update

app.py

@@ -10,6 +10,7 @@ from PIL import Image
 from smolagents import (
     CodeAgent,
     DuckDuckGoSearchTool,
+    GoogleSearchTool,
     InferenceClientModel,
     load_tool,
     OpenAIServerModel,
@@ -26,17 +27,13 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
 
 @tool 
-def extract_table_from_html(html: str, match: str | None = None) -> list:
+def extract_table_from_html(html: str) -> list:
     """
     A tool that extracts HTML tables from HTML content and returns them as pandas DataFrames. 
     Example usecases include extracting tables from Wikipedia pages, HTML emails, or other web content.
     Args:
         html (str): The HTML content containing HTML tables to extract. This can be raw HTML 
                    string content or a URL to a webpage.
-        match (str | None, optional): A string or regular expression pattern to match 
-                                    against table text content. If None, all tables 
-                                    are extracted. Defaults to None. 
-                                    DO NOT use HTML strings / tags in this parameter.
     
     Returns:
         list: A list of pandas DataFrames, where each DataFrame represents a table found 
@@ -46,10 +43,7 @@ def extract_table_from_html(html: str, match: str | None = None) -> list:
     
     try:
         # Extract tables using pandas
-        if match is not None:
-            tables = pd.read_html(html, match=match)
-        else:
-            tables = pd.read_html(html)
+        tables = pd.read_html(html)
         
         # Return the list of DataFrames directly
         return tables if tables else []
@@ -181,7 +175,7 @@ class BasicAgent:
     def __init__(self):
         print("BasicAgent initialized.")
         self.multimodal_agent = CodeAgent(
-            tools=[VisitWebpageTool(), DuckDuckGoSearchTool(), get_file, audio_to_text],
+            tools=[VisitWebpageTool(), GoogleSearchTool("serper"), get_file, audio_to_text],
             model= OpenAIServerModel(model_id="gpt-4o"),
             additional_authorized_imports=[
                 "requests",
@@ -197,7 +191,8 @@ class BasicAgent:
                 "numpy",
                 "json",
                 "whisper",
-                "openpyxl"
+                "openpyxl",
+                "youtube-transcript-api",
             ],
             name="multimodal_agent",
             description="""
@@ -207,7 +202,7 @@ class BasicAgent:
         ) 
         
         self.code_agent = CodeAgent(
-            tools=[VisitWebpageTool(), DuckDuckGoSearchTool(), get_file, audio_to_text, extract_table_from_html],
+            tools=[VisitWebpageTool(), GoogleSearchTool("serper"), get_file, audio_to_text, extract_table_from_html],
             model=InferenceClientModel(
                 model_id="Qwen/Qwen2.5-Coder-32B-Instruct",
             ),
@@ -229,7 +224,8 @@ class BasicAgent:
                 "chess.engine",
                 "json",
                 "whisper",
-                "openpyxl"
+                "openpyxl",
+                "youtube-transcript-api",
             ],
             name="code_agent",
             description="""
@@ -287,8 +283,7 @@ class BasicAgent:
             Answer the following question (question_id is {question_id}):): 
                "{question}""{file}"
             Please follow hints below:
-                1. `wikipedia` Python package is provided to interact with Wikipedia pages. 
-                2. `chess` Python package is provided. Please use it when there is need to solve chess problems.
+                1. `wikipedia` Python package is provided to interact with Wikipedia pages.
                 3. Please take the question literally! Do not add any additional information or assumptions.
         """
         result = self.manager_agent.run(prompt)

requirements.txt

@@ -15,3 +15,4 @@ numpy
 html5lib
 openai-whisper
 openpyxl
+youtube-transcript-api