feat: add schedule scraper for lessons table

Add /schedule command to scrape edu.edu.vn.ua/lessons/table via Playwright. Inline keyboard flow: pick weekday (with today/tomorrow shortcuts), then pick class. Cache 5h per user. Parse subjects/notes/teachers, multi-lesson cells (hr-separated).
2026-09-06 15:46:41 +02:00
parent b752bf88bd
commit d2d4efb0d7
1 changed files with 299 additions and 1 deletions
@@ -34,6 +34,7 @@ EDU_BASE = _env('EDU_URL_BASE', 'https://edu.edu.vn.ua')
 EDU_WEBINAR_PATH = _env('EDU_URL_WEBINAR', '/webinar/useractive')
 WEBINAR_URL = f'{EDU_BASE.rstrip("/")}/{EDU_WEBINAR_PATH.lstrip("/")}'
 DIARY_URL = f'{EDU_BASE.rstrip("/")}/user/diary'
+SCHEDULE_URL = f'{EDU_BASE.rstrip("/")}/lessons/table'

 WEBINAR_CHECK_INTERVAL = int(_env('WEBINAR_CHECK_INTERVAL', 60))
 REDIS_HOST = _env('REDIS_HOST', 'redis')
@@ -270,6 +271,9 @@ def get_admin_keyboard(user_id: int):

 # --- Diary Functions ---

+SCHEDULE_WEEKDAYS_FULL = ['Понеділок', 'Вівторок', 'Середа', 'Четвер', "П'ятниця", 'Субота', 'Неділя']
+SCHEDULE_CACHE_TTL = 18000  # 5 hours
+
 DIARY_MONTH_NAMES = [
    '',
    'Січня',
@@ -478,6 +482,220 @@ async def _get_diary_data(context: ContextTypes.DEFAULT_TYPE) -> dict | None:
    return data


