Mercurial > repos > prog > lcmsmatching

annotate list-file-cols.py @ 7:882f2f20028b draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit 608d9e59a0d2dcf85a037968ddb2c61137fb9bce
author prog
date Sun, 09 Apr 2017 10:44:21 -0400
parents b8f70d8216b3
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
6
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
1 #!/usr/bin/env python
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
2 # vi: fdm=marker
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
3
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
4 import csv
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
5 import re
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
6 import argparse
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
7
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
8 # Get file cols {{{1
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
9 ################################################################
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
10
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
11 def get_file_cols(file, preferred):
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
12
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
13 cols = []
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
14
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
15 with open(file if isinstance(file, str) else file.get_file_name(), 'r') as f:
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
16
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
17 # Read file header
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
18 reader = csv.reader(f, delimiter = "\t", quotechar='"')
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
19 header = reader.next()
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
20
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
21 preferred = preferred.split(',')
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
22
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
23 # Determine default value
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
24 perfect_matches = []
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
25 partial_matches = []
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
26 for p in preferred:
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
27 for c in header:
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
28 if c == p:
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
29 perfect_matches.append(c) # Perfect match !
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
30 elif re.match(p, c):
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
31 partial_matches.append(c) # Keep this partial match in case we find no perfect match
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
32
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
33 ordered_cols = perfect_matches + partial_matches
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
34 for c in header:
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
35 if not c in ordered_cols:
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
36 ordered_cols.append(c)
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
37 ordered_cols.append('NA')
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
38
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
39 default = 0
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
40 if len(perfect_matches) + len(partial_matches) == 0:
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
41 default = len(ordered_cols) - 1
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
42
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
43 # Build list of cols
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
44 for i, c in enumerate(ordered_cols):
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
45 cols.append( (c, c, i == default) )
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
46
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
47 return cols
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
48
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
49 # Main {{{1
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
50 ################################################################
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
51
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
52 if __name__ == '__main__':
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
53
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
54 # Parse command line arguments
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
55 parser = argparse.ArgumentParser(description='Script for getting column names in a csv file.')
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
56 parser.add_argument('-f', help = 'CSV File (separator must be TAB)', dest = 'file', required = True)
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
57 parser.add_argument('-p', help = 'List (comma separated values) of preferred column names for default one.', dest = 'preferred', required = True)
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
58 args = parser.parse_args()
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
59 args_dict = vars(args)
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
60
b8f70d8216b3 planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit b3a4e34cf9356447ae3507cc6fe2ff6a1f24afbc-dirty
prog
parents:
diff changeset
61 print(get_file_cols(**args_dict))