Fix vss-ui, search-ms, dataprep for bugs (open-edge-platform#2057)

Co-authored-by: Yogesh Pandey <yogesh.pandey@intel.com>

Author: bhardwaj-nakul

microservices/visual-data-preparation-for-retrieval/vdms/src/core/embedding/sdk_client.py

@@ -415,13 +415,29 @@ def _store_embeddings(
                         batch_size=batch_size,
                     )
                 except Exception as exc:
-                    logger.error(
-                        "VDMS add_from failed for batch %d-%d: %s",
+                    logger.warning(
+                        "VDMS add_from failed for batch %d-%d. Reinitializing VDMS client and retrying once. Error: %s",
                         start_idx,
                         end_idx - 1,
                         exc,
                     )
-                    raise
+                    self._init_vdms()
+                    try:
+                        inserted_ids = self.video_db.add_from(
+                            texts=batch_texts,
+                            embeddings=batch_embeddings,
+                            metadatas=batch_metadatas,
+                            ids=batch_ids,
+                            batch_size=batch_size,
+                        )
+                    except Exception as retry_exc:
+                        logger.error(
+                            "VDMS add_from retry failed for batch %d-%d: %s",
+                            start_idx,
+                            end_idx - 1,
+                            retry_exc,
+                        )
+                        raise
 
                 if not inserted_ids or len(inserted_ids) != len(batch_ids):
                     raise ValueError(

microservices/visual-data-preparation-for-retrieval/vdms/src/core/embedding/simple_client.py

@@ -230,13 +230,29 @@ def _store_embeddings(
                     batch_size=batch_size,
                 )
             except Exception as exc:
-                logger.error(
-                    "VDMS add_from failed for batch %d-%d: %s",
+                logger.warning(
+                    "VDMS add_from failed for batch %d-%d. Reinitializing VDMS client and retrying once. Error: %s",
                     start_idx,
                     end_idx - 1,
                     exc,
                 )
-                raise
+                self.init_db()
+                try:
+                    inserted_ids = self.video_db.add_from(
+                        texts=batch_texts,
+                        embeddings=batch_embeddings,
+                        metadatas=batch_metadatas,
+                        ids=batch_ids,
+                        batch_size=batch_size,
+                    )
+                except Exception as retry_exc:
+                    logger.error(
+                        "VDMS add_from retry failed for batch %d-%d: %s",
+                        start_idx,
+                        end_idx - 1,
+                        retry_exc,
+                    )
+                    raise
 
             if not inserted_ids or len(inserted_ids) != len(batch_ids):
                 raise ValueError(

sample-applications/video-search-and-summarization/search-ms/server.py

@@ -199,16 +199,34 @@ async def process_query(query_request):
             logger.debug(f"Searching with initial_k={initial_k}")
 
             vdms_start = time.perf_counter()
-            docs_with_score: List[Tuple[Any, float]] = db.similarity_search_with_score(
-                query_request.query,
-                k=initial_k,
-                fetch_k=initial_k + 1,  # ensure fetch_k > k for langchain_vdms
-                filter=vdms_filter,
-                # normalize_distance=True
-            )
+            vdms_attempt = 1
+            try:
+                docs_with_score: List[Tuple[Any, float]] = db.similarity_search_with_score(
+                    query_request.query,
+                    k=initial_k,
+                    fetch_k=initial_k + 1,  # ensure fetch_k > k for langchain_vdms
+                    filter=vdms_filter,
+                    # normalize_distance=True
+                )
+            except Exception as first_error:
+                first_attempt_duration_ms = (time.perf_counter() - vdms_start) * 1000
+                logger.warning(
+                    "VDMS similarity search failed on first attempt after %.2f ms; refreshing client and retrying once. Error: %s",
+                    first_attempt_duration_ms,
+                    first_error,
+                )
+                refreshed_db: VDMS = get_vectordb()
+                vdms_attempt = 2
+                vdms_start = time.perf_counter()
+                docs_with_score = refreshed_db.similarity_search_with_score(
+                    query_request.query,
+                    k=initial_k,
+                    fetch_k=initial_k + 1,
+                    filter=vdms_filter,
+                )
             vdms_duration_ms = (time.perf_counter() - vdms_start) * 1000
             logger.info(
-                f"VDMS similarity search (embedding + retrieval) completed in {vdms_duration_ms:.2f} ms with {len(docs_with_score)} results"
+                f"VDMS similarity search (attempt {vdms_attempt}, embedding + retrieval) completed in {vdms_duration_ms:.2f} ms with {len(docs_with_score)} results"
             )
 
             logger.info(f"Raw search returned {len(docs_with_score)} results")

sample-applications/video-search-and-summarization/search-ms/src/vdms_retriever/retriever.py

