Mercurial > repos > shellac > guppy_basecaller

annotate env/lib/python3.7/site-packages/bagit.py @ 0:26e78fe6e8c4 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
author shellac
date Sat, 02 May 2020 07:14:21 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1 #!/usr/bin/env python
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
2 # encoding: utf-8
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
4 from __future__ import absolute_import, division, print_function, unicode_literals
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
5
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
6 import argparse
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
7 import codecs
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
8 import gettext
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
9 import hashlib
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
10 import logging
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
11 import multiprocessing
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
12 import os
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
13 import re
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
14 import signal
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
15 import sys
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
16 import tempfile
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
17 import unicodedata
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
18 import warnings
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
19 from collections import defaultdict
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
20 from datetime import date
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
21 from functools import partial
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
22 from os.path import abspath, isdir, isfile, join
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
23
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
24 from pkg_resources import DistributionNotFound, get_distribution
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
25
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
26 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
27 from urllib.parse import urlparse
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
28 except ImportError:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
29 from urlparse import urlparse
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
30
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
31
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
32 def find_locale_dir():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
33 for prefix in (os.path.dirname(__file__), sys.prefix):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
34 locale_dir = os.path.join(prefix, "locale")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
35 if os.path.isdir(locale_dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
36 return locale_dir
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
37
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
38
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
39 TRANSLATION_CATALOG = gettext.translation(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
40 "bagit-python", localedir=find_locale_dir(), fallback=True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
41 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
42 if sys.version_info < (3,):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
43 _ = TRANSLATION_CATALOG.ugettext
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
44 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
45 _ = TRANSLATION_CATALOG.gettext
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
46
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
47 MODULE_NAME = "bagit" if __name__ == "__main__" else __name__
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
48
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
49 LOGGER = logging.getLogger(MODULE_NAME)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
50
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
51 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
52 VERSION = get_distribution(MODULE_NAME).version
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
53 except DistributionNotFound:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
54 VERSION = "0.0.dev0"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
55
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
56 PROJECT_URL = "https://github.com/LibraryOfCongress/bagit-python"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
57
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
58 __doc__ = (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
59 _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
60 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
61 BagIt is a directory, filename convention for bundling an arbitrary set of
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
62 files with a manifest, checksums, and additional metadata. More about BagIt
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
63 can be found at:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
64
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
65 http://purl.org/net/bagit
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
66
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
67 bagit.py is a pure python drop in library and command line tool for creating,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
68 and working with BagIt directories.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
69
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
70
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
71 Command-Line Usage:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
72
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
73 Basic usage is to give bagit.py a directory to bag up:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
74
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
75 $ bagit.py my_directory
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
76
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
77 This does a bag-in-place operation where the current contents will be moved
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
78 into the appropriate BagIt structure and the metadata files will be created.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
79
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
80 You can bag multiple directories if you wish:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
81
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
82 $ bagit.py directory1 directory2
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
83
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
84 Optionally you can provide metadata which will be stored in bag-info.txt:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
85
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
86 $ bagit.py --source-organization "Library of Congress" directory
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
87
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
88 You can also select which manifest algorithms will be used:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
89
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
90 $ bagit.py --sha1 --md5 --sha256 --sha512 directory
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
91
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
92
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
93 Using BagIt from your Python code:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
94
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
95 import bagit
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
96 bag = bagit.make_bag('example-directory', {'Contact-Name': 'Ed Summers'})
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
97 print(bag.entries)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
98
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
99 For more information or to contribute to bagit-python's development, please
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
100 visit %(PROJECT_URL)s
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
101 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
102 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
103 % globals()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
104 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
105
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
106 # standard bag-info.txt metadata
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
107 STANDARD_BAG_INFO_HEADERS = [
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
108 "Source-Organization",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
109 "Organization-Address",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
110 "Contact-Name",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
111 "Contact-Phone",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
112 "Contact-Email",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
113 "External-Description",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
114 "External-Identifier",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
115 "Bag-Size",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
116 "Bag-Group-Identifier",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
117 "Bag-Count",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
118 "Internal-Sender-Identifier",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
119 "Internal-Sender-Description",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
120 "BagIt-Profile-Identifier",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
121 # Bagging-Date is autogenerated
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
122 # Payload-Oxum is autogenerated
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
123 ]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
124
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
125 CHECKSUM_ALGOS = hashlib.algorithms_guaranteed
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
126 DEFAULT_CHECKSUMS = ["sha256", "sha512"]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
127
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
128 #: Block size used when reading files for hashing:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
129 HASH_BLOCK_SIZE = 512 * 1024
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
130
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
131 #: Convenience function used everywhere we want to open a file to read text
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
132 #: rather than undecoded bytes:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
133 open_text_file = partial(codecs.open, encoding="utf-8", errors="strict")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
134
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
135 # This is the same as decoding the byte values in codecs.BOM:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
136 UNICODE_BYTE_ORDER_MARK = "\uFEFF"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
137
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
138
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
139 def make_bag(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
140 bag_dir, bag_info=None, processes=1, checksums=None, checksum=None, encoding="utf-8"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
141 ):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
142 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
143 Convert a given directory into a bag. You can pass in arbitrary
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
144 key/value pairs to put into the bag-info.txt metadata file as
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
145 the bag_info dictionary.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
146 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
147
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
148 if checksum is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
149 warnings.warn(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
150 _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
151 "The `checksum` argument for `make_bag` should be replaced with `checksums`"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
152 ),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
153 DeprecationWarning,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
154 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
155 checksums = checksum
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
156
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
157 if checksums is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
158 checksums = DEFAULT_CHECKSUMS
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
159
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
160 bag_dir = os.path.abspath(bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
161 cwd = os.path.abspath(os.path.curdir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
162
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
163 if cwd.startswith(bag_dir) and cwd != bag_dir:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
164 raise RuntimeError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
165 _("Bagging a parent of the current directory is not supported")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
166 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
167
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
168 LOGGER.info(_("Creating bag for directory %s"), bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
169
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
170 if not os.path.isdir(bag_dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
171 LOGGER.error(_("Bag directory %s does not exist"), bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
172 raise RuntimeError(_("Bag directory %s does not exist") % bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
173
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
174 # FIXME: we should do the permissions checks before changing directories
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
175 old_dir = os.path.abspath(os.path.curdir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
176
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
177 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
178 # TODO: These two checks are currently redundant since an unreadable directory will also
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
179 # often be unwritable, and this code will require review when we add the option to
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
180 # bag to a destination other than the source. It would be nice if we could avoid
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
181 # walking the directory tree more than once even if most filesystems will cache it
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
182
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
183 unbaggable = _can_bag(bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
184
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
185 if unbaggable:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
186 LOGGER.error(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
187 _("Unable to write to the following directories and files:\n%s"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
188 unbaggable,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
189 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
190 raise BagError(_("Missing permissions to move all files and directories"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
191
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
192 unreadable_dirs, unreadable_files = _can_read(bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
193
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
194 if unreadable_dirs or unreadable_files:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
195 if unreadable_dirs:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
196 LOGGER.error(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
197 _("The following directories do not have read permissions:\n%s"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
198 unreadable_dirs,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
199 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
200 if unreadable_files:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
201 LOGGER.error(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
