Update main.py

main.py

@@ -1,4 +1,4 @@
-# main.py (Revised: Apify 201 fix + Supadata verify=False TEST)
+# main.py (Revised: Apify 201 fix + Supadata verify=False + Gunicorn/Uvicorn compatible)
 import os
 import re
 import logging
@@ -14,7 +14,7 @@ from telegram.ext import (
     filters,
     ContextTypes,
     CallbackQueryHandler,
-    ApplicationBuilder # Import ApplicationBuilder
+    # ApplicationBuilder is implicitly used by Application.builder()
 )
 from telegram.constants import ParseMode # Import ParseMode explicitly
 
@@ -27,21 +27,23 @@ _apify_token_exists = bool(os.environ.get('APIFY_API_TOKEN'))
 if _apify_token_exists:
     from apify_client import ApifyClient
 else:
-    ApifyClient = None
+    ApifyClient = None # Explicitly set to None if not imported
 
 # --- Logging Setup ---
 logging.basicConfig(
     format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
-    level=logging.DEBUG # Keep DEBUG
+    level=logging.DEBUG # Keep DEBUG for detailed info
 )
+# Reduce noise from libraries
 logging.getLogger("httpx").setLevel(logging.WARNING)
 if ApifyClient: logging.getLogger("apify_client").setLevel(logging.WARNING)
-logging.getLogger("telegram.ext").setLevel(logging.DEBUG)
-logging.getLogger('telegram.bot').setLevel(logging.DEBUG)
-logging.getLogger("urllib3").setLevel(logging.INFO) # Reduce requests noise slightly
-logging.getLogger('gunicorn.error').setLevel(logging.INFO)
+logging.getLogger("telegram.ext").setLevel(logging.INFO) # INFO is usually enough for PTB ext
+logging.getLogger('telegram.bot').setLevel(logging.INFO) # INFO is usually enough for PTB bot API calls
+logging.getLogger("urllib3").setLevel(logging.INFO)
+logging.getLogger('gunicorn.error').setLevel(logging.INFO) # Gunicorn logs
+logging.getLogger('uvicorn.error').setLevel(logging.INFO) # Uvicorn logs
 logger = logging.getLogger(__name__)
-logger.info("Logging configured (DEBUG level).")
+logger.info("Logging configured (Main logger: DEBUG).")
 
 # --- Environment Variable Loading ---
 logger.info("Attempting to load secrets from environment variables...")
@@ -65,16 +67,23 @@ logger.info("Secret loading attempt finished.")
 def is_youtube_url(url):
     """Checks if the URL is a valid YouTube video or shorts URL."""
     youtube_regex = r'(https?://)?(www\.)?(youtube\.com/(watch\?v=|shorts/)|youtu\.be/)([\w-]{11})'
-    return bool(re.search(youtube_regex, url))
+    # Added check for common query params like si= or feature=
+    match = re.search(youtube_regex, url)
+    logger.debug(f"is_youtube_url check for '{url}': {'Match found' if match else 'No match'}")
+    return bool(match)
 
 def extract_youtube_id(url):
     """Extracts the YouTube video ID from a URL."""
-    youtube_id_regex = r'(?:youtube\.com/(?:watch\?v=|shorts/)|youtu\.be/)([\w-]{11})'
+    # Updated regex to better handle query parameters after the ID
+    youtube_id_regex = r'(?:youtube\.com/(?:watch\?v=|shorts/)|youtu\.be/)([\w-]{11})(?:\?|&|\s|$)'
     match = re.search(youtube_id_regex, url)
     if match:
-        return match.group(1)
-    logger.warning(f"Could not extract YouTube ID from URL: {url}")
-    return None
+        video_id = match.group(1)
+        logger.debug(f"Extracted YouTube ID '{video_id}' from URL: {url}")
+        return video_id
+    else:
+        logger.warning(f"Could not extract YouTube ID from URL: {url}")
+        return None
 
 # Supadata Transcript Fetching
 async def get_transcript_via_supadata(video_id: str, api_key: str):
@@ -86,16 +95,16 @@ async def get_transcript_via_supadata(video_id: str, api_key: str):
     params = {"videoId": video_id, "format": "text"}
     headers = {"X-API-Key": api_key}
     try:
-        # ---!!! INSECURE TEST - DISABLES SSL VERIFICATION !!!---
-        logger.warning("[Supadata] Making request with verify=False (INSECURE TEST)")
+        # --- Keep verify=False for testing, but log clearly ---
+        logger.warning("[Supadata] Making request with verify=False (Attempting to bypass SSL verification)")
         response = await asyncio.to_thread(requests.get, api_endpoint, headers=headers, params=params, timeout=30, verify=False)
-        # ---!!! END INSECURE TEST --- (Remember to remove verify=False later) ---
+        # --- End verify=False section ---
 
         logger.debug(f"[Supadata] Received status code {response.status_code} for {video_id}")
         if response.status_code == 200:
-            # (Rest of the success handling code remains the same)
             try:
                 data = response.json()
+                # Handle both direct string response and JSON object response
                 content = data if isinstance(data, str) else data.get("transcript") or data.get("text") or data.get("data")
                 if content and isinstance(content, str):
                     logger.info(f"[Supadata] Successfully fetched transcript for {video_id}. Length: {len(content)}")
@@ -104,6 +113,7 @@ async def get_transcript_via_supadata(video_id: str, api_key: str):
                     logger.warning(f"[Supadata] API success but content empty/invalid for {video_id}. Response: {data}")
                     return None
             except json.JSONDecodeError:
+                # If JSON fails, maybe it's plain text?
                 if response.text:
                     logger.info(f"[Supadata] Successfully fetched transcript (plain text) for {video_id}. Length: {len(response.text)}")
                     return response.text.strip()
@@ -127,9 +137,9 @@ async def get_transcript_via_supadata(video_id: str, api_key: str):
         return None
     except requests.exceptions.RequestException as e:
         logger.error(f"[Supadata] Request error connecting to API for {video_id}: {e}")
-        # Log the specific SSLError if verify=False wasn't the only issue
+        # Log specific SSL Error details even with verify=False
         if isinstance(e, requests.exceptions.SSLError):
-            logger.error(f"[Supadata] SSL Error details: {e}")
+            logger.error(f"[Supadata] SSL Error occurred despite using verify=False. Details: {e}")
         return None
     except Exception as e:
         logger.error(f"[Supadata] Unexpected error during API call for {video_id}: {e}", exc_info=True)
@@ -144,12 +154,13 @@ async def get_transcript_via_apify(video_url: str, api_token: str):
 
     logger.info(f"[Apify] Attempting fetch for URL: {video_url}")
     actor_id = "karamelo~youtube-transcripts"
+    # Using the run-sync-get-dataset-items endpoint for direct results
     api_endpoint = f"https://api.apify.com/v2/acts/{actor_id}/run-sync-get-dataset-items"
     params = {"token": api_token}
     payload = json.dumps({
         "urls": [video_url],
-        "outputFormat": "singleStringText",
-        "maxRetries": 5,
+        "outputFormat": "singleStringText", # Requesting single string directly
+        "maxRetries": 3, # Reduced retries slightly
         "channelHandleBoolean": False,
         "channelNameBoolean": False,
         "datePublishedBoolean": False,
@@ -158,28 +169,36 @@ async def get_transcript_via_apify(video_url: str, api_token: str):
     headers = {"Content-Type": "application/json"}
     try:
         logger.debug(f"[Apify] Sending request to run actor {actor_id} synchronously for {video_url}")
-        response = await asyncio.to_thread(requests.post, api_endpoint, headers=headers, params=params, data=payload, timeout=90)
+        # Use asyncio.to_thread for the blocking requests call
+        response = await asyncio.to_thread(requests.post, api_endpoint, headers=headers, params=params, data=payload, timeout=90) # 90s timeout for potentially long runs
         logger.debug(f"[Apify] Received status code {response.status_code} for {video_url}")
 