@@ -10,7 +10,6 @@
 from src.vdms_retriever.embedding_wrapper import EmbeddingAPI
 
 DEBUG = False
-client = VDMS_Client(settings.VDMS_VDB_HOST, settings.VDMS_VDB_PORT)
 
 
 # Frame-to-Video Aggregation Configuration
@@ -639,20 +638,25 @@ def get_vectordb() -> VDMS:
         tuple: The vector database instance
     """
 
-    embeddings = EmbeddingAPI(
-        api_url=settings.EMBEDDINGS_ENDPOINT,
-        model_name=settings.EMBEDDINGS_MODEL_NAME,
-    )
+    try:
+        embeddings = EmbeddingAPI(
+            api_url=settings.EMBEDDINGS_ENDPOINT,
+            model_name=settings.EMBEDDINGS_MODEL_NAME,
+        )
 
-    vector_dimensions = embeddings.get_embedding_length()
+        vector_dimensions = embeddings.get_embedding_length()
+        client = VDMS_Client(settings.VDMS_VDB_HOST, settings.VDMS_VDB_PORT)
 
-    vector_db = VDMS(
-        client=client,
-        embedding=embeddings,
-        collection_name=settings.INDEX_NAME,
-        distance_strategy=settings.DISTANCE_STRATEGY,
-        embedding_dimensions=vector_dimensions,
-        engine=settings.SEARCH_ENGINE,
-    )
+        vector_db = VDMS(
+            client=client,
+            embedding=embeddings,
+            collection_name=settings.INDEX_NAME,
+            distance_strategy=settings.DISTANCE_STRATEGY,
+            embedding_dimensions=vector_dimensions,
+            engine=settings.SEARCH_ENGINE,
+        )
 
-    return vector_db
+        return vector_db
+    except Exception as exc:
+        logger.error(f"Failed to initialize VDMS vector DB client: {exc}")
+        raise

sample-applications/video-search-and-summarization/ui/react/src/components/VideoGroups/VideoGroupsView.tsx

@@ -5,8 +5,11 @@ import { useTranslation } from 'react-i18next';
 import styled from 'styled-components';
 import { useAppSelector } from '../../redux/store';
 import { Video } from '../../redux/video/video';
+import { SearchResult } from '../../redux/search/search';
 import { videosSelector } from '../../redux/video/videoSlice';
 import { SearchSelector } from '../../redux/search/searchSlice';
+import { ASSETS_ENDPOINT } from '../../config';
+import { resolveVideoUrl } from '../../redux/video/videoUrl';
 
 const VideoGroupsContainer = styled.div`
   padding: 1rem;
@@ -147,6 +150,16 @@ interface TagGroup {
   color: string;
 }
 
+const toPlayableMetadataUrl = (value: unknown): string => {
+  if (typeof value !== 'string') return '';
+  const trimmed = value.trim();
+  if (!trimmed) return '';
+  if (trimmed.includes('/videos/download?video_id=')) return '';
+  if (/^https?:\/\//i.test(trimmed)) return trimmed;
+  if (trimmed.startsWith('/')) return `${ASSETS_ENDPOINT}${trimmed}`;
+  return `${ASSETS_ENDPOINT}/${trimmed}`;
+};
+
 export const VideoGroupsView: FC = () => {
   const { t } = useTranslation();
   const { getVideoUrl } = useAppSelector(videosSelector);
@@ -155,7 +168,7 @@ export const VideoGroupsView: FC = () => {
   // Build a map of relevance scores from search results
   const videoRelevanceMap = useMemo(() => {
     const map = new Map<string, number>();
-    selectedResults?.forEach((result) => {
+    selectedResults?.forEach((result: SearchResult) => {
       const vid = result.metadata?.video_id;
       const score = result.metadata?.relevance_score;
       if (vid && typeof score === 'number') map.set(vid, score);
@@ -167,7 +180,7 @@ export const VideoGroupsView: FC = () => {
   const searchVideos: Video[] = useMemo(() => {
     if (!selectedResults || selectedResults.length === 0) return [];
     return selectedResults
-      .map((result) => {
+      .map((result: SearchResult) => {
         const meta: any = result.metadata ?? {};
         const vid = meta.video_id || meta.id;
         if (!vid) return null;
@@ -195,13 +208,18 @@ export const VideoGroupsView: FC = () => {
         return {
           videoId: vid,
           name: meta.name ?? meta.title ?? vid,
-          url: '',
+          url:
+            resolveVideoUrl(result.video, ASSETS_ENDPOINT) ||
+            toPlayableMetadataUrl(meta.video_rel_url) ||
+            toPlayableMetadataUrl(meta.video_url) ||
+            '',
           tags: tagsArr,
           createdAt: (meta.date_time as string) ?? (meta.date as string) ?? '',
           updatedAt: '',
+          dataStore: result.video?.dataStore,
         } as Video;
       })
-      .filter((v): v is Video => v !== null);
+        .filter((v: Video | null): v is Video => v !== null);
   }, [selectedResults]);
 
