Add time_series_data support for LLM context with historical + forecast data

app.py

@@ -134,6 +134,7 @@ class HeatmapRequest(BaseModel):
     metric: str  # e.g., "soil_moisture", "pest_risk"
     gaussian_sigma: float = 1.5
     show_field_boundary: bool = True
+    time_series_data: Optional[Dict[str, Any]] = None  # Historical + forecast time series from Flutter cache
 
 
 class HeatmapResponse(BaseModel):

llm_analysis.py

@@ -58,7 +58,7 @@ def call_gemini_with_fallback(prompt: str) -> str:
     return call_groq(prompt)
 
 def prepare_indices_context(summary_report: Dict, crop_type: str, farmer_context: Dict, 
-                           temporal_stats: Dict = None) -> str:
+                           temporal_stats: Dict = None, time_series_data: Dict = None) -> str:
     """
     Prepare a comprehensive context string for the LLM including temporal statistics.
     
@@ -126,6 +126,57 @@ VEGETATION INDICES DATA (ALL 13 INDICES):
                 latest_rolling = float(np.nanmean(t_stats['rolling_avg_3'][-1]))
                 context += f"\n  - Latest Rolling Average (3-period): {latest_rolling:.4f}"
     
+    # Add time series data if provided (from Flutter cached data)
+    if time_series_data:
+        context += "\n\nTIME SERIES DATA (HISTORICAL + FORECAST):\n"
+        context += "=" * 45 + "\n"
+        
+        for index_name, ts_data in time_series_data.items():
+            context += f"\n{index_name}:\n"
+            
+            # Historical data summary
+            if 'historical' in ts_data and ts_data['historical']:
+                hist = ts_data['historical']
+                hist_count = len(hist)
+                if hist_count > 0:
+                    # Get first and last values
+                    first_hist = hist[0]
+                    last_hist = hist[-1]
+                    first_val = first_hist.get('value', 0)
+                    last_val = last_hist.get('value', 0)
+                    first_date = first_hist.get('date', 'N/A')
+                    last_date = last_hist.get('date', 'N/A')
+                    
+                    context += f"  Historical ({hist_count} data points):\n"
+                    context += f"    - Start: {first_date} = {first_val:.4f}\n"
+                    context += f"    - End: {last_date} = {last_val:.4f}\n"
+                    context += f"    - Historical Change: {'+' if last_val > first_val else ''}{last_val - first_val:.4f}\n"
+                    
+                    # Calculate average
+                    avg_hist = sum(h.get('value', 0) for h in hist) / hist_count
+                    context += f"    - Average: {avg_hist:.4f}\n"
+            
+            # Forecast data summary
+            if 'forecast' in ts_data and ts_data['forecast']:
+                fcast = ts_data['forecast']
+                fcast_count = len(fcast)
+                if fcast_count > 0:
+                    first_fcast = fcast[0]
+                    last_fcast = fcast[-1]
+                    first_val = first_fcast.get('value', 0)
+                    last_val = last_fcast.get('value', 0)
+                    first_date = first_fcast.get('date', 'N/A')
+                    last_date = last_fcast.get('date', 'N/A')
+                    
+                    context += f"  Forecast ({fcast_count} days ahead):\n"
+                    context += f"    - Start: {first_date} = {first_val:.4f}\n"
+                    context += f"    - End: {last_date} = {last_val:.4f}\n"
+                    context += f"    - Predicted Change: {'+' if last_val > first_val else ''}{last_val - first_val:.4f}\n"
+                    
+                    # Calculate forecast average
+                    avg_fcast = sum(f.get('value', 0) for f in fcast) / fcast_count
+                    context += f"    - Forecast Average: {avg_fcast:.4f}\n"
+    
     return context
 
 def format_stress_context(stress_context: Dict) -> str:
@@ -185,7 +236,8 @@ def format_stress_context(stress_context: Dict) -> str:
 
 def analyze_with_llm(summary_report: Dict, crop_type: str, farmer_context: Dict, 
                      center_lat: float, center_lon: float, field_size_hectares: float,
-                     temporal_stats: Dict = None, stress_context: Dict = None) -> Dict[str, Any]:
+                     temporal_stats: Dict = None, stress_context: Dict = None,
+                     time_series_data: Dict = None) -> Dict[str, Any]:
     """
     Analyze vegetation indices using Gemini LLM and extract soil insights.
     
@@ -198,12 +250,13 @@ def analyze_with_llm(summary_report: Dict, crop_type: str, farmer_context: Dict,
         field_size_hectares: Field size
         temporal_stats: Dictionary with temporal statistics
         stress_context: Dictionary with stress detection results (clustering, anomalies)
+        time_series_data: Dictionary with historical and forecast time series from Flutter cache
     
     Returns:
         Dictionary with structured LLM analysis results
     """
-    # Prepare context with temporal statistics
-    indices_context = prepare_indices_context(summary_report, crop_type, farmer_context, temporal_stats)
+    # Prepare context with temporal statistics and time series data
+    indices_context = prepare_indices_context(summary_report, crop_type, farmer_context, temporal_stats, time_series_data)
     
     # Prepare stress context
     stress_text = format_stress_context(stress_context)