Генерация всевозможных словоформ из списка слов (минус-слов)

allforms.py

import pymorphy2
import os
import sys

def is_russian(word):
    return all('а' <= ch <= 'я' or ch == 'ё' for ch in word)

def get_word_forms(word, morph):
    parses = morph.parse(word)
    word_forms = set()

    for p in parses:
        if 'NOUN' in p.tag or 'ADJF' in p.tag or 'VERB' in p.tag or 'INFN' in p.tag:
            forms = {f.word for f in p.lexeme}
            word_forms.update(forms)

    return word_forms

def process_file(input_path):
    morph = pymorphy2.MorphAnalyzer()
    base_name = os.path.basename(input_path)
    output_path = os.path.join(os.path.dirname(input_path), f'allforms_{base_name}')
    
    input_words = set()
    output_words = set()
    no_forms_russian = []

    with open(input_path, 'r', encoding='utf-8') as infile:
        for line in infile:
            word = line.strip().lstrip('-').lower()
            if word:
                input_words.add(word)

    for word in input_words:
        if is_russian(word):
            forms = get_word_forms(word, morph)
            if forms:
                output_words.update(forms)
            else:
                no_forms_russian.append(word)
                output_words.add(word)
        else:
            output_words.add(word)

    with open(output_path, 'w', encoding='utf-8') as outfile:
        for word in sorted(output_words):
            outfile.write(word + '\n')

    print(f"Было слов: {len(input_words)}")
    print(f"Стало уникальных словоформ: {len(output_words)}")
    if no_forms_russian:
        print(f"Для {len(no_forms_russian)} русских слов не найдены словоформы:")
        for word in no_forms_russian:
            print(f" - {word}")

if __name__ == "__main__":
    if len(sys.argv) != 2:
        print("Использование: python allforms.py input.txt")
    else:
        input_path = sys.argv[1]
        process_file(input_path)

python3 allforms.py minus-kondei.txt
python allforms.py minus-kondei.txt