feat: персональные AI модели, оптимизация задач, фильтрация словаря

- Добавлена поддержка персональных AI моделей для каждого пользователя - Оптимизация создания заданий: батч-запрос к AI вместо N запросов - Фильтрация слов по языку изучения (source_lang) в словаре - Удалены неиспользуемые колонки examples и category из vocabulary - Миграции для ai_model_id и удаления колонок 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
2025-12-08 16:43:08 +03:00
parent 6138af4e63
commit 16a7df0343
13 changed files with 507 additions and 142 deletions
--- a/bot/handlers/import_text.py
+++ b/bot/handlers/import_text.py
@@ -187,7 +187,7 @@ async def import_single_word(callback: CallbackQuery, state: FSMContext):

        # Проверяем, нет ли уже такого слова
        existing = await VocabularyService.get_word_by_original(
-            session, user_id, word_data['word']
+            session, user_id, word_data['word'], source_lang=user.learning_language
        )

        if existing:
@@ -195,10 +195,7 @@ async def import_single_word(callback: CallbackQuery, state: FSMContext):
            return

        # Добавляем слово
-        learn = user.learning_language if user else 'en'
        translation_lang = get_user_translation_lang(user)
-        ctx = word_data.get('context')
-        examples = ([{learn: ctx, translation_lang: ''}] if ctx else [])

        await VocabularyService.add_word(
            session=session,
@@ -208,10 +205,8 @@ async def import_single_word(callback: CallbackQuery, state: FSMContext):
            source_lang=user.learning_language if user else None,
            translation_lang=translation_lang,
            transcription=word_data.get('transcription'),
-            examples=examples,
-            source=WordSource.CONTEXT,
-            category='imported',
-            difficulty_level=data.get('level')
+            difficulty_level=data.get('level'),
+            source=WordSource.CONTEXT
        )

    lang = (user.language_interface if user else 'ru') or 'ru'
@@ -235,7 +230,7 @@ async def import_all_words(callback: CallbackQuery, state: FSMContext):
        for word_data in words:
            # Проверяем, нет ли уже такого слова
            existing = await VocabularyService.get_word_by_original(
-                session, user_id, word_data['word']
+                session, user_id, word_data['word'], source_lang=user.learning_language
            )

            if existing:
@@ -243,10 +238,7 @@ async def import_all_words(callback: CallbackQuery, state: FSMContext):
                continue

            # Добавляем слово
-            learn = user.learning_language if user else 'en'
            translation_lang = get_user_translation_lang(user)
-            ctx = word_data.get('context')
-            examples = ([{learn: ctx, translation_lang: ''}] if ctx else [])

            await VocabularyService.add_word(
                session=session,
@@ -256,10 +248,8 @@ async def import_all_words(callback: CallbackQuery, state: FSMContext):
                source_lang=user.learning_language if user else None,
                translation_lang=translation_lang,
                transcription=word_data.get('transcription'),
-                examples=examples,
-                source=WordSource.CONTEXT,
-                category='imported',
-                difficulty_level=data.get('level')
+                difficulty_level=data.get('level'),
+                source=WordSource.CONTEXT
            )
            added_count += 1

@@ -478,7 +468,7 @@ async def import_file_all_words(callback: CallbackQuery, state: FSMContext):
        for word_data in words:
            # Проверяем, нет ли уже такого слова
            existing = await VocabularyService.get_word_by_original(
-                session, user_id, word_data['word']
+                session, user_id, word_data['word'], source_lang=user.learning_language
            )

            if existing:
--- a/bot/handlers/tasks.py
+++ b/bot/handlers/tasks.py
@@ -180,7 +180,10 @@ async def generate_new_words_tasks(callback: CallbackQuery, state: FSMContext, u
        return

