Mercurial > repos > iuc > query_tabular

annotate filters.py @ 0:3708ff0198b7 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
author iuc
date Tue, 18 Jul 2017 09:07:07 -0400
parents
children 8a33b442ecd9
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
1 #!/usr/binsenv python
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
2
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
3 from __future__ import print_function
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
4
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
5 import re
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
6 import sys
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
7
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
8
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
9 class LineFilter(object):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
10 def __init__(self, source, filter_dict):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
11 self.source = source
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
12 self.filter_dict = filter_dict
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
13 self.func = lambda i, l: l.rstrip('\r\n') if l else None
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
14 self.src_lines = []
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
15 self.src_line_cnt = 0
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
16 if not filter_dict:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
17 return
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
18 if filter_dict['filter'] == 'regex':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
19 rgx = re.compile(filter_dict['pattern'])
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
20 if filter_dict['action'] == 'exclude_match':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
21 self.func = lambda i, l: l if not rgx.match(l) else None
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
22 elif filter_dict['action'] == 'include_match':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
23 self.func = lambda i, l: l if rgx.match(l) else None
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
24 elif filter_dict['action'] == 'exclude_find':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
25 self.func = lambda i, l: l if not rgx.search(l) else None
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
26 elif filter_dict['action'] == 'include_find':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
27 self.func = lambda i, l: l if rgx.search(l) else None
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
28 elif filter_dict['filter'] == 'select_columns':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
29 cols = [int(c) - 1 for c in filter_dict['columns']]
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
30 self.func = lambda i, l: self.select_columns(l, cols)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
31 elif filter_dict['filter'] == 'replace':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
32 p = filter_dict['pattern']
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
33 r = filter_dict['replace']
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
34 c = int(filter_dict['column']) - 1
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
35 self.func = lambda i, l: '\t'.join(
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
36 [x if j != c else re.sub(p, r, x) for j, x in enumerate(l.split('\t'))])
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
37 elif filter_dict['filter'] == 'prepend_line_num':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
38 self.func = lambda i, l: '%d\t%s' % (i, l)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
39 elif filter_dict['filter'] == 'append_line_num':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
40 self.func = lambda i, l: '%s\t%d' % (l.rstrip('\r\n'), i)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
41 elif filter_dict['filter'] == 'prepend_text':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
42 s = filter_dict['column_text']
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
43 self.func = lambda i, l: '%s\t%s' % (s, l)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
44 elif filter_dict['filter'] == 'append_text':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
45 s = filter_dict['column_text']
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
46 self.func = lambda i, l: '%s\t%s' % (l.rstrip('\r\n'), s)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
47 elif filter_dict['filter'] == 'skip':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
48 cnt = filter_dict['count']
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
49 self.func = lambda i, l: l if i > cnt else None
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
50 elif filter_dict['filter'] == 'normalize':
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
51 cols = [int(c) - 1 for c in filter_dict['columns']]
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
52 sep = filter_dict['separator']
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
53 self.func = lambda i, l: self.normalize(l, cols, sep)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
54
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
55 def __iter__(self):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
56 return self
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
57
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
58 def __next__(self):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
59 if not self.src_lines:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
60 self.get_lines()
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
61 if self.src_lines:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
62 return self.src_lines.pop(0)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
63 raise StopIteration
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
64
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
65 next = __next__
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
66
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
67 def select_columns(self, line, cols):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
68 fields = line.split('\t')
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
69 return '\t'.join([fields[x] for x in cols])
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
70
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
71 def normalize(self, line, split_cols, sep):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
72 lines = []
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
73 fields = line.rstrip('\r\n').split('\t')
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
74 split_fields = dict()
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
75 cnt = 0
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
76 for c in split_cols:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
77 if c < len(fields):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
78 split_fields[c] = fields[c].split(sep)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
79 cnt = max(cnt, len(split_fields[c]))
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
80 if cnt == 0:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
81 lines.append('\t'.join(fields))
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
82 else:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
83 for n in range(0, cnt):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
84 flds = [x if c not in split_cols else split_fields[c][n]
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
85 if n < len(split_fields[c])
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
86 else '' for (c, x) in enumerate(fields)]
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
87 lines.append('\t'.join(flds))
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
88 return lines
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
89
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
90 def get_lines(self):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
91 for i, next_line in enumerate(self.source):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
92 self.src_line_cnt += 1
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
93 line = self.func(self.src_line_cnt, next_line)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
94 if line:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
95 if isinstance(line, list):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
96 self.src_lines.extend(line)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
97 else:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
98 self.src_lines.append(line)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
99 return
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
100
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
101
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
102 class TabularReader:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
103 """
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
104 Tabular file iterator. Returns a list
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
105 """
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
106 def __init__(self, input_file, skip=0, comment_char=None, col_idx=None,
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
107 filters=None):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
108 self.skip = skip
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
109 self.comment_char = comment_char
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
110 self.col_idx = col_idx
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
111 self.filters = filters
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
112 self.tsv_file = \
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
113 input_file if hasattr(input_file, 'readline') else open(input_file)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
114 if skip and skip > 0:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
115 for i in range(skip):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
116 if not self.tsv_file.readline():
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
117 break
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
118 source = LineFilter(self.tsv_file, None)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
119 if comment_char:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
120 source = LineFilter(source,
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
121 {"filter": "regex", "pattern": comment_char,
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
122 "action": "exclude_match"})
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
123 if filters:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
124 for f in filters:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
125 source = LineFilter(source, f)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
126 self.source = source
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
127
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
128 def __iter__(self):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
129 return self
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
130
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
131 def __next__(self):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
132 ''' Iteration '''
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
133 for i, line in enumerate(self.source):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
134 fields = line.rstrip('\r\n').split('\t')
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
135 if self.col_idx:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
136 fields = [fields[i] for i in self.col_idx]
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
137 return fields
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
138 raise StopIteration
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
139
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
140 next = __next__
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
141
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
142
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
143 def filter_file(input_file, output, skip=0, comment_char='#', filters=None):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
144 data_lines = 0
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
145 try:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
146 tr = TabularReader(input_file, skip=skip, comment_char=comment_char,
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
147 filters=filters)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
148 for linenum, fields in enumerate(tr):
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
149 data_lines += 1
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
150 try:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
151 output.write('%s\n' % '\t'.join(fields))
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
152 except Exception as e:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
153 print('Failed at line: %d err: %s' % (linenum, e),
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
154 file=sys.stderr)
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
155 except Exception as e:
3708ff0198b7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/query_tabular commit 74915fc9cee746bbce1c4b507e13231259de177d
iuc
parents:
diff changeset
156 exit('Error: %s' % (e))