fed LLM directly with raw data for now.

llm_prompts.py

@@ -241,25 +241,8 @@ def get_synthesis_report_prompt(query, quantitative_data, qualitative_data, plan
     """
     Generates the prompt for synthesizing a final report from the query results.
     """
-    qualitative_prompt_str = ""
-    dimension = plan.get('analysis_dimension', 'N/A')
     query_filter = plan.get('query_filter', 'Not available') # Extract query filter from the plan
-    if qualitative_data and dimension in qualitative_data:
-        for group in qualitative_data.get(dimension, {}).get('groups', []):
-            group_value = group.get('groupValue', 'N/A')
-            if group.get('doclist', {}).get('docs'):
-                doc = group.get('doclist', {}).get('docs', [{}])[0]
-                title = doc.get('abstract', ['No Title'])
-                content_list = doc.get('content', [])
-                content_snip = (' '.join(content_list[0].split()[:40]) + '...') if content_list else 'No content available.'
-                metric_val_raw = doc.get(plan.get('sort_field_for_examples'), 'N/A')
-                metric_val = metric_val_raw[0] if isinstance(metric_val_raw, list) else metric_val_raw
-
-                qualitative_prompt_str += f"- **For category `{group_value}`:**\n"
-                qualitative_prompt_str += f"  - **Top Example Title:** {title}\n"
-                qualitative_prompt_str += f"  - **Metric Value:** {metric_val}\n"
-                qualitative_prompt_str += f"  - **Content Snippet:** {content_snip}\n\n"
-
+    
     return f"""
 You are a top-tier business intelligence analyst. Your task is to write an insightful, data-driven report for an executive. You must synthesize quantitative data (the 'what') with qualitative examples (the 'why') to tell a complete story.
 
@@ -280,8 +263,10 @@ This data shows the high-level aggregates based on the filters above.
 ```
 
 **4. Qualitative Data (The 'Why'):**
-These are the single most significant documents driving the numbers for each category.
-{qualitative_prompt_str}
+These are the most significant documents driving the numbers for each category, presented as raw JSON from Solr's grouping feature.
+```json
+{json.dumps(qualitative_data, indent=2)}
+```
 
 ---
 ### REPORTING INSTRUCTIONS
@@ -298,7 +283,7 @@ Your report must be in clean, professional Markdown and follow this structure pr
 
 `### Key Drivers & Illustrative Examples`
 - **This is the most important section.** Explain the "so what?" behind the numbers.
-- Use the qualitative examples to explain *why* a category is high or low. Reference the top example document for each main category.
+- Use the qualitative examples from the raw JSON to explain *why* a category is high or low. For each main category, reference the top example document's title (e.g., 'abstract' field) and key metrics.
 
 `### Deeper Dive: Suggested Follow-up Analyses`
 - Propose 2-3 logical next questions based on your analysis to uncover deeper trends.
@@ -411,4 +396,4 @@ plt.tight_layout()
 
 Now, generate the raw Python code to create the best possible visualization for the user's goal based on the provided data.
 Do not wrap the code in ```python ... ```.
-"""
+"""