find_terrorist

#! /usr/bin/env python
import argparse
import json
import nltk
import sys

def main(args):
    parser = argparse.ArgumentParser(
        description="Frame someone as a terrorist.",
    )

    # TODO: Make it get data from stdin if this is not supplied
    parser.add_argument(
        '-s', '--source',
        help='Get data from this file.',
        dest='source',
        required=True,
    )

    options = parser.parse_args(args)

    with open(options.source) as fh:
        data = json.load(fh)

    all_sentences = {}
    for pos in range(0,len(data)):
        sentences = nltk.sent_tokenize(data[pos])
        sentences = [sent.strip() for sent in sentences]
        for sentence in sentences:
            if sentence not in all_sentences:
                all_sentences[sentence] = []
            all_sentences[sentence].append(pos)

    # Get the list of suspect terms
    with open('suspect') as fh:
        suspect_terms = fh.readlines()

    suspect_terms = [term.strip() for term in suspect_terms]

    for sentence in all_sentences.keys():
        for term in suspect_terms:
            if term in sentence.lower():
                print(sentence)
                print('-----')

if __name__ == '__main__':
    main(sys.argv[1:])