    # Преобразуем слова в задания нужного типа
-    tasks = await create_tasks_from_words(words, task_type, lang, user.learning_language, translation_lang)
+    tasks = await create_tasks_from_words(
+        words, task_type, lang, user.learning_language, translation_lang,
+        level=level
+    )

    await state.update_data(
        tasks=tasks,
@@ -196,26 +199,68 @@ async def generate_new_words_tasks(callback: CallbackQuery, state: FSMContext, u
    await show_current_task(callback.message, state)


-async def create_tasks_from_words(words: list, task_type: str, lang: str, learning_lang: str, translation_lang: str) -> list:
-    """Создать задания из списка слов в зависимости от типа"""
+async def create_tasks_from_words(
+    words: list,
+    task_type: str,
+    lang: str,
+    learning_lang: str,
+    translation_lang: str,
+    level: str = None
+) -> list:
+    """Создать задания из списка слов в зависимости от типа (оптимизировано - 1 запрос к AI)"""
    import random
-    tasks = []

+    # 1. Определяем типы заданий для всех слов
+    word_tasks = []
    for word in words:
+        if task_type == 'mix':
+            chosen_type = random.choice(['word_translate', 'fill_blank', 'sentence_translate'])
+        else:
+            chosen_type = task_type
+        word_tasks.append({
+            'word_data': word,
+            'chosen_type': chosen_type
+        })
+
+    # 2. Собираем задания, требующие генерации предложений
+    ai_tasks = []
+    ai_task_indices = []  # Индексы в word_tasks для сопоставления результатов
+
+    for i, wt in enumerate(word_tasks):
+        if wt['chosen_type'] in ('fill_blank', 'sentence_translate'):
+            ai_tasks.append({
+                'word': wt['word_data'].get('word', ''),
+                'task_type': wt['chosen_type']
+            })
+            ai_task_indices.append(i)
+
+    # 3. Один запрос к AI для всех предложений (если нужно)
+    ai_results = []
+    if ai_tasks:
+        ai_results = await ai_service.generate_task_sentences_batch(
+            ai_tasks,
+            learning_lang=learning_lang,
+            translation_lang=translation_lang
+        )
+
+    # Создаём маппинг: индекс в word_tasks -> результат AI
+    ai_results_map = {}
+    for idx, result in zip(ai_task_indices, ai_results):
+        ai_results_map[idx] = result
+
+    # 4. Собираем финальные задания
+    tasks = []
+    for i, wt in enumerate(word_tasks):
+        word = wt['word_data']
+        chosen_type = wt['chosen_type']
+
        word_text = word.get('word', '')
        translation = word.get('translation', '')
        transcription = word.get('transcription', '')
        example = word.get('example', '')
        example_translation = word.get('example_translation', '')

-        if task_type == 'mix':
-            # Случайный тип
-            chosen_type = random.choice(['word_translate', 'fill_blank', 'sentence_translate'])
-        else:
-            chosen_type = task_type
-
        if chosen_type == 'word_translate':
-            # Перевод слова
            translate_prompt = t(lang, 'tasks.translate_to', lang_name=t(lang, f'lang.{translation_lang}'))
            tasks.append({
                'type': 'translate',
@@ -224,16 +269,12 @@ async def create_tasks_from_words(words: list, task_type: str, lang: str, learni
                'correct_answer': translation,
                'transcription': transcription,
                'example': example,
-                'example_translation': example_translation
+                'example_translation': example_translation,
+                'difficulty_level': level
            })

        elif chosen_type == 'fill_blank':
-            # Заполнение пропуска - генерируем предложение через AI
-            sentence_data = await ai_service.generate_fill_in_sentence(
-                word_text,
-                learning_lang=learning_lang,
-                translation_lang=translation_lang
-            )
+            sentence_data = ai_results_map.get(i, {})
            if translation_lang == 'en':
                fill_title = "Fill in the blank:"
            elif translation_lang == 'ja':
@@ -243,21 +284,17 @@ async def create_tasks_from_words(words: list, task_type: str, lang: str, learni

