feat: ✨ Working PoC of the Dungeon Masters Vault

src/temp.py

@@ -0,0 +1,31 @@
+from langchain_community.vectorstores import FAISS
+
+from embedding import LocalLMEmbeddings
+
+
+def retrieve_enriched_context(query, db_path="./local_faiss_db"):
+    # 1. Re-initialize the same embedding model
+    embeddings_model = LocalLMEmbeddings(
+        model="text-embedding-qwen3-embedding-8b", base_url="http://192.168.0.49:1234"
+    )
+
+    # 2. Load the index from disk
+    # allow_dangerous_deserialization is required because FAISS uses pickle
+    vectorstore = FAISS.load_local(db_path, embeddings_model, allow_dangerous_deserialization=True)
+
+    # 3. Perform the search
+    # k=4 means "bring back the top 4 most relevant chunks"
+    results_with_scores = vectorstore.similarity_search_with_score(query, k=4)
+
+    return results_with_scores
+
+
+# --- Example Usage ---
+query = "the party get free bread but i cant remember why?"
+hits = retrieve_enriched_context(query)
+
+for doc, score in hits:
+    print(f"\n🎯 [Score: {score:.4f}]")
+    print(f"📄 Content: {doc.page_content[:200]}...")
+    print(f"🛠️ Metadata (Enrichment): {doc.metadata}")
+#    print(f"doc: {doc}")