswith("GNU.volume"): | |
1856 | del tarfile.pax_headers[key] | |
0eb5048f | 1857 | |
7584f5c9 ERE |
1858 | return next |
1859 | ||
be60ffd0 ERE |
1860 | def _proc_gnusparse_00(self, next, pax_headers, buf): |
1861 | """Process a GNU tar extended sparse header, version 0.0. | |
7584f5c9 | 1862 | """ |
be60ffd0 ERE |
1863 | offsets = [] |
1864 | for match in re.finditer(br"\d+ GNU.sparse.offset=(\d+)\n", buf): | |
1865 | offsets.append(int(match.group(1))) | |
1866 | numbytes = [] | |
1867 | for match in re.finditer(br"\d+ GNU.sparse.numbytes=(\d+)\n", buf): | |
1868 | numbytes.append(int(match.group(1))) | |
1869 | next.sparse = list(zip(offsets, numbytes)) | |
7584f5c9 | 1870 | |
be60ffd0 ERE |
1871 | def _proc_gnusparse_01(self, next, pax_headers): |
1872 | """Process a GNU tar extended sparse header, version 0.1. | |
1873 | """ | |
1874 | sparse = [int(x) for x in pax_headers["GNU.sparse.map"].split(",")] | |
1875 | next.sparse = list(zip(sparse[::2], sparse[1::2])) | |
7584f5c9 | 1876 | |
be60ffd0 ERE |
1877 | def _proc_gnusparse_10(self, next, pax_headers, tarfile): |
1878 | """Process a GNU tar extended sparse header, version 1.0. | |
1879 | """ | |
1880 | fields = None | |
1881 | sparse = [] | |
1882 | buf = tarfile.fileobj.read(BLOCKSIZE) | |
1883 | fields, buf = buf.split(b"\n", 1) | |
1884 | fields = int(fields) | |
1885 | while len(sparse) < fields * 2: | |
1886 | if b"\n" not in buf: | |
1887 | buf += tarfile.fileobj.read(BLOCKSIZE) | |
1888 | number, buf = buf.split(b"\n", 1) | |
1889 | sparse.append(int(number)) | |
1890 | next.offset_data = tarfile.fileobj.tell() | |
1891 | next.sparse = list(zip(sparse[::2], sparse[1::2])) | |
7584f5c9 | 1892 | |
be60ffd0 ERE |
1893 | def _apply_pax_info(self, pax_headers, encoding, errors): |
1894 | """Replace fields with supplemental information from a previous | |
1895 | pax extended or global header. | |
1896 | """ | |
1897 | for keyword, value in pax_headers.items(): | |
1898 | if keyword == "GNU.sparse.name": | |
1899 | setattr(self, "path", value) | |
1900 | elif keyword == "GNU.sparse.size": | |
1901 | setattr(self, "size", int(value)) | |
1902 | elif keyword == "GNU.sparse.realsize": | |
1903 | setattr(self, "size", int(value)) | |
1904 | elif keyword in PAX_FIELDS: | |
1905 | if keyword in PAX_NUMBER_FIELDS: | |
1906 | try: | |
1907 | value = PAX_NUMBER_FIELDS[keyword](value) | |
1908 | except ValueError: | |
1909 | value = 0 | |
1910 | if keyword == "path": | |
f0287fb7 | 1911 | value = value.rstrip("/") # pylint: disable=no-member |
be60ffd0 | 1912 | setattr(self, keyword, value) |
7584f5c9 ERE |
1913 | |
1914 | self.pax_headers = pax_headers.copy() | |
1915 | ||
be60ffd0 ERE |
1916 | def _decode_pax_field(self, value, encoding, fallback_encoding, fallback_errors): |
1917 | """Decode a single field from a pax record. | |
1918 | """ | |
1919 | try: | |
1920 | return value.decode(encoding, "strict") | |
1921 | except UnicodeDecodeError: | |
1922 | return value.decode(fallback_encoding, fallback_errors) | |
1923 | ||
7584f5c9 ERE |
1924 | def _block(self, count): |
1925 | """Round up a byte count by BLOCKSIZE and return it, | |
1926 | e.g. _block(834) => 1024. | |
1927 | """ | |
1928 | blocks, remainder = divmod(count, BLOCKSIZE) | |
1929 | if remainder: | |
1930 | blocks += 1 | |
1931 | return blocks * BLOCKSIZE | |
1932 | ||
1933 | def isreg(self): | |
1934 | return self.type in REGULAR_TYPES | |
1935 | def isfile(self): | |
1936 | return self.isreg() | |
1937 | def isdir(self): | |
1938 | return self.type == DIRTYPE | |
1939 | def issym(self): | |
1940 | return self.type == SYMTYPE | |
1941 | def islnk(self): | |
1942 | return self.type == LNKTYPE | |
1943 | def ischr(self): | |
1944 | return self.type == CHRTYPE | |
1945 | def isblk(self): | |
1946 | return self.type == BLKTYPE | |
1947 | def isfifo(self): | |
1948 | return self.type == FIFOTYPE | |
1949 | def issparse(self): | |
be60ffd0 | 1950 | return self.sparse is not None |
7584f5c9 ERE |
1951 | def isdev(self): |
1952 | return self.type in (CHRTYPE, BLKTYPE, FIFOTYPE) | |
68ddf955 | 1953 | def ismultivol(self): |
c04e0751 ERE |
1954 | return self.type == GNUTYPE_MULTIVOL or self.volume_offset > 0 or\ |
1955 | "GNU.volume.offset" in self.pax_headers | |
7584f5c9 ERE |
1956 | # class TarInfo |
1957 | ||
1958 | class TarFile(object): | |
1959 | """The TarFile Class provides an interface to tar archives. | |
1960 | """ | |
1961 | ||
1962 | debug = 0 # May be set from 0 (no msgs) to 3 (all msgs) | |
1963 | ||
1964 | dereference = False # If true, add content of linked file to the | |
1965 | # tar file, else the link. | |
1966 | ||
1967 | ignore_zeros = False # If true, skips empty or invalid blocks and | |
1968 | # continues processing. | |
1969 | ||
83f2d71e | 1970 | max_volume_size = None # If different from None, establishes maximum |
68ddf955 ERE |
1971 | # size of tar volumes |
1972 | ||
1973 | new_volume_handler = None # function handler to be executed before when | |
1974 | # a new volume is needed | |
1975 | ||
1976 | volume_number = 0 # current volume number, used for multi volume | |
1977 | # support | |
1978 | ||
7584f5c9 ERE |
1979 | errorlevel = 1 # If 0, fatal errors only appear in debug |
1980 | # messages (if debug >= 0). If > 0, errors | |
1981 | # are passed to the caller as exceptions. | |
1982 | ||
1983 | format = DEFAULT_FORMAT # The format to use when creating an archive. | |
1984 | ||
1985 | encoding = ENCODING # Encoding for 8-bit character strings. | |
1986 | ||
1987 | errors = None # Error handler for unicode conversion. | |
1988 | ||
1989 | tarinfo = TarInfo # The default TarInfo class to use. | |
1990 | ||
be60ffd0 | 1991 | fileobject = ExFileObject # The file-object for extractfile(). |
7584f5c9 | 1992 | |
d1c38f40 PG |
1993 | arcmode = ARCMODE_PLAIN # Object processing mode (“concat”, encryption, |
1994 | # compression) | |
5fdff89f | 1995 | |
ea625b04 ERE |
1996 | save_to_members = True # If new members are saved. This can be disabled |
1997 | # if you manage lots of files and don't want | |
1998 | # to have high memory usage | |
1999 | ||
9ef1fb87 TJ |
2000 | cache_uid2user = {} # cache to avoid getpwuid calls. It always parses /etc/passwd. |
2001 | cache_gid2group = {} # same cache for groups | |
2002 | ||
7584f5c9 ERE |
2003 | def __init__(self, name=None, mode="r", fileobj=None, format=None, |
2004 | tarinfo=None, dereference=None, ignore_zeros=None, encoding=None, | |
be60ffd0 | 2005 | errors="surrogateescape", pax_headers=None, debug=None, |
548bb8d5 | 2006 | errorlevel=None, max_volume_size=None, new_volume_handler=None, |
d1c38f40 | 2007 | concat=False, nacl=None, |
c7c736b6 | 2008 | save_to_members=True): |
7584f5c9 ERE |
2009 | """Open an (uncompressed) tar archive `name'. `mode' is either 'r' to |
2010 | read from an existing archive, 'a' to append data to an existing | |
2011 | file or 'w' to create a new file overwriting an existing one. `mode' | |
2012 | defaults to 'r'. | |
2013 | If `fileobj' is given, it is used for reading or writing data. If it | |
2014 | can be determined, `mode' is overridden by `fileobj's mode. | |
2015 | `fileobj' is not closed, when TarFile is closed. | |
2016 | """ | |
2017 | if len(mode) > 1 or mode not in "raw": | |
2018 | raise ValueError("mode must be 'r', 'a' or 'w'") | |
2019 | self.mode = mode | |
d1c38f40 | 2020 | self.arcmode = arcmode_set (concat) |
c7c736b6 | 2021 | self.nacl = nacl |
7584f5c9 ERE |
2022 | self._mode = {"r": "rb", "a": "r+b", "w": "wb"}[mode] |
2023 | ||
2024 | if not fileobj: | |
2025 | if self.mode == "a" and not os.path.exists(name): | |
2026 | # Create nonexistent files in append mode. | |
2027 | self.mode = "w" | |
2028 | self._mode = "wb" | |
2029 | fileobj = bltn_open(name, self._mode) | |
2030 | self._extfileobj = False | |
2031 | else: | |
2032 | if name is None and hasattr(fileobj, "name"): | |
2033 | name = fileobj.name | |
d5361dac | 2034 | # when fileobj is a gzip.GzipFile, fileobj.mode is an int (not valid for us) |
be60ffd0 | 2035 | if hasattr(fileobj, "mode") and isinstance(fileobj.mode, str): |
7584f5c9 ERE |
2036 | self._mode = fileobj.mode |
2037 | self._extfileobj = True | |
be60ffd0 | 2038 | self.name = os.path.abspath(name) if name else None |
2f854e77 | 2039 | self.base_name = self.name = os.path.abspath(name) if name else None |
7584f5c9 ERE |
2040 | self.fileobj = fileobj |
2041 | ||
2042 | # Init attributes. | |
2043 | if format is not None: | |
2044 | self.format = format | |
2045 | if tarinfo is not None: | |
2046 | self.tarinfo = tarinfo | |
2047 | if dereference is not None: | |
2048 | self.dereference = dereference | |
2049 | if ignore_zeros is not None: | |
2050 | self.ignore_zeros = ignore_zeros | |
2051 | if encoding is not None: | |
2052 | self.encoding = encoding | |
2053 | ||
be60ffd0 | 2054 | self.errors = errors |
7584f5c9 ERE |
2055 | |
2056 | if pax_headers is not None and self.format == PAX_FORMAT: | |
2057 | self.pax_headers = pax_headers | |
2058 | else: | |
2059 | self.pax_headers = {} | |
2060 | ||
2061 | if debug is not None: | |
2062 | self.debug = debug | |
2063 | if errorlevel is not None: | |
2064 | self.errorlevel = errorlevel | |
2065 | ||
2066 | # Init datastructures. | |
ae48acc8 | 2067 | if max_volume_size and max_volume_size < 3*BLOCKSIZE: |
0c818a18 | 2068 | raise ValueError("max_volume_size needs to be at least %d" % (3*BLOCKSIZE)) |
ae48acc8 ERE |
2069 | if max_volume_size and not callable(new_volume_handler): |
2070 | raise ValueError("new_volume_handler needs to be set and be callable for multivolume support") | |
5ab3f8f9 CH |
2071 | if max_volume_size: |
2072 | self.max_volume_size = int(max_volume_size) | |
2073 | else: | |
2074 | self.max_volume_size = None | |
ae48acc8 | 2075 | |
ea625b04 | 2076 | self.save_to_members = save_to_members |
68ddf955 | 2077 | self.new_volume_handler = new_volume_handler |
7584f5c9 ERE |