202 _("The following files do not have read permissions:\n%s"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
203 unreadable_files,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
204 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
205 raise BagError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
206 _("Read permissions are required to calculate file fixities")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
207 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
208 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
209 LOGGER.info(_("Creating data directory"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
210
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
211 # FIXME: if we calculate full paths we won't need to deal with changing directories
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
212 os.chdir(bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
213 cwd = os.getcwd()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
214 temp_data = tempfile.mkdtemp(dir=cwd)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
215
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
216 for f in os.listdir("."):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
217 if os.path.abspath(f) == temp_data:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
218 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
219 new_f = os.path.join(temp_data, f)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
220 LOGGER.info(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
221 _("Moving %(source)s to %(destination)s"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
222 {"source": f, "destination": new_f},
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
223 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
224 os.rename(f, new_f)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
225
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
226 LOGGER.info(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
227 _("Moving %(source)s to %(destination)s"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
228 {"source": temp_data, "destination": "data"},
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
229 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
230 os.rename(temp_data, "data")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
231
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
232 # permissions for the payload directory should match those of the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
233 # original directory
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
234 os.chmod("data", os.stat(cwd).st_mode)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
235
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
236 total_bytes, total_files = make_manifests(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
237 "data", processes, algorithms=checksums, encoding=encoding
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
238 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
239
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
240 LOGGER.info(_("Creating bagit.txt"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
241 txt = """BagIt-Version: 0.97\nTag-File-Character-Encoding: UTF-8\n"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
242 with open_text_file("bagit.txt", "w") as bagit_file:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
243 bagit_file.write(txt)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
244
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
245 LOGGER.info(_("Creating bag-info.txt"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
246 if bag_info is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
247 bag_info = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
248
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
249 # allow 'Bagging-Date' and 'Bag-Software-Agent' to be overidden
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
250 if "Bagging-Date" not in bag_info:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
251 bag_info["Bagging-Date"] = date.strftime(date.today(), "%Y-%m-%d")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
252 if "Bag-Software-Agent" not in bag_info:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
253 bag_info["Bag-Software-Agent"] = "bagit.py v%s <%s>" % (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
254 VERSION,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
255 PROJECT_URL,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
256 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
257
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
258 bag_info["Payload-Oxum"] = "%s.%s" % (total_bytes, total_files)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
259 _make_tag_file("bag-info.txt", bag_info)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
260
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
261 for c in checksums:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
262 _make_tagmanifest_file(c, bag_dir, encoding="utf-8")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
263 except Exception:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
264 LOGGER.exception(_("An error occurred creating a bag in %s"), bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
265 raise
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
266 finally:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
267 os.chdir(old_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
268
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
269 return Bag(bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
270
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
271
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
272 class Bag(object):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
273 """A representation of a bag."""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
274
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
275 valid_files = ["bagit.txt", "fetch.txt"]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
276 valid_directories = ["data"]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
277
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
278 def __init__(self, path=None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
279 super(Bag, self).__init__()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
280 self.tags = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
281 self.info = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
282 #: Dictionary of manifest entries and the checksum values for each
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
283 #: algorithm:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
284 self.entries = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
285
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
286 # To reliably handle Unicode normalization differences, we maintain
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
287 # lookup dictionaries in both directions for the filenames read from
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
288 # the filesystem and the manifests so we can handle cases where the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
289 # normalization form changed between the bag being created and read.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
290 # See https://github.com/LibraryOfCongress/bagit-python/issues/51.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
291
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
292 #: maps Unicode-normalized values to the raw value from the filesystem
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
293 self.normalized_filesystem_names = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
294
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
295 #: maps Unicode-normalized values to the raw value in the manifest
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
296 self.normalized_manifest_names = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
297
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
298 self.algorithms = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
299 self.tag_file_name = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
300 self.path = abspath(path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
301 if path:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
302 # if path ends in a path separator, strip it off
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
303 if path[-1] == os.sep:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
304 self.path = path[:-1]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
305 self._open()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
306
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
307 def __str__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
308 # FIXME: develop a more informative string representation for a Bag
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
309 return self.path
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
310
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
311 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
312 def algs(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
313 warnings.warn(_("Use Bag.algorithms instead of Bag.algs"), DeprecationWarning)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
314 return self.algorithms
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
315
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
316 @property
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
317 def version(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
318 warnings.warn(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
319 _("Use the Bag.version_info tuple instead of Bag.version"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
320 DeprecationWarning,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
321 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
322 return self._version
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
323
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
324 def _open(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
325 # Open the bagit.txt file, and load any tags from it, including
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
326 # the required version and encoding.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
327 bagit_file_path = os.path.join(self.path, "bagit.txt")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
328
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
329 if not isfile(bagit_file_path):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
330 raise BagError(_("Expected bagit.txt does not exist: %s") % bagit_file_path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
331
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
332 self.tags = tags = _load_tag_file(bagit_file_path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
333
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
334 required_tags = ("BagIt-Version", "Tag-File-Character-Encoding")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
335 missing_tags = [i for i in required_tags if i not in tags]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
336 if missing_tags:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
337 raise BagError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
338 _("Missing required tag in bagit.txt: %s") % ", ".join(missing_tags)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
339 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
340
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
341 # To avoid breaking existing code we'll leave self.version as the string
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
342 # and parse it into a numeric version_info tuple. In version 2.0 we can
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
343 # break that.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
344
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
345 self._version = tags["BagIt-Version"]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
346
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
347 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
348 self.version_info = tuple(int(i) for i in self._version.split(".", 1))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
349 except ValueError:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
350 raise BagError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
351 _("Bag version numbers must be MAJOR.MINOR numbers, not %s")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
352 % self._version
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
353 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
354
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
355 if (0, 93) <= self.version_info <= (0, 95):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
356 self.tag_file_name = "package-info.txt"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
357 elif (0, 96) <= self.version_info < (2,):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
358 self.tag_file_name = "bag-info.txt"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
359 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
360 raise BagError(_("Unsupported bag version: %s") % self._version)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
361
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
362 self.encoding = tags["Tag-File-Character-Encoding"]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
363
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
364 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
365 codecs.lookup(self.encoding)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
366 except LookupError:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
367 raise BagValidationError(_("Unsupported encoding: %s") % self.encoding)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
368
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
369 info_file_path = os.path.join(self.path, self.tag_file_name)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
370 if os.path.exists(info_file_path):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
371 self.info = _load_tag_file(info_file_path, encoding=self.encoding)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
372
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
373 self._load_manifests()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
374
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
375 def manifest_files(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
376 for filename in ["manifest-%s.txt" % a for a in CHECKSUM_ALGOS]:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
377 f = os.path.join(self.path, filename)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
378 if isfile(f):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
379 yield f
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
380
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
381 def tagmanifest_files(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
382 for filename in ["tagmanifest-%s.txt" % a for a in CHECKSUM_ALGOS]:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
383 f = os.path.join(self.path, filename)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
384 if isfile(f):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
385 yield f
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
386
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
387 def compare_manifests_with_fs(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
388 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
389 Compare the filenames in the manifests to the filenames present on the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
390 local filesystem and returns two lists of the files which are only
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
391 present in the manifests and the files which are only present on the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
392 local filesystem, respectively.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
393 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
394
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
395 # We compare the filenames after Unicode normalization so we can
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
396 # reliably detect normalization changes after bag creation:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
397 files_on_fs = set(normalize_unicode(i) for i in self.payload_files())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
398 files_in_manifest = set(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
