def get_word2rank()

in tseval/feature_extraction.py [0:0]

• 9 lines of code
• 3 McCabe index (conditional complexity)

def get_word2rank(vocab_size=50000):
    frequency_table_path = os.path.join(VARIOUS_DIR, 'enwiki_frequency_table.tsv')
    word2rank = {}
    for rank, line in enumerate(yield_lines(frequency_table_path)):
        if (rank+1) > vocab_size:
            break
        word, _ = line.split('\t')
        word2rank[word] = rank
    return word2rank