2078 | self.closed = False |
2079 | self.members = [] # list of members as TarInfo objects | |
2080 | self._loaded = False # flag if all members have been read | |
2081 | self.offset = self.fileobj.tell() | |
2082 | # current position in the archive file | |
2083 | self.inodes = {} # dictionary caching the inodes of | |
2084 | # archive members already added | |
2085 | ||
2086 | try: | |
2087 | if self.mode == "r": | |
2088 | self.firstmember = None | |
2089 | self.firstmember = self.next() | |
2090 | ||
2091 | if self.mode == "a": | |
2092 | # Move to the end of the archive, | |
2093 | # before the first empty block. | |
2094 | while True: | |
2095 | self.fileobj.seek(self.offset) | |
2096 | try: | |
2097 | tarinfo = self.tarinfo.fromtarfile(self) | |
2098 | self.members.append(tarinfo) | |
2099 | except EOFHeaderError: | |
2100 | self.fileobj.seek(self.offset) | |
2101 | break | |
be60ffd0 | 2102 | except HeaderError as e: |
7584f5c9 ERE |
2103 | raise ReadError(str(e)) |
2104 | ||
2105 | if self.mode in "aw": | |
2106 | self._loaded = True | |
2107 | ||
2108 | if self.pax_headers: | |
2109 | buf = self.tarinfo.create_pax_global_header(self.pax_headers.copy()) | |
2110 | self.fileobj.write(buf) | |
2111 | self.offset += len(buf) | |
2112 | except: | |
2113 | if not self._extfileobj: | |
2114 | self.fileobj.close() | |
2115 | self.closed = True | |
2116 | raise | |
2117 | ||
7584f5c9 ERE |
2118 | #-------------------------------------------------------------------------- |
2119 | # Below are the classmethods which act as alternate constructors to the | |
2120 | # TarFile class. The open() method is the only one that is needed for | |
2121 | # public use; it is the "super"-constructor and is able to select an | |
2122 | # adequate "sub"-constructor for a particular compression using the mapping | |
2123 | # from OPEN_METH. | |
2124 | # | |
2125 | # This concept allows one to subclass TarFile without losing the comfort of | |
2126 | # the super-constructor. A sub-constructor is registered and made available | |
2127 | # by adding it to the mapping in OPEN_METH. | |
2128 | ||
2129 | @classmethod | |
2b82f50c | 2130 | def open(cls, name=None, mode="r", fileobj=None, bufsize=RECORDSIZE, |
04f4c7ab PG |
2131 | encryption=None, compresslevel=9, tolerance=TOLERANCE_STRICT, |
2132 | **kwargs): | |
7584f5c9 ERE |
2133 | """Open a tar archive for reading, writing or appending. Return |
2134 | an appropriate TarFile class. | |
2135 | ||
2136 | mode: | |
2137 | 'r' or 'r:*' open for reading with transparent compression | |
2138 | 'r:' open for reading exclusively uncompressed | |
2139 | 'r:gz' open for reading with gzip compression | |
2140 | 'r:bz2' open for reading with bzip2 compression | |
be60ffd0 | 2141 | 'r:xz' open for reading with lzma compression |
7584f5c9 ERE |
2142 | 'a' or 'a:' open for appending, creating the file if necessary |
2143 | 'w' or 'w:' open for writing without compression | |
2144 | 'w:gz' open for writing with gzip compression | |
2145 | 'w:bz2' open for writing with bzip2 compression | |
be60ffd0 | 2146 | 'w:xz' open for writing with lzma compression |
7584f5c9 ERE |
2147 | |
2148 | 'r|*' open a stream of tar blocks with transparent compression | |
2149 | 'r|' open an uncompressed stream of tar blocks for reading | |
2150 | 'r|gz' open a gzip compressed stream of tar blocks | |
2151 | 'r|bz2' open a bzip2 compressed stream of tar blocks | |
be60ffd0 | 2152 | 'r|xz' open an lzma compressed stream of tar blocks |
7584f5c9 ERE |
2153 | 'w|' open an uncompressed stream for writing |
2154 | 'w|gz' open a gzip compressed stream for writing | |
2155 | 'w|bz2' open a bzip2 compressed stream for writing | |
be60ffd0 | 2156 | 'w|xz' open an lzma compressed stream for writing |
85737f48 ERE |
2157 | |
2158 | 'r#gz' open a stream of gzip compressed tar blocks for reading | |
2159 | 'w#gz' open a stream of gzip compressed tar blocks for writing | |
7584f5c9 | 2160 | """ |
7584f5c9 ERE |
2161 | if not name and not fileobj: |
2162 | raise ValueError("nothing to open") | |
2163 | ||
2164 | if mode in ("r", "r:*"): | |
2165 | # Find out which *open() is appropriate for opening the file. | |
2166 | for comptype in cls.OPEN_METH: | |
2167 | func = getattr(cls, cls.OPEN_METH[comptype]) | |
2168 | if fileobj is not None: | |
2169 | saved_pos = fileobj.tell() | |
2170 | try: | |
2171 | return func(name, "r", fileobj, **kwargs) | |
be60ffd0 | 2172 | except (ReadError, CompressionError) as e: |
c7c736b6 | 2173 | # usually nothing exceptional but sometimes is |
7584f5c9 ERE |
2174 | if fileobj is not None: |
2175 | fileobj.seek(saved_pos) | |
2176 | continue | |
2177 | raise ReadError("file could not be opened successfully") | |
2178 | ||
2179 | elif ":" in mode: | |
2180 | filemode, comptype = mode.split(":", 1) | |
2181 | filemode = filemode or "r" | |
2182 | comptype = comptype or "tar" | |
2183 | ||
2184 | # Select the *open() function according to | |
2185 | # given compression. | |
2186 | if comptype in cls.OPEN_METH: | |
2187 | func = getattr(cls, cls.OPEN_METH[comptype]) | |
2188 | else: | |
2189 | raise CompressionError("unknown compression type %r" % comptype) | |
e05f0440 TJ |
2190 | |
2191 | # Pass on compression level for gzip / bzip2. | |
2192 | if comptype == 'gz' or comptype == 'bz2': | |
2193 | kwargs['compresslevel'] = compresslevel | |
2194 | ||
7a2b9329 CH |
2195 | if 'max_volume_size' in kwargs: |
2196 | if comptype != 'tar' and filemode in 'wa' \ | |
2197 | and kwargs['max_volume_size']: | |
2198 | import warnings | |
2199 | warnings.warn('Only the first volume will be compressed ' | |
2200 | 'for modes with "w:"!') | |
2201 | ||
e05f0440 | 2202 | return func(name, filemode, fileobj, **kwargs) |
7584f5c9 ERE |
2203 | |
2204 | elif "|" in mode: | |
2205 | filemode, comptype = mode.split("|", 1) | |
2206 | filemode = filemode or "r" | |
2207 | comptype = comptype or "tar" | |
2208 | ||
2209 | if filemode not in "rw": | |
2210 | raise ValueError("mode must be 'r' or 'w'") | |
2211 | ||
2212 | t = cls(name, filemode, | |
2b82f50c ERE |
2213 | _Stream(name, filemode, comptype, fileobj, bufsize, |
2214 | compresslevel=compresslevel), | |
7584f5c9 ERE |
2215 | **kwargs) |
2216 | t._extfileobj = False | |
2217 | return t | |
2218 | ||
5fdff89f ERE |
2219 | elif "#" in mode: |
2220 | filemode, comptype = mode.split("#", 1) | |
2221 | filemode = filemode or "r" | |
5fdff89f ERE |
2222 | |
2223 | if filemode not in "rw": | |
5faea0e1 PG |
2224 | raise ValueError ("mode %s not compatible with concat " |
2225 | "archive; must be 'r' or 'w'" % mode) | |
5fdff89f | 2226 | |
be60ffd0 | 2227 | stream = _Stream(name, filemode, comptype, fileobj, bufsize, |
d1c38f40 | 2228 | concat=True, encryption=encryption, |
04f4c7ab | 2229 | compresslevel=compresslevel, tolerance=tolerance) |
d1c38f40 | 2230 | kwargs ["concat"] = True |
be60ffd0 ERE |
2231 | try: |
2232 | t = cls(name, filemode, stream, **kwargs) | |
c7c736b6 | 2233 | except: # XXX except what? |
be60ffd0 | 2234 | stream.close() |
c7c736b6 | 2235 | raise # XXX raise what? |
5fdff89f ERE |
2236 | t._extfileobj = False |
2237 | return t | |
2238 | ||
7584f5c9 ERE |
2239 | elif mode in "aw": |
2240 | return cls.taropen(name, mode, fileobj, **kwargs) | |
2241 | ||
133d30da | 2242 | raise ValueError("undiscernible mode %r" % mode) |
7584f5c9 | 2243 | |
d39d4cbf PG |
2244 | |
2245 | @classmethod | |
2246 | def open_at_offset(cls, offset, *a, **kwa): | |
2247 | """ | |
2248 | Same as ``.open()``, but start reading at the given offset. Assumes a | |
2249 | seekable file object. | |
2250 | """ | |
2251 | fileobj = kwa.get ("fileobj") | |
2252 | if fileobj is not None: | |
2253 | fileobj.seek (offset) | |
2254 | return cls.open (*a, **kwa) | |
2255 | ||
2256 | ||
7584f5c9 ERE |
2257 | @classmethod |
2258 | def taropen(cls, name, mode="r", fileobj=None, **kwargs): | |
2259 | """Open uncompressed tar archive name for reading or writing. | |
2260 | """ | |
2261 | if len(mode) > 1 or mode not in "raw": | |
2262 | raise ValueError("mode must be 'r', 'a' or 'w'") | |
2263 | return cls(name, mode, fileobj, **kwargs) | |
2264 | ||
2265 | @classmethod | |
2266 | def gzopen(cls, name, mode="r", fileobj=None, compresslevel=9, **kwargs): | |
2267 | """Open gzip compressed tar archive name for reading or writing. | |
2268 | Appending is not allowed. | |
2269 | """ | |
2270 | if len(mode) > 1 or mode not in "rw": | |
2271 | raise ValueError("mode must be 'r' or 'w'") | |
2272 | ||
2273 | try: | |
2274 | import gzip | |
2275 | gzip.GzipFile | |
2276 | except (ImportError, AttributeError): | |
2277 | raise CompressionError("gzip module is not available") | |
2278 | ||
be60ffd0 | 2279 | extfileobj = fileobj is not None |
7584f5c9 | 2280 | try: |
be60ffd0 ERE |
2281 | fileobj = gzip.GzipFile(name, mode + "b", compresslevel, fileobj) |
2282 | t = cls.taropen(name, mode, fileobj, **kwargs) | |
2283 | except OSError: | |
2284 | if not extfileobj and fileobj is not None: | |
2285 | fileobj.close() | |
2286 | if fileobj is None: | |
2287 | raise | |
7584f5c9 | 2288 | raise ReadError("not a gzip file") |
be60ffd0 ERE |
2289 | except: |
2290 | if not extfileobj and fileobj is not None: | |
2291 | fileobj.close() | |
2292 | raise | |
2293 | t._extfileobj = extfileobj | |
7584f5c9 ERE |
2294 | return t |
2295 | ||
2296 | @classmethod | |
2297 | def bz2open(cls, name, mode="r", fileobj=None, compresslevel=9, **kwargs): | |
2298 | """Open bzip2 compressed tar archive name for reading or writing. | |
2299 | Appending is not allowed. | |
2300 | """ | |
2301 | if len(mode) > 1 or mode not in "rw": | |
2302 | raise ValueError("mode must be 'r' or 'w'.") | |
2303 | ||
2304 | try: | |
2305 | import bz2 | |
2306 | except ImportError: | |
2307 | raise CompressionError("bz2 module is not available") | |
2308 | ||
be60ffd0 ERE |
2309 | fileobj = bz2.BZ2File(fileobj or name, mode, |