399 normalize_unicode(i) for i in self.payload_entries().keys()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
400 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
401
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
402 if self.version_info >= (0, 97):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
403 files_in_manifest.update(self.missing_optional_tagfiles())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
404
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
405 only_on_fs = list()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
406 only_in_manifest = list()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
407
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
408 for i in files_on_fs.difference(files_in_manifest):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
409 only_on_fs.append(self.normalized_filesystem_names[i])
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
410
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
411 for i in files_in_manifest.difference(files_on_fs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
412 only_in_manifest.append(self.normalized_manifest_names[i])
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
413
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
414 return only_in_manifest, only_on_fs
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
415
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
416 def compare_fetch_with_fs(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
417 """Compares the fetch entries with the files actually
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
418 in the payload, and returns a list of all the files
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
419 that still need to be fetched.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
420 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
421
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
422 files_on_fs = set(self.payload_files())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
423 files_in_fetch = set(self.files_to_be_fetched())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
424
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
425 return list(files_in_fetch - files_on_fs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
426
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
427 def payload_files(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
428 """Returns a list of filenames which are present on the local filesystem"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
429 payload_dir = os.path.join(self.path, "data")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
430
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
431 for dirpath, _, filenames in os.walk(payload_dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
432 for f in filenames:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
433 # Jump through some hoops here to make the payload files are
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
434 # returned with the directory structure relative to the base
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
435 # directory rather than the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
436 normalized_f = os.path.normpath(f)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
437 rel_path = os.path.relpath(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
438 os.path.join(dirpath, normalized_f), start=self.path
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
439 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
440
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
441 self.normalized_filesystem_names[normalize_unicode(rel_path)] = rel_path
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
442 yield rel_path
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
443
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
444 def payload_entries(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
445 """Return a dictionary of items """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
446 # Don't use dict comprehension (compatibility with Python < 2.7)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
447 return dict(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
448 (key, value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
449 for (key, value) in self.entries.items()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
450 if key.startswith("data" + os.sep)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
451 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
452
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
453 def save(self, processes=1, manifests=False):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
454 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
455 save will persist any changes that have been made to the bag
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
456 metadata (self.info).
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
457
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
458 If you have modified the payload of the bag (added, modified,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
459 removed files in the data directory) and want to regenerate manifests
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
460 set the manifests parameter to True. The default is False since you
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
461 wouldn't want a save to accidentally create a new manifest for
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
462 a corrupted bag.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
463
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
464 If you want to control the number of processes that are used when
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
465 recalculating checksums use the processes parameter.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
466 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
467 # Error checking
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
468 if not self.path:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
469 raise BagError(_("Bag.save() called before setting the path!"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
470
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
471 if not os.access(self.path, os.R_OK | os.W_OK | os.X_OK):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
472 raise BagError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
473 _("Cannot save bag to non-existent or inaccessible directory %s")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
474 % self.path
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
475 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
476
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
477 unbaggable = _can_bag(self.path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
478 if unbaggable:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
479 LOGGER.error(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
480 _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
481 "Missing write permissions for the following directories and files:\n%s"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
482 ),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
483 unbaggable,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
484 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
485 raise BagError(_("Missing permissions to move all files and directories"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
486
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
487 unreadable_dirs, unreadable_files = _can_read(self.path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
488 if unreadable_dirs or unreadable_files:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
489 if unreadable_dirs:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
490 LOGGER.error(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
491 _("The following directories do not have read permissions:\n%s"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
492 unreadable_dirs,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
493 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
494 if unreadable_files:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
495 LOGGER.error(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
496 _("The following files do not have read permissions:\n%s"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
497 unreadable_files,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
498 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
499 raise BagError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
500 _("Read permissions are required to calculate file fixities")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
501 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
502
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
503 # Change working directory to bag directory so helper functions work
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
504 old_dir = os.path.abspath(os.path.curdir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
505 os.chdir(self.path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
506
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
507 # Generate new manifest files
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
508 if manifests:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
509 total_bytes, total_files = make_manifests(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
510 "data", processes, algorithms=self.algorithms, encoding=self.encoding
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
511 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
512
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
513 # Update Payload-Oxum
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
514 LOGGER.info(_("Updating Payload-Oxum in %s"), self.tag_file_name)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
515 self.info["Payload-Oxum"] = "%s.%s" % (total_bytes, total_files)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
516
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
517 _make_tag_file(self.tag_file_name, self.info)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
518
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
519 # Update tag-manifest for changes to manifest & bag-info files
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
520 for alg in self.algorithms:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
521 _make_tagmanifest_file(alg, self.path, encoding=self.encoding)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
522
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
523 # Reload the manifests
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
524 self._load_manifests()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
525
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
526 os.chdir(old_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
527
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
528 def tagfile_entries(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
529 return dict(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
530 (key, value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
531 for (key, value) in self.entries.items()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
532 if not key.startswith("data" + os.sep)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
533 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
534
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
535 def missing_optional_tagfiles(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
536 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
537 From v0.97 we need to validate any tagfiles listed
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
538 in the optional tagmanifest(s). As there is no mandatory
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
539 directory structure for additional tagfiles we can
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
540 only check for entries with missing files (not missing
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
541 entries for existing files).
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
542 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
543 for tagfilepath in self.tagfile_entries().keys():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
544 if not os.path.isfile(os.path.join(self.path, tagfilepath)):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
545 yield tagfilepath
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
546
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
547 def fetch_entries(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
548 """Load fetch.txt if present and iterate over its contents
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
549
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
550 yields (url, size, filename) tuples
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
551
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
552 raises BagError for errors such as an unsafe filename referencing
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
553 data outside of the bag directory
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
554 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
555
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
556 fetch_file_path = os.path.join(self.path, "fetch.txt")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
557
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
558 if isfile(fetch_file_path):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
559 with open_text_file(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
560 fetch_file_path, "r", encoding=self.encoding
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
561 ) as fetch_file:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
562 for line in fetch_file:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
563 url, file_size, filename = line.strip().split(None, 2)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
564
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
565 if self._path_is_dangerous(filename):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
566 raise BagError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
567 _('Path "%(payload_file)s" in "%(source_file)s" is unsafe')
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
568 % {
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
569 "payload_file": filename,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
570 "source_file": os.path.join(self.path, "fetch.txt"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
571 }
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
572 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
573
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
574 yield url, file_size, filename
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
575
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
576 def files_to_be_fetched(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
577 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
578 Convenience wrapper for fetch_entries which returns only the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
579 local filename
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
580 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
581
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
582 for url, file_size, filename in self.fetch_entries():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
583 yield filename
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
584
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
585 def has_oxum(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
586 return "Payload-Oxum" in self.info
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
587
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
588 def validate(self, processes=1, fast=False, completeness_only=False):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
589 """Checks the structure and contents are valid.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
590
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
591 If you supply the parameter fast=True the Payload-Oxum (if present) will
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
592 be used to check that the payload files are present and accounted for,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
593 instead of re-calculating fixities and comparing them against the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
594 manifest. By default validate() will re-calculate fixities (fast=False).
