Mercurial > repos > shellac > guppy_basecaller
annotate env/lib/python3.7/site-packages/galaxy/util/checkers.py @ 2:6af9afd405e9 draft
"planemo upload commit 0a63dd5f4d38a1f6944587f52a8cd79874177fc1"
author | shellac |
---|---|
date | Thu, 14 May 2020 14:56:58 -0400 |
parents | 26e78fe6e8c4 |
children |
rev | line source |
---|---|
0
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1 import gzip |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
2 import re |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
3 import sys |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
4 import tarfile |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
5 import zipfile |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
6 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
7 from six import BytesIO |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
8 from six.moves import filter |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
9 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
10 from galaxy import util |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
11 from galaxy.util.image_util import image_type |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
12 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
13 if sys.version_info < (3, 3): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
14 gzip.GzipFile.read1 = gzip.GzipFile.read # workaround for https://bugs.python.org/issue12591 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
15 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
16 import bz2file as bz2 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
17 except ImportError: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
18 # If bz2file is unavailable, just fallback to not having pbzip2 support. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
19 import bz2 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
20 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
21 import bz2 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
22 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
23 HTML_CHECK_LINES = 100 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
24 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
25 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
26 def check_html(file_path, chunk=None): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
27 if chunk is None: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
28 temp = open(file_path, mode='rb') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
29 elif hasattr(chunk, "splitlines"): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
30 temp = chunk.splitlines() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
31 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
32 temp = chunk |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
33 regexp1 = re.compile(r"<A\s+[^>]*HREF[^>]+>", re.I) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
34 regexp2 = re.compile(r"<IFRAME[^>]*>", re.I) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
35 regexp3 = re.compile(r"<FRAMESET[^>]*>", re.I) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
36 regexp4 = re.compile(r"<META[\W][^>]*>", re.I) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
37 regexp5 = re.compile(r"<SCRIPT[^>]*>", re.I) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
38 lineno = 0 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
39 # TODO: Potentially reading huge lines into string here, this should be |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
40 # reworked. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
41 for line in temp: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
42 line = util.unicodify(line) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
43 lineno += 1 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
44 matches = regexp1.search(line) or regexp2.search(line) or regexp3.search(line) or regexp4.search(line) or regexp5.search(line) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
45 if matches: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
46 if chunk is None: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
47 temp.close() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
48 return True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
49 if HTML_CHECK_LINES and (lineno > HTML_CHECK_LINES): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
50 break |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
51 if chunk is None: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
52 temp.close() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
53 return False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
54 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
55 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
56 def check_binary(name, file_path=True): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
57 # Handles files if file_path is True or text if file_path is False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
58 if file_path: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
59 temp = open(name, "rb") |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
60 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
61 temp = BytesIO(name) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
62 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
63 return util.is_binary(temp.read(1024)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
64 finally: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
65 temp.close() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
66 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
67 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
68 def check_gzip(file_path, check_content=True): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
69 # This method returns a tuple of booleans representing ( is_gzipped, is_valid ) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
70 # Make sure we have a gzipped file |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
71 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
72 with open(file_path, "rb") as temp: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
73 magic_check = temp.read(2) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
74 if magic_check != util.gzip_magic: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
75 return (False, False) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
76 except Exception: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
77 return (False, False) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
78 # We support some binary data types, so check if the compressed binary file is valid |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
79 # If the file is Bam, it should already have been detected as such, so we'll just check |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
80 # for sff format. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
81 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
82 with gzip.open(file_path, 'rb') as fh: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
83 header = fh.read(4) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
84 if header == b'.sff': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
85 return (True, True) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
86 except Exception: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
87 return(False, False) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
88 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
89 if not check_content: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
90 return (True, True) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
91 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
92 CHUNK_SIZE = 2 ** 15 # 32Kb |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
93 gzipped_file = gzip.GzipFile(file_path, mode='rb') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
94 chunk = gzipped_file.read(CHUNK_SIZE) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
95 gzipped_file.close() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
96 # See if we have a compressed HTML file |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
97 if check_html(file_path, chunk=chunk): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
98 return (True, False) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
99 return (True, True) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
100 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
101 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
102 def check_bz2(file_path, check_content=True): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
103 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
104 with open(file_path, "rb") as temp: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
105 magic_check = temp.read(3) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
106 if magic_check != util.bz2_magic: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
107 return (False, False) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
108 except Exception: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
109 return(False, False) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
110 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
111 if not check_content: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
112 return (True, True) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
113 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
114 CHUNK_SIZE = 2 ** 15 # reKb |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
115 bzipped_file = bz2.BZ2File(file_path, mode='rb') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
116 chunk = bzipped_file.read(CHUNK_SIZE) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
117 bzipped_file.close() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
118 # See if we have a compressed HTML file |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
119 if check_html(file_path, chunk=chunk): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
120 return (True, False) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
121 return (True, True) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
122 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
123 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
124 def check_zip(file_path, check_content=True, files=1): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
125 if not zipfile.is_zipfile(file_path): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
126 return (False, False) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
127 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
128 if not check_content: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
129 return (True, True) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
130 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
131 CHUNK_SIZE = 2 ** 15 # 32Kb |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
132 chunk = None |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
133 for filect, member in enumerate(iter_zip(file_path)): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
134 handle, name = member |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
135 chunk = handle.read(CHUNK_SIZE) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
136 if chunk and check_html(file_path, chunk): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
137 return (True, False) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
138 if filect >= files: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
139 break |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
140 return (True, True) |
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 def is_bz2(file_path): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
144 is_bz2, is_valid = check_bz2(file_path, check_content=False) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
145 return is_bz2 |
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 def is_gzip(file_path): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
149 is_gzipped, is_valid = check_gzip(file_path, check_content=False) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
150 return is_gzipped |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
151 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
152 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
153 def is_zip(file_path): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
154 is_zipped, is_valid = check_zip(file_path, check_content=False) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
155 return is_zipped |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
156 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
157 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
158 def is_single_file_zip(file_path): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
159 for i, member in enumerate(iter_zip(file_path)): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
160 if i > 1: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
161 return False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
162 return True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
163 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
164 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
165 def is_tar(file_path): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
166 return tarfile.is_tarfile(file_path) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
167 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
168 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
169 def iter_zip(file_path): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
170 with zipfile.ZipFile(file_path) as z: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
171 for f in filter(lambda x: not x.endswith('/'), z.namelist()): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
172 yield (z.open(f), f) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
173 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
174 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
175 def check_image(file_path): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
176 """ Simple wrapper around image_type to yield a True/False verdict """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
177 if image_type(file_path): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
178 return True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
179 return False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
180 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
181 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
182 __all__ = ( |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
183 'check_binary', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
184 'check_bz2', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
185 'check_gzip', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
186 'check_html', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
187 'check_image', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
188 'check_zip', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
189 'is_gzip', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
190 'is_bz2', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
191 'is_zip', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
192 ) |