2310 | compresslevel=compresslevel) | |
7584f5c9 ERE |
2311 | |
2312 | try: | |
2313 | t = cls.taropen(name, mode, fileobj, **kwargs) | |
be60ffd0 ERE |
2314 | except (OSError, EOFError): |
2315 | fileobj.close() | |
7584f5c9 ERE |
2316 | raise ReadError("not a bzip2 file") |
2317 | t._extfileobj = False | |
2318 | return t | |
2319 | ||
be60ffd0 ERE |
2320 | @classmethod |
2321 | def xzopen(cls, name, mode="r", fileobj=None, preset=None, **kwargs): | |
2322 | """Open lzma compressed tar archive name for reading or writing. | |
2323 | Appending is not allowed. | |
2324 | """ | |
2325 | if mode not in ("r", "w"): | |
2326 | raise ValueError("mode must be 'r' or 'w'") | |
2327 | ||
2328 | try: | |
2329 | import lzma | |
2330 | except ImportError: | |
2331 | raise CompressionError("lzma module is not available") | |
2332 | ||
2333 | fileobj = lzma.LZMAFile(fileobj or name, mode, preset=preset) | |
2334 | ||
2335 | try: | |
2336 | t = cls.taropen(name, mode, fileobj, **kwargs) | |
2337 | except (lzma.LZMAError, EOFError): | |
2338 | fileobj.close() | |
2339 | raise ReadError("not an lzma file") | |
2340 | t._extfileobj = False | |
2341 | return t | |
2342 | ||
7584f5c9 ERE |
2343 | # All *open() methods are registered here. |
2344 | OPEN_METH = { | |
2345 | "tar": "taropen", # uncompressed tar | |
2346 | "gz": "gzopen", # gzip compressed tar | |
be60ffd0 ERE |
2347 | "bz2": "bz2open", # bzip2 compressed tar |
2348 | "xz": "xzopen" # lzma compressed tar | |
7584f5c9 ERE |
2349 | } |
2350 | ||
2351 | #-------------------------------------------------------------------------- | |
2352 | # The public methods which TarFile provides: | |
2353 | ||
2354 | def close(self): | |
2355 | """Close the TarFile. In write-mode, two finishing zero blocks are | |
fd2f01f2 PG |
2356 | appended to the archive. A special case are empty archives which are |
2357 | initialized accordingly so the two mandatory blocks of zeros are | |
2358 | written abiding by the requested encryption and compression settings. | |
7584f5c9 ERE |
2359 | """ |
2360 | if self.closed: | |
2361 | return | |
2362 | ||
2363 | if self.mode in "aw": | |
fd2f01f2 PG |
2364 | if self.arcmode & ARCMODE_CONCAT and self.fileobj.tell () == 0: |
2365 | self.fileobj.next ("") | |
7584f5c9 ERE |
2366 | self.fileobj.write(NUL * (BLOCKSIZE * 2)) |
2367 | self.offset += (BLOCKSIZE * 2) | |
2368 | # fill up the end with zero-blocks | |
2369 | # (like option -b20 for tar does) | |
2370 | blocks, remainder = divmod(self.offset, RECORDSIZE) | |
2371 | if remainder > 0: | |
2372 | self.fileobj.write(NUL * (RECORDSIZE - remainder)) | |
7584f5c9 ERE |
2373 | if not self._extfileobj: |
2374 | self.fileobj.close() | |
2375 | self.closed = True | |
2376 | ||
2377 | def getmember(self, name): | |
2378 | """Return a TarInfo object for member `name'. If `name' can not be | |
2379 | found in the archive, KeyError is raised. If a member occurs more | |
2380 | than once in the archive, its last occurrence is assumed to be the | |
2381 | most up-to-date version. | |
2382 | """ | |
2383 | tarinfo = self._getmember(name) | |
2384 | if tarinfo is None: | |
2385 | raise KeyError("filename %r not found" % name) | |
2386 | return tarinfo | |
2387 | ||
2388 | def getmembers(self): | |
2389 | """Return the members of the archive as a list of TarInfo objects. The | |
2390 | list has the same order as the members in the archive. | |
2391 | """ | |
2392 | self._check() | |
2393 | if not self._loaded: # if we want to obtain a list of | |
2394 | self._load() # all members, we first have to | |
2395 | # scan the whole archive. | |
2396 | return self.members | |
2397 | ||
ad4402e8 ERE |
2398 | def get_last_member_offset(self): |
2399 | """Return the last member offset. Usually this is self.fileobj.tell(), | |
2400 | but when there's encryption or concat compression going on it's more | |
2401 | complicated than that. | |
2402 | """ | |
b8fc2f5d | 2403 | return self.last_block_offset |
ad4402e8 | 2404 | |
7584f5c9 ERE |
2405 | def getnames(self): |
2406 | """Return the members of the archive as a list of their names. It has | |
2407 | the same order as the list returned by getmembers(). | |
2408 | """ | |
2409 | return [tarinfo.name for tarinfo in self.getmembers()] | |
2410 | ||
2411 | def gettarinfo(self, name=None, arcname=None, fileobj=None): | |
2412 | """Create a TarInfo object for either the file `name' or the file | |
2413 | object `fileobj' (using os.fstat on its file descriptor). You can | |
2414 | modify some of the TarInfo's attributes before you add it using | |
2415 | addfile(). If given, `arcname' specifies an alternative name for the | |
2416 | file in the archive. | |
2417 | """ | |
2418 | self._check("aw") | |
2419 | ||
2420 | # When fileobj is given, replace name by | |
2421 | # fileobj's real name. | |
2422 | if fileobj is not None: | |
2423 | name = fileobj.name | |
2424 | ||
2425 | # Building the name of the member in the archive. | |
2426 | # Backward slashes are converted to forward slashes, | |
2427 | # Absolute paths are turned to relative paths. | |
2428 | if arcname is None: | |
2429 | arcname = name | |
2430 | drv, arcname = os.path.splitdrive(arcname) | |
be60ffd0 | 2431 | arcname = arcname.replace(os.sep, "/") |
7584f5c9 ERE |
2432 | arcname = arcname.lstrip("/") |
2433 | ||
2434 | # Now, fill the TarInfo object with | |
2435 | # information specific for the file. | |
2436 | tarinfo = self.tarinfo() | |
2437 | tarinfo.tarfile = self | |
2438 | ||
2439 | # Use os.stat or os.lstat, depending on platform | |
2440 | # and if symlinks shall be resolved. | |
2441 | if fileobj is None: | |
2442 | if hasattr(os, "lstat") and not self.dereference: | |
2443 | statres = os.lstat(name) | |
2444 | else: | |
2445 | statres = os.stat(name) | |
2446 | else: | |
2447 | statres = os.fstat(fileobj.fileno()) | |
2448 | linkname = "" | |
2449 | ||
2450 | stmd = statres.st_mode | |
2451 | if stat.S_ISREG(stmd): | |
2452 | inode = (statres.st_ino, statres.st_dev) | |
2453 | if not self.dereference and statres.st_nlink > 1 and \ | |
2454 | inode in self.inodes and arcname != self.inodes[inode]: | |
2455 | # Is it a hardlink to an already | |
2456 | # archived file? | |
2457 | type = LNKTYPE | |
2458 | linkname = self.inodes[inode] | |
2459 | else: | |
2460 | # The inode is added only if its valid. | |
2461 | # For win32 it is always 0. | |
2462 | type = REGTYPE | |
6f422b65 | 2463 | if inode[0] and self.save_to_members: |
7584f5c9 ERE |
2464 | self.inodes[inode] = arcname |
2465 | elif stat.S_ISDIR(stmd): | |
2466 | type = DIRTYPE | |
2467 | elif stat.S_ISFIFO(stmd): | |
2468 | type = FIFOTYPE | |
2469 | elif stat.S_ISLNK(stmd): | |
2470 | type = SYMTYPE | |
2471 | linkname = os.readlink(name) | |
2472 | elif stat.S_ISCHR(stmd): | |
2473 | type = CHRTYPE | |
2474 | elif stat.S_ISBLK(stmd): | |
2475 | type = BLKTYPE | |
2476 | else: | |
2477 | return None | |
2478 | ||
2479 | # Fill the TarInfo object with all | |
2480 | # information we can get. | |
2481 | tarinfo.name = arcname | |
2482 | tarinfo.mode = stmd | |
2483 | tarinfo.uid = statres.st_uid | |
2484 | tarinfo.gid = statres.st_gid | |
2485 | if type == REGTYPE: | |
2486 | tarinfo.size = statres.st_size | |
2487 | else: | |
be60ffd0 | 2488 | tarinfo.size = 0 |
7584f5c9 ERE |
2489 | tarinfo.mtime = statres.st_mtime |
2490 | tarinfo.type = type | |
2491 | tarinfo.linkname = linkname | |
2492 | if pwd: | |
9ef1fb87 TJ |
2493 | if tarinfo.uid in self.cache_uid2user: |
2494 | tarinfo.uname = self.cache_uid2user[tarinfo.uid] | |
2495 | else: | |
2496 | try: | |
2497 | tarinfo.uname = pwd.getpwuid(tarinfo.uid)[0] | |
2498 | self.cache_uid2user[tarinfo.uid] = tarinfo.uname | |
2499 | except KeyError: | |
2500 | # remember user does not exist: | |
2501 | # same default value as in tarinfo class | |
2502 | self.cache_uid2user[tarinfo.uid] = "" | |
7584f5c9 | 2503 | if grp: |
9ef1fb87 TJ |
2504 | if tarinfo.gid in self.cache_gid2group: |
2505 | tarinfo.gname = self.cache_gid2group[tarinfo.gid] | |
2506 | else: | |
2507 | try: | |
2508 | tarinfo.gname = grp.getgrgid(tarinfo.gid)[0] | |
2509 | self.cache_gid2group[tarinfo.gid] = tarinfo.gname | |
2510 | except KeyError: | |
2511 | # remember group does not exist: | |
2512 | # same default value as in tarinfo class | |
2513 | self.cache_gid2group[tarinfo.gid] = "" | |
7584f5c9 ERE |
2514 | |
2515 | if type in (CHRTYPE, BLKTYPE): | |
2516 | if hasattr(os, "major") and hasattr(os, "minor"): | |
2517 | tarinfo.devmajor = os.major(statres.st_rdev) | |
2518 | tarinfo.devminor = os.minor(statres.st_rdev) | |
2519 | return tarinfo | |
2520 | ||
2521 | def list(self, verbose=True): | |
2522 | """Print a table of contents to sys.stdout. If `verbose' is False, only | |
2523 | the names of the members are printed. If it is True, an `ls -l'-like | |
2524 | output is produced. | |
2525 | """ | |
2526 | self._check() | |
2527 | ||
2528 | for tarinfo in self: | |
2529 | if verbose: | |
be60ffd0 ERE |
2530 | print(stat.filemode(tarinfo.mode), end=' ') |
2531 | print("%s/%s" % (tarinfo.uname or tarinfo.uid, | |
2532 | tarinfo.gname or tarinfo.gid), end=' ') | |
7584f5c9 | 2533 | if tarinfo.ischr() or tarinfo.isblk(): |
be60ffd0 ERE |
2534 | print("%10s" % ("%d,%d" \ |
2535 | % (tarinfo.devmajor, tarinfo.devminor)), end=' ') | |
7584f5c9 | 2536 | else: |
be60ffd0 ERE |
2537 | print("%10d" % tarinfo.size, end=' ') |
2538 | print("%d-%02d-%02d %02d:%02d:%02d" \ | |
2539 | % time.localtime(tarinfo.mtime)[:6], end=' ') | |
7584f5c9 | 2540 | |
be60ffd0 | 2541 | print(tarinfo.name + ("/" if tarinfo.isdir() else ""), end=' ') |
7584f5c9 ERE |
2542 | |
2543 | if verbose: | |
2544 | if tarinfo.issym(): | |
be60ffd0 | 2545 | print("->", tarinfo.linkname, end=' ') |
7584f5c9 | 2546 | if tarinfo.islnk(): |
be60ffd0 ERE |
2547 | print("link to", tarinfo.linkname, end=' ') |
2548 | print() | |
7584f5c9 | 2549 | |
