Update app.py

app.py

@@ -3,6 +3,12 @@ import gradio as gr
 import requests
 import inspect
 import pandas as pd
+from smolagents import CodeAgent,DuckDuckGoSearchTool, HfApiModel,Tool
+from transformers import load_tool, HfAgent
+import pdfplumber
+import pytesseract
+from PIL import Image
+from io import BytesIO
 
 # (Keep Constants as is)
 # --- Constants ---
@@ -53,14 +59,78 @@ class SimplifyAnswerTool(Tool):
 
 # --- Basic Agent Definition ---
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
-class BasicAgent:
+def extract_text_from_pdf(pdf_url: str) -> str:
+    try:
+        r = requests.get(pdf_url)
+        r.raise_for_status()
+        with pdfplumber.open(BytesIO(r.content)) as pdf:
+            text = "\n".join(page.extract_text() or "" for page in pdf.pages)
+        return text
+    except Exception as e:
+        print(f"Error extracting PDF text: {e}")
+        return ""
+
+# Helper to extract text from image url using OCR
+def extract_text_from_image(image_url: str) -> str:
+    try:
+        r = requests.get(image_url)
+        r.raise_for_status()
+        img = Image.open(BytesIO(r.content))
+        text = pytesseract.image_to_string(img)
+        return text.strip()
+    except Exception as e:
+        print(f"Error extracting image text: {e}")
+        return ""
+
+class OpalAgent:
     def __init__(self):
-        print("BasicAgent initialized.")
-    def __call__(self, question: str) -> str:
-        print(f"Agent received question (first 50 chars): {question[:50]}...")
-        fixed_answer = "This is a default answer."
-        print(f"Agent returning fixed answer: {fixed_answer}")
-        return fixed_answer
+        # Use a small, fast model to save tokens/costs
+        model_name = "tiiuae/falcon-7b-instruct"  # Falcon-7B-Instruct is open and efficient
+        
+        self.llm = HfApiModel(model_name, token=HF_TOKEN)
+        
+        # Load tools - search, QA, image captioning
+        self.tools = [
+            DuckDuckGoSearchTool(),
+            load_tool("question-answering"),     # for text QA
+            load_tool("image-captioning"),       # for images
+            load_tool("image-question-answering")# for image QA
+        ]
+        
+        # Add answer simplification tool
+        self.simplify_tool = SimplifyAnswerTool(self.llm)
+        
+        self.agent = CodeAgent(
+            llm=self.llm,
+            tools=self.tools + [self.simplify_tool],
+            verbose=False,
+        )
+    
+    def __call__(self, question: str, extra_data: dict = None) -> str:
+        context = question
+        
+        # If PDF or image in extra data, preprocess and add to context
+        if extra_data:
+            if "pdf_url" in extra_data:
+                pdf_text = extract_text_from_pdf(extra_data["pdf_url"])
+                context += f"\n\n[PDF CONTENT]: {pdf_text[:1000]}"  # limit length
+            
+            if "image_url" in extra_data:
+                # Use OCR text to add context or just mention image URL
+                img_text = extract_text_from_image(extra_data["image_url"])
+                if img_text:
+                    context += f"\n\n[IMAGE TEXT]: {img_text}"
+                else:
+                    context += f"\n\n[IMAGE URL]: {extra_data['image_url']}"
+        
+        try:
+            raw_answer = self.agent(context)
+            # Simplify final answer
+            simple_answer = self.simplify_tool(question=question, answer=raw_answer)
+            return simple_answer.strip()
+        except Exception as e:
+            return f"AGENT ERROR: {e}"
+
 
 def run_and_submit_all( profile: gr.OAuthProfile | None):
     """