FIle clean

clean.py

import re


def solve_specific(line: str):
    matched = re.match(r'\[(.*)\]\[(.*?)\]', line)
    name = matched.group(1)
    label = matched.group(2)

    return '\n'.join(list(map(lambda x: '{} B-{}'.format(x[1], label) if x[0] == 0 else '{} I-{}'.format(x[1], label), enumerate(name))))


def solve(line: str):
    if line == '':
        return ''
    elif line[0] == '[':
        return solve_specific(line)
    else:
        return '\n'.join(list(map(lambda x: '{} 0'.format(x), line)))


def clean(line: str)->str:
    split_result = filter(None, re.split(
        r'(\[.*?\]\[.*?\])', line, flags=re.S))
    return '\n'.join(list(map(solve, split_result)))


with open('../data/train.txt', 'w') as out_file:
    with open("../data/temp", "r") as in_file:
        for line in in_file.readlines():
            out_file.write(clean(line[:-1]))
            out_file.write('\n\n')