all events after 18 are now pub events

crawler.py

@@ -76,40 +76,63 @@ def send_to_discord_webhook(data, webhook_url):
         except requests.exceptions.HTTPError as e:
             print(f"Discord webhook error: {e}\nPayload: {payload}")
             return None
-    # Send data in batches of up to 40 events per message
-    batch_size = 40
-    total_batches = (len(data) + batch_size - 1) // batch_size
-    for i in range(0, len(data), batch_size):
-        batch = data[i:i+batch_size]
-        content = f"Nationer Öppna Idag (batch {i//batch_size+1}/{total_batches}):\n"
-        for item in batch:
+    # Group events
+    pub_keywords = ['pub', "wermlandskällar’n", 'orvars krog']
+    def is_pub_event(e):
+        # Keyword match
+        if any(k in e['event'].lower() for k in pub_keywords):
+            return True
+        # Time match: look for start time after 18:00
+        import re
+        time_match = re.search(r'(\d{2}):(\d{2})', e.get('open_time', ''))
+        if time_match:
+            hour = int(time_match.group(1))
+            if hour >= 18:
+                return True
+        return False
+    pub_events = [e for e in data if is_pub_event(e)]
+    other_events = [e for e in data if not is_pub_event(e)]
+    # Format as plain grouped messages
+    def format_batch(events, title):
+        content = f"**{title}**\n"
+        for item in events:
             content += f"**{item['nation']}**\nEvent: {item['event']}\nTid: {item['open_time']}\n---\n"
         # Discord message limit is 2000 characters
         if len(content) > 1900:
             content = content[:1900] + "... (truncated)"
-        payload = {"content": content}
+        return content
+    # Do not send non-pub events to Discord
+    # Send pub events second, as normal message
+    if pub_events:
+        payload = {"content": format_batch(pub_events, "Pub-aktiviteter")}
         try:
             response = requests.post(webhook_url, json=payload)
             response.raise_for_status()
-            print(f"Sent batch {i//batch_size+1} to Discord.")
+            print("Sent pub batch to Discord.")
         except requests.exceptions.HTTPError as e:
             print(f"Discord webhook error: {e}\nPayload: {payload}")
 
 
 def main():
     nationer_data = fetch_nationer_open_times(NATIONER_URL)
-    if nationer_data:
-        send_to_discord_webhook(nationer_data, DISCORD_WEBHOOK_URL)
-    else:
-        send_to_discord_webhook([
-            {"nation": "Inga aktiviteter hittades.", "event": "", "open_time": ""}
-        ], DISCORD_WEBHOOK_URL)
+    send_to_discord_webhook(nationer_data, DISCORD_WEBHOOK_URL)
 
 
 if __name__ == "__main__":
     import time
-    while True:
-        main()
-        print("Sleeping for 24 hours...")
-        time.sleep(86400)
+    import datetime
+
+    # Run once immediately
+    main()
+    print("First run complete. Scheduling next runs at 00:05 daily.")
+    while True:
+        now = datetime.datetime.now()
+        # Calculate next 00:05
+        next_run = now.replace(hour=0, minute=5, second=0, microsecond=0)
+        if now >= next_run:
+            next_run += datetime.timedelta(days=1)
+        sleep_seconds = (next_run - now).total_seconds()
+        print(f"Sleeping until next run at {next_run.strftime('%Y-%m-%d %H:%M:%S')} ({int(sleep_seconds)} seconds)")
+        time.sleep(sleep_seconds)
+        main()