Update agent.py

agent.py

@@ -1,4 +1,6 @@
 import os
+import io
+import pandas as pd
 import requests
 from openai import OpenAI
 
@@ -6,35 +8,48 @@ class GaiaAgent:
     def __init__(self):
         self.client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
         self.instructions = (
-            "You are a highly skilled research assistant solving GAIA benchmark questions. "
-            "You can analyze documents, perform reasoning, and answer with a single factual answer only."
+            "You are an expert assistant solving GAIA benchmark questions. "
+            "You analyze file contents (like CSV), reason step-by-step, and respond with a single factual answer."
         )
         self.api_url = "https://agents-course-unit4-scoring.hf.space"
 
-    def fetch_file_content(self, task_id: str) -> str:
+    def fetch_file_context(self, task_id: str) -> str:
         try:
             url = f"{self.api_url}/files/{task_id}"
             response = requests.get(url, timeout=10)
             response.raise_for_status()
-
             content_type = response.headers.get("Content-Type", "")
-            if "text" in content_type or "csv" in content_type or "json" in content_type:
-                return response.text[:3000]  # Truncate to fit token limit
+
+            if "text/csv" in content_type or url.endswith(".csv"):
+                df = pd.read_csv(io.StringIO(response.text))
+                if df.shape[1] <= 15 and df.shape[0] <= 30:
+                    return f"CSV table preview:\n{df.to_markdown(index=False)}"
+                else:
+                    return f"CSV summary: {df.shape[0]} rows, {df.shape[1]} columns.\nColumns: {', '.join(df.columns[:10])}"
+
+            elif "application/json" in content_type:
+                return f"JSON content:\n{response.text[:2000]}"
+
             elif "application/pdf" in content_type:
-                return "[PDF file content detected. Summarize manually or use tool.]"
+                return "[PDF detected. You may need to request OCR summary.]"
+
+            elif "text/plain" in content_type:
+                return f"File preview:\n{response.text[:2000]}"
+
             else:
                 return f"[Unsupported file type: {content_type}]"
+
         except Exception as e:
-            return f"[Error downloading or reading file: {e}]"
+            return f"[Error downloading or processing file: {e}]"
 
     def __call__(self, question: str, task_id: str = None) -> str:
         file_context = ""
         if task_id:
-            file_context = self.fetch_file_content(task_id)
+            file_context = self.fetch_file_context(task_id)
             if file_context:
-                file_context = f"Here is the related file content:\n{file_context}\n"
+                file_context = f"FILE CONTEXT:\n{file_context}\n"
 
-        prompt = f"{self.instructions}\n\n{file_context}Question: {question}\nAnswer:"
+        prompt = f"{self.instructions}\n\n{file_context}QUESTION:\n{question}\nANSWER:"
 
         response = self.client.chat.completions.create(
             model="gpt-4-turbo",