add some useful tools

app.py

@@ -3,14 +3,20 @@ import gradio as gr
 import requests
 import pandas as pd
 from smolagents import CodeAgent, OpenAIServerModel, HfApiModel
+
+from tools.analyze_image import AnalyzeImageTool
+from tools.excel_reader import ExcelReader
+from tools.file_reader import FileReader
 from tools.final_answer import FinalAnswerTool
 from tools.visit_webpage import VisitWebpageTool
 from tools.web_search import DuckDuckGoSearchTool
-from tools.youtube_transcript import YoutubeTranscriptTool
+from tools.wikipedia_search import WikipediaSearch
+from tools.youtube_transcript import YouTubeTranscript
+from utils import download_files
 
 # (Keep Constants as is)
 # --- Constants ---
-DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
+from consts import DEFAULT_API_URL
 
 # --- Basic Agent Definition ---
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
@@ -20,8 +26,12 @@ class BasicAgent:
         final_answer = FinalAnswerTool()
         visit_webpage = VisitWebpageTool()
         web_search = DuckDuckGoSearchTool()
-        youtube_transcript = YoutubeTranscriptTool()
-        # model = OpenAIServerModel(model_id="gpt-4.1", api_key=os.getenv("OPENAI_API_KEY"))
+        analyze_image = AnalyzeImageTool()
+        excel_reader = ExcelReader()
+        file_reader = FileReader()
+        wikipedia_search = WikipediaSearch()
+        youtube_transcript = YouTubeTranscript()
+        # model = OpenAIServerModel(model_id="gpt-4.1", api_key=os.getenv("OPENAI_API_KEY"), api_base=os.getenv("OPENAI_BASE_URL"))
         model = HfApiModel(
             max_tokens=2096,
             temperature=0.5,
@@ -30,11 +40,11 @@ class BasicAgent:
         )
         self.agent = CodeAgent(
             model=model,
-            tools=[visit_webpage, web_search, final_answer, youtube_transcript],
+            tools=[visit_webpage, web_search, final_answer, analyze_image, excel_reader, file_reader, wikipedia_search, youtube_transcript],
             max_steps=5,
             verbosity_level=1,
             add_base_tools=True,
-            # additional_authorized_imports=['random', 'stat',  'time', 'collections', 'itertools', 'queue', 're', 'datetime', 'statistics',  'math', 'unicodedata', 'csv', 'pandas']
+            additional_authorized_imports=['random', 'stat',  'time', 'collections', 'itertools', 'queue', 're', 'datetime', 'statistics',  'math', 'unicodedata', 'csv', 'pandas']
         )
     def __call__(self, question: str) -> str:
         print(f"Agent received question (first 50 chars): {question}")
@@ -116,6 +126,11 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
     for item in questions_data:
         task_id = item.get("task_id")
         question_text = item.get("question")
+        file_name = item.get('file_name')
+        if file_name:
+            file_path = download_files(task_id, file_name)
+            file_format = file_name.split('.')[-1]
+            question_text = question_text + f"This question has an associated file at path: {file_path}. The file is in the {file_format} format"
         if not task_id or question_text is None:
             print(f"Skipping item with missing task_id or question: {item}")
             continue

consts.py

@@ -0,0 +1 @@
+DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"

requirements.txt

@@ -7,3 +7,9 @@ requests
 duckduckgo_search
 pandas
 youtube-transcript-api
+bs4
+wikipedia
+tabulate
+llama_index
+llama-index-readers-youtube-transcript
+openai

tools/analyze_image.py

@@ -0,0 +1,51 @@
+import base64
+import os
+from smolagents import Tool
+from openai import OpenAI
+
+class AnalyzeImageTool(Tool):
+    name = "analyze_image_tool"
+    description = """This tool performs a custom analysis of the provided image and returns the corresponding result."""
+    inputs = {
+        "image_path": {"type": "string", "description": "Image path"},
+        "task": {"type": "string", "description": "Task to perform on the image, be detailed and clear"},
+    }
+    output_type = "string"
+
+    def __init__(self):
+        super().__init__()
+        self.model_id = "gpt-4.1-mini"
+
+    def forward(self, image_path: str, task: str) -> str:
+        """
+        Analyze the image at `image_path` according to `task` and return the textual result.
+        """
+        header = "Image analysis result:\n\n"
+        llm_instruction = (
+            "You are a highly capable image analysis tool, designed to examine images and deliver detailed descriptions, "
+            "insights, and relevant interpretations based on the task at hand.\n\n"
+            "Approach the task methodically and provide a thorough and well-reasoned response to the following:\n\n---\nTask:\n"
+            f"{task}\n\n"
+        )
+        try:
+            return header + self._analyze_with_openai(image_path, llm_instruction)
+        except Exception as e:
+            return f"Error analyzing image: {e}."
+
+    def _analyze_with_openai(self, image_path: str, task: str) -> str:
+        client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"), base_url=os.getenv("OPENAI_BASE_URL"))
+
+        with open(image_path, "rb") as f:
+            encoded_image = base64.b64encode(f.read()).decode("utf-8")
+
+        payload = [
+            {
+                "role": "user",
+                "content": [
+                    {"type": "input_text", "text": task},
+                    {"type": "input_image", "image_url": f"data:image/jpeg;base64,{encoded_image}"},
+                ],
+            }
+        ]
+        response = client.responses.create(model=self.model_id, input=payload)
+        return response.output[0].content[0].text