            tasks.append({
                'type': 'fill_in',
-                'question': f"{fill_title}\n\n<b>{sentence_data['sentence']}</b>\n\n<i>{sentence_data.get('translation', '')}</i>",
+                'question': f"{fill_title}\n\n<b>{sentence_data.get('sentence', '___')}</b>\n\n<i>{sentence_data.get('translation', '')}</i>",
                'word': word_text,
-                'correct_answer': sentence_data['answer'],
+                'correct_answer': sentence_data.get('answer', word_text),
                'transcription': transcription,
                'example': example,
-                'example_translation': example_translation
+                'example_translation': example_translation,
+                'difficulty_level': level
            })

        elif chosen_type == 'sentence_translate':
-            # Перевод предложения - генерируем предложение через AI
-            sentence_data = await ai_service.generate_sentence_for_translation(
-                word_text,
-                learning_lang=learning_lang,
-                translation_lang=translation_lang
-            )
+            sentence_data = ai_results_map.get(i, {})
            if translation_lang == 'en':
                sentence_title = "Translate the sentence:"
                word_hint = "Word"
@@ -270,12 +307,13 @@ async def create_tasks_from_words(words: list, task_type: str, lang: str, learni

            tasks.append({
                'type': 'sentence_translate',
-                'question': f"{sentence_title}\n\n<b>{sentence_data['sentence']}</b>\n\n📝 {word_hint}: <code>{word_text}</code> — {translation}",
+                'question': f"{sentence_title}\n\n<b>{sentence_data.get('sentence', word_text)}</b>\n\n📝 {word_hint}: <code>{word_text}</code> — {translation}",
                'word': word_text,
-                'correct_answer': sentence_data['translation'],
+                'correct_answer': sentence_data.get('translation', translation),
                'transcription': transcription,
                'example': example,
-                'example_translation': example_translation
+                'example_translation': example_translation,
+                'difficulty_level': level
            })

    return tasks
@@ -468,6 +506,7 @@ async def add_task_word(callback: CallbackQuery, state: FSMContext):
    transcription = task.get('transcription', '')
    example = task.get('example', '')  # Пример использования как контекст
    example_translation = task.get('example_translation', '')  # Перевод примера
+    difficulty_level = task.get('difficulty_level')  # Уровень сложности

    async with async_session_maker() as session:
        user = await UserService.get_user_by_telegram_id(session, callback.from_user.id)
@@ -477,9 +516,10 @@ async def add_task_word(callback: CallbackQuery, state: FSMContext):
            return

        lang = get_user_lang(user)
+        translation_lang = get_user_translation_lang(user)

        # Проверяем, есть ли слово уже в словаре
-        existing = await VocabularyService.get_word_by_original(session, user.id, word)
+        existing = await VocabularyService.get_word_by_original(session, user.id, word, source_lang=user.learning_language)

        if existing:
            await callback.answer(t(lang, 'tasks.word_already_exists', word=word), show_alert=True)
@@ -492,8 +532,9 @@ async def add_task_word(callback: CallbackQuery, state: FSMContext):
            word_original=word,
            word_translation=translation,
            source_lang=user.learning_language,
-            translation_lang=get_user_translation_lang(user),
+            translation_lang=translation_lang,
            transcription=transcription,
+            difficulty_level=difficulty_level,
            source=WordSource.AI_TASK
        )

--- a/bot/handlers/vocabulary.py
+++ b/bot/handlers/vocabulary.py
@@ -56,7 +56,7 @@ async def process_word_addition(message: Message, state: FSMContext, word: str):
            return

