Mon, 26 Nov 2007 22:39:29 -0500
[svn] Update documentation.
1 | 1 | #!/usr/bin/env python |
2 | # | |
19 | 3 | # dtrx -- Intelligently extract various archive types. |
23
039dd321a7d0
[svn] If an archive contains other archives, and the user didn't specify that
brett
parents:
22
diff
changeset
|
4 | # Copyright (c) 2006, 2007 Brett Smith <brettcsmith@brettcsmith.org>. |
1 | 5 | # |
6 | # This program is free software; you can redistribute it and/or modify it | |
7 | # under the terms of the GNU General Public License as published by the | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
8 | # Free Software Foundation; either version 3 of the License, or (at your |
1 | 9 | # option) any later version. |
10 | # | |
11 | # This program is distributed in the hope that it will be useful, but | |
12 | # WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General | |
14 | # Public License for more details. | |
15 | # | |
16 | # You should have received a copy of the GNU General Public License along | |
42 | 17 | # with this program; if not, see <http://www.gnu.org/licenses/>. |
1 | 18 | |
5 | 19 | import errno |
12
5d202467c589
[svn] Introduce a real logging system. Right now all this really gets us is the
brett
parents:
11
diff
changeset
|
20 | import logging |
1 | 21 | import mimetypes |
6
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
22 | import optparse |
1 | 23 | import os |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
24 | import re |
15
28dbd52a8bb8
[svn] Add a -f/--flat option, which will extract the archive contents into the
brett
parents:
14
diff
changeset
|
25 | import stat |
1 | 26 | import subprocess |
27 | import sys | |
28 | import tempfile | |
20
69c93c3e6972
[svn] If the archive contains one directory with the "wrong" name, ask the user
brett
parents:
19
diff
changeset
|
29 | import textwrap |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
30 | import traceback |
1 | 31 | |
36
4bf2508d9b9e
[svn] Small optimization to be nice to the system: don't try a given extractor
brett
parents:
35
diff
changeset
|
32 | from sets import Set |
4bf2508d9b9e
[svn] Small optimization to be nice to the system: don't try a given extractor
brett
parents:
35
diff
changeset
|
33 | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
34 | VERSION = "5.0" |
19 | 35 | VERSION_BANNER = """dtrx version %s |
23
039dd321a7d0
[svn] If an archive contains other archives, and the user didn't specify that
brett
parents:
22
diff
changeset
|
36 | Copyright (c) 2006, 2007 Brett Smith <brettcsmith@brettcsmith.org> |
6
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
37 | |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
38 | This program is free software; you can redistribute it and/or modify it |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
39 | under the terms of the GNU General Public License as published by the |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
40 | Free Software Foundation; either version 3 of the License, or (at your |
6
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
41 | option) any later version. |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
42 | |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
43 | This program is distributed in the hope that it will be useful, but |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
44 | WITHOUT ANY WARRANTY; without even the implied warranty of |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
45 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
46 | Public License for more details.""" % (VERSION,) |
1 | 47 | |
48 | MATCHING_DIRECTORY = 1 | |
22
b240777ae53e
[svn] Improve the way we check archive contents. If all the entries look like
brett
parents:
20
diff
changeset
|
49 | ONE_ENTRY = 2 |
1 | 50 | BOMB = 3 |
51 | EMPTY = 4 | |
22
b240777ae53e
[svn] Improve the way we check archive contents. If all the entries look like
brett
parents:
20
diff
changeset
|
52 | ONE_ENTRY_KNOWN = 5 |
1 | 53 | |
20
69c93c3e6972
[svn] If the archive contains one directory with the "wrong" name, ask the user
brett
parents:
19
diff
changeset
|
54 | EXTRACT_HERE = 1 |
69c93c3e6972
[svn] If the archive contains one directory with the "wrong" name, ask the user
brett
parents:
19
diff
changeset
|
55 | EXTRACT_WRAP = 2 |
69c93c3e6972
[svn] If the archive contains one directory with the "wrong" name, ask the user
brett
parents:
19
diff
changeset
|
56 | EXTRACT_RENAME = 3 |
69c93c3e6972
[svn] If the archive contains one directory with the "wrong" name, ask the user
brett
parents:
19
diff
changeset
|
57 | |
23
039dd321a7d0
[svn] If an archive contains other archives, and the user didn't specify that
brett
parents:
22
diff
changeset
|
58 | RECURSE_ALWAYS = 1 |
039dd321a7d0
[svn] If an archive contains other archives, and the user didn't specify that
brett
parents:
22
diff
changeset
|
59 | RECURSE_ONCE = 2 |
039dd321a7d0
[svn] If an archive contains other archives, and the user didn't specify that
brett
parents:
22
diff
changeset
|
60 | RECURSE_NOT_NOW = 3 |
039dd321a7d0
[svn] If an archive contains other archives, and the user didn't specify that
brett
parents:
22
diff
changeset
|
61 | RECURSE_NEVER = 4 |
039dd321a7d0
[svn] If an archive contains other archives, and the user didn't specify that
brett
parents:
22
diff
changeset
|
62 | |
6
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
63 | mimetypes.encodings_map.setdefault('.bz2', 'bzip2') |
34
a8f875e02c83
[svn] Add support for LZMA compression. Holy crap that was easy.
brett
parents:
33
diff
changeset
|
64 | mimetypes.encodings_map.setdefault('.lzma', 'lzma') |
41
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
65 | mimetypes.types_map.setdefault('.gem', 'application/x-ruby-gem') |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
66 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
67 | logger = logging.getLogger('dtrx-log') |
6
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
68 | |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
69 | def run_command(command, description, stdout=None, stderr=None, stdin=None): |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
70 | process = subprocess.Popen(command, stdin=stdin, stdout=stdout, |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
71 | stderr=stderr) |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
72 | status = process.wait() |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
73 | for pipe in (process.stdout, process.stderr): |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
74 | try: |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
75 | pipe.close() |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
76 | except AttributeError: |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
77 | pass |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
78 | if status != 0: |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
79 | return ("%s error: '%s' returned status code %s" % |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
80 | (description, ' '.join(command), status)) |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
81 | return None |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
82 | |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
83 | class FilenameChecker(object): |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
84 | def __init__(self, original_name): |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
85 | self.original_name = original_name |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
86 | |
17
481a2b4be471
[svn] Lots of tests for various boundary cases, and slightly better handling for
brett
parents:
16
diff
changeset
|
87 | def is_free(self, filename): |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
88 | return not os.path.exists(filename) |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
89 | |
41
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
90 | def create(self): |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
91 | fd, filename = tempfile.mkstemp(prefix=self.original_name + '.', |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
92 | dir='.') |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
93 | os.close(fd) |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
94 | return filename |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
95 | |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
96 | def check(self): |
41
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
97 | if self.is_free(self.original_name): |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
98 | return self.original_name |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
99 | for suffix in [''] + ['.%s' % (x,) for x in range(1, 10)]: |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
100 | filename = '%s%s' % (self.original_name, suffix) |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
101 | if self.is_free(filename): |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
102 | return filename |
41
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
103 | return self.create() |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
104 | |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
105 | |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
106 | class DirectoryChecker(FilenameChecker): |
17
481a2b4be471
[svn] Lots of tests for various boundary cases, and slightly better handling for
brett
parents:
16
diff
changeset
|
107 | def is_free(self, filename): |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
108 | try: |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
109 | os.mkdir(filename) |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
110 | except OSError, error: |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
111 | if error.errno == errno.EEXIST: |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
112 | return False |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
113 | raise |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
114 | return True |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
115 | |
41
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
116 | def create(self): |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
117 | return tempfile.mkdtemp(prefix=self.original_name + '.', dir='.') |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
118 | |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
119 | |
1 | 120 | class ExtractorError(Exception): |
121 | pass | |
122 | ||
123 | ||
124 | class BaseExtractor(object): | |
34
a8f875e02c83
[svn] Add support for LZMA compression. Holy crap that was easy.
