Added snippets only in urls

twitter2bsky_daemon.py

@@ -13,6 +13,7 @@ from dotenv import load_dotenv
 from atproto import Client, client_utils, models
 from playwright.sync_api import sync_playwright
 from moviepy import VideoFileClip
+from bs4 import BeautifulSoup
 
 # --- Configuration ---
 LOG_PATH = "twitter2bsky.log"
@@ -28,8 +29,7 @@ BSKY_TEXT_MAX_LENGTH = 275
 #   server-side proxy/PDS body-size caps.
 # - Custom PDSes such as eurosky.social may accept images fine but fail on
 #   larger video blob uploads.
-# - The public video limits discussed in Bluesky tooling references are useful,
-#   but in practice the safest approach is to:
+# - The safest approach is to:
 #     1. cap duration
 #     2. compress aggressively
 #     3. log final file size
@@ -41,6 +41,7 @@ BSKY_BLOB_UPLOAD_MAX_RETRIES = 5
 BSKY_BLOB_UPLOAD_BASE_DELAY = 10
 BSKY_BLOB_UPLOAD_MAX_DELAY = 300
 MEDIA_DOWNLOAD_TIMEOUT = 30
+LINK_METADATA_TIMEOUT = 10
 DEFAULT_BSKY_BASE_URL = "https://bsky.social"
 
 # --- Logging Setup ---
@@ -302,6 +303,62 @@ def get_blob_from_file(file_path, client):
         return None
 
 
+def fetch_link_metadata(url, http_client):
+    """
+    Fetch metadata used to build a Bluesky external link card.
+    """
+    try:
+        r = http_client.get(url, timeout=LINK_METADATA_TIMEOUT, follow_redirects=True)
+        r.raise_for_status()
+        soup = BeautifulSoup(r.text, "html.parser")
+
+        title = (soup.find("meta", property="og:title") or soup.find("title"))
+        desc = (
+            soup.find("meta", property="og:description")
+            or soup.find("meta", attrs={"name": "description"})
+        )
+        image = (
+            soup.find("meta", property="og:image")
+            or soup.find("meta", attrs={"name": "twitter:image"})
+        )
+
+        return {
+            "title": title["content"] if title and title.has_attr("content") else (title.text.strip() if title and title.text else ""),
+            "description": desc["content"] if desc and desc.has_attr("content") else "",
+            "image": image["content"] if image and image.has_attr("content") else None,
+        }
+
+    except Exception as e:
+        logging.warning(f"Could not fetch link metadata for {url}: {repr(e)}")
+        return {}
+
+
+def build_external_link_embed(url, client, http_client, fallback_title="Link"):
+    """
+    Build a Bluesky external embed from a URL.
+
+    This should only be used when the post has no image/video embed, because
+    Bluesky posts can only carry one embed type.
+    """
+    link_metadata = fetch_link_metadata(url, http_client)
+
+    thumb_blob = None
+    if link_metadata.get("image"):
+        thumb_blob = get_blob_from_url(link_metadata["image"], client, http_client)
+
+    if link_metadata.get("title") or link_metadata.get("description") or thumb_blob:
+        return models.AppBskyEmbedExternal.Main(
+            external=models.AppBskyEmbedExternal.External(
+                uri=url,
+                title=link_metadata.get("title") or fallback_title,
+                description=link_metadata.get("description") or "",
+                thumb=thumb_blob,
+            )
+        )
+
+    return None
+
+
 def prepare_post_text(text):
     """
     Prepare the final public text exactly as it should be posted to Bluesky.
@@ -699,7 +756,7 @@ def build_dynamic_alt(raw_text):
     if len(dynamic_alt) > 150:
         dynamic_alt = dynamic_alt[:147] + "..."
     elif not dynamic_alt:
-        dynamic_alt = "Vídeo o imatge adjunta al tuit"
+        dynamic_alt = "Attached video or image from tweet"
 
     return dynamic_alt
 
@@ -986,18 +1043,6 @@ def extract_video_url_from_tweet_page(context, tweet_url):
 def download_and_crop_video(video_url, output_path):
     """
     Download, trim, and compress video before upload.
-
-    Practical comments based on Bluesky video limits and real-world custom PDS behavior:
-    - Duration alone is not enough; final file size matters a lot.
-    - A 90-second 1080x1920 video can still be too large for alternate services.
-    - We therefore:
-        1. download the source
-        2. trim to VIDEO_MAX_DURATION_SECONDS
-        3. re-encode with tighter settings
-        4. scale down to max width 720
-        5. log final file size
-    - This improves compatibility with services like eurosky.social that may have
-      stricter body-size or timeout limits than bsky.social.
     """
     temp_input = output_path.replace(".mp4", "_source.mp4")
     temp_trimmed = output_path.replace(".mp4", "_trimmed.mp4")
@@ -1279,6 +1324,7 @@ def sync_feeds(args):
 
                 image_embeds = []
                 video_embed = None
+                external_embed = None
                 media_upload_failures = []
 
                 if tweet.media:
@@ -1330,6 +1376,20 @@ def sync_feeds(args):
                                 if os.path.exists(temp_video_path):
                                     os.remove(temp_video_path)
 
+                # Only create an external link card if no image/video embed will be used.
+                if not video_embed and not image_embeds and candidate["canonical_non_x_urls"]:
+                    first_non_x_url = sorted(candidate["canonical_non_x_urls"])[0]
+                    external_embed = build_external_link_embed(
+                        first_non_x_url,
+                        bsky_client,
+                        media_http_client,
+                        fallback_title="Link"
+                    )
+                    if external_embed:
+                        logging.info(f"🔗 Built external link card for URL: {first_non_x_url}")
+                    else:
+                        logging.info(f"ℹ️ No external link card metadata available for URL: {first_non_x_url}")
+
                 try:
                     post_result = None
                     post_mode = "text"
@@ -1341,6 +1401,9 @@ def sync_feeds(args):
                         embed = models.AppBskyEmbedImages.Main(images=image_embeds)
                         post_result = bsky_client.send_post(text=rich_text, embed=embed, langs=["ca"])
                         post_mode = f"images:{len(image_embeds)}"
+                    elif external_embed:
+                        post_result = bsky_client.send_post(text=rich_text, embed=external_embed, langs=["ca"])
+                        post_mode = "external_link_card"
                     else:
                         post_result = bsky_client.send_post(text=rich_text, langs=["ca"])
                         post_mode = "text_only"