Update app.py

app.py

@@ -3,6 +3,21 @@ import gradio as gr
 import requests
 import inspect
 import pandas as pd
+from smolagents import CodeAgent, InferenceClientModel, tool
+from smolagents import ActionStep, PlanningStep, ToolCall, Tool
+from smolagents import (
+    CodeAgent,
+    DuckDuckGoSearchTool,
+    VisitWebpageTool,
+    WikipediaSearchTool,
+    OpenAIServerModel,
+    SpeechToTextTool,
+    FinalAnswerTool,
+)
+import yaml, importlib, requests, json, os, base64, re
+import wikipediaapi
+from typing import List, Dict, Any, Union, Optional
+from openai import OpenAI
 
 # (Keep Constants as is)
 # --- Constants ---
@@ -10,14 +25,187 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
 # --- Basic Agent Definition ---
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
-class BasicAgent:
-    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
+model = OpenAIServerModel(
+                model_id="gpt-4.1-mini",
+                api_key=userdata.get('OPENAI_API_KEY'),
+        )
+# PROMPTS
+system_prompt = """You are a general AI assistant. I will ask you a question.
+YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings.
+If your answer is a number and you are not explicitly asked for a string, write it in numerals instead of words, and don't use comma to write your number nor use units such as $ or percent sign unless specified otherwise.
+If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise.
+If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string.
+
+Answer questions as literally as you can, making as few assumptions as possible. Restrict the answer to the narrowest definition that still satifies the question.
+If you are provied with a video, please watch and summarize the entire video before answering the question. The correct answer may be present only in a few frames of the video.
+If you are asked to prove something, first state your assumptions and think step by step before giving your final answer.
+"""
+req_instruction = (
+    "You are a highly capable and autonomous agent named {{name}}, designed to solve complex tasks efficiently.\n"
+    "A valued client has assigned you the following task:\n"
+    "---\n"
+    "Task:\n"
+    "{{task}}\n"
+    "---\n"
+    "To complete this task successfully, follow these steps carefully:\n"
+    "    1. Comprehend the task and identify the intended goal.\n"
+    "    2. Break the task into clear, logical steps.\n"
+    "    3. Select and prepare the tools or resources you need.\n"
+    "    - If a tool does not return useful results on the first attempt, consider retrying it with a simpler, more general, or slightly modified input.\n"
+    "       Avoid switching to a different tool too quickly unless clearly necessary.\n"
+    "    4. Set up the required environment or context.\n"
+    "    5. Execute each step methodically.\n"
+    "    6. Monitor outcomes and identify any deviations.\n"
+    "    7. Revise your plan if necessary based on feedback.\n"
+    "    8. Maintain internal state and track progress.\n"
+    "    9. Verify that the goal has been fully achieved.\n"
+    "   10. Present the final result clearly and concisely.\n"
+    "If you succeed, you will be rewarded with a significant bonus.\n\n"
+    "Your final_answer MUST be:\n"
+    "- a number (retain its original type; do not include units),\n"
+    "- a concise phrase,\n"
+    "- or a comma-separated list of numbers or strings, with a space after each comma (e.g., \"1, 2, 3\", not \"1,2,3\"; do not include articles or abbreviations).\n\n"
+    "Only the content passed to the final_answer tool will be preserved. Any other content will be discarded."
+)
+
+prompts = yaml.safe_load(
+    importlib.resources.files("smolagents.prompts").joinpath("code_agent.yaml").read_text()
+)
+prompts['managed_agent']['task'] = req_instruction
+prompts['managed_agent']['report'] = "{{final_answer}}"
+
+# Tools
+@tool
+def wikipedia_df_tool(query: str) -> List[pd.DataFrame]:
+    """
+    Use this tool first for Wikipedia searches, before switching to text-based tools, and retry this tool if no results are found.
+
+    Retrieve useful tabular data from English Wikipedia.
+
+    This tool searches for HTML tables on the Wikipedia page matching the given query,
+    and returns them as a list of Pandas DataFrames. It always returns a list — if no tables
+    are found, the list will be empty.