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
595 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
596
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
597 self._validate_structure()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
598 self._validate_bagittxt()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
599
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
600 self.validate_fetch()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
601
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
602 self._validate_contents(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
603 processes=processes, fast=fast, completeness_only=completeness_only
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
604 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
605
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
606 return True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
607
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
608 def is_valid(self, fast=False, completeness_only=False):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
609 """Returns validation success or failure as boolean.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
610 Optional fast parameter passed directly to validate().
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
611 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
612
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
613 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
614 self.validate(fast=fast, completeness_only=completeness_only)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
615 except BagError:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
616 return False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
617
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
618 return True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
619
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
620 def _load_manifests(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
621 self.entries = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
622 manifests = list(self.manifest_files())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
623
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
624 if self.version_info >= (0, 97):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
625 # v0.97+ requires that optional tagfiles are verified.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
626 manifests += list(self.tagmanifest_files())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
627
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
628 for manifest_filename in manifests:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
629 if not manifest_filename.find("tagmanifest-") is -1:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
630 search = "tagmanifest-"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
631 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
632 search = "manifest-"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
633 alg = (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
634 os.path.basename(manifest_filename)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
635 .replace(search, "")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
636 .replace(".txt", "")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
637 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
638 if alg not in self.algorithms:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
639 self.algorithms.append(alg)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
640
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
641 with open_text_file(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
642 manifest_filename, "r", encoding=self.encoding
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
643 ) as manifest_file:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
644 if manifest_file.encoding.startswith("UTF"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
645 # We'll check the first character to see if it's a BOM:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
646 if manifest_file.read(1) == UNICODE_BYTE_ORDER_MARK:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
647 # We'll skip it either way by letting line decoding
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
648 # happen at the new offset but we will issue a warning
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
649 # for UTF-8 since the presence of a BOM is contrary to
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
650 # the BagIt specification:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
651 if manifest_file.encoding == "UTF-8":
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
652 LOGGER.warning(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
653 _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
654 "%s is encoded using UTF-8 but contains an unnecessary"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
655 " byte-order mark, which is not in compliance with the"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
656 " BagIt RFC"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
657 ),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
658 manifest_file.name,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
659 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
660 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
661 manifest_file.seek(0) # Pretend the first read never happened
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
662
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
663 for line in manifest_file:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
664 line = line.strip()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
665
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
666 # Ignore blank lines and comments.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
667 if line == "" or line.startswith("#"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
668 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
669
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
670 entry = line.split(None, 1)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
671
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
672 # Format is FILENAME *CHECKSUM
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
673 if len(entry) != 2:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
674 LOGGER.error(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
675 _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
676 "%(bag)s: Invalid %(algorithm)s manifest entry: %(line)s"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
677 ),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
678 {"bag": self, "algorithm": alg, "line": line},
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
679 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
680 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
681
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
682 entry_hash = entry[0]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
683 entry_path = os.path.normpath(entry[1].lstrip("*"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
684 entry_path = _decode_filename(entry_path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
685
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
686 if self._path_is_dangerous(entry_path):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
687 raise BagError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
688 _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
689 'Path "%(payload_file)s" in manifest "%(manifest_file)s" is unsafe'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
690 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
691 % {
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
692 "payload_file": entry_path,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
693 "manifest_file": manifest_file.name,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
694 }
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
695 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
696
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
697 entry_hashes = self.entries.setdefault(entry_path, {})
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
698
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
699 if alg in entry_hashes:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
700 warning_ctx = {
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
701 "bag": self,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
702 "algorithm": alg,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
703 "filename": entry_path,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
704 }
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
705 if entry_hashes[alg] == entry_hash:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
706 msg = _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
707 "%(bag)s: %(algorithm)s manifest lists %(filename)s"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
708 " multiple times with the same value"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
709 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
710 if self.version_info >= (1,):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
711 raise BagError(msg % warning_ctx)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
712 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
713 LOGGER.warning(msg, warning_ctx)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
714 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
715 raise BagError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
716 _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
717 "%(bag)s: %(algorithm)s manifest lists %(filename)s"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
718 " multiple times with conflicting values"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
719 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
720 % warning_ctx
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
721 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
722
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
723 entry_hashes[alg] = entry_hash
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
724
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
725 self.normalized_manifest_names.update(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
726 (normalize_unicode(i), i) for i in self.entries.keys()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
727 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
728
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
729 def _validate_structure(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
730 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
731 Checks the structure of the bag to determine whether it conforms to the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
732 BagIt spec. Returns true on success, otherwise it will raise a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
733 BagValidationError exception.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
734 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
735
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
736 self._validate_structure_payload_directory()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
737 self._validate_structure_tag_files()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
738
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
739 def _validate_structure_payload_directory(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
740 data_dir_path = os.path.join(self.path, "data")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
741
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
742 if not isdir(data_dir_path):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
743 raise BagValidationError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
744 _("Expected data directory %s does not exist") % data_dir_path
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
745 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
746
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
747 def _validate_structure_tag_files(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
748 # Note: we deviate somewhat from v0.96 of the spec in that it allows
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
749 # other files and directories to be present in the base directory
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
750
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
751 if not list(self.manifest_files()):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
752 raise BagValidationError(_("No manifest files found"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
753 if "bagit.txt" not in os.listdir(self.path):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
754 raise BagValidationError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
755 _('Expected %s to contain "bagit.txt"') % self.path
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
756 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
757
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
758 def validate_fetch(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
759 """Validate the fetch.txt file
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
760
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
761 Raises `BagError` for errors and otherwise returns no value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
762 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
763
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
764 for url, file_size, filename in self.fetch_entries():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
765 # fetch_entries will raise a BagError for unsafe filenames
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
766 # so at this point we will check only that the URL is minimally
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
767 # well formed:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
768 parsed_url = urlparse(url)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
769
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
770 if not all((parsed_url.scheme, parsed_url.netloc)):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
771 raise BagError(_("Malformed URL in fetch.txt: %s") % url)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
772
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
773 def _validate_contents(self, processes=1, fast=False, completeness_only=False):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
774 if fast and not self.has_oxum():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
775 raise BagValidationError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
776 _("Fast validation requires bag-info.txt to include Payload-Oxum")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
777 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
778
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
779 # Perform the fast file count + size check so we can fail early:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
780 self._validate_oxum()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
781
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
782 if fast:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
783 return
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
784
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
785 self._validate_completeness()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
786
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
787 if completeness_only:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
788 return
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
789
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
790 self._validate_entries(processes)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
791
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
792 def _validate_oxum(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
793 oxum = self.info.get("Payload-Oxum")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
794
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
795 if oxum is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
796 return
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
797
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
798 # If multiple Payload-Oxum tags (bad idea)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
799 # use the first listed in bag-info.txt
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
800 if isinstance(oxum, list):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
801 LOGGER.warning(_("bag-info.txt defines multiple Payload-Oxum values!"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
802 oxum = oxum[0]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
803
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
804 oxum_byte_count, oxum_file_count = oxum.split(".", 1)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
805
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
806 if not oxum_byte_count.isdigit() or not oxum_file_count.isdigit():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
807 raise BagError(_("Malformed Payload-Oxum value: %s") % oxum)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
808
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
809 oxum_byte_count = int(oxum_byte_count)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
810 oxum_file_count = int(oxum_file_count)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
811 total_bytes = 0
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
812 total_files = 0
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
813
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
814 for payload_file in self.payload_files():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
815 payload_file = os.path.join(self.path, payload_file)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
816 total_bytes += os.stat(payload_file).st_size
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
817 total_files += 1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
818
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
819 if oxum_file_count != total_files or oxum_byte_count != total_bytes:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
820 raise BagValidationError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
821 _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
822 "Payload-Oxum validation failed."