brett
parents:
33
diff
changeset
|
125 | decoders = {'bzip2': 'bzcat', 'gzip': 'zcat', 'compress': 'zcat', |
a8f875e02c83
[svn] Add support for LZMA compression. Holy crap that was easy.
brett
parents:
33
diff
changeset
|
126 | 'lzma': 'lzcat'} |
1 | 127 | |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
128 | name_checker = DirectoryChecker |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
129 | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
130 | def __init__(self, filename, encoding): |
6
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
131 | if encoding and (not self.decoders.has_key(encoding)): |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
132 | raise ValueError("unrecognized encoding %s" % (encoding,)) |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
133 | self.filename = os.path.realpath(filename) |
1 | 134 | self.encoding = encoding |
6
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
135 | self.included_archives = [] |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
136 | self.target = None |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
137 | self.content_type = None |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
138 | self.content_name = None |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
139 | self.pipes = [] |
5 | 140 | try: |
141 | self.archive = open(filename, 'r') | |
142 | except (IOError, OSError), error: | |
143 | raise ExtractorError("could not open %s: %s" % | |
144 | (filename, error.strerror)) | |
1 | 145 | if encoding: |
146 | self.pipe([self.decoders[encoding]], "decoding") | |
147 | self.prepare() | |
148 | ||
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
149 | def pipe(self, command, description="extraction"): |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
150 | self.pipes.append((command, description)) |
1 | 151 | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
152 | def run_pipes(self, final_stdout=None): |
41
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
153 | if not self.pipes: |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
154 | return |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
155 | elif final_stdout is None: |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
156 | # FIXME: Buffering this might be dumb. |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
157 | final_stdout = tempfile.TemporaryFile() |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
158 | num_pipes = len(self.pipes) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
159 | last_pipe = num_pipes - 1 |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
160 | processes = [] |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
161 | for index, command in enumerate([pipe[0] for pipe in self.pipes]): |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
162 | if index == 0: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
163 | stdin = self.archive |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
164 | else: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
165 | stdin = processes[-1].stdout |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
166 | if index == last_pipe: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
167 | stdout = final_stdout |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
168 | else: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
169 | stdout = subprocess.PIPE |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
170 | processes.append(subprocess.Popen(command, stdin=stdin, |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
171 | stdout=stdout, |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
172 | stderr=subprocess.PIPE)) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
173 | exit_codes = [pipe.wait() for pipe in processes] |
1 | 174 | self.archive.close() |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
175 | for index in range(last_pipe): |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
176 | processes[index].stdout.close() |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
177 | processes[index].stderr.close() |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
178 | for index, status in enumerate(exit_codes): |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
179 | if status != 0: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
180 | raise ExtractorError("%s error: '%s' returned status code %s" % |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
181 | (self.pipes[index][1], |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
182 | ' '.join(self.pipes[index][0]), status)) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
183 | self.archive = final_stdout |
1 | 184 | |
185 | def prepare(self): | |
186 | pass | |
187 | ||
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
188 | def check_included_archives(self, filenames): |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
189 | for filename in filenames: |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
190 | if (ExtractorBuilder.try_by_mimetype(filename) or |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
191 | ExtractorBuilder.try_by_extension(filename)): |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
192 | self.included_archives.append(filename) |
22
b240777ae53e
[svn] Improve the way we check archive contents. If all the entries look like
brett
parents:
20
diff
changeset
|
193 | |
b240777ae53e
[svn] Improve the way we check archive contents. If all the entries look like
brett
parents:
20
diff
changeset
|
194 | def check_contents(self): |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
195 | filenames = os.listdir('.') |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
196 | if not filenames: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
197 | self.content_type = EMPTY |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
198 | elif len(filenames) == 1: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
199 | if self.basename() == filenames[0]: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
200 | self.content_type = MATCHING_DIRECTORY |
22
b240777ae53e
[svn] Improve the way we check archive contents. If all the entries look like
brett
parents:
20
diff
changeset
|
201 | else: |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
202 | self.content_type = ONE_ENTRY |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
203 | self.content_name = filenames[0] |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
204 | if os.path.isdir(filenames[0]): |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
205 | self.content_name += '/' |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
206 | else: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
207 | self.content_type = BOMB |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
208 | self.check_included_archives(filenames) |
1 | 209 | |
210 | def basename(self): | |
5 | 211 | pieces = os.path.basename(self.filename).split('.') |
2
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
212 | extension = '.' + pieces[-1] |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
213 | if mimetypes.encodings_map.has_key(extension): |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
214 | pieces.pop() |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
215 | extension = '.' + pieces[-1] |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
216 | if (mimetypes.types_map.has_key(extension) or |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
217 | mimetypes.common_types.has_key(extension) or |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
218 | mimetypes.suffix_map.has_key(extension)): |
1 | 219 | pieces.pop() |
220 | return '.'.join(pieces) | |
221 | ||
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
222 | def extract(self): |
40
ee6a869f8da1
[svn] Be a little nicer about explaining that we can't extract to the current
brett
parents:
39
diff
changeset
|
223 | try: |
ee6a869f8da1
[svn] Be a little nicer about explaining that we can't extract to the current
brett
parents:
39
diff
changeset
|
224 | self.target = tempfile.mkdtemp(prefix='.dtrx-', dir='.') |
ee6a869f8da1
[svn] Be a little nicer about explaining that we can't extract to the current
brett
parents:
39
diff
changeset
|
225 | except (OSError, IOError), error: |
ee6a869f8da1
[svn] Be a little nicer about explaining that we can't extract to the current
brett
