annotate mergeCols.py @ 0:28ca7552e884

Uploaded merge_cols tarball.
author devteam
date Tue, 04 Dec 2012 11:00:46 -0500
parents
children dd40b1e9eebe
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
1 import sys, re
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
2
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
3 def stop_err( msg ):
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
4 sys.stderr.write( msg )
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
5 sys.exit()
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
6
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
7 def __main__():
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
8 try:
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
9 infile = open ( sys.argv[1], 'r')
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
10 outfile = open ( sys.argv[2], 'w')
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
11 except:
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
12 stop_err( 'Cannot open or create a file\n' )
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
13
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
14 if len( sys.argv ) < 4:
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
15 stop_err( 'No columns to merge' )
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
16 else:
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
17 cols = sys.argv[3:]
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
18
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
19 skipped_lines = 0
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
20
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
21 for line in infile:
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
22 line = line.rstrip( '\r\n' )
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
23 if line and not line.startswith( '#' ):
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
24 fields = line.split( '\t' )
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
25 line += '\t'
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
26 for col in cols:
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
27 try:
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
28 line += fields[ int( col ) -1 ]
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
29 except:
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
30 skipped_lines += 1
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
31
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
32 print >>outfile, line
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
33
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
34 if skipped_lines > 0:
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
35 print 'Skipped %d invalid lines' % skipped_lines
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
36
28ca7552e884 Uploaded merge_cols tarball.
devteam
parents:
diff changeset
37 if __name__ == "__main__" : __main__()