   // If no search results, show a helpful empty state
@@ -229,13 +247,19 @@ export const VideoGroupsView: FC = () => {
       if (validTags.length === 0) {
         // Video has no valid tags - add to Untagged group
         if (!groups.has('Untagged')) groups.set('Untagged', []);
-        groups.get('Untagged')!.push(video);
+        const untaggedVideos = groups.get('Untagged')!;
+        if (!untaggedVideos.some((existing) => existing.videoId === video.videoId)) {
+          untaggedVideos.push(video);
+        }
       } else {
         // Video has valid tags - add to each tag group
         validTags.forEach((tag) => {
           const trimmedTag = tag.trim();
           if (!groups.has(trimmedTag)) groups.set(trimmedTag, []);
-          groups.get(trimmedTag)!.push(video);
+          const tagVideos = groups.get(trimmedTag)!;
+          if (!tagVideos.some((existing) => existing.videoId === video.videoId)) {
+            tagVideos.push(video);
+          }
         });
       }
     });
@@ -277,11 +301,12 @@ export const VideoGroupsView: FC = () => {
 
           <VideoGrid>
             {group.videos.map((video) => {
-              const videoUrl = getVideoUrl ? getVideoUrl(video.videoId) : video.url;
+              const reduxVideoUrl = getVideoUrl ? getVideoUrl(video.videoId) : null;
+              const videoUrl = reduxVideoUrl || video.url;
               const relevanceScore = videoRelevanceMap.get(video.videoId) ?? 0;
 
               return (
-                <VideoCard key={video.videoId}>
+                <VideoCard key={`${group.tag}-${video.videoId}`}>
                   <VideoCardWrapper>
                     {videoUrl ? (
                       <VideoPlayer controls preload="metadata">

sample-applications/video-search-and-summarization/ui/react/src/redux/search/VideoTile.tsx

@@ -5,6 +5,7 @@ import { useTranslation } from 'react-i18next';
 import { ASSETS_ENDPOINT } from '../../config';
 import { useAppSelector } from '../store';
 import { SearchSelector } from './searchSlice';
+import { resolveVideoUrl } from '../video/videoUrl';
 
 export interface VideoTileProps {
   resultIndex: number; // Index in the selectedResults array
@@ -20,14 +21,7 @@ export const VideoTile: FC<VideoTileProps> = ({ resultIndex }) => {
 
   const { metadata, video } = searchResult || {};
 
-  const getVideoUrl = () => {
-    if (video?.dataStore) {
-      return `${ASSETS_ENDPOINT}/${video.dataStore.bucket}/${video.url}`;
-    }
-    return null;
-  };
-
-  const videoUrl = getVideoUrl();
+  const videoUrl = resolveVideoUrl(video, ASSETS_ENDPOINT);
 
   useEffect(() => {
     if (videoRef.current && metadata?.timestamp && typeof metadata.timestamp === 'number') {

sample-applications/video-search-and-summarization/ui/react/src/redux/video/videoSlice.ts

@@ -6,6 +6,7 @@ import { RootState } from '../store';
 import axios from 'axios';
 import { APP_URL, ASSETS_ENDPOINT } from '../../config';
 import { StateActionStatus } from '../summary/summary';
+import { resolveVideoUrl } from './videoUrl';
 
 const initialState: VideosState = {
   videos: [],
@@ -58,14 +59,7 @@ export const videosSelector = createSelector([selectVideoState], (videosState) =
   },
   getVideoUrl: (videoId: string) => {
     const video = videosState.videos.find((video) => video.videoId === videoId);
-
-    console.log(video);
-
-    if (video?.dataStore) {
-      return `${ASSETS_ENDPOINT}/${video.dataStore.bucket}/${video.url}`;
-    }
-
-    return null;
+    return resolveVideoUrl(video, ASSETS_ENDPOINT);
   },
 }));
 

sample-applications/video-search-and-summarization/ui/react/src/redux/video/videoUrl.ts

@@ -0,0 +1,17 @@
+// Copyright (C) 2025 Intel Corporation
+// SPDX-License-Identifier: Apache-2.0
+import { Video } from './video';
+
+export const resolveVideoUrl = (video?: Video | null, assetsEndpoint = ''): string | null => {
+  if (!video) return null;
+
+  if (video.dataStore?.bucket && video.url) {
+    return `${assetsEndpoint}/${video.dataStore.bucket}/${video.url}`;
+  }
+
+  if (video.url) {
+    return video.url;
+  }
+
+  return null;
+};