parents:
39
diff
changeset
|
226 | raise ExtractorError("cannot extract here: %s" % (error.strerror,)) |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
227 | old_path = os.path.realpath(os.curdir) |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
228 | os.chdir(self.target) |
33
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
229 | try: |
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
230 | self.archive.seek(0, 0) |
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
231 | self.extract_archive() |
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
232 | self.check_contents() |
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
233 | except ExtractorError: |
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
234 | os.chdir(old_path) |
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
235 | subprocess.call(['rm', '-rf', self.target]) |
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
236 | raise |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
237 | os.chdir(old_path) |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
238 | |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
239 | def get_filenames(self): |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
240 | self.run_pipes() |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
241 | self.archive.seek(0, 0) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
242 | while True: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
243 | line = self.archive.readline() |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
244 | if not line: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
245 | self.archive.close() |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
246 | return |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
247 | yield line.rstrip('\n') |
1 | 248 | |
249 | ||
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
250 | class CompressionExtractor(BaseExtractor): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
251 | name_checker = FilenameChecker |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
252 | |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
253 | def basename(self): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
254 | pieces = os.path.basename(self.filename).split('.') |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
255 | extension = '.' + pieces[-1] |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
256 | if mimetypes.encodings_map.has_key(extension): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
257 | pieces.pop() |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
258 | return '.'.join(pieces) |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
259 | |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
260 | def get_filenames(self): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
261 | yield self.basename() |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
262 | |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
263 | def extract(self): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
264 | self.content_type = ONE_ENTRY_KNOWN |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
265 | self.content_name = self.basename() |
40
ee6a869f8da1
[svn] Be a little nicer about explaining that we can't extract to the current
brett
parents:
39
diff
changeset
|
266 | try: |
ee6a869f8da1
[svn] Be a little nicer about explaining that we can't extract to the current
brett
parents:
39
diff
changeset
|
267 | output_fd, self.target = tempfile.mkstemp(prefix='.dtrx-', dir='.') |
ee6a869f8da1
[svn] Be a little nicer about explaining that we can't extract to the current
brett
parents:
39
diff
changeset
|
268 | except (OSError, IOError), error: |
ee6a869f8da1
[svn] Be a little nicer about explaining that we can't extract to the current
brett
parents:
39
diff
changeset
|
269 | raise ExtractorError("cannot extract here: %s" % (error.strerror,)) |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
270 | try: |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
271 | self.run_pipes(output_fd) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
272 | except ExtractorError: |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
273 | os.close(output_fd) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
274 | os.unlink(self.target) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
275 | raise |
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
276 | os.close(output_fd) |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
277 | |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
278 | |
1 | 279 | class TarExtractor(BaseExtractor): |
280 | def get_filenames(self): | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
281 | self.pipe(['tar', '-t'], "listing") |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
282 | return BaseExtractor.get_filenames(self) |
1 | 283 | |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
284 | def extract_archive(self): |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
285 | self.pipe(['tar', '-x']) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
286 | self.run_pipes() |
1 | 287 | |
288 | ||
289 | class CpioExtractor(BaseExtractor): | |
290 | def get_filenames(self): | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
291 | self.pipe(['cpio', '-t'], "listing") |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
292 | return BaseExtractor.get_filenames(self) |
1 | 293 | |
294 | def extract_archive(self): | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
295 | self.pipe(['cpio', '-i', '--make-directories', |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
296 | '--no-absolute-filenames']) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
297 | self.run_pipes() |
1 | 298 | |
299 | ||
300 | class RPMExtractor(CpioExtractor): | |
301 | def prepare(self): | |
302 | self.pipe(['rpm2cpio', '-'], "rpm2cpio") | |
303 | ||
304 | def basename(self): | |
9
920417b8acc9
[svn] Fix issues with basename methods. First, string's rsplit method only
brett
parents:
8
diff
changeset
|
305 | pieces = os.path.basename(self.filename).split('.') |
2
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
306 | if len(pieces) == 1: |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
307 | return pieces[0] |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
308 | elif pieces[-1] != 'rpm': |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
309 | return BaseExtractor.basename(self) |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
310 | pieces.pop() |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
311 | if len(pieces) == 1: |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
312 | return pieces[0] |
9
920417b8acc9
[svn] Fix issues with basename methods. First, string's rsplit method only
brett
parents:
8
diff
changeset
|
313 | elif len(pieces[-1]) < 8: |
2
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
314 | pieces.pop() |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
315 | return '.'.join(pieces) |
1 | 316 | |
317 | def check_contents(self): | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
318 | self.check_included_archives(os.listdir('.')) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
319 | self.content_type = BOMB |
6
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
320 | |
1 | 321 | |
322 | class DebExtractor(TarExtractor): | |
323 | def prepare(self): | |
324 | self.pipe(['ar', 'p', self.filename, 'data.tar.gz'], | |
325 | "data.tar.gz extraction") | |
326 | self.pipe(['zcat'], "data.tar.gz decompression") | |
327 | ||
328 | def basename(self): | |
9
920417b8acc9
[svn] Fix issues with basename methods. First, string's rsplit method only
brett
parents:
8
diff
changeset
|
329 | pieces = os.path.basename(self.filename).split('_') |
2
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
330 | if len(pieces) == 1: |
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
331 | return pieces[0] |
9
920417b8acc9
[svn] Fix issues with basename methods. First, string's rsplit method only
brett
parents:
8
diff
changeset
|
332 | last_piece = pieces.pop() |
920417b8acc9
[svn] Fix issues with basename methods. First, string's rsplit method only
brett
parents:
8
diff
changeset
|
333 | if (len(last_piece) > 10) or (not last_piece.endswith('.deb')): |
2
1570351bf863
[svn] Fix a small bug that would crash the program if an archive was empty.
