utilities_language_bert/rus_main_workflow_bert.py

import datetime
from io import StringIO
from typing import Union
from random import sample
from collections import defaultdict
from streamlit.runtime.uploaded_file_manager import UploadedFile
from utilities_language_bert.rus_sentence_bert import TASK, SENTENCE
from utilities_language_general.rus_utils import compute_frequency_dict, prepare_tasks, prepare_target_words
from utilities_language_general.rus_constants import st, load_bert, load_classifiers, nlp, summarization, BAD_USER_TARGET_WORDS, MINIMUM_SETS


def main_workflow(
        file: Union[UploadedFile, None],
        text: str,
        logs,
        progress,
        progress_d,
        level: str,
        tw_mode_automatic_mode: str,
        target_words: str,
        num_distractors: int,
        save_name: str,
        global_bad_target_words=BAD_USER_TARGET_WORDS):

    # Clear bad target_words each time
    if global_bad_target_words:
        global_bad_target_words = []

    # Define main global variables
    GLOBAL_DISTRACTORS = set()
    MAX_FREQUENCY = 0

    logs.update(label='Загружаем языковые модели и другие данные', state='running')
    pos_dict, scaler, classifier = load_classifiers('model3')
    mask_filler = load_bert()

    # Get input text
    if file is not None:
        stringio = StringIO(file.getvalue().decode("utf-8"))
        current_text = stringio.read()
    elif text != '':
        current_text = text
    else:
        st.warning('Вы ни текст не вставили, ни файл не выбрали 😢')
        current_text = ''
        st.stop()

    # Process target words
    if tw_mode_automatic_mode == 'Самостоятельно':
        if target_words == '':
            st.warning('Вы не ввели целевые слова')
            st.stop()
        # Cannot make up paradigm, so only USER_TARGET_WORDS is used
        USER_TARGET_WORDS = prepare_target_words(target_words)
        tw_mode_automatic_mode = False
    else:
        USER_TARGET_WORDS = None
        tw_mode_automatic_mode = True

    # Text preprocessing
    original_text = current_text
    current_text = current_text.replace('.', '. ').replace('. . .', '...').replace('  ', ' ').replace('…', '...') \
        .replace('…', '...').replace('—', '-').replace('\u2014', '-').replace('—', '-').replace('-\n', '') \
        .replace('\n', '%^&*')
    current_text_sentences = [sent.text.strip() for sent in nlp(current_text).sents]
    logs.update(label='Получили Ваш текст!', state='running')
    progress.progress(10)

    # Compute frequency dict
    FREQ_DICT = compute_frequency_dict(current_text)

    # Get maximum frequency (top 5% barrier)
    _frequency_barrier_percent = 0.05
    for j, tp in enumerate(FREQ_DICT.items()):
        if j < len(FREQ_DICT) * _frequency_barrier_percent:
            MAX_FREQUENCY = tp[1]
    MAX_FREQUENCY = 3 if MAX_FREQUENCY < 3 else MAX_FREQUENCY
    logs.update(label="Посчитали немного статистики!", state='running')
    progress.progress(15)

    # Choose necessary language minimum according to user's input
    if level:
        target_minimum, distractor_minimum = MINIMUM_SETS[level]
    else:
        target_minimum = None
        distractor_minimum = None
        logs.error('Вы не выбрали языковой уровень!')
        st.stop()

    # Start generation process
    workflow = [SENTENCE(original=sent.strip(), n_sentence=num, max_num_distractors=num_distractors)
                for num, sent in enumerate(current_text_sentences)]
    logs.update(label="Запускаем процесс генерации заданий!", state='running')
    progress.progress(20)

    # Define summary length
    text_length = len(current_text_sentences)
    if text_length <= 15:
        summary_length = text_length
    elif text_length <= 25:
        summary_length = 15
    else:
        n = (text_length - 20) // 5
        summary_length = 15 + 2 * n
    round_summary_length = summary_length - (summary_length % - 10)

    # Get summary. May choose between round_summary_length and summary_length
    SUMMARY = summarization(current_text, num_sentences=round_summary_length)
    logs.update('Нашли интересные предложения. Пригодятся!')
    progress.progress(25)

    for sentence in workflow:
        sentence.lemmatize_sentence()

    for sentence in workflow:
        sentence.bind_phrases()
    logs.update(label="Подготовили предложения для дальнейшей работы!", state='running')
    progress.progress(30)

    for j, sentence in enumerate(workflow):
        sentence.search_target_words(target_words_automatic_mode=tw_mode_automatic_mode,
                                     target_minimum=target_minimum,
                                     user_target_words=USER_TARGET_WORDS,
                                     frequency_dict=FREQ_DICT, 
                                     summary=SUMMARY)
        progress.progress(int(30 + (j * (20 / len(workflow)))))
    progress.progress(50)
    DUPLICATE_TARGET_WORDS = defaultdict(list)
    for sentence in workflow:
        for target_word in sentence.target_words:
            DUPLICATE_TARGET_WORDS[target_word['lemma']].append(target_word)
    RESULT_TW = []
    for tw_lemma, tw_data in DUPLICATE_TARGET_WORDS.items():
        RESULT_TW.append(sample(tw_data, 1)[0])
    for sentence in workflow:
        for target_word in sentence.target_words:
            if target_word not in RESULT_TW:
                global_bad_target_words.append(target_word['original_text'])
                sentence.target_words.remove(target_word)
    progress.progress(55)
    logs.update(label='Выбрали слова-пропуски!', state='running')

    for sentence in workflow:
        for i, target_word in enumerate(sentence.target_words):
            temp = current_text_sentences[:]
            temp[sentence.n_sentence] = target_word['masked_sentence']
            sentence.text_with_masked_task = ' '.join(temp).replace('%^&*', '\n')
            sentence.target_words[i]['text_with_masked_task'] = ' '.join(temp).replace('%^&*', '\n')

    for sentence in workflow:
        sentence.filter_target_words(target_words_automatic_mode=tw_mode_automatic_mode)
    progress.progress(60)

