full interface set up

Author: Shakiran-Nannyombi

backend/api/models/loader.py

@@ -52,19 +52,15 @@ def _load_feature_names(self):
         """Load feature names"""
         try:
             # Try v2 first
-            feature_path = self.local_model_dir / 'feature_names_v2_20251103.pkl'
+            feature_path = self.local_model_dir / 'feature_names_v2_20251120.pkl'
             if not feature_path.exists():
-                feature_path = self.local_model_dir / 'xgboost_feature_names_20251103.pkl'
+                feature_path = self.local_model_dir / 'xgboost_feature_names_20251120.pkl'
 
             with open(feature_path, 'rb') as f:
                 self.feature_names = pickle.load(f)
 
-            # Fix VitaminA field name (µ/μ -> ug for ASCII compatibility)
-            self.feature_names = [
-                'VitaminA_ug_per_serving' if 'VitaminA_' in name and 'g_per_serving' in name
-                else name
-                for name in self.feature_names
-            ]
+            # Normalize feature names to lowercase
+            self.feature_names = [name.lower() for name in self.feature_names]
 
             logger.info(f"Loaded {len(self.feature_names)} feature names")
         except Exception as e:
@@ -75,7 +71,7 @@ def _load_feature_names(self):
     def _load_offline_model(self):
         """Load lightweight HistGradient model"""
         try:
-            model_path = self.local_model_dir / 'baseline_nutrition_model_v2_20251103.pkl'
+            model_path = self.local_model_dir / 'baseline_nutrition_model_v2_20251120.pkl'
             if not model_path.exists():
                 logger.info("Offline model not found in local uploads")
                 self.models['offline'] = {'available': False}
@@ -105,7 +101,7 @@ def _load_offline_model(self):
     def _load_local_xgboost(self):
         """Load local XGBoost model"""
         try:
-            model_path = self.local_model_dir / 'xgboost_nutrition_model_20251103.pkl'
+            model_path = self.local_model_dir / 'xgboost_nutrition_model_20251120.pkl'
             if not model_path.exists():
                 logger.info("Local XGBoost model not found in local uploads")
                 self.models['local_xgboost'] = {'available': False}
@@ -329,10 +325,30 @@ def predict(
             elif self.models.get('offline', {}).get('available'):
                 model_key = 'offline'
             else:
+                # No trained models are available in this environment.
+                # Provide a lightweight heuristic fallback so the API remains usable for testing.
+                try:
+                    # Basic heuristic: estimate daily caloric needs as Energy_kcal_per_serving * meals per day
+                    meals = int(input_data.get('mealsPerDay') or input_data.get('meals_per_day') or 3)
+                    energy = float(input_data.get('Energy_kcal_per_serving') or input_data.get('energy') or 250)
+                    caloric_needs = float(max(800.0, energy * meals))
+                except Exception:
+                    caloric_needs = 2000.0
+
                 return {
-                    'success': False,
-                    'error': 'No models available',
-                    'status': 'error'
+                    'success': True,
+                    'prediction': {
+                        'caloric_needs': float(caloric_needs),
+                        'unit': 'kcal/day',
+                        'model': 'heuristic-fallback',
+                        'accuracy': 'heuristic: approximate'
+                    },
+                    'model_info': {
+                        'type': 'heuristic',
+                        'size': 'n/a',
+                        'mode': 'fallback'
+                    },
+                    'status': 'fallback'
                 }
         else:
             model_key = model_preference
@@ -350,15 +366,36 @@ def predict(
         try:
             # Prepare input
             df = pd.DataFrame([input_data])
+            
+            # Normalize column names to match model expectations
+            column_mapping = {
+                'energy_kcal_per_serving': 'energy_kcal_per_serving',
+                'protein_g_per_serving': 'protein_g_per_serving',
+                'fat_g_per_serving': 'fat_g_per_serving',
+                'carbohydrates_g_per_serving': 'carbohydrate_g_per_serving',
+                'fiber_g_per_serving': 'fiber_g_per_serving',
+                'calcium_mg_per_serving': 'calcium_mg_per_serving',
+                'iron_mg_per_serving': 'iron_mg_per_serving',
+                'zinc_mg_per_serving': 'zinc_mg_per_serving',
+                'vitamina_ug_per_serving': 'vitamin_a_mcg_per_serving',
+                'vitaminc_mg_per_serving': 'vitamin_c_mg_per_serving',
+                'potassium_mg_per_serving': 'potassium_mg_per_serving',
+                'region_encoded': 'region_encoded',
+                'condition_encoded': 'condition_encoded',
+                'age_group_encoded': 'age_group_encoded',
+                'season_encoded': 'season_encoded'
+            }
+            df.columns = [col.lower() for col in df.columns]
+            df = df.rename(columns=column_mapping)
 
