nlp/vrt-creator

#!/usr/bin/env python3.7
# coding=utf-8

from argparse import ArgumentParser
from stand_off_data import StandOffData
import hashlib
import json


parser = ArgumentParser(
    description='Convert plain text and JSON stand off to a CWB vrt file'
)
parser.add_argument(
    '-s', '--stand-off-data-file',
    help='JSON stand off data input file',
    required=True
)
parser.add_argument(
    '-t', '--text-file',
    help='Plain text input file',
    required=True
)
parser.add_argument(
    '-o', '--output-file',
    help='Output file',
    required=True
)
args = parser.parse_args()


with open(args.stand_off_data_file) as stand_of_data_file:
    stand_off_data = StandOffData(json.load(stand_of_data_file))

with open(args.text_file, "rb") as text_file:
    text_md5 = hashlib.md5()
    for chunk in iter(lambda: text_file.read(128 * text_md5.block_size), b''):
        text_md5.update(chunk)
    if text_md5.hexdigest() != stand_off_data.meta['file']['md5']:
        raise Exception('md5 not equal')

with open(args.text_file, encoding=stand_off_data.meta['file']['encoding']) as text_file:  # noqa
    text = text_file.read()

with open(args.output_file, 'w') as vrt_file:
    vrt_file.write(stand_off_data.to_vrt(text))