    RESULT_TASKS = []
    for sentence in workflow:
        for target_word in sentence.target_words:
            task = TASK(task_data=target_word, max_num_distractors=num_distractors)
            RESULT_TASKS.append(task)

    for num, task in enumerate(RESULT_TASKS):
        task.attach_distractors_to_target_word(model=mask_filler, 
                                               scaler=scaler,
                                               classifier=classifier,
                                               pos_dict=pos_dict,
                                               level_name=level,
                                               global_distractors=GLOBAL_DISTRACTORS,
                                               distractor_minimum=distractor_minimum,
                                               max_frequency=MAX_FREQUENCY)
        progress_d.progress(num / len(RESULT_TASKS))
        logs.update(label=f'Обработали {num}/{len(RESULT_TASKS)} целевых слов!', state='running')
    logs.update(label=f'Обработали {len(RESULT_TASKS)}/{len(RESULT_TASKS)} целевых слов!', state='running')
    progress_d.progress(100)
    progress.progress(70)
    logs.update(label='Подобрали неправильные варианты!', state='running')

    for task in RESULT_TASKS:
        task.inflect_distractors()
    progress.progress(80)
    logs.update(label='Просклоняли и проспрягали неправильные варианты!', state='running')

    for task in RESULT_TASKS:
        task.sample_distractors(num_distractors=num_distractors)
    progress.progress(85)
    RESULT_TASKS = list(filter(lambda t: not t.bad_target_word, RESULT_TASKS))

    for task in RESULT_TASKS[::-1]:
        if task.bad_target_word:
            RESULT_TASKS.remove(task)

    # Compute number of final tasks
    if len(RESULT_TASKS) >= 20:
        NUMBER_TASKS = 20
    else:
        if len(RESULT_TASKS) >= 15:
            NUMBER_TASKS = 15
        else:
            if len(RESULT_TASKS) >= 10:
                NUMBER_TASKS = 10
            else:
                NUMBER_TASKS = len(RESULT_TASKS)
    RESULT_TASKS_in_summary = list(filter(lambda task: task.in_summary, RESULT_TASKS))
    RESULT_TASTS_not_in_summary = list(filter(lambda task: not task.in_summary, RESULT_TASKS))
    if len(RESULT_TASKS_in_summary) >= NUMBER_TASKS:
        RESULT_TASKS = RESULT_TASKS_in_summary
    else:
        RESULT_TASKS = RESULT_TASKS_in_summary + sample(RESULT_TASTS_not_in_summary, NUMBER_TASKS - len(RESULT_TASKS_in_summary))
    RESULT_TASKS = sorted(RESULT_TASKS, key=lambda t: (t.sentence_number, t.position_in_sentence))

    for task in RESULT_TASKS:
        task.compile_task(max_num_distractors=num_distractors)
    progress.progress(90)
    logs.update(label='Отобрали лучшие задания!', state='running')

    TEXT_WITH_GAPS = []
    VARIANTS = []
    tasks_counter = 1
    for i, sentence in enumerate(current_text_sentences):
        for task in RESULT_TASKS:
            if task.sentence_text == sentence:
                sentence = sentence.replace(task.original_text, f'__________({tasks_counter})')
                VARIANTS.append(task.variants)
                tasks_counter += 1
        TEXT_WITH_GAPS.append(sentence)
    del RESULT_TASKS

    TEXT_WITH_GAPS = ' '.join([sentence for sentence in TEXT_WITH_GAPS]).replace('%^&*', '\n')
    PREPARED_TASKS = prepare_tasks(VARIANTS)
    STUDENT_OUT = f'{TEXT_WITH_GAPS}\n\n{"=" * 70}\n\n{PREPARED_TASKS["TASKS_STUDENT"]}'
    TEACHER_OUT = f'{TEXT_WITH_GAPS}\n\n{"=" * 70}\n\n{PREPARED_TASKS["TASKS_TEACHER"]}\n\n{"=" * 70}\n\n' \
                  f'{PREPARED_TASKS["KEYS_ONLY"]}'
    TOTAL_OUT = f'{original_text}\n\n{"$" * 70}\n\n{STUDENT_OUT}\n\n{"=" * 70}\n\n{PREPARED_TASKS["TASKS_TEACHER"]}' \
                f'\n\n{"$" * 70}\n\n{PREPARED_TASKS["KEYS_ONLY"]}'
    logs.update(label='Сейчас все будет готово!', state='running')
    progress.progress(95)
    save_name = save_name if save_name != '' else f'{str(datetime.datetime.now())[:-7]}_{original_text[:20]}'
    out = {
        'name': save_name,
        'STUDENT_OUT': STUDENT_OUT,
        'TEACHER_OUT': TEACHER_OUT,
        'TEXT_WITH_GAPS': TEXT_WITH_GAPS,
        'TASKS_ONLY': PREPARED_TASKS["RAW_TASKS"],
        'KEYS_ONLY': PREPARED_TASKS["KEYS_ONLY"],
        'KEYS_ONLY_RAW': PREPARED_TASKS["RAW_KEYS_ONLY"],
        'TOTAL_OUT': TOTAL_OUT,
        'ORIGINAL': original_text,
        'BAD_USER_TARGET_WORDS': sorted(set(global_bad_target_words))
    }
    return out