Add Excel and Python file loading tools with tests and dependencies

- Added ExcelLoadTool to load and parse Excel files
- Added PythonLoadTool to load Python source code files
- Updated requirements.txt with necessary dependencies (pandas, openpyxl, etc.)
- Added test cases for the new tools in test_tools.py
- Updated prompt configurations in my_prompt_config.py

.gitignore

@@ -4,3 +4,6 @@
 /venv/
 /.idea/
 /questions/
+/train-data/
+/saved_models/
+/chess-pieces/

app.py

@@ -10,28 +10,28 @@ from my_agent import MyAgent
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
 # testing --------------------------------------------
-testing_mode = True
+testing_mode = False
 questions_to_run = [
-    #"8e867cd7-cff9-4e6c-867a-ff5ddc2550be",      # -        3, should be ok
-    #"a1e91b78-d3d8-4675-bb8d-62741b4b68a6",     # -        video
-    #"2d83110e-a098-4ebb-9987-066c06fa42d0",     # OK
-    #"cca530fc-4052-43b2-b130-b30968d8aa44",     # OK
-    "4fc2f1ae-8625-45b5-ab34-ad4433bc21f8",     # -  FunkMonk, wrong answer https://en.wikipedia.org/wiki/Giganotosaurus
-    #"6f37996b-2ac7-44b0-8e68-6d28256631b4",     # -
-    #"9d191bce-651d-4746-be2d-7ef8ecadb9c2",     # -
-    #"cabe07ed-9eca-40ea-8ead-410ef5e83f91",     # -
-    #"3cef3a44-215e-4aed-8e3b-b1e3f08063b7",     # -
-    #"99c9cc74-fdc8-46c6-8f8d-3ce2d3bfeea3",    # -
-    #"305ac316-eef6-4446-960a-92d80d542f82",    # Hubert ?
-    #"f918266a-b3e0-4914-865d-4faa564f1aef",    # -
-    #"3f57289b-8c60-48be-bd80-01f8099ca449",    # 566 ?
-    #"1f975693-876d-457b-a649-393859e79bf3",    # -
-    #"840bfca7-4f7b-481a-8794-c560c340185d",    # 80NSSC20K0910 ?
-    #"bda648d7-d618-4883-88f4-3466eabd860e",    # Saint Petersburg ?
-    #"cf106601-ab4f-4af9-b045-5295fe67b37d",    # -
-    #"a0c07678-e491-4bbc-8f0b-07405144218f",    # -
-    #"7bd855d8-463d-4ed5-93ca-5fe35145f733",    # -
-    #"5a0c1adf-205e-4841-a666-7c3ef95def9d",    # Vladimir ?
+    #"8e867cd7-cff9-4e6c-867a-ff5ddc2550be",
+    #"a1e91b78-d3d8-4675-bb8d-62741b4b68a6",
+    #"2d83110e-a098-4ebb-9987-066c06fa42d0",
+    #"cca530fc-4052-43b2-b130-b30968d8aa44",
+    #"4fc2f1ae-8625-45b5-ab34-ad4433bc21f8",
+    #"6f37996b-2ac7-44b0-8e68-6d28256631b4",
+    #"9d191bce-651d-4746-be2d-7ef8ecadb9c2",
+    #"cabe07ed-9eca-40ea-8ead-410ef5e83f91",
+    #"3cef3a44-215e-4aed-8e3b-b1e3f08063b7",
+    #"99c9cc74-fdc8-46c6-8f8d-3ce2d3bfeea3",
+    #"305ac316-eef6-4446-960a-92d80d542f82",
+    #"f918266a-b3e0-4914-865d-4faa564f1aef",
+    #"3f57289b-8c60-48be-bd80-01f8099ca449",
+    #"1f975693-876d-457b-a649-393859e79bf3",
+    #"840bfca7-4f7b-481a-8794-c560c340185d",
+    #"bda648d7-d618-4883-88f4-3466eabd860e",
+    #"cf106601-ab4f-4af9-b045-5295fe67b37d",
+    #"a0c07678-e491-4bbc-8f0b-07405144218f",
+    #"7bd855d8-463d-4ed5-93ca-5fe35145f733",
+    #"5a0c1adf-205e-4841-a666-7c3ef95def9d",
 ]
 # testing --------------------------------------------
 
