vrt-simple-mend

#! /usr/bin/env python3
# -*- mode: Python; -*-

from vrtargslib import trans_args, trans_main

def parsearguments():
    description = '''

    Mend torn sentences by simply removing sen.ten.ce tags and
    sentence-tag pairs within sen.ten.ce, with no adjustment to
    tokens.

    '''
    parser = trans_args(description = description)
    args = parser.parse_args()
    args.prog = parser.prog
    return args

def main(args, ins, ous):
    
    def issome(line): return not line.isspace()

    keep = True
    look = False
    for line in filter(issome, ins):

        # outside <sen.ten.ce>, keep everything;
        # inside <sen.ten.ce>, drop sentence tags
        # but look for a first sentence start tag;
        # at </sen.ten.ce> close the <sentence>
        # (unless still looking for it to start)

        if line.startswith(('<sen.ten.ce>', '<sen.ten.ce ')):
            keep = False
            look = True
            continue

        if line.startswith('</sen.ten.ce>'):
            look or print('</sentence>', file = ous)
            keep = True 
            look = False
            continue

        if line.startswith(('<sentence>', '<sentence ')):
            (keep or look) and print(line, end = '', file = ous)
            if look: look = False
            continue

        if line.startswith('</sentence>'):
            keep and print(line, end = '', file = ous)
            continue

        print(line, end = '', file = ous)

if __name__ == '__main__':
    trans_main(parsearguments(), main)