rm_duplicates.py 文件源码

def rm_doubled_quotes(entry):
    """Some entries have doubled quotes (translations) within different senses.
    Remove the doubled quotes.
    This function return True, if the entry has been modified."""
    senses = list(findall(entry, 'sense'))
    # add quote elements
    senses = [(cit, q)  for s in senses for cit in findall(s, 'cit')
            for q in findall(cit, 'quote')]
    if len(senses) <= 1:
        return
    changed = False
    # pair each sense with another and compare their content
    for trans1, trans2 in itertools.combinations(senses, 2):
        # could have been removed already, so check:
        cit1, quote1 = trans1
        cit2, quote2 = trans2
        if not cit1.findall(quote1.tag) or not cit2.findall(quote2.tag) \
                and cit1 is not cit2:
            continue # one of them has been removed already
        # text of both quotes match, remove second quote
        if quote1.text == quote2.text:
            cit2.remove(quote2)
            changed = True
    return changed