Add model selection dropdown and fix model detection logic

app.py

@@ -1,4 +1,4 @@
-# Force redeploy trigger - version 1.6
+# Force redeploy trigger - version 1.7
 import streamlit as st
 from utils.config import config
 import requests
@@ -19,6 +19,9 @@ if 'model_status' not in st.session_state:
 if 'available_models' not in st.session_state:
     st.session_state.available_models = []
 
+if 'selected_model' not in st.session_state:
+    st.session_state.selected_model = config.local_model_name
+
 # Sidebar for user selection
 st.sidebar.title("🧘 AI Life Coach")
 user = st.sidebar.selectbox("Select User", ["Rob", "Sarah"])
@@ -30,14 +33,13 @@ ngrok_input = st.sidebar.text_input("Ngrok URL", value=st.session_state.ngrok_ur
 if st.sidebar.button("Update Ngrok URL"):
     st.session_state.ngrok_url = ngrok_input
     st.session_state.model_status = "checking"
+    st.session_state.available_models = []
     st.sidebar.success("Ngrok URL updated!")
     st.experimental_rerun()
 
+# Model selection
 st.sidebar.markdown("---")
-
-# Get environment info
-BASE_URL = os.environ.get("SPACE_ID", "")  # Will be set in HF Spaces
-IS_HF_SPACE = bool(BASE_URL)
+st.sidebar.subheader("Model Selection")
 
 # Headers to skip ngrok browser warning
 NGROK_HEADERS = {
@@ -45,6 +47,46 @@ NGROK_HEADERS = {
     "User-Agent": "AI-Life-Coach-App"
 }
 
+# Fetch available models when we have a valid connection
+if st.session_state.ngrok_url and st.session_state.model_status != "unreachable":
+    try:
+        response = requests.get(
+            f"{st.session_state.ngrok_url}/api/tags", 
+            headers=NGROK_HEADERS,
+            timeout=5
+        )
+        if response.status_code == 200:
+            models_data = response.json().get("models", [])
+            model_names = [m.get("name") for m in models_data]
+            if model_names:
+                st.session_state.available_models = model_names
+                # If current selected model not in list, select the first one
+                if st.session_state.selected_model not in model_names:
+                    st.session_state.selected_model = model_names[0]
+    except Exception as e:
+        pass  # Silently fail, we'll handle this in the main logic
+
+# Model selector dropdown
+if st.session_state.available_models:
+    selected_model = st.sidebar.selectbox(
+        "Select Model", 
+        st.session_state.available_models,
+        index=st.session_state.available_models.index(st.session_state.selected_model) 
+              if st.session_state.selected_model in st.session_state.available_models 
+              else 0
+    )
+    st.session_state.selected_model = selected_model
+else:
+    st.sidebar.warning("No models available - check Ollama connection")
+    model_input = st.sidebar.text_input("Or enter model name", value=st.session_state.selected_model)
+    st.session_state.selected_model = model_input
+
+st.sidebar.markdown("---")
+
+# Get environment info
+BASE_URL = os.environ.get("SPACE_ID", "")  # Will be set in HF Spaces
+IS_HF_SPACE = bool(BASE_URL)
+
 # Fetch Ollama status
 def get_ollama_status(ngrok_url):
     try:
@@ -60,11 +102,14 @@ def get_ollama_status(ngrok_url):
             st.session_state.available_models = model_names
             
             if models:
+                # Check if our selected model is available
+                selected_model_available = st.session_state.selected_model in model_names
                 return {
                     "running": True, 
-                    "model_loaded": models[0].get("name"),
+                    "model_loaded": st.session_state.selected_model if selected_model_available else model_names[0],
                     "remote_host": ngrok_url,
-                    "available_models": model_names
+                    "available_models": model_names,
+                    "selected_model_available": selected_model_available
                 }
             else:
                 st.session_state.model_status = "no_models"
@@ -84,29 +129,6 @@ def get_ollama_status(ngrok_url):
             "remote_host": ngrok_url
         }
 
