Code 2

[stestagg]

import json
import operator
from collections import defaultdict

from detect import get_top_bigrams, parse_sample


def check_common_bigrams(bigrams, bigram_data):
    langs = {}
    for lang, lang_bigram in bigram_data.items():
        match = 0
        for bigram in bigrams:
            if bigram in lang_bigram:
                match += 1
        langs[lang] = match
    return langs


def main(sample):
    counts = defaultdict(int)
    parse_sample(counts, text=sample)
    bigrams = get_top_bigrams(counts)
    trained = json.load(open('output.json'))

    results = (check_common_bigrams(bigrams, trained))
    print(sorted(results.items(), key=operator.itemgetter(1), reverse=True)[0], repr(sample))


if __name__ == "__main__":
    for line in open("test_100.json"):
        main(json.loads(line)['text'])