ultimos ajustes

src/api/routes.py

@@ -315,27 +315,29 @@ class LoadKnowledgeRequest(BaseModel):
 
 @router.post("/knowledge/load")
 async def load_knowledge(body: LoadKnowledgeRequest) -> dict[str, Any]:
-    """Load markdown docs from a directory into the knowledge base."""
+    """Load markdown docs from a directory into the knowledge base.
+
+    Generates embeddings for semantic search via OpenAI text-embedding-3-small.
+    """
     memory = _deps.get("memory_store")
     if not memory:
         raise HTTPException(status_code=501, detail="Memory store not available")
 
     docs_dir = pathlib.Path(body.docs_path)
     if not docs_dir.is_absolute():
-        # Resolve relative to project root
         docs_dir = pathlib.Path(__file__).resolve().parent.parent.parent / body.docs_path
 
     if not docs_dir.is_dir():
         raise HTTPException(status_code=400, detail=f"Directory not found: {docs_dir}")
 
-    loaded = []
+    # Read all docs
+    docs_data: list[tuple[str, str, str, str, list[str]]] = []  # (id, title, content, summary, tags)
     for md_file in sorted(docs_dir.glob("*.md")):
         content = md_file.read_text(encoding="utf-8")
         doc_id = md_file.stem
-
-        # Build a summary from the first ~500 chars
         lines = content.strip().splitlines()
         title = lines[0].lstrip("#").strip() if lines else doc_id
+
         summary_lines = []
         for line in lines[:30]:
             line = line.strip()
@@ -345,12 +347,30 @@ async def load_knowledge(body: LoadKnowledgeRequest) -> dict[str, Any]:
                 break
         summary = " ".join(summary_lines)[:500]
 
-        # Extract tags from headings
         tags = []
         for line in lines:
             if line.startswith("## "):
                 tags.append(line.lstrip("#").strip().lower()[:30])
 
+        docs_data.append((doc_id, title, content, summary, tags[:10]))
+
+    # Generate embeddings in batch
+    from ..memory.embeddings import EmbeddingService
+    embed_service = EmbeddingService()
+    embed_texts = [f"{title}\n{summary}\n{content[:2000]}" for _, title, content, summary, _ in docs_data]
+
+    try:
+        embeddings = await embed_service.embed_batch(embed_texts)
+        has_embeddings = True
+        logger.info("Generated %d embeddings for knowledge base", len(embeddings))
+    except Exception as e:
+        logger.warning("Failed to generate embeddings: %s — loading without semantic search", e)
+        embeddings = [None] * len(docs_data)
+        has_embeddings = False
+
+    # Store docs + embeddings
+    loaded = []
+    for i, (doc_id, title, content, summary, tags) in enumerate(docs_data):
         doc = MemoryDocument(
             memory_id=doc_id,
             memory_type=MemoryType.DOCUMENT,
@@ -358,20 +378,26 @@ async def load_knowledge(body: LoadKnowledgeRequest) -> dict[str, Any]:
             title=title,
             content=content,
             summary=summary,
-            tags=tags[:10],
+            tags=tags,
         )
         await memory.store_document(doc)
+
+        if embeddings[i] is not None:
+            await memory.store_embedding(doc_id, embeddings[i], namespace="knowledge")
+
         loaded.append({
             "id": doc_id,
             "title": title,
             "chars": len(content),
             "tags": tags[:5],
+            "embedded": embeddings[i] is not None,
         })
 
-    logger.info("Loaded %d knowledge documents from %s", len(loaded), docs_dir)
+    logger.info("Loaded %d knowledge documents from %s (embeddings: %s)", len(loaded), docs_dir, has_embeddings)
     return {
         "status": "loaded",
         "count": len(loaded),
+        "embeddings": has_embeddings,
         "documents": loaded,
     }