Create src/chimera/utils/data_processing.py

src/chimera/utils/data_processing.py

@@ -0,0 +1,42 @@
+# src/chimera/utils/data_processing.py
+from .logging_config import logger
+import json
+
+def format_api_data_for_llm(api_results: dict) -> str:
+    """
+    Formats the collected data from various APIs into a string suitable for an LLM prompt.
+    """
+    formatted_string = ""
+
+    if "serp" in api_results and isinstance(api_results["serp"], dict):
+        serp_data = api_results["serp"]
+        if "error" in serp_data:
+             formatted_string += f"SERP API Error: {serp_data['error']}\n\n"
+        elif serp_data.get("organic_results"):
+            formatted_string += "Recent Search Results:\n"
+            for i, result in enumerate(serp_data["organic_results"][:5]): # Limit results
+                title = result.get("title", "N/A")
+                link = result.get("link", "N/A")
+                snippet = result.get("snippet", "N/A")
+                formatted_string += f"- Title: {title}\n  Snippet: {snippet}\n  Link: {link}\n"
+            formatted_string += "\n"
+        else:
+            formatted_string += "SERP: No relevant organic results found or data format unexpected.\n\n"
+
+
+    # Add formatting for other APIs (weather, finance, etc.) similarly
+    # Example for a hypothetical weather API result:
+    # if "weather" in api_results and isinstance(api_results["weather"], dict):
+    #     weather_data = api_results["weather"]
+    #     if "error" in weather_data:
+    #            formatted_string += f"Weather API Error: {weather_data['error']}\n\n"
+    #     else:
+    #         temp = weather_data.get('temperature', 'N/A')
+    #         desc = weather_data.get('description', 'N/A')
+    #         location = weather_data.get('location', 'N/A')
+    #         formatted_string += f"Weather in {location}:\n - Temperature: {temp}\n - Conditions: {desc}\n\n"
+
+    if not formatted_string:
+        return "No structured data was successfully retrieved from APIs."
+
+    return formatted_string.strip()