@@ -97,7 +97,6 @@ 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")
-        question_file_name = item.get("file_name")
 
         if not task_id or question_text is None:
             print(f"Skipping item with missing task_id or question: {item}")
@@ -108,11 +107,6 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
                 continue
 
         try:
-            #if question_file_name is not None:
-            #    ext = question_file_name[-4:]
-            #    if ext == ".png":
-            #        question_text = question_text + (f" Use one of available tools to load an image associated with "
-            #                                         f"task_id: '{task_id}'")
             question_text = question_text + f"   task_id: '{task_id}'"
 
             submitted_answer = agent(question_text)
@@ -128,6 +122,7 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
 
     # 4. Prepare Submission 
     submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
+    print(f"Submission data is:\n{submission_data}")
     status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
     print(status_update)
 

my_agent.py

@@ -1,6 +1,14 @@
-from smolagents import CodeAgent, InferenceClientModel, WebSearchTool, FinalAnswerTool
-#from smolagents import ApiModel
-#from mistralai import Mistral
+from smolagents import (
+    CodeAgent,
+    InferenceClientModel,
+    PythonInterpreterTool,  # python_interpreter
+    FinalAnswerTool,        # final_answer
+    VisitWebpageTool,       # visit_webpage
+    WikipediaSearchTool,    # wikipedia_search
+    WebSearchTool,          # web_search
+    PythonInterpreterTool,  # python_interpreter
+    SpeechToTextTool,
+)
 import os
 # --------- my tools
 from my_tool_reverse_string import ReverseStringTool
@@ -15,6 +23,8 @@ from my_tool_wiki_featured_articles import MyWikiFeaturedArticles
 from my_tools_libretexts import MyLibreTextsBookshelvesTool, MyLibreTextsBooksTool
 from my_tools_libretexts import MyLibreTextsBookSectionsTool, MyLibreTextsBookSectionParagraphsTool
 from my_tools_libretexts import MyLibreTextsParagraphContentsTool
+from my_tool_excel_load import ExcelLoadTool
+from my_tool_python_load import PythonLoadTool
 from dotenv import load_dotenv
 
 # https://huggingface.co/docs/transformers/model_doc/mistral?usage=Pipeline
@@ -23,12 +33,17 @@ from dotenv import load_dotenv
 # --- Basic Agent Definition ---
 # ----- THIS IS WHERE YOU CAN BUILD WHAT YOU WANT ------
 class MyAgent:
-    #MODEL_REASONING = "Qwen/Qwen2.5-Coder-32B-Instruct"
-    #MODEL_REASONING = "deepseek-ai/DeepSeek-R1"
-    MODEL_REASONING = "Qwen/Qwen3-235B-A22B"        # test more
-    #MODEL_REASONING = "Qwen/Qwen3-235B-A22B-Thinking-2507"
-    #MODEL_REASONING = "Qwen/Qwen3-235B-A22B-Thinking-2507-FP8"
-    #MODEL_REASONING = "mistralai/Mixtral-8x22B-v0.1"    # not available via HuggingFace
+    #MODEL_REASONING = "deepseek-ai/DeepSeek-V3.1"           # better
+
+    # check again as there were problems with duck duck go
+    MODEL_REASONING = "deepseek-ai/DeepSeek-V3.1-Terminus"  # much worse
+
+    #MODEL_REASONING = "openai/gpt-oss-120b"                 # not working at all
+
+    # check again as there were problems with duck duck go
+    # MODEL_REASONING = "Qwen/Qwen3-Coder-480B-A35B-Instruct"   # worked rather quick
+    #MODEL_REASONING = "deepseek-ai/DeepSeek-R1-0528"
+
     chess_board_model_name = "my_chess_pieces_recognition.pth"
     chess_board_model_dir = "/mnt/c/Users/krzsa/IdeaProjects/Agents-Course-Assignment/saved_models"
 
