Mercurial > repos > bgruening > split_file_to_collection

annotate split_file_to_collection.py @ 0:de3c2c88e710 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
author bgruening
date Tue, 17 Jul 2018 14:37:13 -0400
parents
children d150ac3d853d
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
1 #!/usr/bin/env python
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
2
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
3 import argparse
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
4 import os
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
5 import re
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
6 import random
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
7 import math
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
8
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
9
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
10 """
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
11 regexes that indicate the *beginning* of a record
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
12 new file types can be added by appending to this dict,
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
13 updating the parser, and adding a new type option in the Galaxy wrapper
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
14 """
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
15 FILETYPES = {'fasta': '^>',
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
16 'fastq': '^@',
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
17 'tabular': '^.*',
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
18 'mgf': '^BEGIN IONS'}
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
19
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
20
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
21 def main():
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
22 ps = parser_cli()
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
23 args = vars(ps.parse_args())
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
24
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
25 # get args and validate
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
26 in_file = args["in"]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
27 if not os.path.isfile(args["in"]):
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
28 raise FileNotFoundError('Input file does not exist')
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
29
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
30 out_dir = args["out_dir"]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
31 if not os.path.isdir(args["out_dir"]):
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
32 raise FileNotFoundError('out_dir is not a directory')
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
33
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
34 top = args["top"]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
35 if top < 0:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
36 raise ValueError("Number of header lines cannot be negative")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
37
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
38 ftype = args["ftype"]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
39
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
40 if args["ftype"] == "tabular" and args["by"] == "col":
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
41 args["match"] = replace_mapped_chars(args["match"])
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
42 args["sub"] = replace_mapped_chars(args["sub"])
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
43 split_by_column(args, in_file, out_dir, top)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
44
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
45 else:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
46 split_by_record(args, in_file, out_dir, top, ftype)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
47
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
48
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
49 def parser_cli():
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
50 parser = argparse.ArgumentParser(description="split a file into multiple files. " +
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
51 "Can split on the column of a tabular file, " +
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
52 "with custom and useful names based on column value.")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
53 parser.add_argument('--in', '-i', required=True, help="The input file")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
54 parser.add_argument('--out_dir', '-o', default=os.getcwd(), help="The output directory", required=True)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
55 parser.add_argument('--file_names', '-a', help="If not splitting by column, the base name of the new files")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
56 parser.add_argument('--file_ext', '-e', help="If not splitting by column," +
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
57 " the extension of the new files (without a period)")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
58 parser.add_argument('--ftype', '-f', help="The type of the file to split", required = True,
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
59 choices=["mgf", "fastq", "fasta", "tabular"])
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
60 parser.add_argument('--by', '-b', help="Split by line or by column (tabular only)",
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
61 default = "row", choices = ["col", "row"])
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
62 parser.add_argument('--top', '-t', type=int, default=0, help="Number of header lines to carry over to new files. " +
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
63 "(tabular only).")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
64 parser.add_argument('--rand', '-r', help="Divide records randomly into new files", action='store_true')
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
65 parser.add_argument('--seed', '-x', help="Provide a seed for the random number generator. " +
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
66 "If not provided and args[\"rand\"]==True, then date is used", type=int)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
67 parser.add_argument('--numnew', '-n', type=int, default = 1,
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
68 help="Number of output files desired. Not valid for splitting on a column")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
69 parser.add_argument('--batch', action='store_true',
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
70 help="Distribute files to collection while maintaining order. Ignored if splitting on column.")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
71
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
72 bycol = parser.add_argument_group('If splitting on a column')
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
73 bycol.add_argument('--match', '-m', default = "(.*)", help="The regular expression to match id column entries")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
74 bycol.add_argument('--sub', '-s', default = r'\1',
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
75 help="The regular expression to substitute in for the matched pattern.")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
76 bycol.add_argument('--id_column', '-c', default="1",
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
77 help="Column that is used to name output files. Indexed starting from 1.", type=int)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
78 return parser
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
79
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
80
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
81 def close_files(file_list):
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
82 # finally, close all files
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
83 for open_file in file_list:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
84 open_file.close()
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
85
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
86
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
87 def replace_mapped_chars(pattern):
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
88 """
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
89 handles special escaped characters when coming from galaxy
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
90 """
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
91 mapped_chars = {'\'': '__sq__', '\\': '__backslash__'}
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
92 for key, value in mapped_chars.items():
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
93 pattern = pattern.replace(value, key)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
94 return pattern
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
95
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
96
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
97 def split_by_record(args, in_file, out_dir, top, ftype):
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
98 # get record separator for given filetype
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
99 sep = re.compile(FILETYPES[ftype])
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
100
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
101 numnew = args["numnew"]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
102
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
103 # random division
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
104 rand = args["rand"]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
105 seed = args["seed"]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
106 if seed:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
107 random.seed(seed)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
108 else:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
109 random.seed()
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
110
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
111 # batched division (maintains order)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
112 batch = args["batch"]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
113 # define n_per_file so we don't get a warning about ref before assignment
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
114 n_per_file = math.inf
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
115 if batch:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
116 # number of records
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
117 with open(in_file) as f:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
118 i = 0
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
119 for line in f:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
120 if re.match(sep, line) is not None:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
121 i+=1
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
122 n_records = i + 1
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
123 if top:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
124 n_records -= top # don't count the top lines
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
