Add all indices timeseries + weather + detailed_analysis to LLM

app.py

@@ -135,7 +135,8 @@ class HeatmapRequest(BaseModel):
     gaussian_sigma: float = 1.5
     show_field_boundary: bool = True
     overlay_mode: bool = False  # If True, generate clean heatmap for Google Maps overlay
-    time_series_data: Optional[Dict[str, Any]] = None  # Historical + forecast time series from Flutter cache
+    time_series_data: Optional[Dict[str, Any]] = None  # Historical + forecast time series for ALL indices
+    weather_data: Optional[Dict[str, Any]] = None  # Weather data (temperature, humidity, precipitation)
 
 
 class HeatmapResponse(BaseModel):
@@ -160,6 +161,7 @@ class HeatmapResponse(BaseModel):
     # LLM analysis (for risk metrics)
     level: Optional[str] = None
     analysis: Optional[str] = None
+    detailed_analysis: Optional[str] = None  # Detailed reasoning for timeseries + stress patterns
     stress_score: Optional[float] = None
     cluster_distribution: Optional[dict] = None
     recommendations: Optional[List[str]] = None
@@ -348,8 +350,9 @@ def generate_heatmap_image(data: np.ndarray, index_type: str, gaussian_sigma: fl
 # ============================================================================
 # LLM ANALYSIS (for risk metrics)
 # ============================================================================
-def run_llm_analysis(metric: str, stress_context: dict, indices_data: dict) -> dict:
-    """Call Groq LLM with full context from stress detection."""
+def run_llm_analysis(metric: str, stress_context: dict, indices_data: dict, 
+                     time_series_data: dict = None, weather_data: dict = None) -> dict:
+    """Call Groq LLM with full context from stress detection, timeseries, and weather."""
     try:
         from groq import Groq
         
@@ -361,33 +364,75 @@ def run_llm_analysis(metric: str, stress_context: dict, indices_data: dict) -> d
         # Format stress context
         stress_text = format_stress_context(stress_context)
         
+        # Format time series data for all indices
+        ts_text = ""
+        if time_series_data:
+            ts_text = "\n\nTIME SERIES DATA (ALL INDICES - HISTORICAL + FORECAST):\n"
+            ts_text += "=" * 50 + "\n"
+            for index_name, ts_data in time_series_data.items():
+                ts_text += f"\n{index_name}:\n"
+                # Historical
+                if ts_data.get('historical'):
+                    hist = ts_data['historical']
+                    if len(hist) > 0:
+                        first_val = hist[0].get('value', 0) if isinstance(hist[0], dict) else 0
+                        last_val = hist[-1].get('value', 0) if isinstance(hist[-1], dict) else 0
+                        ts_text += f"  Historical ({len(hist)} points): from {first_val:.4f} to {last_val:.4f} (change: {last_val-first_val:+.4f})\n"
+                # Forecast
+                if ts_data.get('forecast'):
+                    fcast = ts_data['forecast']
+                    if len(fcast) > 0:
+                        first_val = fcast[0].get('value', 0) if isinstance(fcast[0], dict) else 0
+                        last_val = fcast[-1].get('value', 0) if isinstance(fcast[-1], dict) else 0
+                        ts_text += f"  Forecast ({len(fcast)} days): from {first_val:.4f} to {last_val:.4f} (predicted: {last_val-first_val:+.4f})\n"
+        
+        # Format weather data
+        weather_text = ""
+        if weather_data:
+            weather_text = "\n\nWEATHER CONDITIONS:\n"
+            weather_text += "=" * 30 + "\n"
+            if 'temperature' in weather_data:
+                weather_text += f"- Temperature: {weather_data['temperature']}°C\n"
+            if 'humidity' in weather_data:
+                weather_text += f"- Humidity: {weather_data['humidity']}%\n"
+            if 'precipitation' in weather_data:
+                weather_text += f"- Precipitation: {weather_data['precipitation']} mm\n"
+            if 'wind_speed' in weather_data:
+                weather_text += f"- Wind Speed: {weather_data['wind_speed']} km/h\n"
+            if 'conditions' in weather_data:
+                weather_text += f"- Conditions: {weather_data['conditions']}\n"
+            if 'forecast' in weather_data:
+                weather_text += f"- Forecast: {weather_data['forecast']}\n"
+        
         # Create targeted prompt based on metric
-        prompt = f"""
-CROP STRESS ANALYSIS REQUEST
+        prompt = f"""CROP STRESS ANALYSIS REQUEST
 
 {stress_text}
+{ts_text}
+{weather_text}
 
 METRIC TO ANALYZE: {metric.upper().replace('_', ' ')}
 
-Based on the stress detection results above, provide analysis for {metric}.
+Based on the stress detection results, time series trends, and weather conditions above, provide analysis for {metric}.
 
 Respond with ONLY a valid JSON object (no markdown):
 {{
     "level": "Low" or "Moderate" or "High",
     "analysis": "4-5 words describing the current state",
+    "detailed_analysis": "Two detailed sentences: First sentence explaining the reasoning behind time series index changes (what caused the trends). Second sentence explaining the observed stress patterns in the field over time and their likely causes.",
     "temporal_trend": "Improving" or "Stable" or "Worsening",
-    "recommendations": ["action 1", "action 2"]
+    "recommendations": ["action 1", "action 2", "action 3"]
 }}
 """
         
         chat_completion = client.chat.completions.create(
             messages=[
-                {"role": "system", "content": "You are an expert agricultural AI. Respond with valid JSON only."},
+                {"role": "system", "content": "You are an expert agricultural AI. Provide detailed, data-driven analysis. Respond with valid JSON only."},
                 {"role": "user", "content": prompt}
             ],
             model=GROQ_MODEL,
             temperature=0.7,
-            max_tokens=1024,
+            max_tokens=1500,
         )
         
         response_text = chat_completion.choices[0].message.content.strip()
@@ -404,7 +449,12 @@ Respond with ONLY a valid JSON object (no markdown):
         
     except Exception as e:
         logger.error(f"LLM analysis failed: {e}")
-        return {"level": "Moderate", "analysis": "Analysis unavailable", "recommendations": ["Manual inspection recommended"]}
+        return {
+            "level": "Moderate", 
+            "analysis": "Analysis unavailable", 
+            "detailed_analysis": "Unable to generate detailed analysis due to processing error. Please try refreshing.",
+            "recommendations": ["Manual inspection recommended"]
+        }
 
 # ============================================================================
 # API ENDPOINTS
@@ -585,8 +635,12 @@ async def generate_heatmap(request: HeatmapRequest):
             index_func = INDEX_FUNCTIONS[primary_index]
             index_data = index_func(img_data)
             
-            # Run LLM analysis
-            llm_result = run_llm_analysis(request.metric, stress_context, {'primary': index_data})
+            # Run LLM analysis with timeseries and weather context
+            llm_result = run_llm_analysis(
+                request.metric, stress_context, {'primary': index_data},
+                time_series_data=request.time_series_data,
+                weather_data=request.weather_data
+            )
             
             log_step(6, 6, "Generating heatmap")
             
@@ -628,6 +682,7 @@ async def generate_heatmap(request: HeatmapRequest):
                 colorbar_base64=colorbar_b64,
                 level=llm_result.get('level', 'Unknown'),
                 analysis=llm_result.get('analysis', ''),
+                detailed_analysis=llm_result.get('detailed_analysis', ''),
                 stress_score=float(stress_results['stress_scores'].mean()),
                 cluster_distribution=cluster_dist,
                 recommendations=llm_result.get('recommendations', [])