fix(rss): stop posting after Bluesky createRecord rate limit and add clear post start/finish logs

rss2bsky.py

@@ -41,8 +41,12 @@ BSKY_BLOB_TRANSIENT_ERROR_DELAY = 15
 HTTP_TIMEOUT = 20
 POST_RETRY_DELAY_SECONDS = 2
 
+# Thumbnail upload cooldown state
 THUMB_UPLOAD_COOLDOWN_UNTIL = 0
 
+# Post creation cooldown state
+POST_CREATION_COOLDOWN_UNTIL = 0
+
 # --- Logging ---
 logging.basicConfig(
     format="%(asctime)s %(message)s",
@@ -83,7 +87,8 @@ def strip_trailing_url_punctuation(url):
 def canonicalize_url(url):
     if not url:
         return None
-    return strip_trailing_url_punctuation(url.strip())
+    url = html.unescape(url.strip())
+    return strip_trailing_url_punctuation(url)
 
 
 def clean_whitespace(text):
@@ -137,12 +142,9 @@ def build_post_text_variants(title_text, link):
     1. Full title + blank line + real URL
     Fallbacks:
     2. Full title only
-    3. Truncated title only
 
     Intentionally avoid:
     - truncated title + URL
-
-    That pattern creates ugly posts with "..." before the visible link.
     """
     title_text = clean_whitespace(title_text)
     link = canonicalize_url(link) or link or ""
@@ -161,7 +163,6 @@ def build_post_text_variants(title_text, link):
 
     if title_text:
         add_variant(title_text)
-        add_variant(truncate_text_safely(title_text))
 
     if link and not title_text:
         add_variant(link)
@@ -454,7 +455,7 @@ def make_rich(content):
     return text_builder
 
 
-# --- Blob / image upload helpers ---
+# --- Rate limit helpers ---
 def get_rate_limit_wait_seconds(error_obj, default_delay):
     try:
         headers = getattr(error_obj, "headers", None)
@@ -483,6 +484,11 @@ def get_rate_limit_reset_timestamp(error_obj):
     return None
 
 
+def is_rate_limited_error(error_obj):
+    error_text = str(error_obj)
+    return "429" in error_text or "RateLimitExceeded" in error_text
+
+
 def is_transient_blob_error(error_obj):
     error_text = repr(error_obj)
     transient_signals = [
@@ -499,11 +505,30 @@ def is_transient_blob_error(error_obj):
     return any(signal in error_text for signal in transient_signals)
 
 
-def is_rate_limited_error(error_obj):
+# --- Post cooldown helpers ---
-    error_text = str(error_obj)
+def activate_post_creation_cooldown_from_error(error_obj):
-    return "429" in error_text or "RateLimitExceeded" in error_text
+    global POST_CREATION_COOLDOWN_UNTIL
+
+    reset_ts = get_rate_limit_reset_timestamp(error_obj)
+    if reset_ts:
+        if reset_ts > POST_CREATION_COOLDOWN_UNTIL:
+            POST_CREATION_COOLDOWN_UNTIL = reset_ts
+            logging.error(
+                f"=== BSKY POST STOPPED: RATE LIMITED === Posting disabled until "
+                f"{time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(reset_ts))}"
+            )
+    else:
+        fallback_reset = int(time.time()) + 3600
+        if fallback_reset > POST_CREATION_COOLDOWN_UNTIL:
+            POST_CREATION_COOLDOWN_UNTIL = fallback_reset
+            logging.error("=== BSKY POST STOPPED: RATE LIMITED === Posting disabled for 1 hour.")
 
 
+def is_post_creation_cooldown_active():
+    return int(time.time()) < POST_CREATION_COOLDOWN_UNTIL
+
+
+# --- Thumbnail cooldown helpers ---
 def activate_thumb_upload_cooldown_from_error(error_obj):
     global THUMB_UPLOAD_COOLDOWN_UNTIL
 
@@ -526,6 +551,7 @@ def is_thumb_upload_cooldown_active():
     return int(time.time()) < THUMB_UPLOAD_COOLDOWN_UNTIL
 
 
+# --- Blob / image upload helpers ---
 def upload_blob_with_retry(client, binary_data, media_label="media", optional=False, cooldown_on_rate_limit=False):
     last_exception = None
     transient_attempts = 0
@@ -787,6 +813,12 @@ def is_probable_length_error(exc):
 
 
 def try_send_post_with_variants(client, text_variants, embed, post_lang):
+    global POST_CREATION_COOLDOWN_UNTIL
+
+    if is_post_creation_cooldown_active():
+        reset_str = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(POST_CREATION_COOLDOWN_UNTIL))
+        raise RuntimeError(f"Posting skipped because post creation cooldown is active until {reset_str}")
+
     last_exception = None
 
     for idx, variant in enumerate(text_variants, start=1):
@@ -800,6 +832,10 @@ def try_send_post_with_variants(client, text_variants, embed, post_lang):
             last_exception = e
             logging.warning(f"Post variant {idx} failed: {repr(e)}")
 
+            if is_rate_limited_error(e):
+                activate_post_creation_cooldown_from_error(e)
+                raise
+
             if not is_probable_length_error(e):
                 raise
 
@@ -894,11 +930,17 @@ def main():
 
     with httpx.Client() as http_client:
         for candidate in entries_to_post:
+            if is_post_creation_cooldown_active():
+                reset_str = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(POST_CREATION_COOLDOWN_UNTIL))
+                logging.error(f"=== BSKY POST STOPPED: RATE LIMITED === Skipping remaining entries until {reset_str}")
+                break
+
             title_text = candidate["title_text"]
             canonical_link = candidate["canonical_link"]
             text_variants = candidate["post_text_variants"]
 
             logging.info(f"Preparing to post RSS entry: {canonical_link or title_text}")
+            logging.info(f"=== BSKY POST START === {canonical_link or title_text}")
 
             embed = None
             if canonical_link:
@@ -933,11 +975,17 @@ def main():
                 recent_bsky_posts = recent_bsky_posts[:DEDUPE_BSKY_LIMIT]
 
                 noves_entrades += 1
+                logging.info(f"=== BSKY POST SUCCESS === {canonical_link or title_text}")
                 logging.info(f"Posted RSS entry to Bluesky: {canonical_link or title_text}")
                 time.sleep(POST_RETRY_DELAY_SECONDS)
 
-            except Exception:
+            except Exception as e:
-                logging.exception(f"Failed to post RSS entry {canonical_link or title_text}")
+                logging.exception(f"=== BSKY POST FAILED === {canonical_link or title_text}")
+
+                if is_rate_limited_error(e) or is_post_creation_cooldown_active():
+                    reset_str = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(POST_CREATION_COOLDOWN_UNTIL))
+                    logging.error(f"=== BSKY POST STOPPED: RATE LIMITED === Ending publish loop until {reset_str}")
+                    break
 
     if noves_entrades > 0:
         logging.info(f"🎉 Execution finished: published {noves_entrades} new entries to Bluesky.")
@@ -946,4 +994,4 @@ def main():
 
 
 if __name__ == "__main__":
-    main()
+    main()