Update app.py

app.py

@@ -5,61 +5,67 @@ import pandas as pd
 from datetime import datetime
 from typing import Optional
 from smolagents import CodeAgent, HfApiModel, tool
+from duckduckgo_search import DDGS
+from PIL import Image
+import markdownify
 
 # === Tool 1: Current Time Tool ===
 @tool
 def get_current_time_in_timezone(timezone: Optional[str] = None) -> str:
-    """
-    Returns the current time. If a timezone is provided, it is noted in the message.
-    Args:
-        timezone: An optional timezone name or location.
-    Returns:
-        A string with the current time.
-    """
     now = datetime.utcnow()
     if timezone:
         return f"The current time in {timezone} is approximately {now.strftime('%Y-%m-%d %H:%M:%S')} UTC."
-    return f"The current UTC time is {now.strftime('%Y-%m-%d %H:%M:%S')}."
+    return f"The current UTC time is {now.strftime('%Y-%m-%d %H:%M:%S')}"
 
 # === Tool 2: CSV Parsing Tool ===
 @tool
 def parse_csv(file_path: str) -> str:
-    """
-    Parses a CSV file and returns the first few rows.
-    Args:
-        file_path: Path to the CSV file to be parsed.
-    Returns:
-        A preview of the first rows.
-    """
     df = pd.read_csv(file_path)
     return df.head(5).to_string(index=False)
 
 # === Tool 3: Audio Transcription Tool ===
 @tool
 def transcribe_audio(audio_path: str) -> str:
-    """
-    Transcribes spoken words in an audio file.
-    Args:
-        audio_path: Path to the .mp3 or .wav file.
-    Returns:
-        A transcription of the audio.
-    """
     import whisper
     model = whisper.load_model("base")
     result = model.transcribe(audio_path)
     return result.get("text", "No transcription result.")
 
-# === Constants ===
+# === Tool 4: Web Search Tool ===
+@tool
+def duckduckgo_search(query: str) -> str:
+    results = []
+    with DDGS() as ddgs:
+        for r in ddgs.text(query):
+            results.append(f"{r['title']} - {r['href']}")
+            if len(results) >= 3:
+                break
+    return "\n".join(results) if results else "No results found."
+
+# === Tool 5: Image Captioning Tool ===
+@tool
+def describe_image_with_blip(image_path: str) -> str:
+    from transformers import BlipProcessor, BlipForConditionalGeneration
+    import torch
+
+    processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
+    model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")
+
+    raw_image = Image.open(image_path).convert('RGB')
+    inputs = processor(raw_image, return_tensors="pt")
+    out = model.generate(**inputs)
+    caption = processor.decode(out[0], skip_special_tokens=True)
+    return caption
+
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
-# === Agent Definition ===
 class BasicAgent:
     def __init__(self):
         model = HfApiModel(model_id="Qwen/Qwen2.5-Coder-32B-Instruct")
         self.agent = CodeAgent(
             model=model,
-            tools=[get_current_time_in_timezone, parse_csv, transcribe_audio],
-            add_base_tools=True,  # Includes vision and web search
+            tools=[get_current_time_in_timezone, parse_csv, transcribe_audio, duckduckgo_search, describe_image_with_blip],
+            add_base_tools=False,
             max_steps=6,
             name="EvalAgent",
             description="Certification agent using vision, search, time, audio, and csv tools"
@@ -69,12 +75,10 @@ class BasicAgent:
     def __call__(self, question: str) -> str:
         response = self.agent.run(question)
         if "Final Answer:" not in response:
-            # Add fallback final answer if missing
             last_line = response.strip().split("\n")[-1]
             response += f"\nFinal Answer: {last_line}"
         return response
 
-# === Evaluation Logic ===
 def run_and_submit_all(profile: gr.OAuthProfile | None):
     space_id = os.getenv("SPACE_ID")
 
@@ -146,13 +150,11 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
             f"Overall Score: {result_data.get('score', 'N/A')}% "
             f"({result_data.get('correct_count', '?')}/{result_data.get('total_attempted', '?')} correct)\n"
             f"Message: {result_data.get('message', 'No message received.')}"
-
         )
         return final_status, pd.DataFrame(results_log)
     except Exception as e:
         return f"Submission Failed: {e}", pd.DataFrame(results_log)
 
-# === Gradio UI ===
 with gr.Blocks() as demo:
     gr.Markdown("# Basic Agent Evaluation Runner")
     gr.Markdown("""