Update app.py

app.py

@@ -30,13 +30,24 @@ df_embeddings = emb_data["df_embeddings"].astype("float32")
 index = faiss.read_index(INDEX_PATH)
 
 # ---------- Secrets ----------
+
 HF_TOKEN = os.getenv("HF_TOKEN")
 SPOTIFY_CLIENT_ID = os.getenv("SPOTIPY_CLIENT_ID")
-SPOTIFY_CLIENT_SECRET = os.getenv("SPOTIPY_CLIENT_SECRET")
+SPOTIPY_CLIENT_SECRET = os.getenv("SPOTIPY_CLIENT_SECRET")
 
 # ---------- Models ----------
 query_embedder = SentenceTransformer("all-mpnet-base-v2")
-hf_client = InferenceClient(model="meta-llama/Llama-2-7b-chat-hf", token=HF_TOKEN)
+
+LLAMA_MODEL_ID = "meta-llama/Llama-2-7b-chat-hf"
+
+# Create a generic client; we'll pass model per call
+hf_client = None
+if HF_TOKEN:
+    try:
+        hf_client = InferenceClient(token=HF_TOKEN)
+    except Exception as e:
+        print("⚠️ Could not initialize HF Inference client:", repr(e))
+        hf_client = None
 
 sp = None
 if SPOTIFY_CLIENT_ID and SPOTIFY_CLIENT_SECRET:
@@ -47,17 +58,48 @@ if SPOTIFY_CLIENT_ID and SPOTIFY_CLIENT_SECRET:
 def encode_query(text):
     return query_embedder.encode([text], convert_to_numpy=True).astype("float32")
 
-def expand_with_llama(query):
-    if not hf_client:
+def expand_with_llama(query: str) -> str:
+    """
+    Enrich the query using LLaMA via HF Inference.
+
+    On HF Spaces, the Inference provider can sometimes be unavailable
+    or misconfigured (giving the StopIteration error you saw). In that
+    case, we log and fall back to the raw query so the UI keeps working.
+    """
+    if hf_client is None or not HF_TOKEN:
+        # No client/token -> behave like "no expansion"
         return query
+
     prompt = f"""You are helping someone search a lyrics catalog.
-If the input looks like lyrics or a singer name, return artist and song titles that match.
-Otherwise, return a short list of lyric-style keywords related to the input sentence.
 
-Input: {query}
-Output:"""
-    response = hf_client.text_generation(prompt, max_new_tokens=96, temperature=0.2, repetition_penalty=1.05)
-    return query + " " + str(response).strip().replace("\n", " ")
+If the input looks like existing song lyrics or a singer name,
+return artist and song titles that match.
+
+Otherwise, return a short list of lyric-style keywords
+that are closely related to the input sentence.
+
+Input:
+{query}
+
+Output (no explanation, just titles or keywords):"""
+
+    try:
+        response = hf_client.text_generation(
+            prompt,
+            model=LLAMA_MODEL_ID,          
+            max_new_tokens=96,
+            temperature=0.2,
+            repetition_penalty=1.05,
+        )
+    except Exception as e:
+        
+        print("LLaMA expansion failed on HF, using raw query:", repr(e))
+        return query
+
+    keywords = str(response).strip().replace("\n", " ")
+    expanded = query + " " + keywords
+    return expanded
+
 
 def distances_to_similarity_pct(dists):
     if len(dists) == 0: return np.array([])