view regex.py @ 3:ae8c4b2488e7 draft

"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/regex_find_replace commit 0595f14f3b95f8b8190b58625217c72e76c649ee"
author galaxyp
date Fri, 08 May 2020 04:01:27 -0400
parents 538933d9fccc
children 399da6b5ec21
line wrap: on
line source

from optparse import OptionParser
import re


def main():
    parser = OptionParser()
    parser.add_option("--input", dest="input")
    parser.add_option("--output", dest="output")
    parser.add_option("--input_display_name", dest="input_display_name")
    parser.add_option("--pattern", dest="patterns", action="append",
                      help="regex pattern for replacement")
    parser.add_option("--replacement", dest="replacements", action="append",
                      help="replacement for regex match")
    parser.add_option("--column", dest="column", default=None)
    (options, args) = parser.parse_args()

    mapped_chars = {'\'': '__sq__', '\\': '__backslash__'}

    column = None
    if options.column is not None:
        # galaxy tabular is 1-based, python array are zero-based
        column = int(options.column) - 1

    with open(options.input, 'r') as input, open(options.output, 'w') as output:
        while True:
            line = input.readline()
            if line == "":
                break
            for (pattern, replacement) in zip(options.patterns, options.replacements):
                for key, value in mapped_chars.items():
                    pattern = pattern.replace(value, key)
                    replacement = replacement.replace(value, key)
                replacement = replacement.replace("#{input_name}", options.input_display_name)
                if column is None:
                    line = re.sub(pattern, replacement, line)
                else:
                    cells = line.split("\t")
                    if cells and len(cells) > column:
                        cell = cells[column]
                        cell = re.sub(pattern, replacement, cell)
                        cells[column] = cell
                        line = "\t".join(cells)
            output.write(line)


if __name__ == "__main__":
    main()