feat: notify on new webinar, storing N last webinars in redis

edu_master/webinar-checker/checker.py

@@ -1,6 +1,7 @@
 import os
 import logging
 import redis
+import json
 
 from telegram import Update, InlineKeyboardButton, InlineKeyboardMarkup
 from telegram.ext import Application, CommandHandler, CallbackQueryHandler, ContextTypes
@@ -28,6 +29,7 @@ KEY_WHITELIST = "bot:whitelist"
 KEY_WHITELIST_ENABLED = "bot:whitelist_enabled"
 KEY_SUBSCRIBERS = "bot:subscribers"
 KEY_PHPSESSID = "EDU_PHPSESSID"
+KEY_WEBINAR_HISTORY = "bot:webinar_history"  # Stores last 5 webinars
 
 # Initialize Redis
 try:
@@ -196,6 +198,30 @@ async def admin_callback(update: Update, context: ContextTypes.DEFAULT_TYPE):
 
 # --- Webinar Checking Job ---
 
+def get_webinar_key(name: str, url: str) -> str:
+    """Generate unique key for a webinar based on name and URL."""
+    return f"{name}|{url}"
+
+def get_stored_webinars() -> list:
+    """Get list of stored webinar keys from Redis."""
+    data = redis_client.get(KEY_WEBINAR_HISTORY)
+    if data:
+        try:
+            return json.loads(data)
+        except Exception as e:
+            logger.error(f"Failed to parse webinar history: {e}")
+    return []
+
+def store_webinars(webinar_keys: list):
+    """Store up to 5 most recent webinar keys in Redis."""
+    # Keep only last 5
+    webinar_keys = webinar_keys[-5:]
+    try:
+        redis_client.set(KEY_WEBINAR_HISTORY, json.dumps(webinar_keys))
+        logger.info(f"Stored {len(webinar_keys)} webinar(s) in history")
+    except Exception as e:
+        logger.error(f"Failed to store webinar history: {e}")
+
 async def check_webinars_job(context: ContextTypes.DEFAULT_TYPE):
     """Background job to check for webinars using Async Playwright.
     
@@ -216,7 +242,7 @@ async def check_webinars_job(context: ContextTypes.DEFAULT_TYPE):
         # ---------------------
         return None
 
-    webinar_details = []
+    current_webinars = []  # List of dicts with name, url, and formatted text
     content = ""
     
     try:
@@ -259,9 +285,26 @@ async def check_webinars_job(context: ContextTypes.DEFAULT_TYPE):
                         count = await rows.count()
                         
                         for i in range(count):
-                            text = await rows.nth(i).inner_text()
+                            row = rows.nth(i)
+                            text = await row.inner_text()
+                            
                             if "Жодного онлайн уроку зараз" not in text:
-                                webinar_details.append(text.strip())
+                                # Extract name (topic) from first column
+                                name_elem = row.locator('td').nth(0)
+                                name = await name_elem.inner_text()
+                                name = name.strip()
+                                
+                                # Extract join URL from fourth column
+                                url_elem = row.locator('td').nth(3).locator('a[href*="/webinar/join/"]').first
+                                url = await url_elem.get_attribute('href')
+                                
+                                if name and url:
+                                    current_webinars.append({
+                                        'name': name,
+                                        'url': url,
+                                        'text': text.strip()
+                                    })
+                                    logger.info(f"Found webinar: {name} -> {url}")
                     else:
                         logger.info("No webinars found (expected message present)")
                         
@@ -287,7 +330,7 @@ async def check_webinars_job(context: ContextTypes.DEFAULT_TYPE):
         return None
     
     # --- DEBUG LOGGING (Saving last response content) ---
-    if not webinar_details and content: #if no webinars found, save the page content
+    if not current_webinars and content: #if no webinars found, save the page content
         try:
             with open('response.html', 'w', encoding='utf-8') as f:
                 f.write(content)
@@ -296,19 +339,49 @@ async def check_webinars_job(context: ContextTypes.DEFAULT_TYPE):
             logger.error(f"Failed to write debug HTML: {e}")
     # -----------------------------------------------------
 
-    # Notify subscribers if webinars found
+    # Check for NEW webinars and notify
-    if webinar_details:
+    if current_webinars:
-        message = "🎓 <b>Найден вебинар!</b>\n\n" + "\n\n".join([f"📌 {d}" for d in webinar_details])
+        # Get stored webinar history
-        subscribers = redis_client.smembers(KEY_SUBSCRIBERS)
+        stored_keys = get_stored_webinars()
+        logger.info(f"Stored webinar keys: {stored_keys}")
         
-        for sub_id in subscribers:
+        # Find new webinars (not in history)
-            try:
+        new_webinars = []
-                await context.bot.send_message(chat_id=sub_id, text=message, parse_mode='HTML')
+        current_keys = []
-                logger.info(f"Notification sent to {sub_id}")
-            except Exception as e:
-                logger.error(f"Failed to send to {sub_id}: {e}")
         
-    return len(webinar_details)
+        for webinar in current_webinars:
+            key = get_webinar_key(webinar['name'], webinar['url'])
+            current_keys.append(key)
+            
+            if key not in stored_keys:
+                new_webinars.append(webinar)
+                logger.info(f"NEW webinar detected: {webinar['name']}")
+        
+        # Update stored history with current webinars
+        # Merge old and new, keeping only last 5
+        updated_keys = stored_keys + [k for k in current_keys if k not in stored_keys]
+        store_webinars(updated_keys)
+        
+        # Notify subscribers ONLY about NEW webinars
+        if new_webinars:
+            message = "🎓 <b>Новий вебінар!</b>\n\n" + "\n\n".join([
+                f"📌 <b>{w['name']}</b>\n🔗 https://edu.edu.vn.ua{w['url']}" 
+                for w in new_webinars
+            ])
+            
+            subscribers = redis_client.smembers(KEY_SUBSCRIBERS)
+            logger.info(f"Sending notification about {len(new_webinars)} new webinar(s) to {len(subscribers)} subscriber(s)")
+            
+            for sub_id in subscribers:
+                try:
+                    await context.bot.send_message(chat_id=sub_id, text=message, parse_mode='HTML')
+                    logger.info(f"Notification sent to {sub_id}")
+                except Exception as e:
+                    logger.error(f"Failed to send to {sub_id}: {e}")
+        else:
+            logger.info(f"Found {len(current_webinars)} webinar(s), but all are already known")
+    
+    return len(current_webinars)
 
 # --- Main ---