be60ffd0 | 2550 | def add(self, name, arcname=None, recursive=True, exclude=None, *, filter=None): |
7584f5c9 ERE |
2551 | """Add the file `name' to the archive. `name' may be any type of file |
2552 | (directory, fifo, symbolic link, etc.). If given, `arcname' | |
2553 | specifies an alternative name for the file in the archive. | |
2554 | Directories are added recursively by default. This can be avoided by | |
2555 | setting `recursive' to False. `exclude' is a function that should | |
2556 | return True for each filename to be excluded. `filter' is a function | |
2557 | that expects a TarInfo object argument and returns the changed | |
2558 | TarInfo object, if it returns None the TarInfo object will be | |
2559 | excluded from the archive. | |
2560 | """ | |
2561 | self._check("aw") | |
2562 | ||
2563 | if arcname is None: | |
2564 | arcname = name | |
2565 | ||
2566 | # Exclude pathnames. | |
2567 | if exclude is not None: | |
2568 | import warnings | |
2569 | warnings.warn("use the filter argument instead", | |
2570 | DeprecationWarning, 2) | |
2571 | if exclude(name): | |
2572 | self._dbg(2, "tarfile: Excluded %r" % name) | |
2573 | return | |
2574 | ||
2575 | # Skip if somebody tries to archive the archive... | |
2576 | if self.name is not None and os.path.abspath(name) == self.name: | |
2577 | self._dbg(2, "tarfile: Skipped %r" % name) | |
2578 | return | |
2579 | ||
2580 | self._dbg(1, name) | |
2581 | ||
2582 | # Create a TarInfo object from the file. | |
2583 | tarinfo = self.gettarinfo(name, arcname) | |
2584 | ||
2585 | if tarinfo is None: | |
2586 | self._dbg(1, "tarfile: Unsupported type %r" % name) | |
2587 | return | |
2588 | ||
2589 | # Change or exclude the TarInfo object. | |
2590 | if filter is not None: | |
2591 | tarinfo = filter(tarinfo) | |
2592 | if tarinfo is None: | |
2593 | self._dbg(2, "tarfile: Excluded %r" % name) | |
2594 | return | |
2595 | ||
2596 | # Append the tar header and data to the archive. | |
2597 | if tarinfo.isreg(): | |
2598 | with bltn_open(name, "rb") as f: | |
2599 | self.addfile(tarinfo, f) | |
2600 | ||
2601 | elif tarinfo.isdir(): | |
2602 | self.addfile(tarinfo) | |
2603 | if recursive: | |
2604 | for f in os.listdir(name): | |
2605 | self.add(os.path.join(name, f), os.path.join(arcname, f), | |
be60ffd0 | 2606 | recursive, exclude, filter=filter) |
7584f5c9 ERE |
2607 | |
2608 | else: | |
2609 | self.addfile(tarinfo) | |
2610 | ||
defc9a22 | 2611 | def _size_left_file(self): |
be60ffd0 | 2612 | """Calculates size left in a volume with a maximum volume size. |
ba5a449e | 2613 | |
be60ffd0 | 2614 | Assumes self.max_volume_size is set. |
ba5a449e | 2615 | If using compression through a _Stream, use _size_left_stream instead |
be60ffd0 | 2616 | """ |
ba5a449e | 2617 | # left-over size = max_size - offset - 2 zero-blocks written in close |
ae48acc8 ERE |
2618 | size_left = self.max_volume_size - 2*BLOCKSIZE - self.offset |
2619 | # limit size left to a discrete number of blocks, because we won't | |
be60ffd0 | 2620 | # write only half a block when writting the end of a volume |
ae48acc8 | 2621 | # and filling with zeros |
defc9a22 CH |
2622 | return BLOCKSIZE * (size_left // BLOCKSIZE) |
2623 | ||
2624 | def _size_left_stream(self): | |
ba5a449e CH |
2625 | """ Calculates size left in a volume if using comression/encryption |
2626 | ||
2627 | Assumes self.max_volume_size is set and self.fileobj is a _Stream | |
2628 | (otherwise use _size_left_file) | |
2629 | """ | |
2630 | # left-over size = max_size - bytes written - 2 zero-blocks (close) | |
defc9a22 CH |
2631 | size_left = self.max_volume_size - self.fileobj.estim_file_size() \ |
2632 | - 2*BLOCKSIZE | |
2633 | return BLOCKSIZE * (size_left // BLOCKSIZE) | |
ae48acc8 | 2634 | |
7584f5c9 ERE |
2635 | def addfile(self, tarinfo, fileobj=None): |
2636 | """Add the TarInfo object `tarinfo' to the archive. If `fileobj' is | |
2637 | given, tarinfo.size bytes are read from it and added to the archive. | |
2638 | You can create TarInfo objects using gettarinfo(). | |
2639 | On Windows platforms, `fileobj' should always be opened with mode | |
2640 | 'rb' to avoid irritation about the file size. | |
2641 | """ | |
2642 | self._check("aw") | |
2643 | ||
2644 | tarinfo = copy.copy(tarinfo) | |
cbf55ffb | 2645 | |
d1c38f40 PG |
2646 | if self.arcmode & ARCMODE_CONCAT: |
2647 | self.last_block_offset = self.fileobj.next (tarinfo.name) | |
11684b1d ERE |
2648 | else: |
2649 | self.last_block_offset = self.fileobj.tell() | |
7584f5c9 ERE |
2650 | |
2651 | buf = tarinfo.tobuf(self.format, self.encoding, self.errors) | |
2652 | self.fileobj.write(buf) | |
2653 | self.offset += len(buf) | |
2654 | ||
ae9c8de2 CH |
2655 | if self.max_volume_size: |
2656 | if isinstance(self.fileobj, _Stream): | |
2657 | _size_left = self._size_left_stream | |
2658 | else: | |
2659 | _size_left = self._size_left_file | |
2660 | else: | |
2661 | _size_left = lambda: tarinfo.size | |
68ddf955 | 2662 | |
29c354ac PG |
2663 | # If there's no data to follow, finish |
2664 | if not fileobj: | |
29c354ac PG |
2665 | if self.save_to_members: |
2666 | self.members.append(tarinfo) | |
2667 | return | |
2668 | ||
2669 | target_size_left = _size_left() | |
2670 | source_size_left = tarinfo.size | |
2671 | assert tarinfo.volume_offset == 0 | |
2672 | ||
2673 | # we only split volumes in the middle of a file, that means we have | |
2674 | # to write at least one block | |
2675 | if target_size_left < BLOCKSIZE: | |
2676 | target_size_left = BLOCKSIZE | |
2677 | ||
ae9c8de2 CH |
2678 | # loop over multiple volumes |
2679 | while source_size_left > 0: | |
ae48acc8 | 2680 | |
ae9c8de2 CH |
2681 | # Write as much data as possble from source into target. |
2682 | # When compressing data, we cannot easily predict how much data we | |
2683 | # can write until target_size_left == 0 --> need to iterate | |
2684 | size_can_write = min(target_size_left, source_size_left) | |
c04e0751 | 2685 | |
ae9c8de2 CH |
2686 | while size_can_write > 0: |
2687 | copyfileobj(fileobj, self.fileobj, size_can_write) | |
2688 | self.offset += size_can_write | |
2689 | source_size_left -= size_can_write | |
2690 | target_size_left = _size_left() | |
2691 | size_can_write = min(target_size_left, source_size_left) | |
68ddf955 | 2692 | |
ae9c8de2 CH |
2693 | # now target_size_left == 0 or source_size_left == 0 |
2694 | ||
2695 | # if there is data left to write, we need to create a new volume | |
2696 | if source_size_left > 0: | |
5f38bff6 PG |
2697 | # Only finalize the crypto entry here if we’re continuing with |
2698 | # another one; otherwise, the encryption must include the block | |
2699 | # padding below. | |
2f854e77 | 2700 | tarinfo.type = GNUTYPE_MULTIVOL |
68ddf955 ERE |
2701 | |
2702 | if not self.new_volume_handler or\ | |
2703 | not callable(self.new_volume_handler): | |
c04e0751 | 2704 | raise Exception("We need to create a new volume and you " |
ae9c8de2 | 2705 | "didn't supply a new_volume_handler") |
68ddf955 | 2706 | |
54128a00 | 2707 | |
68ddf955 ERE |
2708 | # the new volume handler should do everything needed to |
2709 | # start working in a new volume. usually, the handler calls | |
2710 | # to self.open_volume | |
2f854e77 | 2711 | self.volume_number += 1 |
0eb5048f | 2712 | |
ae9c8de2 | 2713 | # set to be used by open_volume, because in the case of a PAX |
0eb5048f ERE |
2714 | # tar it needs to write information about the volume and offset |
2715 | # in the global header | |
ae9c8de2 | 2716 | tarinfo.volume_offset = tarinfo.size - source_size_left |
0eb5048f | 2717 | self.volume_tarinfo = tarinfo |
ae9c8de2 | 2718 | |
a0873dcc PG |
2719 | # the “new_volume_handler” is supposed to call .close() on the |
2720 | # “fileobj” _Stream | |
2f854e77 ERE |
2721 | self.new_volume_handler(self, self.base_name, self.volume_number) |
2722 | ||
0eb5048f ERE |
2723 | self.volume_tarinfo = None |
2724 | ||
d1c38f40 PG |
2725 | if self.arcmode & ARCMODE_CONCAT: |
2726 | self.fileobj.next_volume (tarinfo.name) | |
5f38bff6 | 2727 | |
2f854e77 ERE |
2728 | # write new volume header |
2729 | buf = tarinfo.tobuf(self.format, self.encoding, self.errors) | |
2f854e77 | 2730 | self.fileobj.write(buf) |
ae9c8de2 CH |
2731 | self.offset += len(buf) |
2732 | ||
2733 | # adjust variables; open_volume should have reset self.offset | |
2734 | # --> _size_left should be big again | |
2735 | target_size_left = _size_left() | |
2736 | size_can_write = min(target_size_left, source_size_left) | |
e0da4709 | 2737 | self._dbg(3, 'new volume') |
ae9c8de2 CH |
2738 | |
2739 | # now, all data has been written. We may have to fill up the rest of | |
2740 | # the block in target with 0s | |
2741 | remainder = (tarinfo.size - tarinfo.volume_offset) % BLOCKSIZE | |
2742 | if remainder > 0: | |
2743 | self.fileobj.write(NUL * (BLOCKSIZE - remainder)) | |
2744 | self.offset += BLOCKSIZE - remainder | |
7584f5c9 | 2745 | |
ea625b04 ERE |
2746 | if self.save_to_members: |
2747 | self.members.append(tarinfo) | |
7584f5c9 | 2748 | |
170c6c52 | 2749 | def open_volume(self, name="", fileobj=None, encryption=None): |
68ddf955 | 2750 | ''' |
0eb5048f | 2751 | Called by the user to change this tar file to point to a new volume. |
68ddf955 | 2752 | ''' |
27ee4dd4 | 2753 | |
68ddf955 ERE |
2754 | # open the file using either fileobj or name |
2755 | if not fileobj: | |
2756 | if self.mode == "a" and not os.path.exists(name): | |
2757 | # Create nonexistent files in append mode. | |
2758 | self.mode = "w" | |
2759 | self._mode = "wb" | |
68ddf955 | 2760 | self._extfileobj = False |
26fa5ad5 ERE |
2761 | |
2762 | if isinstance(self.fileobj, _Stream): | |
e0da4709 | 2763 | self._dbg(3, 'open_volume: create a _Stream') |
26fa5ad5 ERE |
2764 | fileobj = _Stream(name=name, |
2765 | mode=self.fileobj.mode, | |
2766 | comptype=self.fileobj.comptype, | |
2767 | fileobj=None, | |
2768 | bufsize=self.fileobj.bufsize, | |