brett
parents:
1
diff
changeset
|
334 | return BaseExtractor.basename(self) |
9
920417b8acc9
[svn] Fix issues with basename methods. First, string's rsplit method only
brett
parents:
8
diff
changeset
|
335 | return '_'.join(pieces) |
1 | 336 | |
337 | def check_contents(self): | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
338 | self.check_included_archives(os.listdir('.')) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
339 | self.content_type = BOMB |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
340 | |
1 | 341 | |
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
342 | class DebMetadataExtractor(DebExtractor): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
343 | def prepare(self): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
344 | self.pipe(['ar', 'p', self.filename, 'control.tar.gz'], |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
345 | "control.tar.gz extraction") |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
346 | self.pipe(['zcat'], "control.tar.gz decompression") |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
347 | |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
348 | |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
349 | class GemExtractor(TarExtractor): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
350 | def prepare(self): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
351 | self.pipe(['tar', '-xO', 'data.tar.gz'], "data.tar.gz extraction") |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
352 | self.pipe(['zcat'], "data.tar.gz decompression") |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
353 | |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
354 | def check_contents(self): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
355 | self.check_included_archives(os.listdir('.')) |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
356 | self.content_type = BOMB |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
357 | |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
358 | |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
359 | class GemMetadataExtractor(CompressionExtractor): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
360 | def prepare(self): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
361 | self.pipe(['tar', '-xO', 'metadata.gz'], "metadata.gz extraction") |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
362 | self.pipe(['zcat'], "metadata.gz decompression") |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
363 | |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
364 | def basename(self): |
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
365 | return os.path.basename(self.filename) + '-metadata.txt' |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
366 | |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
367 | |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
368 | class NoPipeExtractor(BaseExtractor): |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
369 | # Some extraction tools won't accept the archive from stdin. With |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
370 | # these, the piping infrastructure we normally set up generally doesn't |
39
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
371 | # work, at least at first. We can still use most of it; we just don't |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
372 | # want to seed self.archive with the archive file, since that sucks up |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
373 | # memory. So instead we seed it with /dev/null, and specify the |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
374 | # filename on the command line as necessary. We also open the actual |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
375 | # file with os.open, to make sure we can actually do it (permissions |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
376 | # are good, etc.). This class doesn't do anything by itself; it's just |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
377 | # meant to be a base class for extractors that rely on these dumb |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
378 | # tools. |
32 | 379 | def __init__(self, filename, encoding): |
39
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
380 | os.close(os.open(filename, os.O_RDONLY)) |
32 | 381 | BaseExtractor.__init__(self, '/dev/null', None) |
382 | self.filename = os.path.realpath(filename) | |
383 | ||
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
384 | |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
385 | class ZipExtractor(NoPipeExtractor): |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
386 | def get_filenames(self): |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
387 | self.pipe(['zipinfo', '-1', self.filename], "listing") |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
388 | return BaseExtractor.get_filenames(self) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
389 | |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
390 | def extract_archive(self): |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
391 | self.pipe(['unzip', '-q', self.filename]) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
392 | self.run_pipes() |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
393 | |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
394 | |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
395 | class SevenExtractor(NoPipeExtractor): |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
396 | border_re = re.compile('^[- ]+$') |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
397 | |
32 | 398 | def get_filenames(self): |
399 | self.pipe(['7z', 'l', self.filename], "listing") | |
400 | self.run_pipes() | |
401 | self.archive.seek(0, 0) | |
402 | fn_index = None | |
403 | for line in self.archive: | |
404 | if self.border_re.match(line): | |
405 | if fn_index is not None: | |
406 | break | |
407 | else: | |
408 | fn_index = line.rindex(' ') + 1 | |
409 | elif fn_index is not None: | |
410 | yield line[fn_index:-1] | |
411 | self.archive.close() | |
412 | ||
413 | def extract_archive(self): | |
414 | self.pipe(['7z', 'x', self.filename]) | |
415 | self.run_pipes() | |
416 | ||
417 | ||
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
418 | class CABExtractor(NoPipeExtractor): |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
419 | border_re = re.compile(r'^[-\+]+$') |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
420 | |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
421 | def get_filenames(self): |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
422 | self.pipe(['cabextract', '-l', self.filename], "listing") |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
423 | self.run_pipes() |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
424 | self.archive.seek(0, 0) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
425 | fn_index = None |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
426 | for line in self.archive: |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
427 | if self.border_re.match(line): |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
428 | break |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
429 | for line in self.archive: |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
430 | try: |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
431 | yield line.split(' | ', 2)[2].rstrip('\n') |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
432 | except IndexError: |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
433 | break |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
434 | self.archive.close() |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
435 | |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
436 | def extract_archive(self): |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
437 | self.pipe(['cabextract', '-q', self.filename]) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
438 | self.run_pipes() |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
439 | |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
440 | |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
441 | class BaseHandler(object): |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
442 | def __init__(self, extractor, options): |
8
97388f5ff770
[svn] Make ExtractorApplication suck less. Now the strategies for handling
brett
parents:
7
diff
changeset
|
443 | self.extractor = extractor |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
444 | self.options = options |
17
481a2b4be471
[svn] Lots of tests for various boundary cases, and slightly better handling for
brett
parents:
16
diff
changeset
|
445 | self.target = None |
16
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
446 | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
447 | def handle(self): |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
448 | command = 'find' |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
449 | status = subprocess.call(['find', self.extractor.target, '-type', 'd', |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
450 | '-exec', 'chmod', 'u+rwx', '{}', ';']) |
8
97388f5ff770
[svn] Make ExtractorApplication suck less. Now the strategies for handling
brett
parents:
7
diff
changeset
|
451 | if status == 0: |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
452 | command = 'chmod' |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
453 | status = subprocess.call(['chmod', '-R', 'u+rwX', |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
454 | self.extractor.target]) |
8
97388f5ff770
[svn] Make ExtractorApplication suck less. Now the strategies for handling
brett
parents:
7
diff
changeset
|
455 | if status != 0: |
97388f5ff770
[svn] Make ExtractorApplication suck less. Now the strategies for handling
brett
parents:
7
diff
changeset
|
456 | return "%s returned with exit status %s" % (command, status) |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
457 | return self.organize() |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
458 | |
41
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
459 | def set_target(self, target, checker): |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
460 | self.target = checker(target).check() |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
461 | if self.target != target: |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
462 | logger.warning("extracting %s to %s" % |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
463 | (self.extractor.filename, self.target)) |
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
464 | |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
465 | |
16
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
466 | # The "where to extract" table, with options and archive types. |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
467 | # This dictates the contents of each can_handle method. |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
468 | # |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
469 | # Flat Overwrite None |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
470 | # File basename basename FilenameChecked |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
471 | # Match . . tempdir + checked |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
472 | # Bomb . basename DirectoryChecked |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
473 | |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
474 | class FlatHandler(BaseHandler): |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
475 | def can_handle(contents, options): |
22
b240777ae53e
[svn] Improve the way we check archive contents. If all the entries look like
brett
parents:
20
diff
changeset
|
476 | return ((options.flat and (contents != ONE_ENTRY_KNOWN)) or |
16
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
