fix: reliable ArXiv ID lookup and uniformized paper ID prefixes

app/api/routes/discovery.py

@@ -92,6 +92,7 @@ async def discovery_search(request: SearchRequest, db: Session = Depends(get_db)
         arxiv_data = arxiv_service.fetch_papers(request.query, request.max_results)
         for p in arxiv_data:
             p["source"] = "ArXiv"
+            p["id"] = f"arxiv_{p['id']}" # Uniformize with prefixes
             p["has_pdf"] = True  # ArXiv always has PDF
 
         # 2. Fetch from DOAJ

app/services/analysis_service.py

@@ -94,7 +94,10 @@ class AnalysisService:
                             svc = PubMedService(); remote_data = svc.fetch_papers(pid.replace("pubmed_", ""), 1)
                         elif "openalex_" in pid:
                             svc = OpenAlexService(); remote_data = svc.fetch_papers(pid.replace("openalex_", ""), 1)
+                        elif "arxiv_" in pid:
+                            svc = ArXivService(); remote_data = svc.fetch_papers(pid.replace("arxiv_", ""), 1)
                         else:
+                            # Fallback to ArXiv if no prefix (for old saved papers without prefix)
                             svc = ArXivService(); remote_data = svc.fetch_papers(pid, 1)
                         
                         if remote_data:
@@ -111,6 +114,8 @@ class AnalysisService:
                             db.commit()
                             db.refresh(paper)
                             logger.info(f"Saved remote paper {pid} to project {project_id}")
+                        else:
+                            logger.warning(f"No remote data found for {pid}")
                     except Exception as fetch_err:
                         logger.error(f"Failed to fetch/save remote paper {pid}: {fetch_err}")
 

app/services/arxiv_service.py

@@ -17,8 +17,13 @@ class ArXivService:
         Search ArXiv for papers (Metadata only).
         """
         logger.info(f"Searching ArXiv for: {query} (max: {max_results})")
+        
+        # Check if query looks like an ArXiv ID to use id_list (more reliable)
+        is_id = "." in query or "/" in query
+        
         search = arxiv.Search(
-            query=query,
+            query="" if is_id else query,
+            id_list=[query] if is_id else [],
             max_results=max_results,
             sort_by=arxiv.SortCriterion.Relevance
         )