Upload app.py with huggingface_hub

app.py

@@ -111,7 +111,6 @@ def _build_messages(
     companion_name: str = "Lumi",
     companion_desc: str = "a warm and patient AI companion",
 ) -> list[dict]:
-    history = history or []
     prompt = build_system_prompt(
         PATIENT_ID, PATIENT_NAME,
         companion_name=companion_name,
@@ -230,12 +229,10 @@ def _save_wav_np(audio_tuple) -> str | None:
 
 
 def lumi_api(message: str, history: list[dict], profile_id: str, session_id: str | None = None):
-    print(f"[DEBUG] lumi_api called with:")
-    print(f"  message: {message}")
-    print(f"  history: {history}")
-    print(f"  profile_id: {profile_id}")
-    print(f"  session_id: {session_id}")
-    
+    """
+    Unified API for the React frontend.
+    Returns: (response_text, audio_b64_data_uri, avatar_tag, final_session_id)
+    """
     profile = get_profile(profile_id)
     cname   = profile["display_name"]
     cdesc   = profile["description"]
@@ -243,21 +240,16 @@ def lumi_api(message: str, history: list[dict], profile_id: str, session_id: str
     is_scam, deflection = check_and_deflect(message)
     if is_scam:
         audio_path, _ = _tts_and_video(deflection, profile_id)
-        return deflection, _audio_b64(audio_path), "gentle", session_id or "", []
+        return deflection, _audio_b64(audio_path), "gentle", session_id
 
     messages = _build_messages(history, cname, cdesc) + [{"role": "user", "content": message}]
     try:
-        print(f"[API] Calling LLM...")
         raw = _call_llm(messages)
-        print(f"[API] LLM Response received. Parsing...")
         parsed = parse_structured_output(raw)
         response_text = parsed["full_response"]
         avatar_tag = parsed["avatar_tag"]
-        
-        print(f"[API] Generating audio...")
         audio_path, _ = _tts_and_video(response_text, profile_id)
         
-        print(f"[API] Saving session...")
         # Real-time saving
         new_history = history + [
             {"role": "user", "content": message},
@@ -268,13 +260,12 @@ def lumi_api(message: str, history: list[dict], profile_id: str, session_id: str
                                 f"Chat - {len(new_history)//2} turns", 
                                 new_history, session_id)
 
-        print(f"[API] Success. final_id={final_id}")
-        return response_text, _audio_b64(audio_path), avatar_tag, final_id, parsed.get("actions", [])
+        return response_text, _audio_b64(audio_path), avatar_tag, final_id
     except Exception as e:
         print(f"[API Error] {e}")
         err_msg = "I'm sorry, I'm having a little trouble. Could you try again?"
         audio_path, _ = _tts_and_video(err_msg, profile_id)
-        return err_msg, _audio_b64(audio_path), "concerned", session_id or "", []
+        return err_msg, _audio_b64(audio_path), "concerned", session_id
 
 
 def voice_submit(audio_tuple, history: list[dict], profile_id: str):
@@ -833,14 +824,11 @@ Built for the AMD Developer Hackathon 2026 · Fine-Tuning Track
     _resp  = gr.Textbox(visible=False)
     _audio = gr.Textbox(visible=False)
     _tag   = gr.Textbox(visible=False)
-    _session_id = gr.Textbox(visible=False)
-    _final_id   = gr.Textbox(visible=False)
-    _actions    = gr.JSON(visible=False)
-    _btn        = gr.Button(visible=False)
+    _btn   = gr.Button(visible=False)
     _btn.click(
         fn=lumi_api,
-        inputs=[_msg, _hist, _prof, _session_id],
-        outputs=[_resp, _audio, _tag, _final_id, _actions],
+        inputs=[_msg, _hist, _prof],
+        outputs=[_resp, _audio, _tag],
         api_name="lumi_api",
     )
 
@@ -860,31 +848,6 @@ async def fetch_summaries():
     # For the hackathon, we use the global PATIENT_ID.
     return get_all_summaries(PATIENT_ID)
 
-@_fastapi.get("/api/diag")
-async def diagnostic():
-    results = {}
-    # 1. Test LLM
-    try:
-        start = time.time()
-        test_resp = llm.chat.completions.create(
-            model=MODEL_NAME,
-            messages=[{"role": "user", "content": "ping"}],
-            max_tokens=5
-        )
-        results["llm"] = {"status": "ok", "latency": time.time() - start, "response": test_resp.choices[0].message.content}
-    except Exception as e:
-        results["llm"] = {"status": "error", "message": str(e)}
-    
-    # 2. Test ChromaDB
-    try:
-        from pipeline.memory import _get_collection
-        col = _get_collection()
-        results["chroma"] = {"status": "ok", "count": col.count()}
-    except Exception as e:
-        results["chroma"] = {"status": "error", "message": str(e)}
-        
-    return results
-
 @_fastapi.get("/api/whoami")
 async def whoami(request: Request):
     # Check multiple possible HF headers (request-specific)