feat: integrate automatic VOD ID resolution in chat_mod_worker.py

local_worker/chat_mod_worker.py

@@ -350,9 +350,123 @@ def mod_action_sender():
         time.sleep(1)
 
 # =========================================================================
-# VOD CHAT BACKFILLING LOOP (NO AUDIO)
+# VOD ID AUTO-RESOLUTION & CHAT BACKFILLING LOOP (NO AUDIO)
 # =========================================================================
 
+def get_recent_twitch_vods(channel_name):
+    """Fetch recent Twitch VODs of a user using public Twitch GQL API"""
+    url = "https://gql.twitch.tv/gql"
+    headers = {
+        "Client-Id": "kimne78kx3ncx6brgo4mv6wki5h1ko",
+        "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64)"
+    }
+    payload = {
+        "query": """
+        query($login: String!) {
+            user(login: $login) {
+                videos(first: 10, type: ARCHIVE) {
+                    edges {
+                        node {
+                            id
+                            title
+                            createdAt
+                            lengthSeconds
+                        }
+                    }
+                }
+            }
+        }
+        """,
+        "variables": {
+            "login": channel_name
+        }
+    }
+    try:
+        res = requests.post(url, json=payload, headers=headers, timeout=10)
+        if res.status_code == 200:
+            data = res.json()
+            edges = data.get("data", {}).get("user", {}).get("videos", {}).get("edges", [])
+            return [edge.get("node") for edge in edges if edge and edge.get("node")]
+    except Exception as e:
+        print(f"[VOD Resolve] Error fetching Twitch VODs: {e}")
+    return []
+
+def find_matching_vod(stream_start_time_iso, recent_vods):
+    try:
+        from datetime import datetime
+        clean_stream = stream_start_time_iso.replace('Z', '+00:00')
+        stream_dt = datetime.fromisoformat(clean_stream)
+    except Exception as e:
+        print(f"[Match] Error parsing stream start time: {e}")
+        return None
+
+    best_match = None
+    min_diff = float('inf')
+
+    for node in recent_vods:
+        vod_id = node.get("id")
+        created_at_raw = node.get("createdAt")
+        if not vod_id or not created_at_raw:
+            continue
+        try:
+            clean_vod = created_at_raw.replace('Z', '+00:00')
+            vod_dt = datetime.fromisoformat(clean_vod)
+        except Exception:
+            continue
+        
+        diff = abs((stream_dt - vod_dt).total_seconds())
+        # If the start times are within 2.5 hours (9000 seconds)
+        if diff < 9000 and diff < min_diff:
+            min_diff = diff
+            best_match = vod_id
+
+    return best_match
+
+def resolve_missing_vods():
+    """Find and resolve twitch_vod_id for pending streams that are missing it"""
+    headers = {"x-api-key": API_KEY}
+    try:
+        # 1. Fetch missing VOD streams from backend
+        res = requests.get(f"{API_URL}/api/streams/missing-vod", headers=headers, timeout=10)
+        if res.status_code != 200:
+            return
+        
+        streams = res.json().get("streams", [])
+        if not streams:
+            return
+            
+        print(f"[VOD Resolve] Found {len(streams)} pending stream(s) lacking VOD ID.")
+        
+        # 2. Get recent VODs from Twitch
+        recent_vods = get_recent_twitch_vods(TWITCH_CHANNEL)
+        if not recent_vods:
+            print("[VOD Resolve] Could not retrieve recent Twitch VODs. Skipping resolve.")
+            return
+            
+        # 3. Match each stream to a Twitch VOD
+        for stream in streams:
+            stream_id = stream.get("id")
+            start_time_str = stream.get("start_time")
+            if not stream_id or not start_time_str:
+                continue
+                
+            matched_vod_id = find_matching_vod(start_time_str, recent_vods)
+            if matched_vod_id:
+                print(f"[VOD Resolve] Stream ID {stream_id} ({start_time_str}) matched with Twitch VOD {matched_vod_id}.")
+                # Send update to server
+                up_res = requests.post(f"{API_URL}/api/streams/{stream_id}/resolve-vod", 
+                                       json={"twitchVodId": matched_vod_id}, 
+                                       headers={"x-api-key": API_KEY, "Content-Type": "application/json"},
+                                       timeout=10)
+                if up_res.status_code == 200:
+                    print(f"[VOD Resolve] Successfully updated VOD ID for stream {stream_id}!")
+                else:
+                    print(f"[VOD Resolve] Failed to update VOD ID: {up_res.status_code}")
+            else:
+                print(f"[VOD Resolve] No matching VOD found for stream ID {stream_id} ({start_time_str}) within timeframe.")
+    except Exception as e:
+        print(f"[VOD Resolve] Exception during resolve loop: {e}")
+
 def load_processed_streams():
     """Load successfully processed stream IDs from a local file"""
     if not os.path.exists(PROCESSED_FILE):
@@ -376,6 +490,8 @@ def run_backfill_loop():
     """Main loop to check and backfill VOD chat comments ONLY"""
     
     while not stop_flag.is_set():
+        # Try to resolve any missing VOD IDs before checking pending backfills
+        resolve_missing_vods()
         print(f"[Backfill] Checking for pending VOD backfill tasks...")
         processed_streams = load_processed_streams()