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
823 " Expected %(oxum_file_count)d files and %(oxum_byte_count)d bytes"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
824 " but found %(found_file_count)d files and %(found_byte_count)d bytes"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
825 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
826 % {
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
827 "found_file_count": total_files,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
828 "found_byte_count": total_bytes,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
829 "oxum_file_count": oxum_file_count,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
830 "oxum_byte_count": oxum_byte_count,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
831 }
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
832 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
833
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
834 def _validate_completeness(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
835 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
836 Verify that the actual file manifests match the files in the data directory
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
837 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
838 errors = list()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
839
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
840 # First we'll make sure there's no mismatch between the filesystem
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
841 # and the list of files in the manifest(s)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
842 only_in_manifests, only_on_fs = self.compare_manifests_with_fs()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
843 for path in only_in_manifests:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
844 e = FileMissing(path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
845 LOGGER.warning(force_unicode(e))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
846 errors.append(e)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
847 for path in only_on_fs:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
848 e = UnexpectedFile(path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
849 LOGGER.warning(force_unicode(e))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
850 errors.append(e)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
851
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
852 if errors:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
853 raise BagValidationError(_("Bag validation failed"), errors)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
854
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
855 def _validate_entries(self, processes):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
856 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
857 Verify that the actual file contents match the recorded hashes stored in the manifest files
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
858 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
859 errors = list()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
860
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
861 if os.name == "posix":
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
862 worker_init = posix_multiprocessing_worker_initializer
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
863 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
864 worker_init = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
865
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
866 args = (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
867 (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
868 self.path,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
869 self.normalized_filesystem_names.get(rel_path, rel_path),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
870 hashes,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
871 self.algorithms,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
872 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
873 for rel_path, hashes in self.entries.items()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
874 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
875
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
876 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
877 if processes == 1:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
878 hash_results = [_calc_hashes(i) for i in args]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
879 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
880 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
881 pool = multiprocessing.Pool(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
882 processes if processes else None, initializer=worker_init
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
883 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
884 hash_results = pool.map(_calc_hashes, args)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
885 finally:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
886 pool.terminate()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
887
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
888 # Any unhandled exceptions are probably fatal
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
889 except:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
890 LOGGER.exception(_("Unable to calculate file hashes for %s"), self)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
891 raise
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
892
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
893 for rel_path, f_hashes, hashes in hash_results:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
894 for alg, computed_hash in f_hashes.items():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
895 stored_hash = hashes[alg]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
896 if stored_hash.lower() != computed_hash:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
897 e = ChecksumMismatch(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
898 rel_path, alg, stored_hash.lower(), computed_hash
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
899 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
900 LOGGER.warning(force_unicode(e))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
901 errors.append(e)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
902
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
903 if errors:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
904 raise BagValidationError(_("Bag validation failed"), errors)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
905
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
906 def _validate_bagittxt(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
907 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
908 Verify that bagit.txt conforms to specification
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
909 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
910 bagit_file_path = os.path.join(self.path, "bagit.txt")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
911
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
912 # Note that we are intentionally opening this file in binary mode so we can confirm
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
913 # that it does not start with the UTF-8 byte-order-mark
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
914 with open(bagit_file_path, "rb") as bagit_file:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
915 first_line = bagit_file.read(4)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
916 if first_line.startswith(codecs.BOM_UTF8):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
917 raise BagValidationError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
918 _("bagit.txt must not contain a byte-order mark")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
919 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
920
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
921 def _path_is_dangerous(self, path):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
922 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
923 Return true if path looks dangerous, i.e. potentially operates
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
924 outside the bagging directory structure, e.g. ~/.bashrc, ../../../secrets.json,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
925 \\?\c:\, D:\sys32\cmd.exe
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
926 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
927 if os.path.isabs(path):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
928 return True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
929 if os.path.expanduser(path) != path:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
930 return True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
931 if os.path.expandvars(path) != path:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
932 return True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
933 real_path = os.path.realpath(os.path.join(self.path, path))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
934 real_path = os.path.normpath(real_path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
935 bag_path = os.path.realpath(self.path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
936 bag_path = os.path.normpath(bag_path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
937 common = os.path.commonprefix((bag_path, real_path))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
938 return not (common == bag_path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
939
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
940
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
941 class BagError(Exception):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
942 pass
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
943
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
944
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
945 class BagValidationError(BagError):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
946 def __init__(self, message, details=None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
947 super(BagValidationError, self).__init__()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
948
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
949 if details is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
950 details = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
951
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
952 self.message = message
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
953 self.details = details
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
954
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
955 def __str__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
956 if len(self.details) > 0:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
957 details = "; ".join([force_unicode(e) for e in self.details])
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
958 return "%s: %s" % (self.message, details)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
959 return self.message
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
960
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
961
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
962 class ManifestErrorDetail(BagError):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
963 def __init__(self, path):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
964 super(ManifestErrorDetail, self).__init__()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
965
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
966 self.path = path
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
967
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
968
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
969 class ChecksumMismatch(ManifestErrorDetail):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
970 def __init__(self, path, algorithm=None, expected=None, found=None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
971 super(ChecksumMismatch, self).__init__(path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
972
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
973 self.path = path
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
974 self.algorithm = algorithm
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
975 self.expected = expected
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
976 self.found = found
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
977
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
978 def __str__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
979 return _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
980 '%(path)s %(algorithm)s validation failed: expected="%(expected)s" found="%(found)s"'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
981 ) % {
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
982 "path": force_unicode(self.path),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
983 "algorithm": self.algorithm,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
984 "expected": self.expected,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
985 "found": self.found,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
986 }
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
987
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
988
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
989 class FileMissing(ManifestErrorDetail):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
990 def __str__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
991 return _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
992 "%s exists in manifest but was not found on filesystem"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
993 ) % force_unicode(self.path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
994
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
995
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
996 class UnexpectedFile(ManifestErrorDetail):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
997 def __str__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
998 return _("%s exists on filesystem but is not in the manifest") % self.path
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
999
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1000
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1001 class FileNormalizationConflict(BagError):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1002 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1003 Exception raised when two files differ only in normalization and thus
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1004 are not safely portable
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1005 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1006
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1007 def __init__(self, file_a, file_b):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1008 super(FileNormalizationConflict, self).__init__()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1009
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1010 self.file_a = file_a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1011 self.file_b = file_b
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1012
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1013 def __str__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1014 return _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1015 'Unicode normalization conflict for file "%(file_a)s" and "%(file_b)s"'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1016 ) % {"file_a": self.file_a, "file_b": self.file_b}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1017
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1018
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1019 def posix_multiprocessing_worker_initializer():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1020 """Ignore SIGINT in multiprocessing workers on POSIX systems"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1021 signal.signal(signal.SIGINT, signal.SIG_IGN)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1022