-        # --- MODIFIED STATUS CODE CHECK ---
-        if response.status_code in [200, 201]: # Accept 200 OK or 201 Created
+        # --- !!! MODIFIED STATUS CODE CHECK !!! ---
+        # Accept 200 OK or 201 Created as success for this endpoint
+        if response.status_code in [200, 201]:
         # --- END MODIFIED STATUS CODE CHECK ---
             try:
                 results = response.json()
+                # The run-sync-get-dataset-items endpoint returns a list of dataset items directly
                 if isinstance(results, list) and len(results) > 0:
-                    item = results[0]
+                    item = results[0] # Get the first (and likely only) item
+                    # Look for the transcript text under common keys
                     content = item.get("text") or item.get("transcript") or item.get("captions_concatenated")
+
+                    # Additional fallback: Check if 'captions' list exists and join text from it
+                    # This handles cases where singleStringText might fail but raw captions are present
                     if not content and item.get("captions") and isinstance(item["captions"], list):
-                         logger.info("[Apify] Processing 'captions' format.")
+                         logger.info("[Apify] Processing 'captions' list format as fallback.")
                          content = " ".join(cap.get("text", "") for cap in item["captions"] if cap.get("text"))
+
                     if content and isinstance(content, str):
-                        logger.info(f"[Apify] Successfully fetched transcript for {video_url} (Status: {response.status_code}). Length: {len(content)}")
+                        logger.info(f"[Apify] Successfully fetched transcript via run-sync for {video_url} (Status: {response.status_code}). Length: {len(content)}")
                         return content.strip()
                     else:
-                        logger.warning(f"[Apify] Actor run successful ({response.status_code}) but content not found/empty for {video_url}. Item: {item}")
+                        logger.warning(f"[Apify] Actor run successful ({response.status_code}) but content not found/empty in result item for {video_url}. Item: {item}")
                         return None
                 else:
-                    logger.warning(f"[Apify] Actor run successful ({response.status_code}) but dataset empty for {video_url}. Response: {results}")
+                    logger.warning(f"[Apify] Actor run successful ({response.status_code}) but dataset result list empty for {video_url}. Response: {results}")
                     return None
             except json.JSONDecodeError:
                 logger.error(f"[Apify] Failed to decode JSON response for {video_url}. Status: {response.status_code}. Resp: {response.text[:200]}...")
@@ -187,15 +206,18 @@ async def get_transcript_via_apify(video_url: str, api_token: str):
             except Exception as e:
                 logger.error(f"[Apify] Error processing successful response ({response.status_code}) for {video_url}: {e}", exc_info=True)
                 return None
+        # Handle specific error codes
         elif response.status_code == 400:
              logger.error(f"[Apify] Bad Request (400) for {video_url}. Check payload. Response: {response.text[:200]}...")
              return None
         elif response.status_code == 401:
             logger.error("[Apify] Authentication error (401). Check API token.")
             return None
-        else: # Catch other non-200/201 codes here
-             logger.error(f"[Apify] Unexpected status code {response.status_code} for {video_url}. Response: {response.text[:200]}...")
+        # Catch other non-200/201 codes here
+        else:
+             logger.error(f"[Apify] Unexpected status code {response.status_code} from Apify API for {video_url}. Response: {response.text[:200]}...")
              return None
+
     except requests.exceptions.Timeout:
         logger.error(f"[Apify] Timeout error running actor for {video_url}")
         return None
@@ -206,12 +228,6 @@ async def get_transcript_via_apify(video_url: str, api_token: str):
         logger.error(f"[Apify] Unexpected error during Apify call for {video_url}: {e}", exc_info=True)
         return None
 
-# (The rest of the functions: get_youtube_transcript, get_website_content_via_requests,
-#  get_website_content_via_urltotext_api, generate_summary, start, help_command,
-#  handle_potential_url, handle_summary_type_callback, error_handler, setup_bot,
-#  webhook, index, and the main execution block remain EXACTLY THE SAME as in the
-#  previous complete code block. Ensure they are included below this point.)
-
 # Combined YouTube Transcript Function (with Fallbacks)
 async def get_youtube_transcript(video_id: str, video_url: str, supadata_key: str | None, apify_token: str | None):
     """Fetches YouTube transcript using library, then Supadata, then Apify."""
@@ -222,52 +238,63 @@ async def get_youtube_transcript(video_id: str, video_url: str, supadata_key: st
     # 1. Primary Method: youtube-transcript-api
     logger.info("[Primary YT] Attempting youtube-transcript-api...")
     try:
+        # Run the blocking IO call in a separate thread
         transcript_list = await asyncio.to_thread(
             YouTubeTranscriptApi.get_transcript,
             video_id,
-            languages=['en', 'en-GB', 'en-US']
+            languages=['en', 'en-GB', 'en-US'] # Prioritize English variants
         )
         if transcript_list:
             transcript_text = " ".join([item['text'] for item in transcript_list if 'text' in item])
+            # Clean up excessive whitespace that might result from joining
+            transcript_text = re.sub(r'\s+', ' ', transcript_text).strip()
             if transcript_text:
                 logger.info(f"[Primary YT] Successfully fetched transcript via library for {video_id} (length: {len(transcript_text)})")
                 return transcript_text
             else:
-                 logger.warning(f"[Primary YT] Joined transcript text is empty for {video_id}")
-                 transcript_text = None
+                 logger.warning(f"[Primary YT] Joined transcript text is empty after cleaning for {video_id}")
+                 transcript_text = None # Explicitly set to None
         else:
-            logger.warning(f"[Primary YT] Transcript list empty for {video_id}")
+            logger.warning(f"[Primary YT] Transcript list was empty for {video_id}")
             transcript_text = None
     except Exception as e:
-        logger.warning(f"[Primary YT] Error getting transcript via library for {video_id}: {e}")
-        if "YouTube is blocking requests" in str(e): logger.warning("[Primary YT] IP likely blocked by YouTube.")
-        elif "No transcript found" in str(e): logger.warning(f"[Primary YT] No transcript found for {video_id}.")
-        elif "disabled" in str(e): logger.warning(f"[Primary YT] Transcripts disabled for {video_id}.")
+        # Log the specific error type for better debugging
+        logger.warning(f"[Primary YT] Error getting transcript via library for {video_id}: {type(e).__name__} - {e}")
+        # Be more specific about common errors
+        if "YouTube is blocking requests" in str(e) or "HTTP Error 429" in str(e):
+            logger.warning("[Primary YT] IP likely blocked by YouTube (Rate Limit / Cloud IP).")
+        elif "No transcript found" in str(e):
+            logger.warning(f"[Primary YT] No transcript available in specified languages for {video_id}.")
+        elif "TranscriptsDisabled" in str(e) or "disabled" in str(e):
+            logger.warning(f"[Primary YT] Transcripts are disabled for {video_id}.")
+        # Ensure transcript_text is None if any exception occurred
         transcript_text = None
 
     # 2. Fallback 1: Supadata API
     if transcript_text is None:
-        logger.info("[Fallback YT 1] Primary method failed. Trying Supadata API...")
+        logger.info("[Fallback YT 1] Primary method failed or yielded no text. Trying Supadata API...")
         if supadata_key:
             transcript_text = await get_transcript_via_supadata(video_id, supadata_key)
             if transcript_text:
-                logger.info(f"[Fallback YT 1] Successfully fetched transcript via Supadata for {video_id}")
-                return transcript_text
+                logger.info(f"[Fallback YT 1] Successfully fetched transcript via Supadata for {video_id} (length: {len(transcript_text)})")
+                return transcript_text # Return immediately on success
             else:
                 logger.warning(f"[Fallback YT 1] Supadata API failed or returned no content for {video_id}.")
+                # transcript_text remains None
         else:
             logger.warning("[Fallback YT 1] Supadata API key not available. Skipping.")
 
     # 3. Fallback 2: Apify API
     if transcript_text is None:
-        logger.info("[Fallback YT 2] Primary & Supadata failed. Trying Apify API...")
+        logger.info("[Fallback YT 2] Primary & Supadata failed or yielded no text. Trying Apify API...")
         if apify_token:
             transcript_text = await get_transcript_via_apify(video_url, apify_token)
             if transcript_text:
-                logger.info(f"[Fallback YT 2] Successfully fetched transcript via Apify for {video_url}")
-                return transcript_text
+                logger.info(f"[Fallback YT 2] Successfully fetched transcript via Apify for {video_url} (length: {len(transcript_text)})")
+                return transcript_text # Return immediately on success
             else:
                 logger.warning(f"[Fallback YT 2] Apify API failed or returned no content for {video_url}.")
+                # transcript_text remains None
         else:
             logger.warning("[Fallback YT 2] Apify API token not available. Skipping.")
 