477 | (options.overwrite and (contents == MATCHING_DIRECTORY))) |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
478 | can_handle = staticmethod(can_handle) |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
479 | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
480 | def organize(self): |
17
481a2b4be471
[svn] Lots of tests for various boundary cases, and slightly better handling for
brett
parents:
16
diff
changeset
|
481 | self.target = '.' |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
482 | for curdir, dirs, filenames in os.walk(self.extractor.target, |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
483 | topdown=False): |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
484 | path_parts = curdir.split(os.sep) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
485 | if path_parts[0] == '.': |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
486 | del path_parts[1] |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
487 | else: |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
488 | del path_parts[0] |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
489 | newdir = os.path.join(*path_parts) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
490 | if not os.path.isdir(newdir): |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
491 | os.makedirs(newdir) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
492 | for filename in filenames: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
493 | os.rename(os.path.join(curdir, filename), |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
494 | os.path.join(newdir, filename)) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
495 | os.rmdir(curdir) |
16
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
496 | |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
497 | |
16
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
498 | class OverwriteHandler(BaseHandler): |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
499 | def can_handle(contents, options): |
22
b240777ae53e
[svn] Improve the way we check archive contents. If all the entries look like
brett
parents:
20
diff
changeset
|
500 | return ((options.flat and (contents == ONE_ENTRY_KNOWN)) or |
16
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
501 | (options.overwrite and (contents != MATCHING_DIRECTORY))) |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
502 | can_handle = staticmethod(can_handle) |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
503 | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
504 | def organize(self): |
16
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
505 | self.target = self.extractor.basename() |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
506 | result = run_command(['rm', '-rf', self.target], |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
507 | "removing %s to overwrite" % (self.target,)) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
508 | if result is None: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
509 | os.rename(self.extractor.target, self.target) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
510 | return result |
16
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
511 | |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
512 | |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
513 | class MatchHandler(BaseHandler): |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
514 | def can_handle(contents, options): |
20
69c93c3e6972
[svn] If the archive contains one directory with the "wrong" name, ask the user
brett
parents:
19
diff
changeset
|
515 | return ((contents == MATCHING_DIRECTORY) or |
22
b240777ae53e
[svn] Improve the way we check archive contents. If all the entries look like
brett
parents:
20
diff
changeset
|
516 | ((contents == ONE_ENTRY) and |
25
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
517 | options.one_entry_policy.ok_for_match())) |
16
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
518 | can_handle = staticmethod(can_handle) |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
519 | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
520 | def organize(self): |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
521 | source = os.path.join(self.extractor.target, |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
522 | os.listdir(self.extractor.target)[0]) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
523 | if os.path.isdir(source): |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
524 | checker = DirectoryChecker |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
525 | else: |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
526 | checker = FilenameChecker |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
527 | if self.options.one_entry_policy == EXTRACT_HERE: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
528 | destination = self.extractor.content_name.rstrip('/') |
20
69c93c3e6972
[svn] If the archive contains one directory with the "wrong" name, ask the user
brett
parents:
19
diff
changeset
|
529 | else: |
69c93c3e6972
[svn] If the archive contains one directory with the "wrong" name, ask the user
brett
parents:
19
diff
changeset
|
530 | destination = self.extractor.basename() |
41
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
531 | self.set_target(destination, checker) |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
532 | if os.path.isdir(self.extractor.target): |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
533 | os.rename(source, self.target) |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
534 | os.rmdir(self.extractor.target) |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
535 | else: |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
536 | os.rename(self.extractor.target, self.target) |
8
97388f5ff770
[svn] Make ExtractorApplication suck less. Now the strategies for handling
brett
parents:
7
diff
changeset
|
537 | |
97388f5ff770
[svn] Make ExtractorApplication suck less. Now the strategies for handling
brett
parents:
7
diff
changeset
|
538 | |
97388f5ff770
[svn] Make ExtractorApplication suck less. Now the strategies for handling
brett
parents:
7
diff
changeset
|
539 | class EmptyHandler(object): |
16
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
540 | def can_handle(contents, options): |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
541 | return contents == EMPTY |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
542 | can_handle = staticmethod(can_handle) |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
543 | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
544 | def __init__(self, extractor, options): pass |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
545 | def handle(self): pass |
8
97388f5ff770
[svn] Make ExtractorApplication suck less. Now the strategies for handling
brett
parents:
7
diff
changeset
|
546 | |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
547 | |
16
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
548 | class BombHandler(BaseHandler): |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
549 | def can_handle(contents, options): |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
550 | return True |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
551 | can_handle = staticmethod(can_handle) |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
552 | |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
553 | def organize(self): |
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
554 | basename = self.extractor.basename() |
41
e3675644bbb6
[svn] Minor clean-ups. The most important of these is that we now have a better
brett
parents:
40
diff
changeset
|
555 | self.set_target(basename, self.extractor.name_checker) |
28
4d88f2231d33
[svn] Change all the license notices from GPLv2 to GPLv3.
brett
parents:
27
diff
changeset
|
556 | os.rename(self.extractor.target, self.target) |
16
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
557 | |
29794d4d41aa
[svn] There's now an entirely new object hierarchy for handlers, because the
brett
parents:
15
diff
changeset
|
558 | |
25
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
559 | class BasePolicy(object): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
560 | def __init__(self, options): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
561 | self.current_policy = None |
26 | 562 | if options.batch: |
563 | self.permanent_policy = self.answers[''] | |
564 | else: | |
565 | self.permanent_policy = None | |
25
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
566 | |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
567 | def ask_question(self, question): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
568 | question = textwrap.wrap(question) + self.choices |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
569 | while True: |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
570 | print "\n".join(question) |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
571 | try: |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
572 | answer = raw_input(self.prompt) |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
573 | except EOFError: |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
574 | return self.answers[''] |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
575 | try: |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
576 | return self.answers[answer.lower()] |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
577 | except KeyError: |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
578 | |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
579 | |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
580 | def __cmp__(self, other): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
581 | return cmp(self.current_policy, other) |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
582 | |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
583 | |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
584 | class OneEntryPolicy(BasePolicy): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
585 | answers = {'h': EXTRACT_HERE, 'i': EXTRACT_WRAP, 'r': EXTRACT_RENAME, |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
586 | '': EXTRACT_WRAP} |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
587 | choices = ["You can:", |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
588 | " * extract it Inside another directory", |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
589 | " * extract it and Rename the directory", |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
590 | " * extract it Here"] |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
591 | prompt = "What do you want to do? (I/r/h) " |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
592 | |
26 | 593 | def prep(self, archive_filename, entry_name): |
594 | question = ("%s contains one entry: %s." % | |
595 | (archive_filename, entry_name)) | |
596 | self.current_policy = (self.permanent_policy or | |
597 | self.ask_question(question)) | |
25
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
598 | |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
599 | def ok_for_match(self): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
600 | return self.current_policy in (EXTRACT_RENAME, EXTRACT_HERE) |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
601 | |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
602 | |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
603 | class RecursionPolicy(BasePolicy): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
604 | answers = {'o': RECURSE_ONCE, 'a': RECURSE_ALWAYS, 'n': RECURSE_NOT_NOW, |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
605 | 'v': RECURSE_NEVER, '': RECURSE_NOT_NOW} |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
606 | choices = ["You can:", |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