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1023
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1024 # The Unicode normalization form used here doesn't matter – all we care about
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1025 # is consistency since the input value will be preserved:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1026
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1027
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1028 def normalize_unicode_py3(s):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1029 return unicodedata.normalize("NFC", s)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1030
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1031
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1032 def normalize_unicode_py2(s):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1033 if isinstance(s, str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1034 s = s.decode("utf-8")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1035 return unicodedata.normalize("NFC", s)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1036
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1037
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1038 if sys.version_info > (3, 0):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1039 normalize_unicode = normalize_unicode_py3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1040 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1041 normalize_unicode = normalize_unicode_py2
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1042
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1043
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1044 def build_unicode_normalized_lookup_dict(filenames):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1045 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1046 Return a dictionary mapping unicode-normalized filenames to as-encoded
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1047 values to efficiently detect conflicts between the filesystem and manifests.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1048
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1049 This is necessary because some filesystems and utilities may automatically
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1050 apply a different Unicode normalization form to filenames than was applied
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1051 when the bag was originally created.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1052
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1053 The best known example of this is when a bag is created using a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1054 normalization form other than NFD and then transferred to a Mac where the
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1055 HFS+ filesystem will transparently normalize filenames to a variant of NFD
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1056 for every call:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1057
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1058 https://developer.apple.com/legacy/library/technotes/tn/tn1150.html#UnicodeSubtleties
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1059
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1060 Windows is documented as storing filenames exactly as provided:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1061
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1062 https://msdn.microsoft.com/en-us/library/windows/desktop/aa365247%28v=vs.85%29.aspx
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1063
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1064 Linux performs no normalization in the kernel but it is technically
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1065 valid for a filesystem to perform normalization, such as when an HFS+
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1066 volume is mounted.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1067
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1068 See http://www.unicode.org/reports/tr15/ for a full discussion of
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1069 equivalence and normalization in Unicode.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1070 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1071
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1072 output = dict()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1073
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1074 for filename in filenames:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1075 normalized_filename = normalize_unicode(filename)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1076 if normalized_filename in output:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1077 raise FileNormalizationConflict(filename, output[normalized_filename])
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1078 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1079 output[normalized_filename] = filename
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1080
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1081 return output
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1082
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1083
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1084 def get_hashers(algorithms):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1085 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1086 Given a list of algorithm names, return a dictionary of hasher instances
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1087
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1088 This avoids redundant code between the creation and validation code where in
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1089 both cases we want to avoid reading the same file more than once. The
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1090 intended use is a simple for loop:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1091
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1092 for block in file:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1093 for hasher in hashers.values():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1094 hasher.update(block)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1095 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1096
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1097 hashers = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1098
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1099 for alg in algorithms:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1100 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1101 hasher = hashlib.new(alg)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1102 except ValueError:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1103 LOGGER.warning(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1104 _("Disabling requested hash algorithm %s: hashlib does not support it"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1105 alg,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1106 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1107 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1108
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1109 hashers[alg] = hasher
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1110
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1111 if not hashers:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1112 raise ValueError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1113 _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1114 "Unable to continue: hashlib does not support any of the requested algorithms!"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1115 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1116 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1117
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1118 return hashers
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1119
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1120
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1121 def _calc_hashes(args):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1122 # auto unpacking of sequences illegal in Python3
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1123 (base_path, rel_path, hashes, algorithms) = args
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1124 full_path = os.path.join(base_path, rel_path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1125
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1126 # Create a clone of the default empty hash objects:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1127 f_hashers = dict((alg, hashlib.new(alg)) for alg in hashes if alg in algorithms)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1128
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1129 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1130 f_hashes = _calculate_file_hashes(full_path, f_hashers)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1131 except BagValidationError as e:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1132 f_hashes = dict((alg, force_unicode(e)) for alg in f_hashers.keys())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1133
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1134 return rel_path, f_hashes, hashes
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1135
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1136
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1137 def _calculate_file_hashes(full_path, f_hashers):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1138 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1139 Returns a dictionary of (algorithm, hexdigest) values for the provided
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1140 filename
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1141 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1142 LOGGER.info(_("Verifying checksum for file %s"), full_path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1143
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1144 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1145 with open(full_path, "rb") as f:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1146 while True:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1147 block = f.read(HASH_BLOCK_SIZE)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1148 if not block:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1149 break
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1150 for i in f_hashers.values():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1151 i.update(block)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1152 except (OSError, IOError) as e:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1153 raise BagValidationError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1154 _("Could not read %(filename)s: %(error)s")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1155 % {"filename": full_path, "error": force_unicode(e)}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1156 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1157
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1158 return dict((alg, h.hexdigest()) for alg, h in f_hashers.items())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1159
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1160
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1161 def _load_tag_file(tag_file_name, encoding="utf-8-sig"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1162 with open_text_file(tag_file_name, "r", encoding=encoding) as tag_file:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1163 # Store duplicate tags as list of vals
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1164 # in order of parsing under the same key.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1165 tags = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1166 for name, value in _parse_tags(tag_file):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1167 if name not in tags:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1168 tags[name] = value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1169 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1170
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1171 if not isinstance(tags[name], list):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1172 tags[name] = [tags[name], value]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1173 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1174 tags[name].append(value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1175
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1176 return tags
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1177
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1178
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1179 def _parse_tags(tag_file):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1180 """Parses a tag file, according to RFC 2822. This
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1181 includes line folding, permitting extra-long
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1182 field values.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1183
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1184 See http://www.faqs.org/rfcs/rfc2822.html for
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1185 more information.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1186 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1187
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1188 tag_name = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1189 tag_value = None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1190
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1191 # Line folding is handled by yielding values only after we encounter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1192 # the start of a new tag, or if we pass the EOF.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1193 for num, line in enumerate(tag_file):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1194 # Skip over any empty or blank lines.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1195 if len(line) == 0 or line.isspace():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1196 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1197 elif line[0].isspace() and tag_value is not None: # folded line
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1198 tag_value += line
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1199 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1200 # Starting a new tag; yield the last one.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1201 if tag_name:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1202 yield (tag_name, tag_value.strip())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1203
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1204 if ":" not in line:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1205 raise BagValidationError(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1206 _("%(filename)s contains invalid tag: %(line)s")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1207 % {
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1208 "line": line.strip(),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1209 "filename": os.path.basename(tag_file.name),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1210 }
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1211 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1212
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1213 parts = line.strip().split(":", 1)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1214 tag_name = parts[0].strip()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1215 tag_value = parts[1]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1216
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1217 # Passed the EOF. All done after this.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1218 if tag_name:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1219 yield (tag_name, tag_value.strip())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1220
