frequency_order.py

#!/usr/bin/env python3

from collections import defaultdict

from pysblgnt import morphgnt_rows

count_by_item = defaultdict(int)
total_item_count = 0

for book_num in range(1, 28):
    for row in morphgnt_rows(book_num):
        item = row["lemma"]  # "lemma" or "norm"
        count_by_item[item] += 1
        total_item_count += 1

cummulative_count = 0
items_learnt = 0
for item, count in sorted(count_by_item.items(), key=lambda element: element[1], reverse=True):
    print(item, count)
    cummulative_count += count
    items_learnt += 1
    if cummulative_count > total_item_count * 0.50:  # @@@
        break