@@ -39,17 +54,6 @@ class MyAgent:
         self.__create_agents__()
 
     def __create_agents__(self):
-        # --- Mistral -----------------------------------------------------------------------------------
-        # will need class implementation based on ApiModel
-        # https://github.com/huggingface/smolagents/blob/main/src/smolagents/models.py
-        # https://github.com/mistralai/client-python/tree/main/examples
-        #self.mistral_client = Mistral(api_key=self.mistral_api_key)
-        #self.model = ApiModel(
-        #    model_id="mistral-large-2411",
-        #    client=self.mistral_client
-        #)
-        # --- Mistral -----------------------------------------------------------------------------------
-
         # --- HF Inference ------------------------------------------------------------------------------
         self.model = InferenceClientModel(model_id=self.MODEL_REASONING)
         # --- HF Inference ------------------------------------------------------------------------------
@@ -65,8 +69,6 @@ class MyAgent:
             name="CourseAssistant",
             description="General AI Assistant",
             tools=[
-                WebSearchTool(),
-                # -------
                 ImageLoadTool(),
                 ReverseStringTool(),
                 ChessBoard(self.chess_board_model_name, self.chess_board_model_dir),
@@ -80,6 +82,13 @@ class MyAgent:
                 MyLibreTextsBookSectionsTool(),
                 MyLibreTextsBookSectionParagraphsTool(),
                 MyLibreTextsParagraphContentsTool(),
+                PythonInterpreterTool(),
+                VisitWebpageTool(),
+                WikipediaSearchTool(),
+                WebSearchTool(),
+                SpeechToTextTool(),
+                ExcelLoadTool(),
+                PythonLoadTool(),
                 # ---------
                 FinalAnswerTool(),
             ],
@@ -88,6 +97,8 @@ class MyAgent:
             prompt_templates=MyPromptConfig.PROMPT_TEMPLATES,
             managed_agents=[],
             additional_authorized_imports=[
+                "os",
+                "io"
                 "PIL",
                 "chess",
                 "matplotlib",
@@ -108,7 +119,14 @@ class MyAgent:
                 "my_tool_wiki_table_filter",
                 "my_tool_wiki_featured_articles",
                 "my_tools_libretexts",
-                "my_base_libretexts_api"
+                "my_base_libretexts_api",
+                "my_tool_excel_load",
+                "my_tool_python_load",
+                "smolagents",
+                "transformers",
+                "pandas",
+                "openpyxl",
+                "markdownify"
             ],
         )
         print(f"Main agent initialized: {self.reasoning_agent}")

my_prompt_config.py

@@ -11,13 +11,9 @@ class MyPromptConfig:
             
             Answer the following questions as best you can.
             
-            Describe your initial plan as a set of bullet points. 
-            Each bullet point should describe in one sentence an action which is to be taken in this step.
             Do NOT provide hypothetical examples for the final answer.
             
             Use the tools provided, and only those which are necessary to answer the question. 
-            If you are going to use a tool, describe in detail how you are going
-            to invoke the tool and explain parameters used to invoke the tool.
             
             The list of tools available is provided below. They allow to solve a variety of various questions asked.
             Decide and pick and use only tools which are necessary to answer the question asked.
@@ -37,27 +33,131 @@ class MyPromptConfig:
             * _my_tool_libretexts_book_sections : Get a list of sections in a book in LibreTexts materials
             * _my_tool_libretexts_book_section_paragraphs : Get a list of section paragraphs in a book in LibreTexts materials
             * _my_tool_libretexts_paragraph_contents : Get contents of a paragraph in a book in LibreTexts materials
+            * _my_tool_excel_load : load Excel file for given task_id
+            * _my_tool_python_load : load Python file for given task_id
+            * python_interpreter : This is a tool that evaluates python code. It can be used to perform calculations.
+            * visit_webpage : Visits a webpage at the given url and reads its content as a markdown string. Use this to browse webpages.
+            * wikipedia_search : Search Wikipedia and return the summary or full text of the requested article, along with the page URL.
+            * web_search : Performs a web search for your query then returns a string of the top search results.
+            
+            YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of
+            numbers and/or strings.
+            If you are asked for a number, don’t use comma to write your number neither 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
             