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1221
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1222 def _make_tag_file(bag_info_path, bag_info):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1223 headers = sorted(bag_info.keys())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1224 with open_text_file(bag_info_path, "w") as f:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1225 for h in headers:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1226 values = bag_info[h]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1227 if not isinstance(values, list):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1228 values = [values]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1229 for txt in values:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1230 # strip CR, LF and CRLF so they don't mess up the tag file
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1231 txt = re.sub(r"\n|\r|(\r\n)", "", force_unicode(txt))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1232 f.write("%s: %s\n" % (h, txt))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1233
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1234
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1235 def make_manifests(data_dir, processes, algorithms=DEFAULT_CHECKSUMS, encoding="utf-8"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1236 LOGGER.info(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1237 _("Using %(process_count)d processes to generate manifests: %(algorithms)s"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1238 {"process_count": processes, "algorithms": ", ".join(algorithms)},
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1239 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1240
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1241 manifest_line_generator = partial(generate_manifest_lines, algorithms=algorithms)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1242
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1243 if processes > 1:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1244 pool = multiprocessing.Pool(processes=processes)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1245 checksums = pool.map(manifest_line_generator, _walk(data_dir))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1246 pool.close()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1247 pool.join()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1248 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1249 checksums = [manifest_line_generator(i) for i in _walk(data_dir)]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1250
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1251 # At this point we have a list of tuples which start with the algorithm name:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1252 manifest_data = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1253 for batch in checksums:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1254 for entry in batch:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1255 manifest_data.setdefault(entry[0], []).append(entry[1:])
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1256
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1257 # These will be keyed on the algorithm name so we can perform sanity checks
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1258 # below to catch failures in the hashing process:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1259 num_files = defaultdict(lambda: 0)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1260 total_bytes = defaultdict(lambda: 0)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1261
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1262 for algorithm, values in manifest_data.items():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1263 manifest_filename = "manifest-%s.txt" % algorithm
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1264
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1265 with open_text_file(manifest_filename, "w", encoding=encoding) as manifest:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1266 for digest, filename, byte_count in values:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1267 manifest.write("%s %s\n" % (digest, _encode_filename(filename)))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1268 num_files[algorithm] += 1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1269 total_bytes[algorithm] += byte_count
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1270
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1271 # We'll use sets of the values for the error checks and eventually return the payload oxum values:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1272 byte_value_set = set(total_bytes.values())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1273 file_count_set = set(num_files.values())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1274
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1275 # allow a bag with an empty payload
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1276 if not byte_value_set and not file_count_set:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1277 return 0, 0
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1278
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1279 if len(file_count_set) != 1:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1280 raise RuntimeError(_("Expected the same number of files for each checksum"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1281
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1282 if len(byte_value_set) != 1:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1283 raise RuntimeError(_("Expected the same number of bytes for each checksums"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1284
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1285 return byte_value_set.pop(), file_count_set.pop()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1286
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1287
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1288 def _make_tagmanifest_file(alg, bag_dir, encoding="utf-8"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1289 tagmanifest_file = join(bag_dir, "tagmanifest-%s.txt" % alg)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1290 LOGGER.info(_("Creating %s"), tagmanifest_file)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1291
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1292 checksums = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1293 for f in _find_tag_files(bag_dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1294 if re.match(r"^tagmanifest-.+\.txt$", f):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1295 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1296 with open(join(bag_dir, f), "rb") as fh:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1297 m = hashlib.new(alg)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1298 while True:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1299 block = fh.read(HASH_BLOCK_SIZE)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1300 if not block:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1301 break
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1302 m.update(block)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1303 checksums.append((m.hexdigest(), f))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1304
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1305 with open_text_file(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1306 join(bag_dir, tagmanifest_file), mode="w", encoding=encoding
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1307 ) as tagmanifest:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1308 for digest, filename in checksums:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1309 tagmanifest.write("%s %s\n" % (digest, filename))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1310
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1311
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1312 def _find_tag_files(bag_dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1313 for dir in os.listdir(bag_dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1314 if dir != "data":
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1315 if os.path.isfile(dir) and not dir.startswith("tagmanifest-"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1316 yield dir
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1317 for dir_name, _, filenames in os.walk(dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1318 for filename in filenames:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1319 if filename.startswith("tagmanifest-"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1320 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1321 # remove everything up to the bag_dir directory
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1322 p = join(dir_name, filename)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1323 yield os.path.relpath(p, bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1324
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1325
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1326 def _walk(data_dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1327 for dirpath, dirnames, filenames in os.walk(data_dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1328 # if we don't sort here the order of entries is non-deterministic
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1329 # which makes it hard to test the fixity of tagmanifest-md5.txt
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1330 filenames.sort()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1331 dirnames.sort()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1332 for fn in filenames:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1333 path = os.path.join(dirpath, fn)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1334 # BagIt spec requires manifest to always use '/' as path separator
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1335 if os.path.sep != "/":
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1336 parts = path.split(os.path.sep)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1337 path = "/".join(parts)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1338 yield path
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1339
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1340
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1341 def _can_bag(test_dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1342 """Scan the provided directory for files which cannot be bagged due to insufficient permissions"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1343 unbaggable = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1344
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1345 if not os.access(test_dir, os.R_OK):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1346 # We cannot continue without permission to read the source directory
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1347 unbaggable.append(test_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1348 return unbaggable
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1349
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1350 if not os.access(test_dir, os.W_OK):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1351 unbaggable.append(test_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1352
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1353 for dirpath, dirnames, filenames in os.walk(test_dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1354 for directory in dirnames:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1355 full_path = os.path.join(dirpath, directory)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1356 if not os.access(full_path, os.W_OK):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1357 unbaggable.append(full_path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1358
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1359 return unbaggable
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1360
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1361
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1362 def _can_read(test_dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1363 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1364 returns ((unreadable_dirs), (unreadable_files))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1365 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1366 unreadable_dirs = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1367 unreadable_files = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1368
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1369 if not os.access(test_dir, os.R_OK):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1370 unreadable_dirs.append(test_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1371 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1372 for dirpath, dirnames, filenames in os.walk(test_dir):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1373 for dn in dirnames:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1374 full_path = os.path.join(dirpath, dn)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1375 if not os.access(full_path, os.R_OK):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1376 unreadable_dirs.append(full_path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1377 for fn in filenames:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1378 full_path = os.path.join(dirpath, fn)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1379 if not os.access(full_path, os.R_OK):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1380 unreadable_files.append(full_path)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1381 return (tuple(unreadable_dirs), tuple(unreadable_files))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1382
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1383
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1384 def generate_manifest_lines(filename, algorithms=DEFAULT_CHECKSUMS):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1385 LOGGER.info(_("Generating manifest lines for file %s"), filename)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1386
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1387 # For performance we'll read the file only once and pass it block
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1388 # by block to every requested hash algorithm:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1389 hashers = get_hashers(algorithms)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1390
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1391 total_bytes = 0
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1392
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1393 with open(filename, "rb") as f:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1394 while True:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1395 block = f.read(HASH_BLOCK_SIZE)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1396
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1397 if not block:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1398 break
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1399
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1400 total_bytes += len(block)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1401 for hasher in hashers.values():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1402 hasher.update(block)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1403
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1404 decoded_filename = _decode_filename(filename)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1405
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1406 # We'll generate a list of results in roughly manifest format but prefixed with the algorithm:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1407 results = [