607 | " * Always extract included archives", |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
608 | " * extract included archives this Once", |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
609 | " * choose Not to extract included archives", |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
610 | " * neVer extract included archives"] |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
611 | prompt = "What do you want to do? (a/o/N/v) " |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
612 | |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
613 | def __init__(self, options): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
614 | BasePolicy.__init__(self, options) |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
615 | if options.show_list: |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
616 | self.permanent_policy = RECURSE_NEVER |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
617 | elif options.recursive: |
25
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
618 | self.permanent_policy = RECURSE_ALWAYS |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
619 | |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
620 | def prep(self, current_filename, included_archives): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
621 | archive_count = len(included_archives) |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
622 | if (self.permanent_policy is not None) or (archive_count == 0): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
623 | self.current_policy = self.permanent_policy or RECURSE_NOT_NOW |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
624 | return |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
625 | elif archive_count > 1: |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
626 | question = ("%s contains %s other archive files." % |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
627 | (current_filename, archive_count)) |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
628 | else: |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
629 | question = ("%s contains another archive: %s." % |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
630 | (current_filename, included_archives[0])) |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
631 | self.current_policy = self.ask_question(question) |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
632 | if self.current_policy in (RECURSE_ALWAYS, RECURSE_NEVER): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
633 | self.permanent_policy = self.current_policy |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
634 | |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
635 | def ok_to_recurse(self): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
636 | return self.current_policy in (RECURSE_ALWAYS, RECURSE_ONCE) |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
637 | |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
638 | |
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
639 | class ExtractorBuilder(object): |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
640 | extractor_map = {'tar': (TarExtractor, None), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
641 | 'zip': (ZipExtractor, None), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
642 | 'deb': (DebExtractor, DebMetadataExtractor), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
643 | 'rpm': (RPMExtractor, None), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
644 | 'cpio': (CpioExtractor, None), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
645 | 'gem': (GemExtractor, GemMetadataExtractor), |
32 | 646 | 'compress': (CompressionExtractor, None), |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
647 | '7z': (SevenExtractor, None), |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
648 | 'cab': (CABExtractor, None)} |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
649 | |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
650 | mimetype_map = {} |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
651 | for mapping in (('tar', 'x-tar'), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
652 | ('zip', 'x-msdos-program', 'zip'), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
653 | ('deb', 'x-debian-package'), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
654 | ('rpm', 'x-redhat-package-manager', 'x-rpm'), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
655 | ('cpio', 'x-cpio'), |
32 | 656 | ('gem', 'x-ruby-gem'), |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
657 | ('7z', 'x-7z-compressed'), |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
658 | ('cab', 'x-cab')): |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
659 | for mimetype in mapping[1:]: |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
660 | if '/' not in mimetype: |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
661 | mimetype = 'application/' + mimetype |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
662 | mimetype_map[mimetype] = mapping[0] |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
663 | |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
664 | magic_mime_map = {} |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
665 | for mapping in (('deb', 'Debian binary package'), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
666 | ('cpio', 'cpio archive'), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
667 | ('tar', 'POSIX tar archive'), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
668 | ('zip', 'Zip archive'), |
32 | 669 | ('rpm', 'RPM'), |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
670 | ('7z', '7-zip archive'), |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
671 | ('cab', 'Microsoft Cabinet archive')): |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
672 | for pattern in mapping[1:]: |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
673 | magic_mime_map[re.compile(pattern)] = mapping[0] |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
674 | |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
675 | magic_encoding_map = {} |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
676 | for mapping in (('bzip2', 'bzip2 compressed'), |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
677 | ('gzip', 'gzip compressed')): |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
678 | for pattern in mapping[1:]: |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
679 | magic_encoding_map[re.compile(pattern)] = mapping[0] |
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
680 | |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
681 | extension_map = {} |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
682 | for mapping in (('tar', 'bzip2', 'tar.bz2'), |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
683 | ('tar', 'gzip', 'tar.gz', 'tgz'), |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
684 | ('tar', None, 'tar'), |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
685 | ('zip', None, 'zip', 'exe'), |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
686 | ('deb', None, 'deb'), |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
687 | ('rpm', None, 'rpm'), |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
688 | ('cpio', None, 'cpio'), |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
689 | ('gem', None, 'gem'), |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
690 | ('compress', 'gzip', 'Z', 'gz'), |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
691 | ('compress', 'bzip2', 'bz2'), |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
692 | ('compress', 'lzma', 'lzma'), |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
693 | ('7z', None, '7z'), |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
694 | ('cab', None, 'cab', 'exe')): |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
695 | for extension in mapping[2:]: |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
696 | extension_map.setdefault(extension, []).append(mapping[:2]) |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
697 | |
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
698 | def __init__(self, filename, options): |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
699 | self.filename = filename |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
700 | self.options = options |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
701 | |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
702 | def build_extractor(self, archive_type, encoding): |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
703 | extractors = self.extractor_map[archive_type] |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
704 | if self.options.metadata and (extractors[1] is not None): |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
705 | extractor = extractors[1] |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
706 | else: |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
707 | extractor = extractors[0] |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
708 | return extractor(self.filename, encoding) |
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
709 | |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
710 | def get_extractor(self): |
36
4bf2508d9b9e
[svn] Small optimization to be nice to the system: don't try a given extractor
brett
parents:
35
diff
changeset
|
711 | tried_types = Set() |
4bf2508d9b9e
[svn] Small optimization to be nice to the system: don't try a given extractor
brett
parents:
35
diff
changeset
|
712 | # As smart as it is, the magic test can't go first, because at least |
4bf2508d9b9e
[svn] Small optimization to be nice to the system: don't try a given extractor
brett
parents:
35
diff
changeset
|
713 | # on my system it just recognizes gem files as tar files. I guess |
4bf2508d9b9e
[svn] Small optimization to be nice to the system: don't try a given extractor
brett
parents:
35
diff
changeset
|
714 | # it's possible for the opposite problem to occur -- where the mimetype |
4bf2508d9b9e
[svn] Small optimization to be nice to the system: don't try a given extractor
brett
parents:
35
diff
changeset
|
715 | # or extension suggests something less than ideal -- but it seems less |
4bf2508d9b9e
[svn] Small optimization to be nice to the system: don't try a given extractor
brett
parents:
35
diff
changeset
|
716 | # likely so I'm sticking with this. |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
717 | for func_name in ('mimetype', 'extension', 'magic'): |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
718 | logger.debug("getting extractors by %s" % (func_name,)) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
719 | extractor_types = \ |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
720 | getattr(self, 'try_by_' + func_name)(self.filename) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
721 | logger.debug("done getting extractors") |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
722 | for ext_args in extractor_types: |
36
4bf2508d9b9e
[svn] Small optimization to be nice to the system: don't try a given extractor
brett
parents:
35
diff
changeset
|
723 | if ext_args in tried_types: |
4bf2508d9b9e
[svn] Small optimization to be nice to the system: don't try a given extractor
brett
parents:
35
diff
changeset
|
724 | continue |
4bf2508d9b9e
[svn] Small optimization to be nice to the system: don't try a given extractor
brett
parents:
35
diff
changeset
|
725 | tried_types.add(ext_args) |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
726 | logger.debug("trying %s extractor from %s" % |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
727 | (ext_args, func_name)) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
728 | yield self.build_extractor(*ext_args) |
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
