Merge pull request #23 from aar0np/main

reworking recommended videos...again?

Author: aar0np

app/api/v1/endpoints/video_catalog.py

@@ -366,9 +366,6 @@ async def get_related_videos_for_video(
     ] = 5,
 ):
     """Return a list of videos related to the given video.
-
-    The underlying implementation is currently stubbed out and will return the
-    latest videos (excluding the source video) with a random relevance score.
     """
 
     related_items = await recommendation_service.get_related_videos(

app/services/recommendation_service.py

@@ -16,13 +16,6 @@
 async def get_related_videos(
     video_id: VideoID, limit: int = 10
 ) -> List[RecommendationItem]:
-    """Return a stubbed *related videos* list.
-
-    In a future iteration this will call into a real recommendation engine that
-    analyses the content of the referenced video to find similar items. For the
-    moment we simply return the latest videos (excluding the reference video)
-    and assign each a random relevance score.
-    """
 
     from opentelemetry import trace
     import time
@@ -42,18 +35,28 @@ async def get_related_videos(
         if target_video is None:
             return []
 
-        latest_summaries, _total = await video_service.list_latest_videos(
-            page=1, page_size=limit + 5
+        latest_summaries, _total = await video_service.get_recommended_videos(
+            query_vector=target_video.content_features, page=1, page_size=limit + 5
         )
 
         related_items: List[RecommendationItem] = []
+        unique_video_names: List[str] = []
 
         for summary in latest_summaries:
             if summary.videoId == video_id:
                 # Skip the source video itself
                 continue
+
+            if summary.title in unique_video_names:
+                # Skip if we've already added this video to the list
+                # ...sometimes we get duplicate rows
+                continue
+            
             if len(related_items) >= limit:
                 break
+
+            unique_video_names.append(summary.title)
+
             related_items.append(
                 RecommendationItem(
                     videoId=summary.videoId,

app/services/video_service.py

@@ -7,6 +7,7 @@
 
 from __future__ import annotations
 
+from bdb import effective
 import re
 from datetime import datetime, timezone, timedelta
 import asyncio
@@ -529,13 +530,33 @@ async def list_latest_videos(
         source_table_name=VIDEOS_TABLE_NAME,
     )
 
+async def get_recommended_videos(
+    query_vector: Optional[List[float]],
+    page: int,
+    page_size: int,
+    db_table: Optional[AstraDBCollection] = None,
+) -> Tuple[List[VideoSummary], int]:
+    
+    if query_vector is None:
+        return [], 0
+
+    return await list_videos_with_query(
+        {}, 
+        page, 
+        page_size=page_size,
+        sort_options={"content_features": query_vector},
+        db_table=db_table, 
+        source_table_name=VIDEOS_TABLE_NAME
+    )
+
 
 async def list_videos_by_tag(
     tag: str,
     page: int,
     page_size: int,
     db_table: Optional[AstraDBCollection] = None,
 ) -> Tuple[List[VideoSummary], int]:
+
     query_filter = {
         "tags": {"$in": [tag]},
     }