Mercurial > repos > yating-l > jbrowsearchivecreator
annotate util/santitizer.py @ 50:804e7217f27c draft default tip
planemo upload for repository https://github.com/goeckslab/jbrowse-archive-creator.git commit 3afab1eb0da4823419945c523a1d37cd5973e7f6-dirty
| author | yating-l | 
|---|---|
| date | Fri, 20 Jul 2018 11:38:01 -0400 | 
| parents | 31a41ce128cc | 
| children | 
| rev | line source | 
|---|---|
| 25 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 1 #!/usr/bin/python | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 2 # -*- coding: utf8 -*- | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 3 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 4 """ | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 5 This class handles the subprocess calls of the different tools used | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 6 in HubArchiveCreator | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 7 """ | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 8 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 9 import logging | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 10 import os | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 11 import subprocess | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 12 import sys | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 13 import string | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 14 import tempfile | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 15 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 16 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 17 def prefixTrackName(filename): | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 18 """ | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 19 santitize trackName. Because track name must begin with a letter and | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 20 contain only the following chars: [a-zA-Z0-9_]. | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 21 See the "track" Common settings at: | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 22 https://genome.ucsc.edu/goldenpath/help/trackDb/trackDbHub.html#bigPsl_-_Pairwise_Alignments | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 23 skip the santitization for cytoBandIdeo track | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 24 """ | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 25 if filename == 'cytoBandIdeo': | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 26 return filename | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 27 valid_chars = "_%s%s" % (string.ascii_letters, string.digits) | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 28 sanitize_name = ''.join([c if c in valid_chars else '_' for c in filename]) | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 29 sanitize_name = "gonramp_" + sanitize_name | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 30 return sanitize_name | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 31 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 32 def sanitize_name_input(string_to_sanitize): | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 33 """ | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 34 Sanitize the string passed in parameter by replacing '/' and ' ' by '_' | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 35 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 36 :param string_to_sanitize: | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 37 :return : | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 38 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 39 :Example: | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 40 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 41 >>> sanitize_name_input('this/is an//example') | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 42 this_is_an__example | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 43 """ | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 44 return string_to_sanitize \ | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 45 .replace("/", "_") \ | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 46 .replace(" ", "_") | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 47 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 48 def sanitize_name_inputs(inputs_data): | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 49 """ | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 50 Sanitize value of the keys "name" of the dictionary passed in parameter. | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 51 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 52 Because sometimes output from Galaxy, or even just file name, from user inputs, have spaces. | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 53 Also, it can contain '/' character and could break the use of os.path function. | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 54 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 55 :param inputs_data: dict[string, dict[string, string]] | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 56 """ | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 57 for key in inputs_data: | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 58 inputs_data[key]["name"] = sanitize_name_input(inputs_data[key]["name"]) | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 59 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 60 def sanitize_group_name(group_name): | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 61 return group_name.lower().replace(' ', '_') | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 62 | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 63 def sanitize_name(input_name): | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 64 """ | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 65 Galaxy will name all the files and dirs as *.dat, | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 66 the function can replace '.' to '_' for the dirs | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 67 """ | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 68 validChars = "_-%s%s" % (string.ascii_letters, string.digits) | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 69 sanitized_name = ''.join([c if c in validChars else '_' for c in input_name]) | 
| 
31a41ce128cc
planemo upload for repository https://github.com/Yating-L/jbrowse-archive-creator.git commit 691e5366893905d30943a3cb8cdfb6341f0f5362-dirty
 yating-l parents: diff
changeset | 70 return "gonramp_" + sanitized_name | 