        # Проверяем, есть ли уже такое слово
-        existing_word = await VocabularyService.find_word(session, user.id, word)
+        existing_word = await VocabularyService.find_word(session, user.id, word, source_lang=user.learning_language)
        if existing_word:
            lang = get_user_lang(user)
            await message.answer(t(lang, 'add.exists', word=word, translation=existing_word.word_translation))
@@ -107,7 +107,6 @@ async def process_word_addition(message: Message, state: FSMContext, word: str):
    card_text = (
        f"📝 <b>{word_data['word']}</b>\n"
        f"🔊 [{word_data.get('transcription', '')}]\n\n"
-        f"{t(lang, 'add.category_label')}: {word_data.get('category', '')}\n"
        f"{t(lang, 'add.level_label')}: {word_data.get('difficulty', 'A1')}"
        f"{translations_text}"
        f"{t(lang, 'add.confirm_question')}"
@@ -153,7 +152,6 @@ async def confirm_add_word(callback: CallbackQuery, state: FSMContext):
            source_lang=source_lang,
            translation_lang=translation_lang,
            transcription=word_data.get("transcription"),
-            category=word_data.get("category"),
            difficulty_level=word_data.get("difficulty"),
            source=WordSource.MANUAL
        )
--- a/bot/handlers/words.py
+++ b/bot/handlers/words.py
@@ -158,22 +158,16 @@ async def add_single_word(callback: CallbackQuery, state: FSMContext):
        user = await UserService.get_user_by_telegram_id(session, callback.from_user.id)
        # Проверяем, нет ли уже такого слова
        existing = await VocabularyService.get_word_by_original(
-            session, user_id, word_data['word']
+            session, user_id, word_data['word'], source_lang=user.learning_language
        )

        if existing:
-            async with async_session_maker() as session:
-                user = await UserService.get_user_by_telegram_id(session, callback.from_user.id)
-            lang = (user.language_interface if user else 'ru') or 'ru'
+            lang = get_user_lang(user)
            await callback.answer(t(lang, 'words.already_exists', word=word_data['word']), show_alert=True)
            return

        # Добавляем слово
-        # Формируем examples с учётом языков
-        learn = user.learning_language if user else 'en'
        translation_lang = get_user_translation_lang(user)
-        ex = word_data.get('example')
-        examples = ([{learn: ex, translation_lang: ''}] if ex else [])

        await VocabularyService.add_word(
            session=session,
@@ -183,10 +177,8 @@ async def add_single_word(callback: CallbackQuery, state: FSMContext):
            source_lang=user.learning_language if user else None,
            translation_lang=translation_lang,
            transcription=word_data.get('transcription'),
-            examples=examples,
-            source=WordSource.SUGGESTED,
-            category=data.get('theme', 'general'),
-            difficulty_level=data.get('level')
+            difficulty_level=data.get('level'),
+            source=WordSource.SUGGESTED
        )

    async with async_session_maker() as session:
@@ -203,7 +195,6 @@ async def add_all_words(callback: CallbackQuery, state: FSMContext):
    data = await state.get_data()
    words = data.get('words', [])
    user_id = data.get('user_id')
-    theme = data.get('theme', 'general')

    added_count = 0
    skipped_count = 0
@@ -213,7 +204,7 @@ async def add_all_words(callback: CallbackQuery, state: FSMContext):
        for word_data in words:
            # Проверяем, нет ли уже такого слова
            existing = await VocabularyService.get_word_by_original(
-                session, user_id, word_data['word']
+                session, user_id, word_data['word'], source_lang=user.learning_language
            )

            if existing:
@@ -221,10 +212,7 @@ async def add_all_words(callback: CallbackQuery, state: FSMContext):
                continue

            # Добавляем слово
-            learn = user.learning_language if user else 'en'
            translation_lang = get_user_translation_lang(user)
-            ex = word_data.get('example')
-            examples = ([{learn: ex, translation_lang: ''}] if ex else [])

            await VocabularyService.add_word(
                session=session,
@@ -234,10 +222,8 @@ async def add_all_words(callback: CallbackQuery, state: FSMContext):
                source_lang=user.learning_language if user else None,
                translation_lang=translation_lang,
                transcription=word_data.get('transcription'),
-                examples=examples,
-                source=WordSource.SUGGESTED,
-                category=theme,
-                difficulty_level=data.get('level')
+                difficulty_level=data.get('level'),
+                source=WordSource.SUGGESTED
            )
            added_count += 1