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1408 (alg, hasher.hexdigest(), decoded_filename, total_bytes)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1409 for alg, hasher in hashers.items()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1410 ]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1411
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1412 return results
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1413
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1414
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1415 def _encode_filename(s):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1416 s = s.replace("\r", "%0D")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1417 s = s.replace("\n", "%0A")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1418 return s
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1419
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1420
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1421 def _decode_filename(s):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1422 s = re.sub(r"%0D", "\r", s, re.IGNORECASE)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1423 s = re.sub(r"%0A", "\n", s, re.IGNORECASE)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1424 return s
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1425
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1426
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1427 def force_unicode_py2(s):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1428 """Reliably return a Unicode string given a possible unicode or byte string"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1429 if isinstance(s, str):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1430 return s.decode("utf-8")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1431 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1432 return unicode(s)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1433
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1434
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1435 if sys.version_info > (3, 0):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1436 force_unicode = str
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1437 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1438 force_unicode = force_unicode_py2
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1439
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1440 # following code is used for command line program
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1441
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1442
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1443 class BagArgumentParser(argparse.ArgumentParser):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1444 def __init__(self, *args, **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1445 self.bag_info = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1446 argparse.ArgumentParser.__init__(self, *args, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1447
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1448
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1449 class BagHeaderAction(argparse.Action):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1450 def __call__(self, parser, _, values, option_string=None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1451 opt = option_string.lstrip("--")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1452 opt_caps = "-".join([o.capitalize() for o in opt.split("-")])
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1453 parser.bag_info[opt_caps] = values
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1454
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1455
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1456 def _make_parser():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1457 parser = BagArgumentParser(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1458 formatter_class=argparse.RawDescriptionHelpFormatter,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1459 description="bagit-python version %s\n\n%s\n" % (VERSION, __doc__.strip()),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1460 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1461 parser.add_argument(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1462 "--processes",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1463 type=int,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1464 dest="processes",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1465 default=1,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1466 help=_(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1467 "Use multiple processes to calculate checksums faster (default: %(default)s)"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1468 ),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1469 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1470 parser.add_argument("--log", help=_("The name of the log file (default: stdout)"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1471 parser.add_argument(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1472 "--quiet",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1473 action="store_true",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1474 help=_("Suppress all progress information other than errors"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1475 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1476 parser.add_argument(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1477 "--validate",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1478 action="store_true",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1479 help=_(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1480 "Validate existing bags in the provided directories instead of"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1481 " creating new ones"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1482 ),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1483 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1484 parser.add_argument(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1485 "--fast",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1486 action="store_true",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1487 help=_(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1488 "Modify --validate behaviour to only test whether the bag directory"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1489 " has the number of files and total size specified in Payload-Oxum"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1490 " without performing checksum validation to detect corruption."
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1491 ),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1492 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1493 parser.add_argument(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1494 "--completeness-only",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1495 action="store_true",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1496 help=_(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1497 "Modify --validate behaviour to test whether the bag directory"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1498 " has the expected payload specified in the checksum manifests"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1499 " without performing checksum validation to detect corruption."
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1500 ),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1501 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1502
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1503 checksum_args = parser.add_argument_group(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1504 _("Checksum Algorithms"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1505 _(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1506 "Select the manifest algorithms to be used when creating bags"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1507 " (default=%s)"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1508 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1509 % ", ".join(DEFAULT_CHECKSUMS),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1510 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1511
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1512 for i in CHECKSUM_ALGOS:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1513 alg_name = re.sub(r"^([A-Z]+)(\d+)$", r"\1-\2", i.upper())
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1514 checksum_args.add_argument(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1515 "--%s" % i,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1516 action="append_const",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1517 dest="checksums",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1518 const=i,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1519 help=_("Generate %s manifest when creating a bag") % alg_name,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1520 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1521
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1522 metadata_args = parser.add_argument_group(_("Optional Bag Metadata"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1523 for header in STANDARD_BAG_INFO_HEADERS:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1524 metadata_args.add_argument(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1525 "--%s" % header.lower(), type=str, action=BagHeaderAction
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1526 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1527
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1528 parser.add_argument(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1529 "directory",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1530 nargs="+",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1531 help=_(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1532 "Directory which will be converted into a bag in place"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1533 " by moving any existing files into the BagIt structure"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1534 " and creating the manifests and other metadata."
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1535 ),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1536 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1537
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1538 return parser
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1539
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1540
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1541 def _configure_logging(opts):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1542 log_format = "%(asctime)s - %(levelname)s - %(message)s"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1543 if opts.quiet:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1544 level = logging.ERROR
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1545 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1546 level = logging.INFO
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1547 if opts.log:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1548 logging.basicConfig(filename=opts.log, level=level, format=log_format)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1549 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1550 logging.basicConfig(level=level, format=log_format)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1551
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1552
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1553 def main():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1554 if "--version" in sys.argv:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1555 print(_("bagit-python version %s") % VERSION)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1556 sys.exit(0)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1557
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1558 parser = _make_parser()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1559 args = parser.parse_args()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1560
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1561 if args.processes < 0:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1562 parser.error(_("The number of processes must be 0 or greater"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1563
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1564 if args.fast and not args.validate:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1565 parser.error(_("--fast is only allowed as an option for --validate!"))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1566
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1567 _configure_logging(args)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1568
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1569 rc = 0
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1570 for bag_dir in args.directory:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1571 # validate the bag
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1572 if args.validate:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1573 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1574 bag = Bag(bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1575 # validate throws a BagError or BagValidationError
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1576 bag.validate(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1577 processes=args.processes,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1578 fast=args.fast,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1579 completeness_only=args.completeness_only,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1580 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1581 if args.fast:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1582 LOGGER.info(_("%s valid according to Payload-Oxum"), bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1583 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1584 LOGGER.info(_("%s is valid"), bag_dir)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1585 except BagError as e:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1586 LOGGER.error(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1587 _("%(bag)s is invalid: %(error)s"), {"bag": bag_dir, "error": e}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1588 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1589 rc = 1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1590
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1591 # make the bag
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1592 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1593 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1594 make_bag(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1595 bag_dir,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1596 bag_info=parser.bag_info,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1597 processes=args.processes,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1598 checksums=args.checksums,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1599 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1600 except Exception as exc:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1601 LOGGER.error(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1602 _("Failed to create bag in %(bag_directory)s: %(error)s"),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1603 {"bag_directory": bag_dir, "error": exc},
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1604 exc_info=True,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1605 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1606 rc = 1
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1607
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1608 sys.exit(rc)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1609
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1610
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1611 if __name__ == "__main__":
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff changeset
1612 main()