cea130ec | 2769 | encryption=encryption or self.fileobj.encryption, |
27ee4dd4 PG |
2770 | concat=self.fileobj.arcmode & ARCMODE_CONCAT, |
2771 | tolerance=self.fileobj.tolerance) | |
26fa5ad5 | 2772 | else: |
7a2b9329 | 2773 | # here, we lose information about compression/encryption! |
e0da4709 | 2774 | self._dbg(3, 'open_volume: builtin open') |
26fa5ad5 | 2775 | fileobj = bltn_open(name, self._mode) |
68ddf955 ERE |
2776 | else: |
2777 | if name is None and hasattr(fileobj, "name"): | |
2778 | name = fileobj.name | |
2779 | if hasattr(fileobj, "mode"): | |
2780 | self._mode = fileobj.mode | |
2781 | self._extfileobj = True | |
1027433a | 2782 | self._dbg(3, 'open_volume: using external fileobj {}', fileobj) |
68ddf955 ERE |
2783 | self.name = os.path.abspath(name) if name else None |
2784 | self.fileobj = fileobj | |
2785 | ||
2786 | # init data structures | |
2787 | self.closed = False | |
2788 | self.members = [] # list of members as TarInfo objects | |
2789 | self._loaded = False # flag if all members have been read | |
2790 | self.offset = self.fileobj.tell() | |
2791 | # current position in the archive file | |
2792 | self.inodes = {} # dictionary caching the inodes of | |
2793 | # archive members already added | |
2794 | ||
2795 | try: | |
2796 | if self.mode == "r": | |
2797 | self.firstmember = None | |
2798 | self.firstmember = self.next() | |
2799 | ||
2800 | if self.mode == "a": | |
2801 | # Move to the end of the archive, | |
2802 | # before the first empty block. | |
2803 | while True: | |
2804 | self.fileobj.seek(self.offset) | |
2805 | try: | |
2806 | tarinfo = self.tarinfo.fromtarfile(self) | |
2807 | self.members.append(tarinfo) | |
2808 | except EOFHeaderError: | |
2809 | self.fileobj.seek(self.offset) | |
2810 | break | |
be60ffd0 | 2811 | except HeaderError as e: |
68ddf955 ERE |
2812 | raise ReadError(str(e)) |
2813 | ||
2814 | if self.mode in "aw": | |
2815 | self._loaded = True | |
2816 | ||
c04e0751 ERE |
2817 | if self.format == PAX_FORMAT: |
2818 | volume_info = { | |
be60ffd0 ERE |
2819 | "GNU.volume.filename": str(self.volume_tarinfo.name), |
2820 | "GNU.volume.size": str(self.volume_tarinfo.size - self.volume_tarinfo.volume_offset), | |
2821 | "GNU.volume.offset": str(self.volume_tarinfo.volume_offset), | |
c04e0751 | 2822 | } |
0eb5048f | 2823 | |
c04e0751 ERE |
2824 | self.pax_headers.update(volume_info) |
2825 | ||
a0873dcc PG |
2826 | if isinstance(self.fileobj, _Stream): |
2827 | self.fileobj._init_write_gz () | |
c04e0751 ERE |
2828 | buf = self.tarinfo.create_pax_global_header(volume_info.copy()) |
2829 | self.fileobj.write(buf) | |
2830 | self.offset += len(buf) | |
54128a00 | 2831 | except Exception as exn: |
68ddf955 ERE |
2832 | if not self._extfileobj: |
2833 | self.fileobj.close() | |
2834 | self.closed = True | |
2835 | raise | |
2836 | ||
e5f5681b | 2837 | def extractall(self, path=".", members=None, filter=None): |
7584f5c9 ERE |
2838 | """Extract all members from the archive to the current working |
2839 | directory and set owner, modification time and permissions on | |
2840 | directories afterwards. `path' specifies a different directory | |
2841 | to extract to. `members' is optional and must be a subset of the | |
2842 | list returned by getmembers(). | |
2843 | """ | |
2844 | directories = [] | |
2845 | ||
2846 | if members is None: | |
2847 | members = self | |
2848 | ||
2849 | for tarinfo in members: | |
c474439c ERE |
2850 | if self.volume_number > 0 and tarinfo.ismultivol(): |
2851 | continue | |
2852 | ||
974408b5 | 2853 | if filter and not filter(tarinfo): |
e5f5681b ERE |
2854 | continue |
2855 | ||
7584f5c9 ERE |
2856 | if tarinfo.isdir(): |
2857 | # Extract directories with a safe mode. | |
2858 | directories.append(tarinfo) | |
2859 | tarinfo = copy.copy(tarinfo) | |
be60ffd0 ERE |
2860 | tarinfo.mode = 0o0700 |
2861 | # Do not set_attrs directories, as we will do that further down | |
2862 | self.extract(tarinfo, path, set_attrs=not tarinfo.isdir()) | |
7584f5c9 ERE |
2863 | |
2864 | # Reverse sort directories. | |
be60ffd0 | 2865 | directories.sort(key=lambda a: a.name) |
7584f5c9 ERE |
2866 | directories.reverse() |
2867 | ||
2868 | # Set correct owner, mtime and filemode on directories. | |
2869 | for tarinfo in directories: | |
2870 | dirpath = os.path.join(path, tarinfo.name) | |
2871 | try: | |
2872 | self.chown(tarinfo, dirpath) | |
2873 | self.utime(tarinfo, dirpath) | |
2874 | self.chmod(tarinfo, dirpath) | |
be60ffd0 | 2875 | except ExtractError as e: |
7584f5c9 ERE |
2876 | if self.errorlevel > 1: |
2877 | raise | |
2878 | else: | |
2879 | self._dbg(1, "tarfile: %s" % e) | |
2880 | ||
786addd6 | 2881 | def extract(self, member, path="", set_attrs=True, symlink_cb=None): |
7584f5c9 ERE |
2882 | """Extract a member from the archive to the current working directory, |
2883 | using its full name. Its file information is extracted as accurately | |
2884 | as possible. `member' may be a filename or a TarInfo object. You can | |
be60ffd0 ERE |
2885 | specify a different directory using `path'. File attributes (owner, |
2886 | mtime, mode) are set unless `set_attrs' is False. | |
786addd6 PG |
2887 | ``symlink_cb`` is a hook accepting a function that is passed the |
2888 | ``member``, ``path``, and ``set_attrs`` arguments if the tarinfo for | |
2889 | ``member`` indicates a symlink in which case only the callback | |
9b13f5c4 PG |
2890 | passed will be applied, skipping the actual extraction. In case the |
2891 | callback is invoked, its return value is passed on to the caller. | |
7584f5c9 ERE |
2892 | """ |
2893 | self._check("r") | |
2894 | ||
be60ffd0 | 2895 | if isinstance(member, str): |
7584f5c9 ERE |
2896 | tarinfo = self.getmember(member) |
2897 | else: | |
2898 | tarinfo = member | |
2899 | ||
2900 | # Prepare the link target for makelink(). | |
2901 | if tarinfo.islnk(): | |
2902 | tarinfo._link_target = os.path.join(path, tarinfo.linkname) | |
2903 | ||
9b13f5c4 | 2904 | if symlink_cb is not None and tarinfo.issym(): |
83f5fd71 | 2905 | return symlink_cb(member, path, set_attrs) |
786addd6 | 2906 | |
7584f5c9 | 2907 | try: |
be60ffd0 ERE |
2908 | self._extract_member(tarinfo, os.path.join(path, tarinfo.name), |
2909 | set_attrs=set_attrs) | |
2910 | except EnvironmentError as e: | |
7584f5c9 ERE |
2911 | if self.errorlevel > 0: |
2912 | raise | |
2913 | else: | |
2914 | if e.filename is None: | |
2915 | self._dbg(1, "tarfile: %s" % e.strerror) | |
2916 | else: | |
2917 | self._dbg(1, "tarfile: %s %r" % (e.strerror, e.filename)) | |
be60ffd0 | 2918 | except ExtractError as e: |
7584f5c9 ERE |
2919 | if self.errorlevel > 1: |
2920 | raise | |
2921 | else: | |
2922 | self._dbg(1, "tarfile: %s" % e) | |
2923 | ||
2924 | def extractfile(self, member): | |
2925 | """Extract a member from the archive as a file object. `member' may be | |
be60ffd0 ERE |
2926 | a filename or a TarInfo object. If `member' is a regular file or a |
2927 | link, an io.BufferedReader object is returned. Otherwise, None is | |
2928 | returned. | |
7584f5c9 ERE |
2929 | """ |
2930 | self._check("r") | |
2931 | ||
be60ffd0 | 2932 | if isinstance(member, str): |
7584f5c9 ERE |
2933 | tarinfo = self.getmember(member) |
2934 | else: | |
2935 | tarinfo = member | |
2936 | ||
be60ffd0 ERE |
2937 | if tarinfo.isreg() or tarinfo.ismultivol() or\ |
2938 | tarinfo.type not in SUPPORTED_TYPES: | |
7584f5c9 ERE |
2939 | # If a member's type is unknown, it is treated as a |
2940 | # regular file. | |
2941 | return self.fileobject(self, tarinfo) | |
2942 | ||
2943 | elif tarinfo.islnk() or tarinfo.issym(): | |
2944 | if isinstance(self.fileobj, _Stream): | |
2945 | # A small but ugly workaround for the case that someone tries | |
2946 | # to extract a (sym)link as a file-object from a non-seekable | |
2947 | # stream of tar blocks. | |
2948 | raise StreamError("cannot extract (sym)link as file object") | |
2949 | else: | |
2950 | # A (sym)link's file object is its target's file object. | |
2951 | return self.extractfile(self._find_link_target(tarinfo)) | |
2952 | else: | |
2953 | # If there's no data associated with the member (directory, chrdev, | |
2954 | # blkdev, etc.), return None instead of a file object. | |
2955 | return None | |
2956 | ||
be60ffd0 | 2957 | def _extract_member(self, tarinfo, targetpath, set_attrs=True): |
7584f5c9 ERE |
2958 | """Extract the TarInfo object tarinfo to a physical |
2959 | file called targetpath. | |
2960 | """ | |
2961 | # Fetch the TarInfo object for the given name | |
2962 | # and build the destination pathname, replacing | |
2963 | # forward slashes to platform specific separators. | |
2964 | targetpath = targetpath.rstrip("/") | |
2965 | targetpath = targetpath.replace("/", os.sep) | |
2966 | ||
2967 | # Create all upper directories. | |
2968 | upperdirs = os.path.dirname(targetpath) | |
2969 | if upperdirs and not os.path.exists(upperdirs): | |
2970 | # Create directories that are not part of the archive with | |
2971 | # default permissions. | |
2972 | os.makedirs(upperdirs) | |
2973 | ||
2974 | if tarinfo.islnk() or tarinfo.issym(): | |
2975 | self._dbg(1, "%s -> %s" % (tarinfo.name, tarinfo.linkname)) | |
2976 | else: | |
2977 | self._dbg(1, tarinfo.name) | |
2978 | ||
2979 | if tarinfo.isreg(): | |
2980 | self.makefile(tarinfo, targetpath) | |
2981 | elif tarinfo.isdir(): | |
2982 | self.makedir(tarinfo, targetpath) | |
2983 | elif tarinfo.isfifo(): | |
2984 | self.makefifo(tarinfo, targetpath) | |
2985 | elif tarinfo.ischr() or tarinfo.isblk(): | |
2986 | self.makedev(tarinfo, targetpath) | |
2987 | elif tarinfo.islnk() or tarinfo.issym(): | |
2988 | self.makelink(tarinfo, targetpath) | |
2989 | elif tarinfo.type not in SUPPORTED_TYPES: | |
2990 | self.makeunknown(tarinfo, targetpath) | |
2991 | else: | |
2992 | self.makefile(tarinfo, targetpath) | |
2993 | ||
be60ffd0 ERE |
2994 | if set_attrs: |
2995 | self.chown(tarinfo, targetpath) | |
