annotate util/santitizer.py @ 85:a65e52152476 draft default tip

planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1a81ebd0ddea950b84af3fc830e9267a4814b29f
author yating-l
date Mon, 06 Nov 2017 13:13:01 -0500
parents 39a32cb31623
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
66
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
1 #!/usr/bin/python
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
2 # -*- coding: utf8 -*-
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
3
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
4 """
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
5 This class handles the subprocess calls of the different tools used
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
6 in HubArchiveCreator
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
7 """
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
8
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
9 import logging
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
10 import os
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
11 import subprocess
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
12 import sys
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
13 import string
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
14 import tempfile
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
15
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
16
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
17 def prefixTrackName(filename):
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
18 """
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
19 santitize trackName. Because track name must begin with a letter and
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
20 contain only the following chars: [a-zA-Z0-9_].
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
21 See the "track" Common settings at:
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
22 https://genome.ucsc.edu/goldenpath/help/trackDb/trackDbHub.html#bigPsl_-_Pairwise_Alignments
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
23 skip the santitization for cytoBandIdeo track
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
24 """
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
25 if filename == 'cytoBandIdeo':
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
26 return filename
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
27 valid_chars = "_%s%s" % (string.ascii_letters, string.digits)
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
28 sanitize_name = ''.join([c if c in valid_chars else '_' for c in filename])
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
29 sanitize_name = "gonramp_" + sanitize_name
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
30 return sanitize_name
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
31
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
32 def sanitize_name_input(string_to_sanitize):
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
33 """
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
34 Sanitize the string passed in parameter by replacing '/' and ' ' by '_'
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
35
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
36 :param string_to_sanitize:
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
37 :return :
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
38
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
39 :Example:
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
40
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
41 >>> sanitize_name_input('this/is an//example')
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
42 this_is_an__example
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
43 """
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
44 return string_to_sanitize \
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
45 .replace("/", "_") \
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
46 .replace(" ", "_")
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
47
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
48 def sanitize_name_inputs(inputs_data):
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
49 """
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
50 Sanitize value of the keys "name" of the dictionary passed in parameter.
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
51
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
52 Because sometimes output from Galaxy, or even just file name, from user inputs, have spaces.
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
53 Also, it can contain '/' character and could break the use of os.path function.
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
54
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
55 :param inputs_data: dict[string, dict[string, string]]
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
56 """
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
57 for key in inputs_data:
4ca7cbf2d9b8 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents:
diff changeset
58 inputs_data[key]["name"] = sanitize_name_input(inputs_data[key]["name"])
70
39a32cb31623 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents: 66
diff changeset
59
39a32cb31623 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents: 66
diff changeset
60 def sanitize_group_name(group_name):
39a32cb31623 planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 159730602ff500b59aefc7080fb49b726c88d655-dirty
yating-l
parents: 66
diff changeset
61 return group_name.lower().replace(' ', '_')