Mercurial > repos > devteam > subtract_query
annotate subtract_query.py @ 3:f58ba0382c26 draft default tip
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
| author | devteam |
|---|---|
| date | Thu, 22 Jun 2017 18:51:44 -0400 |
| parents | df4bce472400 |
| children |
| rev | line source |
|---|---|
| 0 | 1 #!/usr/bin/env python |
| 2 # Greg Von Kuster | |
| 3 """ | |
| 4 Subtract an entire query from another query | |
| 5 usage: %prog in_file_1 in_file_2 begin_col end_col output | |
| 6 --ignore-empty-end-cols: ignore empty end columns when subtracting | |
| 7 """ | |
|
3
f58ba0382c26
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
1
diff
changeset
|
8 from __future__ import print_function |
|
f58ba0382c26
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
1
diff
changeset
|
9 |
|
1
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
10 import sys |
|
3
f58ba0382c26
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
1
diff
changeset
|
11 |
| 0 | 12 from bx.cookbook import doc_optparse |
| 13 | |
| 14 # Older py compatibility | |
| 15 try: | |
| 16 set() | |
| 17 except: | |
| 18 from sets import Set as set | |
| 19 | |
| 20 assert sys.version_info[:2] >= ( 2, 4 ) | |
| 21 | |
|
1
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
22 |
| 0 | 23 def get_lines(fname, begin_col='', end_col='', ignore_empty_end_cols=False): |
| 24 lines = set([]) | |
| 25 i = 0 | |
| 26 for i, line in enumerate(file(fname)): | |
| 27 line = line.rstrip('\r\n') | |
| 28 if line and not line.startswith('#'): | |
| 29 if begin_col and end_col: | |
| 30 """Both begin_col and end_col must be integers at this point.""" | |
| 31 try: | |
| 32 line = line.split('\t') | |
|
1
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
33 line = '\t'.join([line[j] for j in range(begin_col - 1, end_col)]) |
| 0 | 34 if ignore_empty_end_cols: |
| 35 # removing empty fields, we do not compare empty fields at the end of a line. | |
| 36 line = line.rstrip() | |
| 37 lines.add( line ) | |
|
1
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
38 except: |
|
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
39 pass |
| 0 | 40 else: |
| 41 if ignore_empty_end_cols: | |
| 42 # removing empty fields, we do not compare empty fields at the end of a line. | |
| 43 line = line.rstrip() | |
| 44 lines.add( line ) | |
|
1
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
45 if i: |
|
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
46 return (i + 1, lines) |
|
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
47 else: |
|
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
48 return (i, lines) |
|
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
49 |
| 0 | 50 |
| 51 def main(): | |
| 52 # Parsing Command Line here | |
| 53 options, args = doc_optparse.parse( __doc__ ) | |
| 54 | |
| 55 try: | |
| 56 inp1_file, inp2_file, begin_col, end_col, out_file = args | |
| 57 except: | |
| 58 doc_optparse.exception() | |
|
1
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
59 |
| 0 | 60 begin_col = begin_col.strip() |
| 61 end_col = end_col.strip() | |
|
1
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
62 |
| 0 | 63 if begin_col != 'None' or end_col != 'None': |
| 64 """ | |
| 65 The user selected columns for restriction. We'll allow default | |
| 66 values for both begin_col and end_col as long as the user selected | |
| 67 at least one of them for restriction. | |
| 68 """ | |
| 69 if begin_col == 'None': | |
| 70 begin_col = end_col | |
| 71 elif end_col == 'None': | |
| 72 end_col = begin_col | |
| 73 begin_col = int(begin_col) | |
| 74 end_col = int(end_col) | |
| 75 """Make sure that begin_col <= end_col (switch if not)""" | |
| 76 if begin_col > end_col: | |
| 77 tmp_col = end_col | |
| 78 end_col = begin_col | |
| 79 begin_col = tmp_col | |
| 80 else: | |
| 81 begin_col = end_col = '' | |
| 82 | |
| 83 try: | |
|
1
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
84 fo = open(out_file, 'w') |
| 0 | 85 except: |
|
3
f58ba0382c26
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
1
diff
changeset
|
86 print("Unable to open output file", file=sys.stderr) |
| 0 | 87 sys.exit() |
| 88 | |
| 89 """ | |
| 90 len1 is the number of lines in inp1_file | |
| 91 lines1 is the set of unique lines in inp1_file | |
| 92 diff1 is the number of duplicate lines removed from inp1_file | |
| 93 """ | |
| 94 len1, lines1 = get_lines(inp1_file, begin_col, end_col, options.ignore_empty_end_cols) | |
| 95 diff1 = len1 - len(lines1) | |
| 96 len2, lines2 = get_lines(inp2_file, begin_col, end_col, options.ignore_empty_end_cols) | |
|
1
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
97 |
| 0 | 98 lines1.difference_update(lines2) |
| 99 """lines1 is now the set of unique lines in inp1_file - the set of unique lines in inp2_file""" | |
| 100 | |
| 101 for line in lines1: | |
|
3
f58ba0382c26
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
1
diff
changeset
|
102 print(line, file=fo) |
| 0 | 103 |
| 104 fo.close() | |
|
1
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
105 |
|
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
106 info_msg = 'Subtracted %d lines. ' % ((len1 - diff1) - len(lines1)) |
|
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
107 |
| 0 | 108 if begin_col and end_col: |
| 109 info_msg += 'Restricted to columns c' + str(begin_col) + ' thru c' + str(end_col) + '. ' | |
| 110 | |
| 111 if diff1 > 0: | |
|
1
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
112 info_msg += 'Eliminated %d duplicate/blank/comment/invalid lines from first query.' % diff1 |
|
df4bce472400
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
113 |
|
3
f58ba0382c26
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
1
diff
changeset
|
114 print(info_msg) |
|
f58ba0382c26
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/subtract_query commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
1
diff
changeset
|
115 |
| 0 | 116 |
| 117 if __name__ == "__main__": | |
| 118 main() |