729 | |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
730 | def try_by_mimetype(cls, filename): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
731 | mimetype, encoding = mimetypes.guess_type(filename) |
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
732 | try: |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
733 | return [(cls.mimetype_map[mimetype], encoding)] |
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
734 | except KeyError: |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
735 | if encoding: |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
736 | return [('compress', encoding)] |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
737 | return [] |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
738 | try_by_mimetype = classmethod(try_by_mimetype) |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
739 | |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
740 | def magic_map_matches(cls, output, magic_map): |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
741 | return [result for regexp, result in magic_map.items() |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
742 | if regexp.search(output)] |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
743 | magic_map_matches = classmethod(magic_map_matches) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
744 | |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
745 | def try_by_magic(cls, filename): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
746 | process = subprocess.Popen(['file', '-z', filename], |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
747 | stdout=subprocess.PIPE) |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
748 | status = process.wait() |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
749 | if status != 0: |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
750 | return [] |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
751 | output = process.stdout.readline() |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
752 | process.stdout.close() |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
753 | if output.startswith('%s: ' % filename): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
754 | output = output[len(filename) + 2:] |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
755 | mimes = cls.magic_map_matches(output, cls.magic_mime_map) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
756 | encodings = cls.magic_map_matches(output, cls.magic_encoding_map) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
757 | if mimes and not encodings: |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
758 | encodings = [None] |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
759 | elif encodings and not mimes: |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
760 | mimes = ['compress'] |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
761 | return [(m, e) for m in mimes for e in encodings] |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
762 | try_by_magic = classmethod(try_by_magic) |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
763 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
764 | def try_by_extension(cls, filename): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
765 | parts = filename.rsplit('.', 2)[1:] |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
766 | results = [] |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
767 | while parts: |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
768 | results.extend(cls.extension_map.get('.'.join(parts), [])) |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
769 | del parts[0] |
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
770 | return results |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
771 | try_by_extension = classmethod(try_by_extension) |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
772 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
773 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
774 | class BaseAction(object): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
775 | def __init__(self, options, filenames): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
776 | self.options = options |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
777 | self.filenames = filenames |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
778 | self.target = None |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
779 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
780 | def report(self, function, *args): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
781 | try: |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
782 | error = function(*args) |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
783 | except (ExtractorError, IOError, OSError), exception: |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
784 | error = str(exception) |
33
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
785 | logger.debug(''.join(traceback.format_exception(*sys.exc_info()))) |
39
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
786 | return error |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
787 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
788 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
789 | class ExtractionAction(BaseAction): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
790 | handlers = [FlatHandler, OverwriteHandler, MatchHandler, EmptyHandler, |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
791 | BombHandler] |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
792 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
793 | def get_handler(self, extractor): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
794 | if extractor.content_type == ONE_ENTRY: |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
795 | self.options.one_entry_policy.prep(self.current_filename, |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
796 | extractor.content_name) |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
797 | for handler in self.handlers: |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
798 | if handler.can_handle(extractor.content_type, self.options): |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
799 | logger.debug("using %s handler" % (handler.__name__,)) |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
800 | self.current_handler = handler(extractor, self.options) |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
801 | break |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
802 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
803 | def run(self, filename, extractor): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
804 | self.current_filename = filename |
39
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
805 | error = (self.report(extractor.extract) or |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
806 | self.report(self.get_handler, extractor) or |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
807 | self.report(self.current_handler.handle)) |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
808 | if not error: |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
809 | self.target = self.current_handler.target |
39
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
810 | return error |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
811 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
812 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
813 | class ListAction(BaseAction): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
814 | def __init__(self, options, filenames): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
815 | BaseAction.__init__(self, options, filenames) |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
816 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
817 | def get_list(self, extractor): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
818 | # Note: The reason I'm getting all the filenames up front is |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
819 | # because if we run into trouble partway through the archive, we'll |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
820 | # try another extractor. So before we display anything we have to |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
821 | # be sure this one is successful. We maybe don't have to be quite |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
822 | # this conservative but this is the easy way out for now. |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
823 | self.filelist = list(extractor.get_filenames()) |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
824 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
825 | def show_list(self, filename): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
826 | if len(self.filenames) != 1: |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
827 | if filename != self.filenames[0]: |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
828 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
829 | print "%s:" % (filename,) |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
830 | print '\n'.join(self.filelist) |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
831 | |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
832 | def run(self, filename, extractor): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
833 | self.current_filename = filename |
39
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
834 | return (self.report(self.get_list, extractor) or |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
835 | self.report(self.show_list, filename)) |
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
836 | |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
837 | |
5 | 838 | class ExtractorApplication(object): |
839 | def __init__(self, arguments): | |
6
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
840 | self.parse_options(arguments) |
12
5d202467c589
[svn] Introduce a real logging system. Right now all this really gets us is the
brett
parents:
11
diff
changeset
|
841 | self.setup_logger() |
5 | 842 | self.successes = [] |
843 | self.failures = [] | |
844 | ||
6
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
845 | def parse_options(self, arguments): |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
846 | parser = optparse.OptionParser( |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
847 | usage="%prog [options] archive [archive2 ...]", |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
848 | description="Intelligent archive extractor", |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
849 | version=VERSION_BANNER |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
850 | ) |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
851 | parser.add_option('-r', '--recursive', dest='recursive', |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
852 | action='store_true', default=False, |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
853 | help='extract archives contained in the ones listed') |
13
0a3ef1b9f6d4
[svn] Add options to tweak the logging level to taste.
brett
parents:
12
diff
changeset
|
854 | parser.add_option('-q', '--quiet', dest='quiet', |
0a3ef1b9f6d4
[svn] Add options to tweak the logging level to taste.
brett
parents:
12
diff
changeset
|
855 | action='count', default=3, |
0a3ef1b9f6d4
[svn] Add options to tweak the logging level to taste.
brett
parents:
12
diff
changeset
|
856 | help='suppress warning/error messages') |
0a3ef1b9f6d4
[svn] Add options to tweak the logging level to taste.
brett
parents:
12
diff
changeset
|
857 | parser.add_option('-v', '--verbose', dest='verbose', |
0a3ef1b9f6d4
[svn] Add options to tweak the logging level to taste.
brett
parents:
12
diff
changeset
|
858 | action='count', default=0, |
0a3ef1b9f6d4
[svn] Add options to tweak the logging level to taste.