@@ -276,6 +303,7 @@ async def get_youtube_transcript(video_id: str, video_url: str, supadata_key: st
         logger.error(f"All methods failed to fetch transcript for video ID: {video_id}")
         return None
 
+    # This line should technically not be reached if logic is correct, but added for safety
     return transcript_text
 
 # Website Content via Requests/BS4 (Primary Method for Simplified Bot)
@@ -284,211 +312,682 @@ async def get_website_content_via_requests(url):
     if not url: logger.error("[Web Scraper - Requests/BS4] called with no URL"); return None
     logger.info(f"[Web Scraper - Requests/BS4] Fetching website content for: {url}")
     try:
+        # Standard headers to mimic a browser
         headers = {
              'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/108.0.0.0 Safari/537.36',
              'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8',
              'Accept-Language': 'en-US,en;q=0.9',
-             'Connection': 'keep-alive', 'DNT': '1', 'Upgrade-Insecure-Requests': '1' }
-        logger.debug(f"[Web Scraper - Requests/BS4] Sending request to {url}")
+             'Connection': 'keep-alive',
+             'DNT': '1', # Do Not Track
+             'Upgrade-Insecure-Requests': '1'
+         }
+        logger.debug(f"[Web Scraper - Requests/BS4] Sending GET request to {url}")
+        # Run blocking requests call in thread
         response = await asyncio.to_thread(requests.get, url, headers=headers, timeout=25, allow_redirects=True)
-        response.raise_for_status()
+        response.raise_for_status() # Raise HTTPError for bad responses (4xx or 5xx)
         logger.debug(f"[Web Scraper - Requests/BS4] Received response {response.status_code} from {url}")
+
         content_type = response.headers.get('content-type', '').lower()
         if 'html' not in content_type:
-            logger.warning(f"[Web Scraper - Requests/BS4] Non-HTML content type received from {url}: {content_type}")
+            logger.warning(f"[Web Scraper - Requests/BS4] Non-HTML content type received from {url}: {content_type}. Skipping parsing.")
+            # Maybe return raw text if it's plain text? Or just None.
+            if 'text/plain' in content_type and response.text:
+                 logger.info(f"[Web Scraper - Requests/BS4] Returning plain text content for {url}")
+                 return response.text.strip()
             return None
+
+        # Parse with BeautifulSoup
         soup = BeautifulSoup(response.text, 'html.parser')
-        for element in soup(["script", "style", "header", "footer", "nav", "aside", "form", "button", "input", "textarea", "select", "option", "label", "iframe", "img", "svg", "link", "meta", "noscript", "figure", "figcaption", "video", "audio"]): element.extract()
-        main_content = soup.find('main') or soup.find('article') or soup.find(id='content') or soup.find(class_='content') or soup.find(id='main-content') or soup.find(class_='main-content') or soup.find(role='main')
+
+        # Remove common unwanted elements
+        for element in soup(["script", "style", "header", "footer", "nav", "aside", "form", "button", "input", "textarea", "select", "option", "label", "iframe", "img", "svg", "link", "meta", "noscript", "figure", "figcaption", "video", "audio"]):
+            element.extract() # Remove the tag from the tree
+
+        # Try to find the main content area (add more selectors if needed)
+        main_content = (
+            soup.find('main') or
+            soup.find('article') or
+            soup.find(id='content') or
+            soup.find(class_='content') or
+            soup.find(id='main-content') or
+            soup.find(class_='main-content') or
+            soup.find(role='main')
+        )
+
+        # Fallback to body if no specific main area found
         target_element = main_content if main_content else soup.body
+
         if not target_element:
             logger.warning(f"[Web Scraper - Requests/BS4] Could not find body or main content container for parsing {url}")
-            return None
+            # Try getting text from the whole soup as a last resort? Could be messy.
+            # raw_text = soup.get_text(separator='\n', strip=True)
+            # if raw_text: return "\n".join(line.strip() for line in raw_text.splitlines() if line.strip())
+            return None # Return None if body itself is missing
+
+        # Extract text, clean up lines, and join
         lines = [line.strip() for line in target_element.get_text(separator='\n', strip=True).splitlines() if line.strip()]
         text = "\n".join(lines)
-        if not text or len(text) < 50:
-             logger.warning(f"[Web Scraper - Requests/BS4] Extracted text is very short or empty after cleaning for {url} (Length: {len(text)})")
+
+        # Check if extracted text is reasonably long
+        MIN_TEXT_LENGTH = 50
+        if not text or len(text) < MIN_TEXT_LENGTH:
+             logger.warning(f"[Web Scraper - Requests/BS4] Extracted text is very short (<{MIN_TEXT_LENGTH} chars) or empty after cleaning for {url} (Length: {len(text)})")
+             # Consider returning None if text is too short, might indicate failed extraction
+             # return None
+
         logger.info(f"[Web Scraper - Requests/BS4] Successfully scraped content for {url} (final length: {len(text)})")
         return text
-    except requests.exceptions.Timeout: logger.error(f"[Web Scraper - Requests/BS4] Timeout error scraping website: {url}"); return None
-    except requests.exceptions.TooManyRedirects: logger.error(f"[Web Scraper - Requests/BS4] Too many redirects error scraping website: {url}"); return None
-    except requests.exceptions.RequestException as e: logger.error(f"[Web Scraper - Requests/BS4] Request error scraping website {url}: {e}"); return None
-    except Exception as e: logger.error(f"[Web Scraper - Requests/BS4] Error scraping or parsing website {url}: {e}", exc_info=True); return None
+
+    except requests.exceptions.Timeout:
+        logger.error(f"[Web Scraper - Requests/BS4] Timeout error scraping website: {url}"); return None
+    except requests.exceptions.TooManyRedirects:
+        logger.error(f"[Web Scraper - Requests/BS4] Too many redirects error scraping website: {url}"); return None
+    except requests.exceptions.RequestException as e: # Catches HTTPError, ConnectionError etc.
+        logger.error(f"[Web Scraper - Requests/BS4] Request error scraping website {url}: {e}"); return None
+    except Exception as e: # Catch potential BS4 errors or others
+        logger.error(f"[Web Scraper - Requests/BS4] Error scraping or parsing website {url}: {e}", exc_info=True); return None
+
 