+# --- Schedule Functions ---
+
+
+def get_schedule_day_keyboard():
+    today = datetime.now()
+    tomorrow = today + timedelta(days=1)
+    keyboard = [
+        [
+            InlineKeyboardButton(f'📌 Сьогодні ({SCHEDULE_WEEKDAYS_FULL[today.weekday()]})', callback_data=f'schedule_day_{SCHEDULE_WEEKDAYS_FULL[today.weekday()]}'),
+            InlineKeyboardButton(f'📌 Завтра ({SCHEDULE_WEEKDAYS_FULL[tomorrow.weekday()]})', callback_data=f'schedule_day_{SCHEDULE_WEEKDAYS_FULL[tomorrow.weekday()]}'),
+        ],
+        [
+            InlineKeyboardButton('Пн', callback_data=f'schedule_day_{SCHEDULE_WEEKDAYS_FULL[0]}'),
+            InlineKeyboardButton('Вт', callback_data=f'schedule_day_{SCHEDULE_WEEKDAYS_FULL[1]}'),
+            InlineKeyboardButton('Ср', callback_data=f'schedule_day_{SCHEDULE_WEEKDAYS_FULL[2]}'),
+        ],
+        [
+            InlineKeyboardButton('Чт', callback_data=f'schedule_day_{SCHEDULE_WEEKDAYS_FULL[3]}'),
+            InlineKeyboardButton('Пт', callback_data=f'schedule_day_{SCHEDULE_WEEKDAYS_FULL[4]}'),
+            InlineKeyboardButton('Сб', callback_data=f'schedule_day_{SCHEDULE_WEEKDAYS_FULL[5]}'),
+        ],
+        [
+            InlineKeyboardButton('Нд', callback_data=f'schedule_day_{SCHEDULE_WEEKDAYS_FULL[6]}'),
+        ],
+    ]
+    return InlineKeyboardMarkup(keyboard)
+
+
+def get_schedule_class_keyboard(classes: list):
+    rows = []
+    for i in range(0, len(classes), 4):
+        row = [InlineKeyboardButton(str(c), callback_data=f'schedule_class_{c}') for c in classes[i:i + 4]]
+        rows.append(row)
+    return InlineKeyboardMarkup(rows)
+
+
+def _parse_schedule_cell(cell_html: str) -> list:
+    """Parse a single schedule cell, returning list of {subject, note, teacher}."""
+    lessons = []
+    parts = re.split(r'<hr\s*/?>', cell_html, flags=re.IGNORECASE)
+    for part in parts:
+        text = re.sub(r'<[^>]+>', '\n', part)
+        text = re.sub(r'&nbsp;', ' ', text)
+        text = text.strip()
+        if not text:
+            continue
+        lines = [l.strip() for l in text.split('\n') if l.strip()]
+        if not lines:
+            continue
+        subject = lines[0] if lines else ''
+        note = ''
+        teacher = ''
+        for line in lines[1:]:
+            if line in ('вебінар', 'онлайн', 'асинхрон', 'асинхронно') or 'Вебінар' in line or 'Асинхронн' in line or 'вебінар' in line.lower() or 'онлайн' in line.lower() or 'асинхрон' in line.lower() or 'А/С урок' in line or 'Онлайн' in line:
+                note = line
+            else:
+                teacher = line
+        lessons.append({'subject': subject, 'note': note, 'teacher': teacher})
+    return lessons
+
+
+def _parse_schedule_html(table_html: str) -> dict:
+    """Parse schedule HTML table into {classes: [...], weekdays: {day: [...]}}."""
+    result = {'classes': [], 'weekdays': {}}
+    rows = re.findall(r'<tr[^>]*>(.*?)</tr>', table_html, re.DOTALL)
+
+    current_day = None
+    classes_header = []
+
+    for row in rows:
+        row_classes_match = re.search(r'class="([^"]*)"', row)
+        row_classes = row_classes_match.group(1) if row_classes_match else ''
+
+        if 'first-row' in row_classes:
+            cells = re.findall(r'<td[^>]*>(.*?)</td>', row, re.DOTALL)
+            if cells:
+                day_text = re.sub(r'<[^>]+>', '', cells[0]).strip()
+                current_day = day_text
+                result['weekdays'][current_day] = []
+                classes_header = []
+                for cell in cells[1:]:
+                    cls = re.sub(r'<[^>]+>', '', cell).strip()
+                    if cls:
+                        try:
+                            classes_header.append(int(cls))
+                        except ValueError:
+                            classes_header.append(cls)
+                result['classes'] = classes_header
+            continue
+
+        if 'second-row' in row_classes:
+            continue
+
+        if current_day is None:
+            continue
+
+        cells = re.findall(r'<td[^>]*>(.*?)</td>', row, re.DOTALL)
+        if len(cells) < 3:
+            continue
+
+        first_cell_text = re.sub(r'<[^>]+>', '', cells[0]).strip()
+        if not re.match(r'^\d+\.', first_cell_text):
+            if 'Позакласне' in first_cell_text or re.search(r'colspan\s*=\s*"?(\d+)"?', row):
+                continue
+
+        lesson_num_match = re.match(r'^(\d+)\.', first_cell_text)