2996 | if not tarinfo.issym(): | |
2997 | self.chmod(tarinfo, targetpath) | |
2998 | self.utime(tarinfo, targetpath) | |
7584f5c9 ERE |
2999 | |
3000 | #-------------------------------------------------------------------------- | |
3001 | # Below are the different file methods. They are called via | |
3002 | # _extract_member() when extract() is called. They can be replaced in a | |
3003 | # subclass to implement other functionality. | |
3004 | ||
3005 | def makedir(self, tarinfo, targetpath): | |
3006 | """Make a directory called targetpath. | |
3007 | """ | |
3008 | try: | |
3009 | # Use a safe mode for the directory, the real mode is set | |
3010 | # later in _extract_member(). | |
be60ffd0 ERE |
3011 | os.mkdir(targetpath, 0o0700) |
3012 | except FileExistsError: | |
3013 | pass | |
7584f5c9 ERE |
3014 | |
3015 | def makefile(self, tarinfo, targetpath): | |
3016 | """Make a file called targetpath. | |
3017 | """ | |
be60ffd0 ERE |
3018 | source = self.fileobj |
3019 | source.seek(tarinfo.offset_data) | |
c7c736b6 | 3020 | decrypt = False |
c474439c ERE |
3021 | iterate = True |
3022 | target = bltn_open(targetpath, "wb") | |
3023 | ||
be60ffd0 ERE |
3024 | if tarinfo.sparse is not None: |
3025 | try: | |
3026 | for offset, size in tarinfo.sparse: | |
3027 | target.seek(offset) | |
3028 | copyfileobj(source, target, size) | |
3029 | target.seek(tarinfo.size) | |
3030 | target.truncate() | |
3031 | finally: | |
3032 | target.close() | |
3033 | return | |
3034 | ||
c474439c ERE |
3035 | while iterate: |
3036 | iterate = False | |
3037 | try: | |
3038 | copyfileobj(source, target, tarinfo.size) | |
aa828cd1 | 3039 | except OSError: |
c474439c ERE |
3040 | source.close() |
3041 | # only if we are extracting a multivolume this can be treated | |
3042 | if not self.new_volume_handler: | |
3043 | target.close() | |
3044 | raise Exception("We need to read a new volume and you" | |
3045 | " didn't supply a new_volume_handler") | |
3046 | ||
3047 | # the new volume handler should do everything needed to | |
3048 | # start working in a new volume. usually, the handler calls | |
3049 | # to self.open_volume | |
3050 | self.volume_number += 1 | |
3051 | self.new_volume_handler(self, self.base_name, self.volume_number) | |
be60ffd0 ERE |
3052 | tarinfo = self.firstmember |
3053 | source = self.fileobj | |
c474439c | 3054 | iterate = True |
c474439c ERE |
3055 | target.close() |
3056 | ||
7584f5c9 ERE |
3057 | |
3058 | def makeunknown(self, tarinfo, targetpath): | |
3059 | """Make a file from a TarInfo object with an unknown type | |
3060 | at targetpath. | |
3061 | """ | |
3062 | self.makefile(tarinfo, targetpath) | |
3063 | self._dbg(1, "tarfile: Unknown file type %r, " \ | |
3064 | "extracted as regular file." % tarinfo.type) | |
3065 | ||
3066 | def makefifo(self, tarinfo, targetpath): | |
3067 | """Make a fifo called targetpath. | |
3068 | """ | |
3069 | if hasattr(os, "mkfifo"): | |
3070 | os.mkfifo(targetpath) | |
3071 | else: | |
3072 | raise ExtractError("fifo not supported by system") | |
3073 | ||
3074 | def makedev(self, tarinfo, targetpath): | |
3075 | """Make a character or block device called targetpath. | |
3076 | """ | |
3077 | if not hasattr(os, "mknod") or not hasattr(os, "makedev"): | |
3078 | raise ExtractError("special devices not supported by system") | |
3079 | ||
3080 | mode = tarinfo.mode | |
3081 | if tarinfo.isblk(): | |
3082 | mode |= stat.S_IFBLK | |
3083 | else: | |
3084 | mode |= stat.S_IFCHR | |
3085 | ||
3086 | os.mknod(targetpath, mode, | |
3087 | os.makedev(tarinfo.devmajor, tarinfo.devminor)) | |
3088 | ||
3089 | def makelink(self, tarinfo, targetpath): | |
3090 | """Make a (symbolic) link called targetpath. If it cannot be created | |
3091 | (platform limitation), we try to make a copy of the referenced file | |
3092 | instead of a link. | |
3093 | """ | |
be60ffd0 | 3094 | try: |
7584f5c9 ERE |
3095 | # For systems that support symbolic and hard links. |
3096 | if tarinfo.issym(): | |
7584f5c9 ERE |
3097 | os.symlink(tarinfo.linkname, targetpath) |
3098 | else: | |
3099 | # See extract(). | |
3100 | if os.path.exists(tarinfo._link_target): | |
7584f5c9 ERE |
3101 | os.link(tarinfo._link_target, targetpath) |
3102 | else: | |
be60ffd0 ERE |
3103 | self._extract_member(self._find_link_target(tarinfo), |
3104 | targetpath) | |
3105 | except symlink_exception: | |
7584f5c9 | 3106 | try: |
be60ffd0 ERE |
3107 | self._extract_member(self._find_link_target(tarinfo), |
3108 | targetpath) | |
7584f5c9 ERE |
3109 | except KeyError: |
3110 | raise ExtractError("unable to resolve link inside archive") | |
3111 | ||
3112 | def chown(self, tarinfo, targetpath): | |
3113 | """Set owner of targetpath according to tarinfo. | |
3114 | """ | |
3115 | if pwd and hasattr(os, "geteuid") and os.geteuid() == 0: | |
3116 | # We have to be root to do so. | |
3117 | try: | |
3118 | g = grp.getgrnam(tarinfo.gname)[2] | |
3119 | except KeyError: | |
3120 | g = tarinfo.gid | |
3121 | try: | |
3122 | u = pwd.getpwnam(tarinfo.uname)[2] | |
3123 | except KeyError: | |
3124 | u = tarinfo.uid | |
3125 | try: | |
3126 | if tarinfo.issym() and hasattr(os, "lchown"): | |
3127 | os.lchown(targetpath, u, g) | |
3128 | else: | |
be60ffd0 ERE |
3129 | os.chown(targetpath, u, g) |
3130 | except OSError as e: | |
7584f5c9 ERE |
3131 | raise ExtractError("could not change owner") |
3132 | ||
3133 | def chmod(self, tarinfo, targetpath): | |
3134 | """Set file permissions of targetpath according to tarinfo. | |
3135 | """ | |
3136 | if hasattr(os, 'chmod'): | |
3137 | try: | |
3138 | os.chmod(targetpath, tarinfo.mode) | |
be60ffd0 | 3139 | except OSError as e: |
7584f5c9 ERE |
3140 | raise ExtractError("could not change mode") |
3141 | ||
3142 | def utime(self, tarinfo, targetpath): | |
3143 | """Set modification time of targetpath according to tarinfo. | |
3144 | """ | |
3145 | if not hasattr(os, 'utime'): | |
3146 | return | |
3147 | try: | |
3148 | os.utime(targetpath, (tarinfo.mtime, tarinfo.mtime)) | |
be60ffd0 | 3149 | except OSError as e: |
7584f5c9 ERE |
3150 | raise ExtractError("could not change modification time") |
3151 | ||
3152 | #-------------------------------------------------------------------------- | |
3153 | def next(self): | |
3154 | """Return the next member of the archive as a TarInfo object, when | |
3155 | TarFile is opened for reading. Return None if there is no more | |
3156 | available. | |
3157 | """ | |
3158 | self._check("ra") | |
3159 | if self.firstmember is not None: | |
3160 | m = self.firstmember | |
3161 | self.firstmember = None | |
3162 | return m | |
3163 | ||
be60ffd0 ERE |
3164 | # Read the next block. |
3165 | self.fileobj.seek(self.offset) | |
7584f5c9 ERE |
3166 | tarinfo = None |
3167 | while True: | |
3168 | try: | |
3169 | tarinfo = self.tarinfo.fromtarfile(self) | |
be60ffd0 | 3170 | except EOFHeaderError as e: |
7584f5c9 ERE |
3171 | if self.ignore_zeros: |
3172 | self._dbg(2, "0x%X: %s" % (self.offset, e)) | |
3173 | self.offset += BLOCKSIZE | |
3174 | continue | |
be60ffd0 | 3175 | except InvalidHeaderError as e: |
7584f5c9 ERE |
3176 | if self.ignore_zeros: |
3177 | self._dbg(2, "0x%X: %s" % (self.offset, e)) | |
3178 | self.offset += BLOCKSIZE | |
3179 | continue | |
3180 | elif self.offset == 0: | |
3181 | raise ReadError(str(e)) | |
3182 | except EmptyHeaderError: | |
3183 | if self.offset == 0: | |
3184 | raise ReadError("empty file") | |
be60ffd0 | 3185 | except TruncatedHeaderError as e: |
7584f5c9 ERE |
3186 | if self.offset == 0: |
3187 | raise ReadError(str(e)) | |
be60ffd0 | 3188 | except SubsequentHeaderError as e: |
7584f5c9 ERE |
3189 | raise ReadError(str(e)) |
3190 | break | |
3191 | ||
3192 | if tarinfo is not None: | |
ea625b04 ERE |
3193 | if self.save_to_members: |
3194 | self.members.append(tarinfo) | |
7584f5c9 ERE |
3195 | else: |
3196 | self._loaded = True | |
3197 | ||
3198 | return tarinfo | |
3199 | ||
3200 | #-------------------------------------------------------------------------- | |
3201 | # Little helper methods: | |
3202 | ||
3203 | def _getmember(self, name, tarinfo=None, normalize=False): | |
3204 | """Find an archive member by name from bottom to top. | |
3205 | If tarinfo is given, it is used as the starting point. | |
3206 | """ | |
3207 | # Ensure that all members have been loaded. | |
3208 | members = self.getmembers() | |
3209 | ||
3210 | # Limit the member search list up to tarinfo. | |
3211 | if tarinfo is not None: | |
3212 | members = members[:members.index(tarinfo)] | |
3213 | ||
3214 | if normalize: | |
3215 | name = os.path.normpath(name) | |
3216 | ||
3217 | for member in reversed(members): | |
3218 | if normalize: | |
3219 | member_name = os.path.normpath(member.name) | |
3220 | else: | |
3221 | member_name = member.name | |
3222 | ||
3223 | if name == member_name: | |
3224 | return member | |
3225 | ||
3226 | def _load(self): | |
3227 | """Read through the entire archive file and look for readable | |
3228 | members. | |
3229 | """ | |
3230 | while True: | |
3231 | tarinfo = self.next() | |
3232 | if tarinfo is None: | |
3233 | break | |
3234 | self._loaded = True | |
3235 | ||
3236 | def _check(self, mode=None): | |
3237 | """Check if TarFile is still open, and if the operation's mode | |
3238 | corresponds to TarFile's mode. | |
3239 | """ | |
3240 | if self.closed: | |
be60ffd0 | 3241 | raise OSError("%s is closed" % self.__class__.__name__) |
7584f5c9 | 3242 | if mode is not None and self.mode not in mode: |
be60ffd0 | 3243 | raise OSError("bad operation for mode %r" % self.mode) |
7584f5c9 ERE |
3244 | |
3245 | def _find_link_target(self, tarinfo): | |
3246 | """Find the target member of a symlink or hardlink member in the | |
3247 | archive. | |
3248 | """ | |
3249 | if tarinfo.issym(): | |
3250 | # Always search the entire archive. | |
3251 | linkname = "/".join(filter(None, (os.path.dirname(tarinfo.name), tarinfo.linkname))) | |