tools/excel_reader.py

@@ -0,0 +1,22 @@
+from smolagents import Tool
+import pandas as pd
+from tabulate import tabulate
+
+class ExcelReader(Tool):
+    name = 'excel_processor'
+    description = "excel reading tool, processed files of .xlsx and .xls format."
+    inputs = {
+        "file_path": {
+            "type": "string",
+            "description": "path to the excel file"
+        }
+    }
+    output_type = "string"
+
+    def forward(self, file_path: str) -> str:
+        try:
+            df = pd.read_excel(file_path)
+            txt_excel = tabulate(df, headers="keys", tablefmt="github", showindex=False)
+            return txt_excel
+        except Exception as e:
+            return f'Error in reading excel file: {str(e)}'

tools/file_reader.py

@@ -0,0 +1,20 @@
+from smolagents import Tool
+
+class FileReader(Tool):
+    name = 'file_reader'
+    description = "reads saved files"
+    inputs = {
+        "file_path": {
+            "type": "string",
+            "description": "path to the file"
+        }
+    }
+    output_type = "string"
+
+    def forward(self, file_path: str) -> str:
+        try:
+            with open(file_path, "r") as file:
+                content = file.read()
+            return content
+        except Exception as e:
+            return f'Error in reading file: {str(e)}'

tools/wikipedia_search.py

@@ -0,0 +1,23 @@
+from smolagents import Tool
+import wikipedia
+from bs4 import BeautifulSoup
+
+class WikipediaSearch(Tool):
+    name = "wikipedia_search"
+    description = "Fetches wikipedia pages."
+    inputs = {
+        "query": {
+            "type": "string",
+            "description": "Query to be searched on wikipedia"
+        }
+        }
+    output_type = "string"
+
+    def forward(self, query:str)->str:
+        try:
+            res = wikipedia.page(query)
+            bs = BeautifulSoup(res.html(), 'html.parser')
+            text_only = bs.get_text()
+            return text_only
+        except Exception as e:
+            return f'Error in wikipedia search: {str(e)}'

tools/youtube_transcript.py

@@ -1,24 +1,22 @@
-from smolagents.tools import Tool
+from smolagents import Tool
+from llama_index.readers.youtube_transcript import YoutubeTranscriptReader
 
-class YoutubeTranscriptTool(Tool):
-    name = "youtube_transcript"
-    description = "This tool allows you to retrieve the transcript/subtitles for a given YouTube video. To use this tool you should pass in the video ID, not the video URL. For a video with the URL https://www.youtube.com/watch?v=12345 the ID is 12345."
-    inputs = {'video_id': {'type': 'string', 'description': 'The Id of YouTube video'}}
-    output_type = "list[dict]"
+class YouTubeTranscript(Tool):
+    name = 'youtube_transcript'
+    description = "a tool that returns a transcript for a youtube video. Youtube videos come from urls containing www.youtube.com"
+    inputs = {
+        "url": {
+            "type": "string",
+            "description": "url to the youtube video, has 'www.youtube.com' in it."
+        }
+    }
+    output_type = "string"
 
-    def __init__(self):
-        super().__init__()
+    def forward(self, url: str) -> str:
         try:
-            from youtube_transcript_api import YouTubeTranscriptApi
-        except ImportError as e:
-            raise ImportError(
-                "You must install package `youtube_transcript_api` to run this tool: for instance run `pip install youtube_transcript_api`."
-            ) from e
-        self.ytt_api = YouTubeTranscriptApi()
-
-    def forward(self, video_id: str) -> list[dict] | str:
-        try:
-            result = self.ytt_api.fetch(video_id)
-            return result.to_raw_data()
+            loader = YoutubeTranscriptReader()
+            documents = loader.load_data(ytlinks=[url])
+            transcript = documents[0].text
+            return transcript
         except Exception as e:
-            return f"An unexpected error occurred: {str(e)}"
+            return f'Error getting youtube transcript: {str(e)}'

utils.py

@@ -0,0 +1,16 @@
+import requests
+import os
+import tempfile
+from pathlib import Path
+from consts import DEFAULT_API_URL
+
+def download_files(task_id, file_name):
+    url = f'{DEFAULT_API_URL}/files/{task_id}'
+    response = requests.get(url, timeout=15)
+    tmp_dir = Path(tempfile.gettempdir()) / "project_files"
+    tmp_dir.mkdir(exist_ok=True)
+    filepath = os.path.join(tmp_dir, file_name)
+    with open(filepath, "wb") as f:
+        f.write(response.content)
+
+    return filepath