+        lesson_num = lesson_num_match.group(1) if lesson_num_match else ''
+        time_text = re.sub(r'<[^>]+>', '', cells[1]).strip()
+
+        entry = {'lesson_num': lesson_num, 'time': time_text, 'classes': {}}
+
+        data_cells = cells[2:]
+        for idx, cell in enumerate(data_cells):
+            cls = classes_header[idx] if idx < len(classes_header) else idx
+            lessons = _parse_schedule_cell(cell)
+            entry['classes'][cls] = lessons
+
+        result['weekdays'][current_day].append(entry)
+
+    return result
+
+
+async def fetch_schedule_data(phpsessid: str) -> dict | None:
+    logger.info('Fetching schedule data via Playwright...')
+    try:
+        async with async_playwright() as p:
+            browser = await p.chromium.connect(PLAYWRIGHT_WS)
+            try:
+                context_browser = await browser.new_context(user_agent=USER_AGENT)
+                await context_browser.add_cookies(
+                    [{'name': 'PHPSESSID', 'value': phpsessid, 'domain': 'edu.edu.vn.ua', 'path': '/'}]
+                )
+                page = await context_browser.new_page()
+
+                try:
+                    await page.goto(SCHEDULE_URL, wait_until='domcontentloaded')
+                    await page.wait_for_selector('table.schedule-table', timeout=10000)
+                    await page.wait_for_timeout(1500)
+
+                    table_html = await page.evaluate("""
+                        () => {
+                            const t = document.querySelector('table.schedule-table');
+                            return t ? t.outerHTML : null;
+                        }
+                    """)
+                    if not table_html:
+                        logger.error('table.schedule-table not found in DOM')
+                        return None
+
+                    with contextlib.suppress(Exception), open('/tmp/schedule_debug.html', 'w', encoding='utf-8') as f:
+                        f.write(table_html)
+
+                    data = _parse_schedule_html(table_html)
+                    total_lessons = sum(len(entry['classes'].get(c, [])) for day in data['weekdays'].values() for entry in day for c in data['classes'])
+                    logger.info(f'Schedule parsed: {len(data["weekdays"])} days, classes={data["classes"]}')
+
+                    return data
+
+                except Exception as e:
+                    logger.error(f'Error parsing schedule: {e}')
+                    return None
+                finally:
+                    await page.close()
+                    await context_browser.close()
+            finally:
+                await browser.close()
+    except Exception as e:
+        logger.error(f'Playwright error in schedule fetch: {e}')
+        return None
+
+
+async def _get_schedule_data(context: ContextTypes.DEFAULT_TYPE) -> dict | None:
+    cached = context.user_data.get('schedule_cache')
+    now_ts = time.time()
+    if cached and (now_ts - cached.get('timestamp', 0)) < SCHEDULE_CACHE_TTL:
+        return cached['data']
+    phpsessid = redis_client.get(KEY_PHPSESSID)
+    if not phpsessid:
+        return None
+    data = await fetch_schedule_data(phpsessid)
+    if data:
+        context.user_data['schedule_cache'] = {'data': data, 'timestamp': now_ts}
+    return data
+
+
+def format_schedule_day(data: dict, weekday: str, class_num) -> str:
+    days = data.get('weekdays', {})
+    entries = days.get(weekday, [])
+    class_str = str(class_num)
+    lines = [f'📅 <b>{weekday} — Клас {class_num}</b>', '─' * 20]
+
+    if not entries:
+        lines.append('Немає уроків')
+    else:
+        for entry in entries:
+            time_str = entry.get('time', '').replace(' ', '–')
+            lines.append(f'\n<b>{entry["lesson_num"]}. </b>({time_str})')
+
+            lessons = entry.get('classes', {}).get(class_num, [])
+            if not lessons:
+                lines.append('  Вільно')
+            else:
+                for i, lesson in enumerate(lessons):
+                    note_part = f' ({lesson["note"]})' if lesson.get('note') else ''
+                    lines.append(f'  📌 <b>{lesson["subject"]}</b>{note_part}')
+                    if lesson.get('teacher'):
+                        lines.append(f'  👨‍🏫 {lesson["teacher"]}')
+                    if i < len(lessons) - 1:
+                        lines.append('  ──')
+
+    lines.append(f'\n🔗 {SCHEDULE_URL}')
+    return '\n'.join(lines)
+
+
 # --- Command Handlers ---