-# Poll for model availability
-def poll_model_status(ngrok_url):
-    if st.session_state.model_status in ["checking", "no_models"]:
-        try:
-            response = requests.get(
-                f"{ngrok_url}/api/tags", 
-                headers=NGROK_HEADERS,
-                timeout=5
-            )
-            if response.status_code == 200:
-                models = response.json().get("models", [])
-                model_names = [m.get("name") for m in models]
-                st.session_state.available_models = model_names
-                
-                if config.local_model_name in model_names:
-                    st.session_state.model_status = "ready"
-                elif models:
-                    st.session_state.model_status = "different_models"
-                else:
-                    st.session_state.model_status = "no_models"
-        except:
-            st.session_state.model_status = "unreachable"
-
 # After user selects name, load conversation history
 def get_conversation_history(user_id):
     try:
@@ -120,8 +142,14 @@ def get_conversation_history(user_id):
 # Check Ollama status with the current ngrok URL
 ollama_status = get_ollama_status(st.session_state.ngrok_url)
 
-# Poll for model status (run once per session)
-poll_model_status(st.session_state.ngrok_url)
+# Update model status
+if ollama_status.get("running", False):
+    if ollama_status.get("available_models"):
+        st.session_state.model_status = "ready"
+    else:
+        st.session_state.model_status = "no_models"
+else:
+    st.session_state.model_status = "unreachable"
 
 # Display Ollama status
 use_fallback = not ollama_status.get("running", False) or config.use_fallback
@@ -131,7 +159,11 @@ if use_fallback:
     if "error" in ollama_status:
         st.sidebar.caption(f"Error: {ollama_status['error'][:50]}...")
 else:
-    st.sidebar.success(f"🧠 Ollama Model: {ollama_status['model_loaded']}")
+    model_status_msg = ollama_status.get('model_loaded', 'Unknown')
+    if ollama_status.get('selected_model_available', True):
+        st.sidebar.success(f"🧠 Ollama Model: {model_status_msg}")
+    else:
+        st.sidebar.warning(f"🧠 Ollama Model: {model_status_msg} (selected model not available)")
     st.sidebar.info(f"Connected to: {ollama_status['remote_host']}")
 
 # Model status indicator
@@ -142,8 +174,6 @@ elif st.session_state.model_status == "checking":
     model_status_container.info("🔍 Checking model...")
 elif st.session_state.model_status == "no_models":
     model_status_container.warning("⚠️ No models found")
-elif st.session_state.model_status == "different_models":
-    model_status_container.warning("⚠️ Different models available")
 else:  # unreachable
     model_status_container.error("❌ Ollama unreachable")
 
@@ -162,20 +192,20 @@ st.markdown("Talk to your personal development assistant.")
 with st.expander("🔍 Connection Status"):
     st.write("Ollama Status:", ollama_status)
     st.write("Model Status:", st.session_state.model_status)
+    st.write("Selected Model:", st.session_state.selected_model)
     st.write("Available Models:", st.session_state.available_models)
     st.write("Environment Info:")
     st.write("- Is HF Space:", IS_HF_SPACE)
     st.write("- Base URL:", BASE_URL or "Not in HF Space")
-    st.write("- Configured Ollama Host:", config.ollama_host)
     st.write("- Current Ngrok URL:", st.session_state.ngrok_url)
     st.write("- Using Fallback:", use_fallback)
     st.write("- Redis Health:", check_redis_health())
 
 # Function to send message to Ollama
-def send_to_ollama(user_input, conversation_history, ngrok_url):
+def send_to_ollama(user_input, conversation_history, ngrok_url, model_name):
     try:
         payload = {
-            "model": config.local_model_name,
+            "model": model_name,
             "messages": conversation_history,
             "stream": False
         }
@@ -253,7 +283,12 @@ if st.button("Send"):
                 ai_response = send_to_hf(user_input, conversation_history)
                 backend_used = "Hugging Face"
             else:
-                ai_response = send_to_ollama(user_input, conversation_history, st.session_state.ngrok_url)
+                ai_response = send_to_ollama(
+                    user_input, 
+                    conversation_history, 
+                    st.session_state.ngrok_url,
+                    st.session_state.selected_model
+                )
                 backend_used = "Ollama"
             
             if ai_response:

diagnose_ollama.py

@@ -13,7 +13,6 @@ def test_ollama_connectivity(custom_ngrok_url=None):
     
     print("=== Ollama Connectivity Diagnostic ===")
     print(f"Testing Ollama Host: {test_url}")
-    print(f"Configured Model: {config.local_model_name}")
     print()
     