             # If we have canonical feature order, apply it. Otherwise try model's
             # feature_names_in_ attribute. If neither present, pass df as-is.
             if self.feature_names:
                 try:
                     df = df[self.feature_names]
-                except Exception:
+                except Exception as e:
                     # columns missing or mismatched; fall back
-                    logger.warning("Feature names present but input missing some columns; falling back to available input columns")
+                    logger.warning(f"Feature names present but input missing some columns: {e}; falling back to available input columns")
             else:
                 if hasattr(model, 'feature_names_in_'):
                     cols = list(getattr(model, 'feature_names_in_'))
@@ -369,19 +406,23 @@ def predict(
 
             # Make prediction
             prediction = model.predict(df)[0]
-            
-            # Determine status
+
+            # Determine status based on model key
             if model_key == 'huggingface':
                 status = 'online'
+            elif model_key == 'local_xgboost':
+                status = 'local'
+            elif model_key == 'offline':
+                status = 'offline'
             else:
                 status = 'unknown'
-            
+
             return {
                 'success': True,
                 'prediction': {
                     'caloric_needs': float(prediction),
                     'unit': 'kcal/day',
-                    'model': f"{model_info['type']} ({status.upper()})",
+                    'model': model_info['type'],
                     'accuracy': model_info['accuracy']
                 },
                 'model_info': {

backend/api/routers/foods.py

@@ -47,7 +47,8 @@ def get_local_foods(limit: int = 500):
     df = pd.read_csv(csv_path)
 
     foods = []
-    for _, row in df.iterrows():
+    seen = set()
+    for i, row in df.iterrows():
         try:
             name = row.get('food_name_english') or row.get('food_name') or ''
             category = _normalize_category(row.get('food_category', 'other'))
@@ -70,8 +71,17 @@ def get_local_foods(limit: int = 500):
             availability_score = float(row.get('availability_score', 0) or 0)
             available = availability_score >= 0.5
 
+            # normalize name for deduplication
+            clean_name = str(name).strip()
+            key = clean_name.lower()
+            if key in seen:
+                # skip duplicates by normalized name
+                continue
+            seen.add(key)
+
             foods.append({
-                'name': str(name),
+                'id': f"food_{i}",
+                'name': clean_name,
                 'category': category,
                 'region': region,
                 'energy': round(energy, 1),

frontend/src/components/layout/AppHeader.vue

@@ -56,15 +56,17 @@ const navItems = computed(() => [
       </nav>
 
       <div class="header-right">
-        <!-- Online/Offline Status -->
+        <!-- Online/Offline Status (click to toggle forced offline in settings) -->
         <div class="status-indicator">
-          <span
+          <button
             class="status-badge"
-            :class="appStore.isOnline ? 'status-online' : 'status-offline'"
+            :class="(settingsStore.offlineMode ? 'status-offline' : (appStore.isOnline ? 'status-online' : 'status-offline'))"
+            @click="settingsStore.toggleOfflineMode()"
+            title="Toggle forced offline mode"
           >
-            <i :class="appStore.isOnline ? 'pi pi-wifi' : 'pi pi-ban'"></i>
-            {{ appStore.isOnline ? t('common.online') : t('common.offline') }}
-          </span>
+            <i :class="(settingsStore.offlineMode ? 'pi pi-ban' : (appStore.isOnline ? 'pi pi-wifi' : 'pi pi-ban'))"></i>
+            {{ settingsStore.offlineMode ? (t('common.offline') + ' (forced)') : (appStore.isOnline ? t('common.online') : t('common.offline')) }}
+          </button>
         </div>
 
         <!-- Language selection removed (single-language mode) -->