feat: 100% live search-only top stories with multi-query parallelization

src/api/routes/top_stories.py

@@ -14,9 +14,8 @@ from fastapi import APIRouter, Query, Depends
 from pydantic import BaseModel
 from datetime import datetime
 
-from src.api.dependencies import get_cache_port, get_live_search_port, get_vector_store_port
+from src.api.dependencies import get_cache_port, get_live_search_port
 from src.core.ports.cache_port import CachePort
-from src.core.ports.vector_store_port import VectorStorePort
 from src.infrastructure.adapters.duckduckgo_adapter import DuckDuckGoAdapter
 
 try:
@@ -232,35 +231,52 @@ async def fetch_live_stories(n: int = 6, adapter: DuckDuckGoAdapter = None) -> L
     if not adapter:
         return []
     
-        # Focus on Ethiopia news
-        query = "Ethiopia breaking news latest"
-        results = await adapter.search(query)
+async def fetch_live_stories(n: int = 6, adapter: DuckDuckGoAdapter = None) -> List[TopStory]:
+    """Fetch N live stories from DuckDuckGo using multiple queries to ensure high yield"""
+    if not adapter:
+        return []
+    
+    try:
+        # Run multiple queries in parallel for better coverage
+        queries = [
+            "Ethiopia news breaking today",
+            "Addis Ababa latest updates",
+            "Ethiopia world news headlines"
+        ]
         
-        # If very few results, try a broader regional search
-        if len(results) < 3:
-            logger.info("Live search yielded few results, trying broader regional query")
-            results += await adapter.search("East Africa breaking news today")
+        search_tasks = [adapter.search(q) for q in queries]
+        all_results_lists = await asyncio.gather(*search_tasks)
         
+        # Flatten and deduplicate
         stories = []
         seen_urls = set()
-        for r in results:
-            url = r.get("url", "#")
-            if url in seen_urls: continue
-            seen_urls.add(url)
-            
-            stories.append(TopStory(
-                title=r.get("title", "Untitled"),
-                url=url,
-                source=r.get("source", "Live News"),
-                published_at=r.get("published_at", datetime.utcnow().isoformat()),
-                category="BREAKING",
-                excerpt=r.get("content", "")[:150],
-                image_url=r.get("image_url") or r.get("thumbnail"),
-                origin="live",
-            ))
+        seen_titles = set()
         
-        logger.info(f"Live top stories: fetched {len(stories)}")
-        return stories
+        for results in all_results_lists:
+            for r in results:
+                url = r.get("url", "#")
+                title = r.get("title", "Untitled")
+                title_key = title.lower().strip()[:60]
+                
+                if url in seen_urls or title_key in seen_titles:
+                    continue
+                
+                seen_urls.add(url)
+                seen_titles.add(title_key)
+                
+                stories.append(TopStory(
+                    title=title,
+                    url=url,
+                    source=r.get("source", "Live News"),
+                    published_at=r.get("published_at", datetime.utcnow().isoformat()),
+                    category="BREAKING",
+                    excerpt=r.get("content", "")[:150],
+                    image_url=r.get("image_url") or r.get("thumbnail"),
+                    origin="live",
+                ))
+        
+        logger.info(f"Multi-query live search: collected {len(stories)} unique stories")
+        return stories[:n]
     except Exception as e:
         logger.error(f"Live top stories error: {e}")
         return []
@@ -272,8 +288,7 @@ async def fetch_live_stories(n: int = 6, adapter: DuckDuckGoAdapter = None) -> L
 async def get_top_stories(
     force_refresh: bool = Query(default=False, description="Force cache refresh"),
     cache: CachePort = Depends(get_cache_port),
-    adapter: DuckDuckGoAdapter = Depends(get_live_search_port),
-    vector_store: VectorStorePort = Depends(get_vector_store_port)
+    adapter: DuckDuckGoAdapter = Depends(get_live_search_port)
 ):
     """
     Get top 6 news stories for the landing page.
@@ -314,41 +329,12 @@ async def get_top_stories(
             seen_titles.add(title_key)
             all_stories.append(story)
 
-    # ── FALLBACK: If still less than 6, pull from Vector DB (Qdrant) ──────────
-    if len(all_stories) < 6:
-        needed = 6 - len(all_stories)
-        logger.info(f"Top stories fallback: only have {len(all_stories)}, pulling {needed} from Vector DB")
-        try:
-            # Try 30 days back to be safe
-            db_res = vector_store.browse(limit=needed * 4, days_back=30)
-            db_articles = db_res.get("articles", [])
-            logger.info(f"Vector DB returned {len(db_articles)} potential fallback articles")
-            
-            for p in db_articles:
-                payload = p.payload or {}
-                title = payload.get("title") or payload.get("content", "")[:100]
-                url = payload.get("url") or "#"
-                
-                title_key = title.lower().strip()[:60]
-                if title_key not in seen_titles and len(all_stories) < 6:
-                    seen_titles.add(title_key)
-                    all_stories.append(TopStory(
-                        title=title,
-                        url=url,
-                        source=payload.get("source", "ARKI Intelligence"),
-                        published_at=payload.get("published_at", datetime.utcnow().isoformat()),
-                        category="UPDATE",
-                        excerpt=payload.get("content", "")[:150],
-                        image_url=payload.get("image_url") or payload.get("thumbnail"),
-                        origin="db"
-                    ))
-        except Exception as e:
-            logger.error(f"Top stories DB fallback failed: {e}")
-
-    logger.info(f"Final top stories count: {len(all_stories)}")
-    now_iso = datetime.utcnow().isoformat()
+    # Ensure exactly 6
     final_stories = all_stories[:6]
     
+    logger.info(f"Final top stories count: {len(final_stories)}")
+    now_iso = datetime.utcnow().isoformat()
+    
     payload = {
         "stories": [s.dict() for s in final_stories],
         "fetched_at": now_iso,