brett
parents:
12
diff
changeset
|
859 | help='be verbose/print debugging information') |
14
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
860 | parser.add_option('-o', '--overwrite', dest='overwrite', |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
861 | action='store_true', default=False, |
6f9e1bb59719
[svn] Add support for just decompressing files that are compressed. So, if you
brett
parents:
13
diff
changeset
|
862 | help='overwrite any existing target directory') |
15
28dbd52a8bb8
[svn] Add a -f/--flat option, which will extract the archive contents into the
brett
parents:
14
diff
changeset
|
863 | parser.add_option('-f', '--flat', '--no-directory', dest='flat', |
28dbd52a8bb8
[svn] Add a -f/--flat option, which will extract the archive contents into the
brett
parents:
14
diff
changeset
|
864 | action='store_true', default=False, |
28dbd52a8bb8
[svn] Add a -f/--flat option, which will extract the archive contents into the
brett
parents:
14
diff
changeset
|
865 | help="don't put contents in their own directory") |
19 | 866 | parser.add_option('-l', '-t', '--list', '--table', dest='show_list', |
867 | action='store_true', default=False, | |
868 | help="list contents of archives on standard output") | |
20
69c93c3e6972
[svn] If the archive contains one directory with the "wrong" name, ask the user
brett
parents:
19
diff
changeset
|
869 | parser.add_option('-n', '--noninteractive', dest='batch', |
69c93c3e6972
[svn] If the archive contains one directory with the "wrong" name, ask the user
brett
parents:
19
diff
changeset
|
870 | action='store_true', default=False, |
69c93c3e6972
[svn] If the archive contains one directory with the "wrong" name, ask the user
brett
parents:
19
diff
changeset
|
871 | help="don't ask how to handle special cases") |
29
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
872 | parser.add_option('-m', '--metadata', dest='metadata', |
5fad99c17221
[svn] Add support for Ruby Gems, and extracting metadata from .deb/.gem files.
brett
parents:
28
diff
changeset
|
873 | action='store_true', default=False, |
35
957b402d4b90
[svn] Add support for extracting CAB archives. Because the CAB archive I was
brett
parents:
34
diff
changeset
|
874 | help="extract metadata from a .deb/.gem") |
6
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
875 | self.options, filenames = parser.parse_args(arguments) |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
876 | if not filenames: |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
877 | parser.error("you did not list any archives") |
25
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
878 | self.options.one_entry_policy = OneEntryPolicy(self.options) |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
879 | self.options.recursion_policy = RecursionPolicy(self.options) |
6
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
880 | self.archives = {os.path.realpath(os.curdir): filenames} |
77043f4e6a9f
[svn] The big thing here is recursive extraction. Find archive files in the
brett
parents:
5
diff
changeset
|
881 | |
12
5d202467c589
[svn] Introduce a real logging system. Right now all this really gets us is the
brett
parents:
11
diff
changeset
|
882 | def setup_logger(self): |
33
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
883 | # WARNING is the default. |
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
884 | log_level = (10 * (self.options.quiet - self.options.verbose)) |
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
885 | logging.getLogger().setLevel(log_level) |
12
5d202467c589
[svn] Introduce a real logging system. Right now all this really gets us is the
brett
parents:
11
diff
changeset
|
886 | handler = logging.StreamHandler() |
33
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
887 | handler.setLevel(log_level) |
19 | 888 | formatter = logging.Formatter("dtrx: %(levelname)s: %(message)s") |
12
5d202467c589
[svn] Introduce a real logging system. Right now all this really gets us is the
brett
parents:
11
diff
changeset
|
889 | handler.setFormatter(formatter) |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
890 | logger.addHandler(handler) |
33
3547e3124729
[svn] Fix some bugs and make things a little more user-friendly now that we can
brett
parents:
32
diff
changeset
|
891 | logger.debug("logger is set up") |
1 | 892 | |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
893 | def recurse(self, filename, extractor, action): |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
894 | archives = extractor.included_archives |
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
895 | self.options.recursion_policy.prep(filename, archives) |
25
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
896 | if self.options.recursion_policy.ok_to_recurse(): |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
897 | for filename in archives: |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
898 | tail_path, basename = os.path.split(filename) |
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
899 | directory = os.path.join(self.current_directory, |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
900 | action.target, tail_path) |
25
ef62f2f55eb8
[svn] Move policy-handling code into a dedicated set of classes. This makes
brett
parents:
23
diff
changeset
|
901 | self.archives.setdefault(directory, []).append(basename) |
8
97388f5ff770
[svn] Make ExtractorApplication suck less. Now the strategies for handling
brett
parents:
7
diff
changeset
|
902 | |
39
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
903 | def check_file(self, filename): |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
904 | try: |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
905 | result = os.stat(filename) |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
906 | except OSError, error: |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
907 | return error.strerror |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
908 | if stat.S_ISDIR(result.st_mode): |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
909 | return "cannot extract a directory" |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
910 | |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
911 | def try_extractors(self, filename, builder): |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
912 | last_error = "could not find a way to extract this" |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
913 | while True: |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
914 | try: |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
915 | extractor = builder.next() |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
916 | except StopIteration: |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
917 | return last_error |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
918 | except (IOError, OSError, ExtractorError), error: |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
919 | return str(error) |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
920 | error = self.action.run(filename, extractor) |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
921 | if error: |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
922 | logger.info("%s: %s" % (filename, error)) |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
923 | last_error = error |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
924 | else: |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
925 | self.recurse(filename, extractor, self.action) |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
926 | return |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
927 | |
19 | 928 | def run(self): |
929 | if self.options.show_list: | |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
930 | action = ListAction |
19 | 931 | else: |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
932 | action = ExtractionAction |
39
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
933 | self.action = action(self.options, self.archives.values()[0]) |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
934 | while self.archives: |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
935 | self.current_directory, self.filenames = self.archives.popitem() |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
936 | os.chdir(self.current_directory) |
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
937 | for filename in self.filenames: |
31
c3a2760d1c3a
[svn] Refactor actions (extract the archive, vs. list the contents) into their
brett
parents:
30
diff
changeset
|
938 | builder = ExtractorBuilder(filename, self.options) |
39
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
939 | error = (self.check_file(filename) or |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
940 | self.try_extractors(filename, builder.get_extractor())) |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
941 | if error: |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
942 | logger.error("%s: %s" % (filename, error)) |
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
943 | self.failures.append(filename) |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
944 | else: |
39
027fcd7ae002
[svn] Improve the error reporting to be more user-friendly, at least in many of
brett
parents:
36
diff
changeset
|
945 | self.successes.append(filename) |
30
1015bbd6dc5e
[svn] If we can't figure out what the file is by mimetype, try using the file
brett
parents:
29
diff
changeset
|
946 | self.options.one_entry_policy.permanent_policy = EXTRACT_WRAP |
5 | 947 | if self.failures: |
948 | return 1 | |
949 | return 0 | |
950 | ||
1 | 951 | |
952 | if __name__ == '__main__': | |
5 | 953 | app = ExtractorApplication(sys.argv[1:]) |
954 | sys.exit(app.run()) |