 # Website Content via URLToText API (Fallback Method)
 async def get_website_content_via_urltotext_api(url: str, api_key: str):
     """Fetches website content using the URLToText API (Fallback)."""
     if not url: logger.error("[Web Scraper - URLToText API] called with no URL"); return None
     if not api_key: logger.error("[Web Scraper - URLToText API] API key is missing."); return None
+
     logger.info(f"[Web Scraper - URLToText API] Attempting to fetch content for: {url}")
     api_endpoint = "https://urltotext.com/api/v1/urltotext/"
-    payload = json.dumps({ "url": url, "output_format": "text", "extract_main_content": True, "render_javascript": True, "residential_proxy": False })
-    headers = { "Authorization": f"Token {api_key}", "Content-Type": "application/json" }
+    # Ensure payload is correctly formatted JSON string
+    payload = json.dumps({
+        "url": url,
+        "output_format": "text",
+        "extract_main_content": True, # Use their main content extraction
+        "render_javascript": True,    # Important for dynamic sites
+        "residential_proxy": False    # Usually not needed unless specifically blocked
+    })
+    headers = {
+        "Authorization": f"Token {api_key}",
+        "Content-Type": "application/json"
+    }
+
     try:
-        logger.debug(f"[Web Scraper - URLToText API] Sending request for {url}")
-        response = await asyncio.to_thread(requests.post, api_endpoint, headers=headers, data=payload, timeout=45)
+        logger.debug(f"[Web Scraper - URLToText API] Sending POST request for {url}")
+        # Run blocking requests call in thread
+        response = await asyncio.to_thread(requests.post, api_endpoint, headers=headers, data=payload, timeout=45) # Slightly longer timeout
         logger.debug(f"[Web Scraper - URLToText API] Received status code {response.status_code} for {url}")
+
         if response.status_code == 200:
             try:
                 data = response.json()
-                content = data.get("data", {}).get("content")
+                content = data.get("data", {}).get("content") # Navigate nested structure
                 credits = data.get("credits_used", "N/A")
-                warning = data.get("data", {}).get("warning")
-                if warning: logger.warning(f"[Web Scraper - URLToText API] Warning for {url}: {warning}")
-                if content: logger.info(f"[Web Scraper - URLToText API] Successfully fetched content via API for {url}. Length: {len(content)}. Credits: {credits}"); return content.strip()
-                else: logger.warning(f"[Web Scraper - URLToText API] API returned success but content was empty for {url}. Response: {data}"); return None
-            except json.JSONDecodeError: logger.error(f"[Web Scraper - URLToText API] Failed to decode JSON response for {url}. Response: {response.text[:500]}..."); return None
-            except Exception as e: logger.error(f"[Web Scraper - URLToText API] Error processing successful API response for {url}: {e}", exc_info=True); return None
-        elif response.status_code in [400, 402, 422, 500]: logger.error(f"[Web Scraper - URLToText API] Error {response.status_code} from API for {url}. Response: {response.text[:200]}..."); return None
-        else: logger.error(f"[Web Scraper - URLToText API] Unexpected status {response.status_code} from API for {url}. Response: {response.text[:200]}..."); return None
-    except requests.exceptions.Timeout: logger.error(f"[Web Scraper - URLToText API] Timeout error connecting to API for {url}"); return None
-    except requests.exceptions.RequestException as e: logger.error(f"[Web Scraper - URLToText API] Request error connecting to API for {url}: {e}"); return None
-    except Exception as e: logger.error(f"[Web Scraper - URLToText API] Unexpected error during API call for {url}: {e}", exc_info=True); return None
+                warning = data.get("data", {}).get("warning") # Check for warnings
+
+                if warning:
+                    logger.warning(f"[Web Scraper - URLToText API] Warning received for {url}: {warning}")
+
+                if content:
+                    logger.info(f"[Web Scraper - URLToText API] Successfully fetched content via API for {url}. Length: {len(content)}. Credits: {credits}")
+                    return content.strip()
+                else:
+                    logger.warning(f"[Web Scraper - URLToText API] API returned success (200) but content was empty for {url}. Response: {data}")
+                    return None # Return None if content is missing despite 200 OK
+
+            except json.JSONDecodeError:
+                logger.error(f"[Web Scraper - URLToText API] Failed to decode JSON response for {url}. Status: {response.status_code}. Response text: {response.text[:500]}...")
+                return None
+            except Exception as e:
+                logger.error(f"[Web Scraper - URLToText API] Error processing successful API response for {url}: {e}", exc_info=True)
+                return None
+
+        # Handle specific error codes from URLToText API docs
+        elif response.status_code == 400: # Bad Request (e.g., invalid URL format)
+            logger.error(f"[Web Scraper - URLToText API] Bad Request (400) from API for {url}. Check URL/payload. Response: {response.text[:200]}...")
+        elif response.status_code == 401: # Unauthorized (Invalid API Key)
+             logger.error(f"[Web Scraper - URLToText API] Unauthorized (401) from API for {url}. Check API Key. Response: {response.text[:200]}...")
+        elif response.status_code == 402: # Payment Required (Credits exhausted)
+             logger.error(f"[Web Scraper - URLToText API] Payment Required (402) from API for {url}. Check credits. Response: {response.text[:200]}...")
+        elif response.status_code == 422: # Unprocessable Entity (e.g., URL cannot be reached)
+             logger.warning(f"[Web Scraper - URLToText API] Unprocessable URL (422) reported by API for {url}. Response: {response.text[:200]}...")
+        elif response.status_code == 500: # Internal Server Error on their end
+             logger.error(f"[Web Scraper - URLToText API] Internal Server Error (500) from API for {url}. Response: {response.text[:200]}...")
+        else: # Catch-all for other unexpected codes
+            logger.error(f"[Web Scraper - URLToText API] Unexpected status {response.status_code} from API for {url}. Response: {response.text[:200]}...")
+        return None # Return None for all error cases
+
+    except requests.exceptions.Timeout:
+        logger.error(f"[Web Scraper - URLToText API] Timeout error connecting to API for {url}")
+        return None
+    except requests.exceptions.RequestException as e:
+        logger.error(f"[Web Scraper - URLToText API] Request error connecting to API for {url}: {e}")
+        return None
+    except Exception as e:
+        logger.error(f"[Web Scraper - URLToText API] Unexpected error during API call for {url}: {e}", exc_info=True)
+        return None
 