3252 | limit = None | |
3253 | else: | |
3254 | # Search the archive before the link, because a hard link is | |
3255 | # just a reference to an already archived file. | |
3256 | linkname = tarinfo.linkname | |
3257 | limit = tarinfo | |
3258 | ||
3259 | member = self._getmember(linkname, tarinfo=limit, normalize=True) | |
3260 | if member is None: | |
3261 | raise KeyError("linkname %r not found" % linkname) | |
3262 | return member | |
3263 | ||
3264 | def __iter__(self): | |
3265 | """Provide an iterator object. | |
3266 | """ | |
3267 | if self._loaded: | |
3268 | return iter(self.members) | |
3269 | else: | |
3270 | return TarIter(self) | |
3271 | ||
1027433a | 3272 | def _dbg(self, level, msg, *args): |
7584f5c9 ERE |
3273 | """Write debugging output to sys.stderr. |
3274 | """ | |
3275 | if level <= self.debug: | |
1027433a | 3276 | print(msg.format(*args), file=sys.stderr) |
7584f5c9 ERE |
3277 | |
3278 | def __enter__(self): | |
3279 | self._check() | |
3280 | return self | |
3281 | ||
3282 | def __exit__(self, type, value, traceback): | |
3283 | if type is None: | |
3284 | self.close() | |
3285 | else: | |
3286 | # An exception occurred. We must not call close() because | |
3287 | # it would try to write end-of-archive blocks and padding. | |
3288 | if not self._extfileobj: | |
3289 | self.fileobj.close() | |
3290 | self.closed = True | |
3291 | # class TarFile | |
3292 | ||
3293 | class TarIter: | |
3294 | """Iterator Class. | |
3295 | ||
3296 | for tarinfo in TarFile(...): | |
3297 | suite... | |
3298 | """ | |
3299 | ||
3300 | def __init__(self, tarfile): | |
3301 | """Construct a TarIter object. | |
3302 | """ | |
3303 | self.tarfile = tarfile | |
3304 | self.index = 0 | |
3305 | def __iter__(self): | |
3306 | """Return iterator object. | |
3307 | """ | |
3308 | return self | |
be60ffd0 | 3309 | def __next__(self): |
7584f5c9 ERE |
3310 | """Return the next item using TarFile's next() method. |
3311 | When all members have been read, set TarFile as _loaded. | |
3312 | """ | |
3313 | # Fix for SF #1100429: Under rare circumstances it can | |
3314 | # happen that getmembers() is called during iteration, | |
3315 | # which will cause TarIter to stop prematurely. | |
3316 | ||
3317 | if self.index == 0 and self.tarfile.firstmember is not None: | |
3318 | tarinfo = self.tarfile.next() | |
3319 | elif self.index < len(self.tarfile.members): | |
3320 | tarinfo = self.tarfile.members[self.index] | |
3321 | elif not self.tarfile._loaded: | |
3322 | tarinfo = self.tarfile.next() | |
3323 | if not tarinfo: | |
3324 | self.tarfile._loaded = True | |
3325 | raise StopIteration | |
3326 | else: | |
3327 | raise StopIteration | |
3328 | self.index += 1 | |
fb27c6e8 | 3329 | |
7584f5c9 ERE |
3330 | return tarinfo |
3331 | ||
6690f5e0 PG |
3332 | #--------------------------------------------------------- |
3333 | # support functionality for rescue mode | |
3334 | #--------------------------------------------------------- | |
3335 | ||
8fc6040c PG |
3336 | TAR_FMT_HDR = (# See tar(5): |
3337 | "<" | |
3338 | "100s" # ← char name[100]; /* 100 */ | |
3339 | "8s" # ← char mode[8]; /* 108 */ | |
3340 | "8s" # ← char uid[8]; /* 116 */ | |
3341 | "8s" # ← char gid[8]; /* 124 */ | |
3342 | "12s" # ← char size[12]; /* 136 */ | |
3343 | "12s" # ← char mtime[12]; /* 148 */ | |
3344 | "8s" # ← char checksum[8]; /* 156 */ | |
3345 | "B" # ← char typeflag[1]; /* 157 */ | |
3346 | "100s" # ← char linkname[100]; /* 257 */ | |
3347 | "6s" # ← char magic[6]; /* 263 */ | |
3348 | "2s" # ← char version[2]; /* 265 */ | |
3349 | "32s" # ← char uname[32]; /* 297 */ | |
3350 | "32s" # ← char gname[32]; /* 329 */ | |
3351 | "8s" # ← char devmajor[8]; /* 337 */ | |
3352 | "8s" # ← char devminor[8]; /* 345 */ | |
3353 | "12s" # ← char atime[12]; /* 357 */ | |
3354 | "12s" # ← char ctime[12]; /* 369 */ | |
3355 | "12s" # ← char offset[12]; /* 381 */ | |
3356 | "4s" # ← char longnames[4]; /* 385 */ | |
3357 | "B" # ← char unused[1]; /* 386 */ | |
3358 | "" # struct { | |
3359 | "12s" # ← char offset[12]; | |
3360 | "12s" # ← char numbytes[12]; | |
3361 | "12s" # ← char offset[12]; | |
3362 | "12s" # ← char numbytes[12]; | |
3363 | "12s" # ← char offset[12]; | |
3364 | "12s" # ← char numbytes[12]; | |
3365 | "12s" # ← char offset[12]; | |
3366 | "12s" # ← char numbytes[12]; | |
3367 | "" # } sparse[4]; /* 482 */ | |
3368 | "B" # ← char isextended[1]; /* 483 */ | |
3369 | "12s" # ← char realsize[12]; /* 495 */ | |
3370 | "17s" # ← char pad[17]; /* 512 */ | |
3371 | ) | |
3372 | ||
3373 | # The “magic” and “version” fields are special: | |
3374 | # | |
3375 | # tar(5) | |
3376 | # magic The magic field holds the five characters “ustar” followed by a | |
3377 | # space. Note that POSIX ustar archives have a trailing null. | |
3378 | # | |
3379 | # however, “tar.h”: | |
3380 | # | |
3381 | # /* OLDGNU_MAGIC uses both magic and version fields, which are contiguous. | |
3382 | # Found in an archive, it indicates an old GNU header format, which will be | |
3383 | # hopefully become obsolescent. With OLDGNU_MAGIC, uname and gname are | |
3384 | # valid, though the header is not truly POSIX conforming. */ | |
3385 | # | |
3386 | # | |
3387 | TAR_FMT_OLDGNU_MAGIC = b"ustar " | |
3388 | ||
3389 | def read_gnu_tar_hdr (data): | |
3390 | if len (data) != BLOCKSIZE: # header requires one complete block | |
3391 | return None | |
65b35c42 | 3392 | |
8fc6040c PG |
3393 | try: |
3394 | name, mode, \ | |
3395 | uid, gid, \ | |
3396 | size, mtime, \ | |
3397 | checksum, \ | |
3398 | typeflag, \ | |
3399 | linkname, \ | |
3400 | magic, \ | |
3401 | version, \ | |
3402 | uname, \ | |
3403 | gname, \ | |
3404 | devmajor, \ | |
3405 | devminor, \ | |
3406 | atime, \ | |
3407 | ctime, \ | |
3408 | offset, \ | |
3409 | longnames, \ | |
3410 | unused, \ | |
3411 | offset1, numbytes1, \ | |
3412 | offset2, numbytes2, \ | |
3413 | offset3, numbytes3, \ | |
3414 | offset4, numbytes4, \ | |
3415 | isextended, \ | |
3416 | realsize, \ | |
3417 | pad = struct.unpack (TAR_FMT_HDR, data) | |
3418 | except struct.error: | |
3419 | return None | |
3420 | ||
3421 | if magic != TAR_FMT_OLDGNU_MAGIC: | |
3422 | return None | |
3423 | ||
3424 | # return all except “unused” and “pad” | |
3425 | return \ | |
3426 | { "name" : name, "mode" : mode | |
3427 | , "uid" : uid , "gid" : gid | |
3428 | , "size" : size, "mtime" : mtime | |
3429 | , "checksum" : checksum | |
3430 | , "typeflag" : typeflag | |
3431 | , "linkname" : linkname | |
3432 | , "magic" : magic | |
3433 | , "version" : version | |
3434 | , "uname" : uname, "gname" : gname | |
3435 | , "devmajor" : devmajor, "devminor" : devminor | |
3436 | , "atime" : atime, "ctime" : ctime | |
3437 | , "offset" : offset | |
3438 | , "longnames" : longnames | |
3439 | , "offset1" : offset1, "numbytes1" : numbytes1 | |
3440 | , "offset2" : offset2, "numbytes2" : numbytes2 | |
3441 | , "offset3" : offset3, "numbytes3" : numbytes3 | |
3442 | , "offset4" : offset4, "numbytes4" : numbytes4 | |
3443 | , "isextended" : isextended | |
3444 | , "realsize" : realsize | |
3445 | } | |
3446 | ||
3447 | ||
3448 | def readable_tar_objects_offsets (ifd): | |
3449 | """ | |
3450 | Traverse blocks in file, trying to extract tar headers. | |
3451 | """ | |
3452 | pos = 0 | |
3453 | offsets = [] | |
3454 | ||
3455 | while True: | |
3456 | blk = os.read (ifd, BLOCKSIZE) | |
3457 | if len (blk) != BLOCKSIZE: | |
3458 | break | |
3459 | hdr = read_gnu_tar_hdr (blk) | |
3460 | if hdr is not None: | |
3461 | offsets.append (pos) | |
3462 | pos += BLOCKSIZE | |
65b35c42 | 3463 | |
8fc6040c | 3464 | return offsets |
65b35c42 PG |
3465 | |
3466 | ||
dfd7865e PG |
3467 | def locate_gz_hdr_candidates (fd): |
3468 | """ | |
3469 | Walk over instances of the GZ magic in the payload, collecting their | |
3470 | positions. If the offset of the first found instance is not zero, the file | |
3471 | begins with leading garbage. | |
3472 | ||
3473 | Note that since the GZ magic consists of only two bytes, we expect a lot of | |
3474 | false positives inside binary data. | |
3475 | ||
3476 | :return: The list of offsets in the file. | |
3477 | """ | |
3478 | pos = 0 | |
3479 | cands = [] | |
3480 | mm = mmap.mmap(fd, 0, mmap.MAP_SHARED, mmap.PROT_READ) | |
3481 | ||
3482 | while True: | |
3483 | pos = mm.find (GZ_MAGIC_BYTES, pos) | |
3484 | if pos == -1: | |
3485 | break | |
3486 | cands.append (pos) | |
3487 | pos += len (GZ_MAGIC_BYTES) | |
3488 | ||
3489 | return cands | |
3490 | ||
3491 | ||
3492 | HDR_CAND_GOOD = 0 # header marks begin of valid object | |
3493 | HDR_CAND_FISHY = 1 # inconclusive | |
3494 | HDR_CAND_JUNK = 2 # not a header / object unreadable | |
3495 | ||
3496 | ||
3497 | def read_cstring (fd, max=-1, encoding=None): | |
3498 | """ | |
3499 | Read one NUL-terminated string from *fd* into a Python string. If *max* is | |
3500 | non-negative, reading will terminate after the specified number of bytes. | |
3501 | ||
3502 | Optionally, an *encoding* may be specified to interpret the data as. | |
3503 | ||
3504 | :returns: *None* if parsing failed or the maximum number of bytes has been | |
3505 | exceeded; a Python string with the data otherwise. | |
3506 | """ | |
3507 | buf = b"" | |
3508 | l = 0 | |
3509 | ||
3510 | while True: | |
3511 | c = os.read (fd, 1) | |
3512 | if c == NUL: | |
3513 | break | |
3514 | if max >= 0 and l > max: | |
3515 | return None | |
3516 | buf += c | |
3517 | l += 1 | |
3518 | if encoding is not None: | |
3519 | buf = buf.decode (encoding) | |
3520 | ||
3521 | return buf | |
3522 | ||
3523 | ||
3524 | def inspect_gz_hdr (fd, off): | |
3525 | """ | |
3526 | Attempt to parse a Gzip header in *fd* at position *off*. The format is | |
3527 | documented as RFC1952. | |
3528 | ||
3529 | Returns a verdict about the quality of that header plus the parsed header | |