     # Headers to skip ngrok browser warning
@@ -39,18 +38,14 @@ def test_ollama_connectivity(custom_ngrok_url=None):
                 models = data.get("models", [])
                 print(f"  Available Models: {len(models)}")
                 for model in models:
-                    print(f"    - {model.get('name', 'Unknown model')}")
+                    print(f"    - {model.get('name', 'Unknown model')} (Size: {model.get('size', 'Unknown')})")
                 
-                # Check if our configured model is available
-                model_names = [m.get('name') for m in models]
-                if config.local_model_name in model_names:
-                    print(f"  ✓ Configured model '{config.local_model_name}' is available")
+                if models:
+                    print("\n  ✅ Ollama is working correctly with models loaded!")
+                    print("  You should be able to use the AI Life Coach app now.")
                 else:
-                    print(f"  ⚠ Configured model '{config.local_model_name}' not found")
-                    if models:
-                        print(f"    Available models: {', '.join(model_names)}")
-                    else:
-                        print(f"    No models found. Try running: ollama pull {config.local_model_name}")
+                    print("\n  ⚠️  Ollama is running but no models are loaded.")
+                    print("  Try pulling a model: ollama pull mistral")
                         
             except Exception as e:
                 print(f"  Error parsing response: {e}")
@@ -68,90 +63,6 @@ def test_ollama_connectivity(custom_ngrok_url=None):
     except Exception as e:
         print(f"  ✗ Unexpected error: {e}")
     
-    print()
-    
-    # Test 2: Test model generation if we can connect
-    print("Test 2: Testing model generation...")
-    try:
-        # First verify we can connect
-        response = requests.get(
-            f"{test_url}/api/tags", 
-            headers=headers, 
-            timeout=10
-        )
-        
-        if response.status_code == 200:
-            data = response.json()
-            models = data.get("models", [])
-            model_names = [m.get('name') for m in models]
-            
-            if config.local_model_name in model_names:
-                print(f"  Testing generation with model: {config.local_model_name}")
-                
-                # Test generation
-                generate_payload = {
-                    "model": config.local_model_name,
-                    "prompt": "Hello, please respond with just the word 'Success' in uppercase.",
-                    "stream": False
-                }
-                
-                generate_response = requests.post(
-                    f"{test_url}/api/generate",
-                    json=generate_payload,
-                    headers=headers,
-                    timeout=30
-                )
-                
-                if generate_response.status_code == 200:
-                    generate_data = generate_response.json()
-                    response_text = generate_data.get("response", "").strip()
-                    print(f"  ✓ Model generation successful")
-                    print(f"  Response: {response_text}")
-                else:
-                    print(f"  ✗ Model generation failed with status {generate_response.status_code}")
-                    print(f"  Response: {generate_response.text[:200]}...")
-            else:
-                print(f"  ⚠ Skipping generation test - model '{config.local_model_name}' not available")
-                if not models:
-                    print(f"    No models found. Try running: ollama pull {config.local_model_name}")
-        else:
-            print(f"  ⚠ Skipping generation test - cannot connect to Ollama host")
-            
-    except requests.exceptions.Timeout:
-        print("  ✗ Generation test timed out (took more than 30 seconds)")
-        print("    This may indicate the model is still loading or the server is under heavy load")
-    except requests.exceptions.ConnectionError as e:
-        print(f"  ✗ Generation test connection error: {e}")
-    except Exception as e:
-        print(f"  ✗ Generation test error: {e}")
-    
-    print()
-    
-    # Test 3: Check localhost as fallback
-    print("Test 3: Checking localhost fallback (if different from configured host)...")
-    if test_url != "http://localhost:11434":
-        try:
-            local_response = requests.get(
-                "http://localhost:11434/api/tags", 
-                headers=headers, 
-                timeout=5
-            )
-            print(f"  Local Status Code: {local_response.status_code}")
-            if local_response.status_code == 200:
-                print("  ✓ Successfully connected to localhost Ollama")
-                try:
-                    local_data = local_response.json()
-                    local_models = local_data.get("models", [])
-                    print(f"  Local Available Models: {len(local_models)}")
-                except Exception as e:
-                    print(f"  Error parsing local response: {e}")
-            else:
-                print(f"  ✗ Local connection failed with status {local_response.status_code}")
-        except Exception as e:
-            print(f"  ✗ Local connection error: {e}")
-    else:
-        print("  Skipping (configured host is already localhost)")
-    
     print()
     print("=== Diagnostic Complete ===")
     print()