 # DeepSeek Summary Function (via OpenRouter)
 async def generate_summary(text: str, summary_type: str, api_key: str) -> str:
     """Generates summary using DeepSeek via OpenRouter API."""
-    logger.info(f"Generating {summary_type} summary using DeepSeek/OpenRouter. Input text length: {len(text)}")
-    if not api_key: logger.error("OpenRouter API key was not provided."); return "Error: AI model config key missing."
+    logger.info(f"Generating '{summary_type}' summary using DeepSeek/OpenRouter. Input text length: {len(text)}")
+    if not api_key:
+        logger.error("OpenRouter API key was not provided.")
+        return "Error: AI model configuration key is missing."
+    if not text:
+        logger.warning("generate_summary called with empty text.")
+        return "Error: No content provided to summarize."
+
     openrouter_api_endpoint = "https://openrouter.ai/api/v1/chat/completions"
+    # Using the free DeepSeek model - be mindful of potential rate limits or changes
     model_name = "deepseek/deepseek-chat:free"
-    if summary_type == "paragraph": prompt = "..." # Keep prompt as before
-    else: prompt = """...""" # Keep prompt as before
-    MAX_INPUT_LENGTH = 500000
-    if len(text) > MAX_INPUT_LENGTH: logger.warning(f"Input text ({len(text)}) > limit ({MAX_INPUT_LENGTH}). Truncating."); text = text[:MAX_INPUT_LENGTH] + "... (Truncated)"
-    full_prompt = f"{prompt}\n\n{text}"
-    headers = { "Authorization": f"Bearer {api_key}", "Content-Type": "application/json", "HTTP-Referer": "https://huggingface.co/spaces/", "X-Title": "Telegram Summary Bot (HF Space)"}
-    payload = json.dumps({ "model": model_name, "messages": [{"role": "user", "content": full_prompt}]})
+
+    # Define prompts based on summary type
+    if summary_type == "paragraph":
+        prompt = "Please provide a concise, well-written paragraph summarizing the key information and main points of the following text. Focus on capturing the essence of the content accurately."
+    elif summary_type == "points":
+        prompt = "Please summarize the following text into clear, distinct bullet points. Each point should highlight a key piece of information, finding, or main topic discussed. Aim for clarity and conciseness."
+    else:
+        logger.error(f"Invalid summary_type '{summary_type}' passed to generate_summary.")
+        return f"Error: Invalid summary type requested ('{summary_type}')."
+
+    # Check and truncate input text if necessary
+    # Max context varies per model, DeepSeek Chat's is large, but setting a practical limit is wise.
+    # Let's use a large but reasonable limit to avoid huge API calls.
+    MAX_INPUT_LENGTH = 500000 # Approx 500k characters
+    if len(text) > MAX_INPUT_LENGTH:
+        logger.warning(f"Input text length ({len(text)}) exceeds maximum limit ({MAX_INPUT_LENGTH}). Truncating.")
+        text = text[:MAX_INPUT_LENGTH] + "... (Content truncated due to length)"
+
+    full_prompt = f"{prompt}\n\n--- Start of Text ---\n\n{text}\n\n--- End of Text ---"
+
+    headers = {
+        "Authorization": f"Bearer {api_key}",
+        "Content-Type": "application/json",
+        # Recommended headers for OpenRouter when calling from identifiable services
+        "HTTP-Referer": "https://huggingface.co/spaces/YOUR_SPACE_NAME", # TODO: Replace with your actual space name if possible
+        "X-Title": "Telegram URL Summarizer Bot (HF Space)" # Or your bot's name
+    }
+
+    payload = json.dumps({
+        "model": model_name,
+        "messages": [
+            {"role": "user", "content": full_prompt}
+        ],
+        # Optional parameters (adjust if needed):
+        # "temperature": 0.7, # Controls randomness (optional)
+        # "max_tokens": 1024, # Limit response length (optional)
+    })
+
     try:
-        logger.debug(f"Sending request to OpenRouter ({model_name})...")
-        response = await asyncio.to_thread(requests.post, openrouter_api_endpoint, headers=headers, data=payload, timeout=60)
+        logger.debug(f"Sending request to OpenRouter (Model: {model_name})...")
+        # Run blocking requests call in thread
+        response = await asyncio.to_thread(requests.post,
+            openrouter_api_endpoint,
+            headers=headers,
+            data=payload,
+            timeout=90 # Increased timeout for potentially long generation
+        )
         logger.debug(f"Received status code {response.status_code} from OpenRouter.")
+
         if response.status_code == 200:
             try:
                 data = response.json()
-                if data.get("choices") and data["choices"]:
+                # Check the response structure carefully
+                if data.get("choices") and isinstance(data["choices"], list) and len(data["choices"]) > 0:
                     message = data["choices"][0].get("message")
                     if message and message.get("content"):
                         summary = message["content"].strip()
-                        logger.info(f"Success generating summary. Len: {len(summary)}")
-                        return summary
-                    else: logger.warning(f"OpenRouter success but empty content. Resp: {data}"); return "Sorry, AI model returned empty summary."
+                        # Basic check for empty or placeholder summary
+                        if summary:
+                           logger.info(f"Successfully generated summary via OpenRouter. Length: {len(summary)}")
+                           return summary
+                        else:
+                           logger.warning(f"OpenRouter returned success (200) but summary content is empty. Response: {data}")
+                           return "Sorry, the AI model returned an empty summary."
+                    else:
+                        logger.warning(f"OpenRouter success (200) but response structure missing expected content. Response: {data}")
+                        return "Sorry, could not parse the AI model's response (missing content)."
+                # Handle cases where 'choices' might be missing or empty, or if there's an error object
+                elif data.get("error"):
+                    error_details = data["error"]
+                    logger.error(f"OpenRouter API Error ({response.status_code}): {error_details}")
+                    return f"Sorry, the AI service reported an error: {error_details.get('message', 'Unknown error')}"
                 else:
-                    if data.get("error"): logger.error(f"OpenRouter API Error: {data['error']}")
-                    else: logger.error(f"Unexpected choices structure: {data.get('choices')}. Resp: {data}")
-                    return "Sorry, could not parse AI response (choices/error)."
-            except json.JSONDecodeError: logger.error(f"Failed JSON decode from OpenRouter. Status: {response.status_code}. Resp: {response.text[:500]}..."); return "Sorry, failed to understand AI response."
-            except Exception as e: logger.error(f"Error processing OpenRouter success resp: {e}", exc_info=True); return "Sorry, error processing AI response."
-        elif response.status_code == 401: logger.error("OpenRouter API key invalid (401). Check HF Secrets."); return "Error: AI model config key invalid."
-        elif response.status_code == 402: logger.error("OpenRouter Payment Required (402). Check OpenRouter account."); return "Sorry, issue with AI service limits/payment."
-        elif response.status_code == 429: logger.warning("OpenRouter Rate Limit (429)."); return "Sorry, AI model busy. Try again."
-        elif response.status_code == 500: logger.error(f"OpenRouter Internal Error (500). Resp: {response.text[:500]}..."); return "Sorry, AI model service error. Try again later."
-        else:
-             logger.error(f"Unexpected status {response.status_code} from OpenRouter. Resp: {response.text[:500]}...")
-             try: error_data = response.json(); error_msg = error_data.get("error", {}).get("message", response.text[:100]); return f"Sorry, AI service error ({response.status_code}): {error_msg}"
-             except: return f"Sorry, AI service returned status {response.status_code}."
-    except requests.exceptions.Timeout: logger.error("Timeout connecting to OpenRouter."); return "Sorry, request to AI model timed out."
-    except requests.exceptions.RequestException as e: logger.error(f"Request error connecting to OpenRouter: {e}"); return "Sorry, error connecting to AI model service."
-    except Exception as e: logger.error(f"Unexpected error in generate_summary: {e}", exc_info=True); return "Sorry, unexpected error generating summary."
+                    logger.error(f"OpenRouter success (200) but unexpected response structure (no choices/error). Response: {data}")
+                    return "Sorry, could not parse the AI model's response (unexpected structure)."
+
+            except json.JSONDecodeError:
+                logger.error(f"Failed to decode JSON response from OpenRouter. Status: {response.status_code}. Response: {response.text[:500]}...")
+                return "Sorry, failed to understand the AI model's response format."
+            except Exception as e:
+                logger.error(f"Error processing successful OpenRouter response: {e}", exc_info=True)
+                return "Sorry, an error occurred while processing the AI model's response."
+
+        # Handle specific HTTP error codes
+        elif response.status_code == 401: # Unauthorized
+            logger.error("OpenRouter API key is invalid or unauthorized (401). Check the key in HF Secrets.")
+            return "Error: AI model configuration key is invalid."
+        elif response.status_code == 402: # Payment Required / Quota Exceeded
+            logger.error("OpenRouter Payment Required / Quota Exceeded (402). Check OpenRouter account status and limits.")
+            return "Sorry, there's an issue with the AI service account (limits or payment)."
+        elif response.status_code == 429: # Rate Limit Exceeded
+            logger.warning("OpenRouter Rate Limit Exceeded (429). Need to wait or slow down requests.")
+            return "Sorry, the AI model service is busy right now. Please try again in a moment."
+        elif response.status_code >= 500: # Server Error on OpenRouter's side
+            logger.error(f"OpenRouter Internal Server Error ({response.status_code}). Response: {response.text[:500]}...")
+            return "Sorry, the AI model service encountered an internal error. Please try again later."
+        else: # Catch-all for other client-side or unexpected errors
+             logger.error(f"Unexpected status code {response.status_code} received from OpenRouter. Response: {response.text[:500]}...")
+             try:
+                 # Try to parse error message from response if possible
+                 error_data = response.json()
+                 error_msg = error_data.get("error", {}).get("message", response.text[:100])
+                 return f"Sorry, the AI service returned an error ({response.status_code}): {error_msg}"
+             except:
+                 # Fallback if response is not JSON or doesn't have expected error structure
+                 return f"Sorry, the AI service returned an unexpected status code ({response.status_code})."
+
+    except requests.exceptions.Timeout:
+        logger.error("Timeout occurred while connecting to OpenRouter.")
+        return "Sorry, the request to the AI model timed out. Please try again."
+    except requests.exceptions.RequestException as e:
+        logger.error(f"Network error connecting to OpenRouter: {e}")
+        return "Sorry, there was a network error connecting to the AI model service."
+    except Exception as e:
+        logger.error(f"Unexpected error occurred in generate_summary function: {e}", exc_info=True)
+        return "Sorry, an unexpected internal error occurred while generating the summary."
+
 