+
+    If no results are found, retry this tool with a more general, simpler, or alternative version of the query.
+    Examples of simplifications: removing terms like 'discography', using only the person's name, or trying keywords like 'albums', 'list', or 'table'.
+
+
+    Args:
+        query: A Wikipedia page title or related phrase (e.g., "Argentina", "Mercedes Sosa discography").
+    """
+    wiki = wikipediaapi.Wikipedia(user_agent='MyProjectName ([email protected])', language='en')
+    wiki_page = wiki.page(query)
+    try:
+      url = wiki_page.fullurl
+    except Exception:
+      return []
+    dfs = pd.read_html(url)
+    return dfs if dfs else []
+
+@tool
+def get_file_from_task(task_id: str, file_name: str) -> str:
+  """
+  Use this tool to download the file content associated with the given task_id if exists.
+  Returns absolute file path.
+  Args:
+        task_id: The unique identifier of the task whose associated file should be downloaded. 
+        This is used to locate the file on the server via the API endpoint.
+        file_name: The desired name (or path) to save the downloaded file locally. 
+        This will be the name of the file written to disk.
+  Returns:
+        The absolute path to the downloaded file saved on the local filesystem.
+  """
+  response = requests.get(f"{DEFAULT_API_URL}/files/{task_id}", timeout=15)
+  response.raise_for_status()
+  with open(file_name, 'wb') as file:
+    file.write(response.content)
+  return os.path.abspath(file_name)
+
+@tool
+def load_text_file(file_path: str) -> str:
+  """
+    Reads and returns the content of a UTF-8 encoded text file.
+
+    Args:
+        file_path (str): Path to the file to be read.
+
+    Returns:
+        str: The content of the file as a string.
+    """
+  with open(file_path, 'r', encoding='utf-8') as file:
+    return file.read()
+
+
+
+@tool
+def analyze_image(image_path: str, task: str) -> str:
+    """
+    Analyzes the image at the given path using OpenAI's vision model,
+    based on the provided task description.
+
+    Args:
+        image_path (str): Path to the image file.
+        task (str): Task to perform on the image (e.g., describe, interpret, extract data).
+
+    Returns:
+        str: Result of the analysis as a string.
+    """
+    client = OpenAI(api_key=userdata.get('OPENAI_API_KEY'))
+
+    with open(image_path, "rb") as f:
+        encoded_image = base64.b64encode(f.read()).decode("utf-8")
+
+    
+
+    prompt = (
+        "You are an expert image analysis tool. Please examine the following image and perform the task:\n\n"
+        f"{task}"
+    )
+
+    response =  client.responses.create(
+        model="gpt-4.1-mini",
+        input=[
+            {"role": "user", "content": [
+                {"type": "input_text", "text": prompt},
+                {"type": "input_image", "image_url": f"data:image/jpeg;base64,{encoded_image}"}
+            ]}
+        ]
+    )
+
+    return response.output_text
+
+def summarize_steps(agent):
+  summary = []
+  for step in agent.memory.steps:
+    if isinstance(step, ActionStep):
+      args = step.tool_calls[0].arguments.strip().replace('\n', ' ')
+      summary.append({'Step': step.step_number, 'Summary': args})
+  return summary
+
+mp3_to_text_tool = Tool.from_space(
+    "mrfakename/fast-whisper-turbo",
+    name="voice_to_text",
+    description="Transcribes an English audio file into text. Returns the transcribed text.",
+    api_name="/transcribe"
+)
+
+agent = CodeAgent(
+        tools=[
+            DuckDuckGoSearchTool(),
+            VisitWebpageTool(),
+            WikipediaSearchTool(),
+            mp3_to_text_tool,
+            FinalAnswerTool(),
+            wikipedia_df_tool,
+            get_file_from_task,
+            load_text_file,
+            analyze_image,
+        ],
+        model=model,
+        prompt_templates=prompts,
+        additional_authorized_imports = ["pandas", "requests", "BeautifulSoap"],
+        name="tikito",
+    )
 
 def run_and_submit_all( profile: gr.OAuthProfile | None):
     """
@@ -40,7 +228,7 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
 
     # 1. Instantiate Agent ( modify this part to create your agent)
     try:
-        agent = BasicAgent()
+        agent = agent
     except Exception as e:
         print(f"Error instantiating agent: {e}")
         return f"Error initializing agent: {e}", None