3530 | when readable. Problematic sizes such as fields running past the EOF are | |
3531 | treated as garbage. Properties in which the header merely doesn’t conform | |
3532 | to the spec (garbage flag bits, bogus timestamp) are considered “fishy”. No | |
3533 | validation is possible on embedded strings because they are single-byte | |
3534 | encoded. | |
3535 | """ | |
3536 | fname = None | |
3537 | flags = 0x00 | |
3538 | dflags = 0x00 | |
3539 | mtime = 0x00000000 | |
3540 | oscode = 0x00 | |
3541 | verdict = HDR_CAND_GOOD | |
3542 | ||
3543 | os.lseek (fd, off, os.SEEK_SET) | |
3544 | if os.lseek (fd, 0, os.SEEK_CUR) != off: | |
3545 | return HDR_CAND_JUNK, None | |
3546 | ||
3547 | raw = os.read (fd, GZ_HEADER_SIZE) | |
3548 | if len (raw) != GZ_HEADER_SIZE: | |
3549 | return HDR_CAND_JUNK, None | |
3550 | ||
3551 | flags = 0x0 | |
3552 | try: | |
3553 | _m1, _m2, meth, flags, mtime, dflags, oscode = \ | |
3554 | struct.unpack (GZ_FMT_HEADER, raw) | |
3555 | if meth != GZ_METHOD_DEFLATE: # only deflate is supported | |
3556 | return HDR_CAND_JUNK, None | |
3557 | except struct.error as exn: | |
3558 | return HDR_CAND_JUNK, None | |
3559 | ||
3560 | if mtime > int (time.time ()): | |
3561 | verdict = HDR_CAND_FISHY | |
3562 | ||
3563 | if dflags != GZ_DEFLATE_FLAGS: | |
3564 | verdict = HDR_CAND_FISHY | |
3565 | ||
3566 | if oscode != GZ_OS_CODE: | |
3567 | verdict = HDR_CAND_FISHY | |
3568 | ||
3569 | if flags & GZ_FLAG_FTEXT: # created by some contrarian | |
3570 | verdict = HDR_CAND_FISHY | |
3571 | if flags & GZ_FLAG_FEXTRA: | |
3572 | xlen = struct.unpack ("<H", os.read (fd, 2)) | |
3573 | xtra = os.read (fd, xlen) | |
3574 | if len (xtra) != xlen: # eof inside header | |
3575 | return HDR_CAND_JUNK, None | |
3576 | if flags & GZ_FLAG_FNAME: | |
3577 | # read up to the next NUL byte, not exceeding the maximum path length | |
3578 | # allowed by tar(5) | |
3579 | fname = read_cstring (fd, max=(LENGTH_NAME + LENGTH_PREFIX), | |
3580 | encoding="iso-8859-1") | |
3581 | if fname is None: | |
3582 | return HDR_CAND_JUNK, None | |
3583 | if flags & GZ_FLAG_FCOMMENT: | |
3584 | fname = read_cstring (fd, max=(LENGTH_NAME + LENGTH_PREFIX), | |
3585 | encoding="iso-8859-1") | |
3586 | if fname is None: | |
3587 | return HDR_CAND_JUNK, None | |
3588 | if flags & GZ_FLAG_FHCRC: # half a CRC32; discarded | |
3589 | crc16 = os.read (fd, 2) | |
3590 | if len (crc16) != 2: # eof inside header | |
3591 | return HDR_CAND_JUNK, None | |
3592 | if flags & GZ_FLAG_RESERVED: | |
3593 | # according to the RFC, these must not be set | |
3594 | verdict = HDR_CAND_FISHY | |
3595 | ||
3596 | hlen = os.lseek (fd, 0, os.SEEK_CUR) - off | |
3597 | ||
3598 | return verdict, \ | |
3599 | { "fname" : fname | |
3600 | , "flags" : flags | |
3601 | , "dflags" : dflags | |
3602 | , "mtime" : mtime | |
3603 | , "oscode" : oscode | |
3604 | , "hlen" : hlen | |
3605 | } | |
3606 | ||
3607 | ||
3608 | def try_decompress (ifd, off, hdr): | |
3609 | """ | |
3610 | Attempt to process the object starting at *off* with gzip. | |
3611 | ||
3612 | :returns: A pair containing the values of the decompressed data and | |
3613 | the length of the input consumed. Note that the latter value | |
3614 | may exceed the length of the compressed data because the | |
3615 | *zlib* module does not provide a means to query how much | |
3616 | of the input it processed before the end of an object. | |
3617 | """ | |
3618 | import zlib | |
3619 | decmp = zlib.decompressobj (-zlib.MAX_WBITS) | |
3620 | pos = off | |
3621 | dlen = 0 # size of decompressed data | |
3622 | ||
3623 | os.lseek (ifd, pos, os.SEEK_SET) | |
3624 | while True: | |
3625 | cnk = os.read (ifd, BUFSIZE) | |
3626 | pos += len (cnk) | |
3627 | try: | |
3628 | data = decmp.decompress (cnk) | |
3629 | except zlib.error as exn: # probably CRC32 mismatch; terminate softly | |
3630 | break # fishy | |
3631 | dlen += len (data) | |
3632 | if decmp.eof is True: | |
3633 | break | |
3634 | if len (cnk) != BUFSIZE: # eof, but not end of decompressed object! | |
3635 | break # fishy | |
3636 | ||
3637 | return dlen, pos - off | |
3638 | ||
3639 | def readable_gz_objects_offsets (ifd, cands): | |
3640 | """ | |
3641 | Inspect header candidates for parseable *ifd* gzipped objects. | |
3642 | """ | |
3643 | good = [] | |
3644 | nobj = 0 | |
3645 | ||
3646 | for cand in cands: | |
3647 | nobj += 1 | |
3648 | vdt, hdr = inspect_gz_hdr (ifd, cand) | |
3649 | if vdt == HDR_CAND_JUNK: | |
3650 | pass # ignore unreadable ones | |
3651 | elif vdt in [HDR_CAND_GOOD, HDR_CAND_FISHY]: | |
3652 | off0 = cand + hdr ["hlen"] | |
3653 | dlen, clen = try_decompress (ifd, off0, hdr) | |
3654 | if dlen > 0 and clen > 0: | |
3655 | good.append (cand) | |
3656 | ||
3657 | return good | |
3658 | ||
3659 | ||
3660 | def reconstruct_offsets_gz (fname): | |
3661 | """ | |
3662 | From the given file, retrieve all GZ header-like offsets (“candidates”). | |
3663 | Then check each of those locations whether they can be processed as | |
3664 | compressed data. | |
3665 | """ | |
3666 | ifd = os.open (fname, os.O_RDONLY) | |
3667 | ||
3668 | try: | |
3669 | cands = locate_gz_hdr_candidates (ifd) | |
3670 | return readable_gz_objects_offsets (ifd, cands) | |
3671 | finally: | |
3672 | os.close (ifd) | |
3673 | ||
3674 | ||
65b35c42 PG |
3675 | def reconstruct_offsets_tar (fname): |
3676 | """ | |
3677 | From the given file, retrieve all tar header-like offsets (“candidates”). | |
3678 | Then check each of those locations whether they can be processed as tar | |
3679 | data. | |
3680 | """ | |
3681 | ifd = os.open (fname, os.O_RDONLY) | |
3682 | ||
3683 | try: | |
8fc6040c | 3684 | return readable_tar_objects_offsets (ifd) |
65b35c42 PG |
3685 | finally: |
3686 | os.close (ifd) | |
3687 | ||
3688 | ||
d39d4cbf PG |
3689 | def read_tarobj_at_offset (fileobj, offset, mode, secret=None): |
3690 | decr = None | |
d39d4cbf | 3691 | |
dfd7865e PG |
3692 | if secret is not None: |
3693 | ks = secret [0] | |
3694 | ||
3695 | if ks == crypto.PDTCRYPT_SECRET_PW: | |
3696 | decr = crypto.Decrypt (password=secret [1]) | |
3697 | elif ks == crypto.PDTCRYPT_SECRET_KEY: | |
3698 | key = binascii.unhexlify (secret [1]) | |
3699 | decr = crypto.Decrypt (key=key) | |
3700 | else: | |
3701 | raise RuntimeError | |
d39d4cbf PG |
3702 | |
3703 | tarobj = \ | |
3704 | TarFile.open_at_offset (offset, | |
3705 | mode=mode, | |
3706 | fileobj=fileobj, | |
3707 | format=GNU_FORMAT, | |
3708 | concat='#' in mode, | |
3709 | encryption=decr, | |
3710 | save_to_members=False, | |
3711 | tolerance=TOLERANCE_RESCUE) | |
3712 | ||
3713 | return tarobj.next () | |
3714 | ||
3715 | ||
2d50b7f7 PG |
3716 | def idxent_of_tarinfo (tarinfo): |
3717 | """ | |
3718 | Scrape the information relevant for the index from a *TarInfo* object. | |
3719 | Keys like the inode number that lack a corresponding field in a TarInfo | |
3720 | will be set to some neutral value. | |
3721 | Example output: | |
3722 | ||
3723 | { "inode" : 0 | |
3724 | , "uid" : 0 | |
3725 | , "path" : "snapshot://annotations.db" | |
3726 | , "offset" : 0 | |
3727 | , "volume" : 0 | |
3728 | , "mode" : 33152 | |
3729 | , "ctime" : 1502798115 | |
3730 | , "mtime" : 1502196423 | |
3731 | , "size" : 144 | |
3732 | , "type" : "file" | |
3733 | , "gid" : 0 | |
3734 | } | |
3735 | ||
3736 | """ | |
3737 | ||
3738 | return \ | |
3739 | { "inode" : 0 # ignored when reading the index | |
3740 | , "uid" : tarinfo.uid | |
3741 | , "gid" : tarinfo.gid | |
3742 | , "path" : tarinfo.name # keeping URI scheme | |
3743 | , "offset" : 0 # to be added by the caller | |
3744 | , "volume" : tarinfo.volume_offset | |
3745 | , "mode" : tarinfo.mode | |
3746 | , "ctime" : tarinfo.mtime | |
3747 | , "mtime" : tarinfo.mtime | |
3748 | , "size" : tarinfo.size | |
3749 | , "type" : tarinfo.type | |
3750 | } | |
3751 | ||
3752 | ||
27ee4dd4 PG |
3753 | def gen_rescue_index (gen_volume_name, mode, maxvol=None, password=None, key=None): |
3754 | infos = [] | |
6690f5e0 PG |
3755 | psidx = [] # pseudo index, return value |
3756 | offsets = None | |
addcec42 | 3757 | secret = crypto.make_secret (password=password, key=key) |
6690f5e0 | 3758 | |
27ee4dd4 | 3759 | nvol = 0 |
dfd7865e | 3760 | |
27ee4dd4 | 3761 | def aux (o, nvol, ti): |
dfd7865e PG |
3762 | ie = idxent_of_tarinfo (ti) |
3763 | ie ["offset"] = o | |
27ee4dd4 | 3764 | ie ["volume"] = nvol |
dfd7865e | 3765 | return ie |
27ee4dd4 PG |
3766 | |
3767 | while True: | |
3768 | vpath = gen_volume_name (nvol) | |
3769 | try: | |
3770 | if secret is not None: | |
3771 | offsets = crypto.reconstruct_offsets (vpath, secret) | |
3772 | elif mode == "#gz": | |
3773 | offsets = reconstruct_offsets_gz (vpath) | |
3774 | elif mode == "#": | |
3775 | offsets = reconstruct_offsets_tar (vpath) | |
3776 | else: | |
3777 | raise TarError ("no rescue handling for mode “%s”" % mode) | |
3778 | except FileNotFoundError as exn: | |
3779 | # volume does not exist | |
3780 | if maxvol is not None and i < maxvol: | |
3781 | continue # explicit volume number specified, ignore missing ones | |
3782 | else: | |
3783 | break | |
3784 | ||
3785 | fileobj = bltn_open (vpath, "rb") | |
3786 | infos += [ (off, nvol, read_tarobj_at_offset (fileobj, off, mode, | |
3787 | secret=secret)) | |
3788 | for off in offsets ] | |
3789 | nvol += 1 | |
3790 | ||
3791 | psidx = [ aux (o, nvol, ti) for o, nvol, ti in infos ] | |
6690f5e0 PG |
3792 | |
3793 | return psidx | |
7584f5c9 ERE |
3794 | |
3795 | #-------------------- | |
3796 | # exported functions | |
3797 | #-------------------- | |
3798 | def is_tarfile(name): | |
3799 | """Return True if name points to a tar archive that we | |
3800 | are able to handle, else return False. | |
3801 | """ | |
3802 | try: | |
3803 | t = open(name) | |
3804 | t.close() | |
3805 | return True | |
3806 | except TarError: | |
3807 | return False | |
3808 | ||
3809 | bltn_open = open | |
3810 | open = TarFile.open |