 # --- Telegram Bot Handlers ---
+
 async def start(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
-    user = update.effective_user; logger.info(f"User {user.id} ({user.username or 'NoUsername'}) used /start.")
+    """Handles the /start command."""
+    user = update.effective_user
+    if not user: return # Should not happen but good practice
+    logger.info(f"User {user.id} ({user.username or 'NoUsername'}) triggered /start.")
+    # Use mention_html for a clickable link if username exists, otherwise just first name
     mention = user.mention_html() if user.username else user.first_name
-    await update.message.reply_html(f"👋 Hello {mention}! I can summarize YouTube links or website URLs.\n\nJust send me a link anytime!")
+    await update.message.reply_html(
+        f"👋 Hello {mention}! I can summarize YouTube links or website URLs.\n\n"
+        "Just send me a valid link (starting with http:// or https://) and I'll ask you how you want it summarized."
+    )
+
 async def help_command(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
-    logger.info(f"User {update.effective_user.id} used /help.")
-    await update.message.reply_text("...", parse_mode=ParseMode.MARKDOWN) # Keep help text
+    """Handles the /help command."""
+    user = update.effective_user
+    logger.info(f"User {user.id if user else 'Unknown'} triggered /help.")
+    help_text = (
+        "**How I Work:**\n\n"
+        "1. Send me a full URL (starting with `http://` or `https://`).\n"
+        "2. I'll detect if it's a YouTube video link or a general website URL.\n"
+        "3. I'll ask if you want a **Paragraph** summary or **Points** summary.\n"
+        "4. Choose your preferred format by clicking the button.\n"
+        "5. I'll fetch the content (transcript for YouTube, text for websites) and use an AI model (via OpenRouter) to generate the summary.\n\n"
+        "**Troubleshooting:**\n"
+        "- **YouTube:** Sometimes transcripts aren't available (private video, no captions, disabled). I use multiple methods (library, Supadata, Apify) to try and get them.\n"
+        "- **Websites:** Complex websites with lots of JavaScript might be difficult to scrape accurately. I use a primary scraping method and a fallback API (URLToText) if needed.\n"
+        "- **AI Errors:** Occasionally, the AI model might be busy or encounter an error. You can try again later.\n\n"
+        "Just send a link to get started!"
+    )
+    await update.message.reply_text(help_text, parse_mode=ParseMode.MARKDOWN)
+
 async def handle_potential_url(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
-    if not update.message or not update.message.text: return
-    url = update.message.text.strip(); user = update.effective_user
-    logger.info(f"User {user.id} ({user.username or 'NoUsername'}) sent potential URL: {url}")
-    if not (url.startswith('http://') or url.startswith('https://')) or '.' not in url[8:]: logger.debug(f"Ignoring non-URL: {url}"); return
-    context.user_data['url_to_summarize'] = url; logger.debug(f"Stored URL '{url}' for user {user.id}")
-    keyboard = [[InlineKeyboardButton("Paragraph", callback_data="paragraph"), InlineKeyboardButton("Points", callback_data="points")]]
-    reply_markup = InlineKeyboardMarkup(keyboard)
-    await update.message.reply_text(f"Link detected:\n{url}\n\nChoose summary type:", reply_markup=reply_markup, disable_web_page_preview=True)
+    """Handles messages containing potential URLs."""
+    if not update.message or not update.message.text:
+        return # Ignore empty messages
+
+    message_text = update.message.text.strip()
+    user = update.effective_user
+    if not user: return
+
+    # Simple check for potential URLs - refine if needed
+    # Looks for http:// or https:// followed by some characters and a dot.
+    url_pattern = r'https?://[^\s/$.?#].[^\s]*'
+    match = re.search(url_pattern, message_text)
+
+    if match:
+        url = match.group(0) # Extract the first matched URL
+        logger.info(f"User {user.id} ({user.username or 'NoUsername'}) sent potential URL: {url}")
+
+        # Store the URL in user_data, associated with the user ID
+        context.user_data['url_to_summarize'] = url
+        logger.debug(f"Stored URL '{url}' in user_data for user {user.id}")
+
+        # Define the inline keyboard buttons
+        keyboard = [
+            [
+                InlineKeyboardButton("Paragraph Summary", callback_data="paragraph"),
+                InlineKeyboardButton("Points Summary", callback_data="points")
+            ]
+        ]
+        reply_markup = InlineKeyboardMarkup(keyboard)
+
+        # Reply to the user, asking for the summary type
+        await update.message.reply_text(
+            f"Okay, I see this link:\n{url}\n\nHow would you like it summarized?",
+            reply_markup=reply_markup,
+            # disable_web_page_preview=True # Good practice to avoid double previews
+            link_preview_options={'is_disabled': True} # Updated way for PTB v20+
+        )
+    else:
+        # If the message doesn't look like a URL, optionally reply or just ignore
+        logger.debug(f"Ignoring non-URL message from user {user.id}: {message_text[:100]}")
+        # Optional: Reply if you want to guide the user
+        # await update.message.reply_text("Please send me a valid URL starting with http:// or https://")
+
+
 async def handle_summary_type_callback(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
-    query = update.callback_query; await query.answer()
-    summary_type = query.data; user = update.effective_user or query.from_user
+    """Handles button presses for choosing the summary type."""
+    query = update.callback_query
+    if not query: return
+    user = query.from_user # Get user from the callback query
+
+    # --- 1. Answer the callback query ---
+    # It's crucial to answer quickly to remove the "loading" state on the button.
+    try:
+        await query.answer()
+        logger.debug(f"Answered callback query {query.id}")
+    except Exception as e:
+        # This can happen if the query is too old, but log it.
+        logger.error(f"Failed to answer callback query {query.id}: {e}")
+        # Don't stop processing, but be aware the button might stay loading for the user.
+
+    # --- 2. Get data and context ---
+    summary_type = query.data # e.g., "paragraph" or "points"
+    # Retrieve the URL stored earlier for this user
     url = context.user_data.get('url_to_summarize')
-    logger.info(f"User {user.id} chose '{summary_type}' for URL '{url}'.")
-    if not url: logger.warning(f"User {user.id} pressed button, NO URL in context."); await query.edit_message_text(text="Context lost. Send link again."); return
-    context.user_data.pop('url_to_summarize', None); logger.debug(f"Cleared URL {url} for user {user.id}")
-    current_openrouter_key = os.environ.get('OPENROUTER_API_KEY'); current_urltotext_key = os.environ.get('URLTOTEXT_API_KEY')
-    current_supadata_key = os.environ.get('SUPADATA_API_KEY'); current_apify_token = os.environ.get('APIFY_API_TOKEN')
-    if not current_openrouter_key: logger.error("OpenRouter key missing."); await context.bot.send_message(chat_id=user.id, text="Error: AI config missing."); await query.delete_message(); return
-    processing_message = f"Working on '{summary_type}' summary for:\n{url}\n..."; message_to_delete_later = None
-    try: await query.edit_message_text(processing_message); logger.debug(f"Edited message query {query.id}")
-    except Exception as e: logger.warning(f"Could not edit message {query.id}: {e}. Sending new."); message_to_delete_later = await context.bot.send_message(chat_id=user.id, text=processing_message)
-    content = None; user_feedback_message = None; success = False; is_youtube = is_youtube_url(url)
+
+    logger.info(f"User {user.id} ({user.username or 'NoUsername'}) chose '{summary_type}' summary. Checking for URL '{url}' in context.")
+
+    if not url:
+        logger.warning(f"User {user.id} pressed button, but NO URL was found in user_data context. Query data: {summary_type}")
+        # Edit the original message to inform the user
+        try:
+            await query.edit_message_text(text="Sorry, I seem to have lost the context. 🤔 Please send the link again.")
+        except Exception as edit_err:
+            logger.error(f"Failed to edit message after lost context error: {edit_err}")
+        return # Stop processing
+
+    # Clear the URL from context once we start processing it
+    context.user_data.pop('url_to_summarize', None)
+    logger.debug(f"Retrieved and cleared URL {url} from user_data for user {user.id}")
+
+    # --- 3. Check for required API keys ---
+    # Re-fetch keys inside handler in case they were updated (less critical here, but good practice)
+    current_openrouter_key = os.environ.get('OPENROUTER_API_KEY')
+    current_urltotext_key = os.environ.get('URLTOTEXT_API_KEY')
+    current_supadata_key = os.environ.get('SUPADATA_API_KEY')
+    current_apify_token = os.environ.get('APIFY_API_TOKEN')
+
+    logger.debug(f"Reading API keys from environment variables within handler...")
+    logger.debug(f"Keys read: OpenRouter={'Yes' if current_openrouter_key else 'No'}, "
+                 f"URLToText={'Yes' if current_urltotext_key else 'No'}, "
+                 f"Supadata={'Yes' if current_supadata_key else 'No'}, "
+                 f"Apify={'Yes' if current_apify_token else 'No'}")
+
+
+    if not current_openrouter_key:
+        logger.error("OpenRouter API key is missing in environment variables. Cannot generate summary.")
+        try:
+            await query.edit_message_text(text="⚠️ Error: The AI summarization service is not configured correctly (missing API key). Please contact the bot admin.")
+        except Exception as edit_err:
+            logger.error(f"Failed to edit message about missing OpenRouter key: {edit_err}")
+        return
+
+    # --- 4. Edit message to show "Processing..." ---
+    processing_message_text = f"Got it! Generating '{summary_type}' summary for:\n{url}\n\nThis might take a moment..."
+    message_to_delete_later_id = None
     try:
-        logger.debug(f"Sending 'typing' action for chat {user.id}"); await context.bot.send_chat_action(chat_id=user.id, action='typing')
-        if is_youtube:
+        # Edit the original message where the buttons were
+        await query.edit_message_text(text=processing_message_text)
+        logger.debug(f"Edited message for query {query.id} to show processing status.")
+    except Exception as e:
+        # If editing fails (e.g., message too old, deleted), send a new message
+        logger.warning(f"Could not edit original message {query.message.message_id if query.message else 'N/A'} (query {query.id}): {e}. Sending a new status message.")
+        try:
+            status_message = await context.bot.send_message(chat_id=user.id, text=processing_message_text)
+            message_to_delete_later_id = status_message.message_id
+            logger.debug(f"Sent new status message {message_to_delete_later_id}")
+        except Exception as send_err:
+            logger.error(f"Failed even to send a new status message: {send_err}")
+            # Can't easily inform the user now, just log.
+
+    # --- 5. Fetch Content (YouTube or Website) ---
+    content = None
+    user_feedback_message = None # Store potential error messages for the user
+    success = False # Track overall success
+
+    try:
+        # Send "typing..." action to indicate activity
+        logger.debug(f"Sending 'typing' action for chat {user.id}")
+        await context.bot.send_chat_action(chat_id=user.id, action='typing')
+
+        is_yt = is_youtube_url(url)
+        logger.debug(f"URL determined to be YouTube: {is_yt}")
+
+        if is_yt:
             video_id = extract_youtube_id(url)
             if video_id:
-                logger.info(f"Fetching YT transcript: {video_id}"); content = await get_youtube_transcript(video_id, url, current_supadata_key, current_apify_token)
-                user_feedback_message = None if content else "Sorry, couldn't get YT transcript."
-                logger.info(f"YT transcript fetch done. Found: {bool(content)}")
-            else: user_feedback_message = "Sorry, couldn't parse YT URL."
+                logger.info(f"Fetching YouTube transcript for video_id: {video_id}")
+                content = await get_youtube_transcript(video_id, url, current_supadata_key, current_apify_token)
+                if not content:
+                    # Provide a more informative error if all YT methods fail
+                    user_feedback_message = "Sorry, I couldn't get the transcript for that YouTube video using any available method (unavailable/private/no captions?)."
+                logger.info(f"YouTube transcript fetch completed. Content found: {bool(content)}")
+            else:
+                logger.warning(f"Failed to extract video ID from supposedly YouTube URL: {url}")
+                user_feedback_message = "Sorry, I couldn't properly identify the YouTube video ID from the link."
         else:
-            logger.info(f"Scraping website (Requests/BS4): {url}"); content = await get_website_content_via_requests(url)
-            if content: logger.info("Website scrape (Requests/BS4) OK."); user_feedback_message = None
+            # Try fetching website content
+            logger.info(f"Attempting website scrape (Requests/BS4) for: {url}")
+            content = await get_website_content_via_requests(url)
+
+            if content:
+                logger.info("Website scrape (Requests/BS4) successful.")
+                user_feedback_message = None # Clear any previous potential message
             else:
-                logger.warning(f"Website scrape failed for {url}. Trying URLToText API.");
+                logger.warning(f"Primary website scrape (Requests/BS4) failed for {url}. Trying fallback API (URLToText)...")
                 if current_urltotext_key:
-                    await context.bot.send_chat_action(chat_id=user.id, action='typing'); content = await get_website_content_via_urltotext_api(url, current_urltotext_key)
-                    if content: logger.info("URLToText API scrape OK."); user_feedback_message = None
-                    else: user_feedback_message = "Sorry, couldn't fetch web content (both methods)."
-                else: user_feedback_message = "Sorry, couldn't fetch web content (fallback not configured)."
+                    # Send typing action again before potentially long API call
+                    await context.bot.send_chat_action(chat_id=user.id, action='typing')
+                    content = await get_website_content_via_urltotext_api(url, current_urltotext_key)
+                    if content:
+                        logger.info("Website scrape fallback (URLToText API) successful.")
+                        user_feedback_message = None
+                    else:
+                        logger.warning(f"Fallback website scrape (URLToText API) also failed for {url}.")
+                        user_feedback_message = "Sorry, I couldn't fetch the content from that website using either the standard method or the fallback API."
+                else:
+                    logger.warning("URLToText API key is not configured. Cannot use fallback.")
+                    user_feedback_message = "Sorry, I couldn't fetch the content from that website (primary method failed, and fallback is not configured)."
+
+        # --- 6. Generate Summary if Content Exists ---
         if content:
-            logger.info("Content found, generating summary."); await context.bot.send_chat_action(chat_id=user.id, action='typing')
+            logger.info("Content retrieved successfully. Proceeding to generate summary.")
+            # Send typing action again before AI call
+            await context.bot.send_chat_action(chat_id=user.id, action='typing')
+
             summary = await generate_summary(content, summary_type, current_openrouter_key)
-            if summary.startswith("Error:") or summary.startswith("Sorry,"): user_feedback_message = summary; logger.warning(f"Summary generation failed: {summary}")
-            else: logger.info("Summary generated OK. Sending."); await context.bot.send_message(chat_id=user.id, text=summary, parse_mode=ParseMode.MARKDOWN, disable_web_page_preview=True); success = True; user_feedback_message = None
-        elif not user_feedback_message: user_feedback_message = "Sorry, couldn't retrieve content."
-        if user_feedback_message and not success: logger.warning(f"Sending failure feedback: {user_feedback_message}"); await context.bot.send_message(chat_id=user.id, text=user_feedback_message)
-    except Exception as e: logger.error(f"Unexpected error in callback processing: {e}", exc_info=True); await context.bot.send_message(chat_id=user.id, text="Oops! Internal error.")
+
+            # Check if the summary generation itself returned an error message
+            if summary.startswith("Error:") or summary.startswith("Sorry,"):
+                user_feedback_message = summary # Use the error from generate_summary
+                logger.warning(f"Summary generation failed or returned an error message: {summary}")
+            else:
+                # Success! Send the summary.
+                logger.info("Summary generated successfully. Sending to user.")
+                await context.bot.send_message(
+                    chat_id=user.id,
+                    text=summary,
+                    parse_mode=ParseMode.MARKDOWN, # Assume summary might contain markdown
+                    # disable_web_page_preview=True
+                    link_preview_options={'is_disabled': True}
+                )
+                success = True # Mark as successful
+                user_feedback_message = None # Clear any potential previous error
+        elif not user_feedback_message:
+            # If content is None and no specific error message was set, provide a generic one
+            logger.error(f"Content fetching resulted in None, but no specific user feedback message was set for URL: {url}")
+            user_feedback_message = "Sorry, I was unable to retrieve any content from the provided link."
+
+        # --- 7. Send Final Feedback (if error occurred) ---
+        if user_feedback_message and not success:
+            logger.warning(f"Sending failure feedback message to user {user.id}: {user_feedback_message}")
+            await context.bot.send_message(
+                chat_id=user.id,
+                text=user_feedback_message
+            )
+
+    except Exception as e:
+        # Catch any unexpected errors during the whole process
+        logger.error(f"An unexpected error occurred in handle_summary_type_callback for user {user.id}, URL {url}: {e}", exc_info=True)
+        try:
+            # Try to inform the user about the internal error
+             await context.bot.send_message(chat_id=user.id, text=" माफी माग्छु ! An unexpected internal error occurred while processing your request. The developers have been notified.") # माफी माग्छु = Oops!
+        except Exception as final_err:
+             logger.error(f"Failed to even send the final error message to user {user.id}: {final_err}")
+
     finally:
-        logger.debug("Cleaning up status message...");
+        # --- 8. Clean up the "Processing..." message ---
+        # This runs whether success or failure, unless an error prevented it
+        logger.debug("Callback handler finished. Cleaning up status message (if possible)...")
         try:
-             if message_to_delete_later: await context.bot.delete_message(chat_id=user.id, message_id=message_to_delete_later.message_id); logger.debug("Deleted separate status msg.")
-             elif query: await query.delete_message(); logger.debug(f"Deleted original message query {query.id}.")
-        except Exception as del_e: logger.warning(f"Could not delete status/button message: {del_e}")
-async def error_handler(update: object, context: ContextTypes.DEFAULT_TYPE) -> None: logger.error(f"Exception while handling update: {context.error}", exc_info=context.error)
+            if message_to_delete_later_id:
+                # If we sent a separate status message, delete it
+                await context.bot.delete_message(chat_id=user.id, message_id=message_to_delete_later_id)
+                logger.debug(f"Deleted separate status message {message_to_delete_later_id}.")
+            elif query.message:
+                 # If we successfully edited the original message, delete it now that we're done
+                 # (or you could edit it to "Summary complete!" or the error message)
+                 # Deleting is cleaner usually.
+                 await query.delete_message()
+                 logger.debug(f"Deleted original message (via query {query.id}, message_id {query.message.message_id}).")
+        except Exception as del_err:
+            # Log if deletion fails, but don't crash
+            logger.warning(f"Could not delete status/button message: {del_err}")
+
+
+async def error_handler(update: object, context: ContextTypes.DEFAULT_TYPE) -> None:
+    """Log Errors caused by Updates."""
+    logger.error(f"Exception while handling an update: {context.error}", exc_info=context.error)
+
+    # Optional: Inform user about generic errors (be cautious with this)
+    # try:
+    #     if isinstance(update, Update) and update.effective_chat:
+    #         await context.bot.send_message(
+    #             chat_id=update.effective_chat.id,
+    #             text="Sorry, something went wrong processing your request."
+    #         )
+    # except Exception as e:
+    #     logger.error(f"Failed to send error message to chat: {e}")
+
 