@@ -675,6 +893,84 @@ async def diary_callback(update: Update, context: ContextTypes.DEFAULT_TYPE):
    await query.edit_message_text(text, parse_mode='HTML', reply_markup=get_diary_keyboard())


+# --- Schedule Command Handlers ---
+
+
+async def schedule_command(update: Update, _context: ContextTypes.DEFAULT_TYPE):
+    user = update.effective_user
+    if not is_whitelisted(user.id):
+        await update.message.reply_text(t(user.id, 'access_denied'))
+        return
+    await update.message.reply_text(
+        '🗓 <b>Розклад уроків</b> — виберіть день:', parse_mode='HTML', reply_markup=get_schedule_day_keyboard()
+    )
+
+
+async def schedule_callback(update: Update, context: ContextTypes.DEFAULT_TYPE):
+    query = update.callback_query
+    user_id = query.from_user.id
+    await query.answer()
+
+    if user_id != ADMIN_ID and not is_whitelisted(user_id):
+        await query.edit_message_text('⛔ Доступ заборонено.')
+        return
+
+    data = query.data
+
+    if data.startswith('schedule_day_'):
+        weekday = data[len('schedule_day_'):]
+        context.user_data['schedule_weekday'] = weekday
+        await query.edit_message_text(f'🗓 <b>{weekday}</b> — виберіть клас:', parse_mode='HTML')
+
+        schedule_data = await _get_schedule_data(context)
+        if not schedule_data:
+            await query.edit_message_text('❌ Не вдалося завантажити розклад.')
+            return
+        classes = schedule_data.get('classes', [])
+        if not classes:
+            await query.edit_message_text('❌ Класи не знайдені в розкладі.')
+            return
+        await query.edit_message_reply_markup(reply_markup=get_schedule_class_keyboard(classes))
+        return
+
+    if data.startswith('schedule_class_'):
+        class_num_str = data[len('schedule_class_'):]
+        weekday = context.user_data.get('schedule_weekday')
+        if not weekday:
+            await query.edit_message_text('❌ Спочатку виберіть день.')
+            return
+
+        await query.edit_message_text('🔄 Завантажую розклад...')
+        schedule_data = await _get_schedule_data(context)
+        if not schedule_data:
+            await query.edit_message_text('❌ Не вдалося завантажити розклад.')
+            return
+
+        try:
+            class_num = int(class_num_str)
+        except ValueError:
+            class_num = class_num_str
+
+        text = format_schedule_day(schedule_data, weekday, class_num)
+        if len(text) > 4096:
+            text = text[:4090] + '\n\n✂️ ...(обрізано)'
+
+        await query.edit_message_text(text, parse_mode='HTML')
+        return
+
+    if data == 'schedule_refresh':
+        context.user_data.pop('schedule_cache', None)
+        await query.edit_message_text('🔄 Завантажую розклад...')
+        schedule_data = await _get_schedule_data(context)
+        if not schedule_data:
+            await query.edit_message_text('❌ Не вдалося завантажити розклад.')
+            return
+        await query.edit_message_text(
+            '🗓 <b>Розклад уроків</b> — виберіть день:', parse_mode='HTML', reply_markup=get_schedule_day_keyboard()
+        )
+        return
+
+
 # --- Admin Callbacks ---


@@ -947,9 +1243,11 @@ def main():
    app.add_handler(CommandHandler('removeuser', remove_user))
    app.add_handler(CommandHandler('clearhistory', clear_history))
    app.add_handler(CommandHandler('diary', diary_command))
+    app.add_handler(CommandHandler('schedule', schedule_command))

-    # Callback handlers - diary first, then language selection, then admin panel
+    # Callback handlers - diary/schedule first, then language selection, then admin panel
    app.add_handler(CallbackQueryHandler(diary_callback, pattern='^diary_'))
+    app.add_handler(CallbackQueryHandler(schedule_callback, pattern='^schedule_'))
    app.add_handler(CallbackQueryHandler(language_callback, pattern='^lang_'))
    app.add_handler(CallbackQueryHandler(admin_callback))