             If the question mentions image or other file, use one of provided tools to load it
             using task_id associated with the question.
+            
+            Use the tools using the examples provided below.
+            
+            <code>
+            loaded_image = _my_tool_image_load(task_id="dummy")
+            </code>
+            
+            <code>
+            loaded_python_file = _my_tool_python_load(task_id="dummy")
+            </code>
+            
+            <code>
+            contents = _my_tool_excel_load(task_id="dummy")
+            </code>
+            
+            <code>
+            reversed_string = _my_tool_reverse_string(input_str="dummy")
+            </code>
+            
+            <code>
+            chess_pieces = _my_tool_chess_board(img=loaded_image)
+            </code>
+            
+            <code>
+            loaded_fen = _my_tool_fen(chest_pieces="dummy")
+            </code>
+            
+            <code>
+            best_answer = _my_tool_chess_analysis(fen=loaded_fen, player_color="dummy")
+            </code>
+            
+            <code>
+            section_contents = _my_tool_wiki_page_section(page_title='aaa', section_title='bbb', version='2022')
+            </code>
+            
+            <code>
+            table_data = _my_tool_wiki_table_filter(section_content='aaa', sub_section_name='bbb', year_start='1998', year_end='2005')
+            </code>
+            
+            <code>
+            featured_articles = _my_tool_wiki_featured_articles(month='January', year=2005)
+            </code>
+            
+            <code>
+            bookshelves = _my_tool_libretexts_bookshelves() 
+            </code>
+            
+            <code>
+            books = _my_tool_libretexts_books(bookshelf_url='https://chem.libretexts.org/Bookshelves/Inorganic_Chemistry')
+            </code>
+            
+            <code>
+            book_sections = _my_tool_libretexts_book_sections(book_url='https://chem.libretexts.org/Bookshelves/Inorganic_Chemistry/Inorganic_Chemistry_(Saito)')
+            </code>
+            
+            <code>
+            book_section_paragraphs = _my_tool_libretexts_book_section_paragraphs(section_url='https://chem.libretexts.org/Bookshelves/Inorganic_Chemistry/Inorganic_Chemistry_(Saito)/03%3A_Reactions')
+            </code>
+            
+            <code>
+            paragraph_contents = _my_tool_libretexts_paragraph_contents(paragraph_url='https://chem.libretexts.org/Bookshelves/Inorganic_Chemistry/Inorganic_Chemistry_(Saito)/03%3A_Reactions/3.01%3A_Thermodynamics')
+            </code>
+            
+            <code>
+            results = web_search("question")
+            </code>
+            
+            <code>
+            result = visit_webpage("https://www.site1.com/page1")
+            </code>
+            
+            <code>
+            import os
+            print("Some information") 
+            </code>
+            
+            <code>
+            import os
+            print(\"\"\"
+                Some information
+                which has
+                many lines
+            \"\"\") 
+            </code>
+            
+            <code>
+            final_answer("1")
+            </code>
+            
+            <code>
+            final_answer("Word")
+            </code>
+            
+            <code>
+            final_answer("list,of,words")
+            </code>
+            
+            
         """,
         planning=PlanningPromptTemplate(
-            initial_plan="""
-                
-            """,
-            update_plan_pre_messages="""
-                
-            """,
-            update_plan_post_messages="""
-                
-            """,
+            initial_plan="",
+            update_plan_pre_messages="",
+            update_plan_post_messages="",
         ),
         managed_agent=ManagedAgentPromptTemplate(task="", report=""),
         final_answer=FinalAnswerPromptTemplate(
-            pre_messages="",
-            post_messages="""
-            Report your answer using final_answer tool
-            
+            pre_messages="""
             YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of
             numbers and/or strings.
             If you are asked for a number, don’t use comma to write your number neither use units such as $ or 
@@ -66,7 +166,16 @@ class MyPromptConfig:
             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
-            """
+            
+            Report your answer using final_answer tool as below
+            
+            <code>
+            final_answer("YOUR FINAL ANSWER")
+            </code>
+            
+
+            """,
+            post_messages="",
         ),
     )
 