 # --- Bot Application Setup Function ---
+# Make setup_bot an async function as Application.initialize() is async
 async def setup_bot():
-    logger.info("Setting up Telegram Application...");
-    if not TELEGRAM_TOKEN: logger.critical("Cannot initialize: TELEGRAM_TOKEN missing."); return None
+    """Initializes and configures the Telegram Bot Application."""
+    logger.info("Setting up Telegram Application...")
+    if not TELEGRAM_TOKEN:
+        logger.critical("CRITICAL: TELEGRAM_TOKEN environment variable not found. Bot cannot start.")
+        return None
+
+    # Use Application.builder() for setup
     application = Application.builder().token(TELEGRAM_TOKEN).build()
-    logger.info("Running application.initialize()..."); await application.initialize(); logger.info("Finished application.initialize().")
-    application.add_handler(CommandHandler("start", start)); application.add_handler(CommandHandler("help", help_command))
-    application.add_handler(MessageHandler(filters.TEXT & ~filters.COMMAND, handle_potential_url)); application.add_handler(CallbackQueryHandler(handle_summary_type_callback))
-    application.add_error_handler(error_handler); logger.info("Telegram handlers registered.")
+
+    # --- Register Handlers ---
+    # Command Handlers
+    application.add_handler(CommandHandler("start", start))
+    application.add_handler(CommandHandler("help", help_command))
+
+    # Message Handler for URLs (ensure it doesn't catch commands)
+    # Using filters.TEXT & ~filters.COMMAND & filters.Entity("url") | filters.Entity("text_link") might be more precise
+    # But a simple text check + regex inside the handler is often robust enough.
+    application.add_handler(MessageHandler(filters.TEXT & ~filters.COMMAND, handle_potential_url))
+
+    # Callback Query Handler for button presses
+    application.add_handler(CallbackQueryHandler(handle_summary_type_callback))
+
+    # Error Handler (registers the function defined above)
+    application.add_error_handler(error_handler)
+
+    logger.info("Telegram handlers registered.")
+
+    # Initialize the application (fetches bot info, etc.)
+    # This is now done automatically when application runs, but explicit call ensures it happens early
+    try:
+        logger.info("Running application.initialize()...")
+        await application.initialize()
+        logger.info("Finished application.initialize(). Bot details: %s", application.bot.username)
+    except Exception as e:
+        logger.critical(f"Failed to initialize Telegram application: {e}", exc_info=True)
+        return None # Indicate failure
+
     return application
-logger.info("Running bot setup..."); ptb_app = asyncio.run(setup_bot()); logger.info(f"Bot setup finished. App instance: {'OK' if ptb_app else 'Failed'}")
 
