Added new yml

twitter2bsky_daemon.py

@@ -181,7 +181,6 @@ def build_video_embed(video_blob, alt_text):
 
 # --- Playwright Scraping ---
 def scrape_tweets_via_playwright(username, password, email, target_handle):
-    """Logs in (or loads session) and scrapes tweets directly from the DOM."""
     tweets = []
     state_file = "twitter_browser_state.json"
 
@@ -333,19 +332,18 @@ def scrape_tweets_via_playwright(username, password, email, target_handle):
 
 def extract_video_url_from_tweet_page(context, tweet_url):
     """
-    Opens a tweet page and captures the best real video URL.
+    Prefer HLS playlist first because it usually contains the full playable stream
+    with audio + video and proper timing metadata.
 
-    Preference order:
-    1. real video .mp4
-    2. .m3u8 playlist
+    Fallback to direct video MP4 only if no HLS playlist is found.
 
-    Ignores:
+    Ignore:
     - .m4s fragments
-    - audio-only mp4 URLs
+    - audio-only MP4 URLs
     """
     page = context.new_page()
-    best_video_mp4_url = None
     best_m3u8_url = None
+    best_video_mp4_url = None
 
     def is_audio_only_mp4(url, content_type):
         url_l = url.lower()
@@ -358,7 +356,7 @@ def extract_video_url_from_tweet_page(context, tweet_url):
         )
 
     def handle_response(response):
-        nonlocal best_video_mp4_url, best_m3u8_url
+        nonlocal best_m3u8_url, best_video_mp4_url
         try:
             url = response.url
             url_l = url.lower()
@@ -368,16 +366,6 @@ def extract_video_url_from_tweet_page(context, tweet_url):
             if ".m4s" in url_l:
                 return
 
-            if (".mp4" in url_l or "video/mp4" in content_type_l or "audio/mp4" in content_type_l) and is_audio_only_mp4(url, content_type):
-                logging.info(f"🔇 Ignoring audio-only MP4: {url}")
-                return
-
-            if ".mp4" in url_l or "video/mp4" in content_type_l:
-                if best_video_mp4_url is None:
-                    best_video_mp4_url = url
-                    logging.info(f"🎥 Found VIDEO MP4 URL: {url}")
-                return
-
             if (
                 ".m3u8" in url_l or
                 "application/vnd.apple.mpegurl" in content_type_l or
@@ -388,6 +376,16 @@ def extract_video_url_from_tweet_page(context, tweet_url):
                     logging.info(f"📺 Found HLS playlist URL: {url}")
                 return
 
+            if (".mp4" in url_l or "video/mp4" in content_type_l or "audio/mp4" in content_type_l) and is_audio_only_mp4(url, content_type):
+                logging.info(f"🔇 Ignoring audio-only MP4: {url}")
+                return
+
+            if ".mp4" in url_l or "video/mp4" in content_type_l:
+                if best_video_mp4_url is None:
+                    best_video_mp4_url = url
+                    logging.info(f"🎥 Found VIDEO MP4 URL: {url}")
+                return
+
         except Exception:
             pass
 
@@ -406,7 +404,10 @@ def extract_video_url_from_tweet_page(context, tweet_url):
             except Exception:
                 pass
 
-        return best_video_mp4_url or best_m3u8_url
+        selected_url = best_m3u8_url or best_video_mp4_url
+        if selected_url:
+            logging.info(f"✅ Selected media URL for download: {selected_url}")
+        return selected_url
 
     except Exception as e:
         logging.warning(f"⚠️ Could not extract video URL from tweet page {tweet_url}: {e}")
@@ -419,7 +420,9 @@ def extract_video_url_from_tweet_page(context, tweet_url):
 def download_and_crop_video(video_url, output_path):
     """
     Downloads a video from MP4 or HLS (.m3u8), then trims it to max 59 seconds.
-    Requires ffmpeg installed on the system.
+    Uses ffmpeg for download and MoviePy for crop.
+
+    HLS is preferred because it usually produces a complete muxed file.
     """
     temp_input = output_path.replace(".mp4", "_source.mp4")
     temp_output = output_path.replace(".mp4", "_cropped.mp4")
@@ -430,10 +433,13 @@ def download_and_crop_video(video_url, output_path):
         download_cmd = [
             "ffmpeg",
             "-y",
+            "-protocol_whitelist", "file,http,https,tcp,tls,crypto",
+            "-allowed_extensions", "ALL",
             "-i", video_url,
             "-c", "copy",
             temp_input,
         ]
+
         download_result = subprocess.run(
             download_cmd,
             capture_output=True,
@@ -480,10 +486,6 @@ def download_and_crop_video(video_url, output_path):
             return None
 
         os.replace(temp_output, output_path)
-
-        if os.path.exists(temp_input):
-            os.remove(temp_input)
-
         logging.info(f"✅ Video cropped to 59 seconds: {output_path}")
         return output_path
 
@@ -543,8 +545,8 @@ def sync_feeds(args):
             for tweet in reversed(tweets):
                 tweet_time = arrow.get(tweet.created_on)
 
-                if tweet_time <= last_bsky_time:
-                #if False:
+                #if tweet_time <= last_bsky_time:
+                if False:
                     continue
 
                 logging.info(f"📝 Found new tweet from {tweet_time}. Posting to Bluesky...")
@@ -636,7 +638,6 @@ def main():
     load_dotenv()
 
     parser = argparse.ArgumentParser(description="Twitter to Bluesky Sync")
-
     parser.add_argument("--twitter-username", help="Your Twitter login username")
     parser.add_argument("--twitter-password", help="Your Twitter login password")
     parser.add_argument("--twitter-email", help="Your Twitter email for security challenges")