125
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
126 # approx. number of lines per file
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
127 n_per_file = n_records // numnew
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
128
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
129 # make new files
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
130 # strip extension of old file and add number
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
131 custom_new_file_name = args["file_names"]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
132 custom_new_file_ext = "." + args["file_ext"]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
133 if custom_new_file_name is None:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
134 new_file_base = os.path.splitext(os.path.basename(in_file))
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
135 else:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
136 new_file_base = [custom_new_file_name, custom_new_file_ext]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
137
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
138 newfiles = [
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
139 open(out_dir + "/" + new_file_base[0] + "_" + str(count) + new_file_base[1], "w")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
140 for count in range(0, numnew)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
141 ]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
142
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
143 # bunch o' counters
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
144 # index to list of new files
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
145 new_file_counter = 0
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
146
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
147 # used for top
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
148 # number of lines read so far
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
149 n_read = 0
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
150 # to contain header specified by top
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
151 header = ""
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
152 # keep track of the files that have been opened so far
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
153 fresh_files = {i for i in range(0, numnew)}
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
154
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
155 # keep track in loop of number of records in each file
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
156 # only used in batch
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
157 records_in_file = 0
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
158
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
159 # open file
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
160 with open(in_file, "r") as file:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
161 record = ""
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
162 for line in file:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
163 n_read += 1
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
164 if n_read <= top:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
165 header += line
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
166 continue
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
167 # check if beginning of line is record sep
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
168 # if beginning of line is record sep, either start record or finish one
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
169 if re.match(sep, line) is not None:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
170 # this only happens first time through
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
171 if record == "":
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
172 record += line
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
173 else:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
174 # if is in fresh_files, write header and drop from freshFiles
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
175 if new_file_counter in fresh_files:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
176 newfiles[new_file_counter].write(header)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
177 fresh_files.remove(new_file_counter)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
178
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
179 # write record to file
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
180 newfiles[new_file_counter].write(record)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
181
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
182 # if not the first time through, we assign the new record
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
183 record = line
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
184
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
185 # change destination file
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
186 if rand:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
187 new_file_counter = int(math.floor(random.random() * numnew))
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
188 elif batch:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
189 # number of records read per file
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
190 records_in_file += 1
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
191 # have we reached the max for each file?
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
192 # if so, switch file
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
193 if records_in_file >= n_per_file:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
194 new_file_counter = (new_file_counter + 1) % numnew
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
195 records_in_file = 0 # reset to 0
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
196 else:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
197 new_file_counter = (new_file_counter + 1) % numnew
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
198 # if beginning of line is not record sep, we must be inside a record
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
199 # so just append
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
200 else:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
201 record += line
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
202 # after loop, write final record to file
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
203 newfiles[new_file_counter].write(record)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
204 # close new files
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
205 close_files(newfiles)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
206
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
207
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
208 def split_by_column(args, in_file, out_dir, top):
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
209
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
210 # shift to 0-based indexing
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
211 id_col = int(args["id_column"]) - 1
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
212
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
213 try:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
214 match = re.compile(args["match"])
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
215 except re.error:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
216 print("ERROR: Match (-m) supplied is not valid regex.")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
217 raise
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
218
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
219 sub = args["sub"]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
220
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
221 # set of file names
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
222 new_files = dict()
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
223
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
224 # keep track of how many lines have been read
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
225 n_read = 0
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
226 header = ""
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
227 with open(in_file) as file:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
228 for line in file:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
229 # if still in top, save to header
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
230 n_read += 1
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
231 if n_read <= top:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
232 header += line
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
233 continue
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
234 # split into columns, on tab
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
235 fields = re.split(r'\t', line.strip('\n'))
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
236
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
237 # get id column value
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
238 id_col_val = fields[id_col]
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
239
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
240 # use regex to get new file name
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
241 out_file_name = re.sub(match, sub, id_col_val)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
242 out_file_path = os.path.join(out_dir, out_file_name)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
243
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
244 # write
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
245 if out_file_name not in new_files.keys():
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
246 # open file (new, so not already open)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
247 current_new_file = open(out_file_path, "w")
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
248 current_new_file.write(header)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
249 current_new_file.write(line)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
250 # add to dict
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
251 new_files[out_file_name] = current_new_file
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
252 else:
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
253 # file is already open, so just write to it
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
254 new_files[out_file_name].write(line)
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
255
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
256 # finally, close all files
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
257 close_files(new_files.values())
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
258
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
259
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
260 if __name__ == "__main__":
de3c2c88e710 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/text_processing/split_file_to_collection commit 85015046a6d8a9dc0f4b54611986676aceeeadd7
bgruening
parents:
diff changeset
261 main()