-# --- Flask App Setup ---
-app = Flask(__name__); logger.info("Flask app created.")
+# --- Global Application Instance ---
+# Run setup_bot once at startup
+logger.info("Running bot setup at startup...")
+# Use asyncio.get_event_loop().run_until_complete() if running outside an existing async context
+# In newer Python versions, asyncio.run() is preferred but might cause issues if nested.
+# Since this is top-level before Flask/Gunicorn starts its loop, it should be okay.
+ptb_app: Application | None = asyncio.run(setup_bot())
+logger.info(f"Bot setup finished. Application instance: {'OK' if ptb_app else 'FAILED'}")
+
+
+# --- Flask App Setup (for Webhook) ---
+app = Flask(__name__)
+logger.info("Flask app created.")
+
+@app.route('/')
+def index():
+    """Basic health check endpoint."""
+    logger.debug("Health check '/' accessed.")
+    bot_status = "Initialized" if ptb_app and ptb_app.bot else "Initialization FAILED"
+    return f"Telegram Bot Webhook Listener ({bot_status}) is running."
 
-# --- Webhook Endpoint ---
 @app.route('/webhook', methods=['POST'])
 async def webhook() -> Response:
-    logger.info("Webhook request received...")
-    if not ptb_app: logger.error("PTB App not initialized."); return Response('Bot not configured', status=500)
+    """Webhook endpoint for Telegram updates."""
+    if not ptb_app:
+        logger.error("Webhook triggered, but Telegram Application (ptb_app) is not initialized.")
+        return Response('Bot not configured correctly.', status=500)
+
+    logger.debug("Webhook request received (POST)...")
     if request.is_json:
         try:
-            update = Update.de_json(request.get_json(), ptb_app.bot); logger.debug(f"Processing update ID: {update.update_id}")
-            logger.debug("Directly awaiting process_update..."); await ptb_app.process_update(update); logger.debug("Finished awaiting process_update.")
-            return Response('ok', status=200)
-        except json.JSONDecodeError: logger.error("Failed JSON decode from Telegram."); return Response('Bad Request', status=400)
-        except Exception as e: logger.error(f"Error processing update in webhook: {e}", exc_info=True); return Response('Internal Server Error', status=500)
-    else: logger.warning("Received non-JSON request to webhook."); return Response('Bad Request', status=400)
-@app.route('/')
-def index(): logger.debug("Health check '/' accessed."); bot_status = "Initialized" if ptb_app else "FAILED Init"; return f"TG Bot Webhook Listener ({bot_status}) running."
+            update_data = request.get_json()
+            update = Update.de_json(update_data, ptb_app.bot)
+            logger.debug(f"Processing update ID: {update.update_id} via webhook")
+
+            # Crucial: Queue the update for processing by PTB's internal mechanisms
+            # This ensures proper handling within PTB's async context
+            # Use await ptb_app.process_update(update) directly as Flask is run by Uvicorn
+            await ptb_app.process_update(update)
+            logger.debug(f"Finished processing update ID: {update.update_id}")
+
+            return Response('ok', status=200) # Acknowledge receipt to Telegram
+
+        except json.JSONDecodeError:
+            logger.error("Failed to decode JSON from Telegram webhook request.")
+            return Response('Bad Request: Invalid JSON', status=400)
+        except Exception as e:
+            logger.error(f"Error processing update in webhook: {e}", exc_info=True)
+            # Avoid sending detailed errors back in the response
+            return Response('Internal Server Error processing update.', status=500)
+    else:
+        logger.warning("Received non-JSON request to webhook endpoint.")
+        return Response('Bad Request: Expected JSON', status=400)
 
-# --- Main Execution Block ---
+# --- Main Execution Block (for local testing ONLY) ---
 if __name__ == '__main__':
-    if not ptb_app: logger.critical("Aborting local Flask start: PTB App failed init.")
-    else: logger.info("Starting Flask server directly (local testing?)..."); port = int(os.environ.get('PORT', 5000)); app.run(host='0.0.0.0', port=port, debug=True)
+    # This block is NOT used when running with Gunicorn/Uvicorn in production
+    logger.warning("Running Flask development server directly (for local testing only).")
+    if not ptb_app:
+        logger.critical("Aborting local Flask start: Telegram App (ptb_app) failed initialization.")
+    else:
+        # For local testing, you might need to set a webhook manually or use polling.
+        # Flask's dev server can work with asyncio now.
+        # Running with debug=True enables auto-reloading and more detailed errors.
+        local_port = int(os.environ.get('PORT', 8080)) # Use a different port locally?
+        logger.info(f"Flask dev server starting on http://0.0.0.0:{local_port}")
+        # Note: Flask's built-in server is NOT recommended for production.
+        # Use Gunicorn + Uvicorn worker as configured in gunicorn.conf.py for deployment.
+        app.run(host='0.0.0.0', port=local_port, debug=True)