my_tool_excel_load.py

@@ -0,0 +1,50 @@
+from smolagents import Tool
+import requests
+import pandas as pd
+from io import BytesIO
+from typing import Any
+
+#AUTHORIZED_TYPES = [
+#    "string",
+#    "boolean",
+#    "integer",
+#    "number",
+#    "image",
+#    "audio",
+#    "array",
+#    "object",
+#    "any",
+#    "null",
+#]
+
+
+class ExcelLoadTool(Tool):
+    name = "_my_tool_excel_load"
+    description = """
+        Load Excel spreadsheet for the provided task id
+        To invoke the tool use code as below
+        <code>
+        contents = _my_tool_excel_load(task_id="dummy")
+        </code>
+    """
+
+    inputs = {
+        "task_id": {
+            "type": "string",
+            "description": "task id to load excel file",
+        }
+    }
+
+    output_type = "any"
+    api_url = "https://agents-course-unit4-scoring.hf.space"
+
+    def forward(self, task_id: str) -> Any:
+        headers = {
+            "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/114.0.0.0 Safari/537.36"
+        }
+        url = f"{self.api_url}/files/{task_id}"
+        response = requests.get(url, headers=headers)
+        bt = BytesIO(response.content)
+        df = pd.read_excel(bt)
+        #print(f"***KS*** Loaded Excel file contents for \n\ttask id: {task_id} \n\timage: {df}")
+        return df

my_tool_python_load.py

@@ -0,0 +1,46 @@
+from smolagents import Tool
+import requests
+from io import BytesIO
+from typing import Any
+
+#AUTHORIZED_TYPES = [
+#    "string",
+#    "boolean",
+#    "integer",
+#    "number",
+#    "image",
+#    "audio",
+#    "array",
+#    "object",
+#    "any",
+#    "null",
+#]
+
+
+class PythonLoadTool(Tool):
+    name = "_my_tool_python_load"
+    description = """
+        Load file with Python source code for the provided task id
+        To invoke the tool use code as below
+        <code>
+        contents = _my_tool_python_load(task_id="dummy")
+        </code>
+    """
+
+    inputs = {
+        "task_id": {
+            "type": "string",
+            "description": "task id to load Python source code file",
+        }
+    }
+
+    output_type = "string"
+    api_url = "https://agents-course-unit4-scoring.hf.space"
+
+    def forward(self, task_id: str) -> Any:
+        headers = {
+            "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/114.0.0.0 Safari/537.36"
+        }
+        url = f"{self.api_url}/files/{task_id}"
+        response = requests.get(url, headers=headers)
+        return response.content.decode('utf-8')

requirements.txt

@@ -1,6 +1,7 @@
 gradio
 requests
 smolagents
+smolagents[transformers]
 gradio[oauth]
 pytest
 matplotlib
@@ -13,4 +14,8 @@ wikitextparser
 mwparserfromhell
 lxml
 cssselect
-html-to-markdown
+html-to-markdown
+transformers
+pandas
+openpyxl
+markdownify

test_tools.py

@@ -5,14 +5,33 @@ from my_tool_fen import FENTool
 from my_tool_chess_analysis import ChessAnalysisTool
 from my_tool_wiki_page_section import MyWikiPageSectionTool
 from my_tool_wiki_filter_tables import MyWikiTableFilterTool
-from my_base_wiki_api import MyWikiAPI
 from my_tool_wiki_featured_articles import MyWikiFeaturedArticles
 from my_base_libretexts_api import MyLibreTextsAPI
+from my_tool_excel_load import ExcelLoadTool
+from my_tool_python_load import PythonLoadTool
 import pytest
 import matplotlib as mp
 
 #pytest --capture=no
 
+@pytest.mark.parametrize("_task_id,_exp", [("f918266a-b3e0-4914-865d-4faa564f1aef", "")])
+def test_tool_python_load(_task_id,_exp):
+    print(f"\nLoading Python file for task id: {_task_id}")
+    t = PythonLoadTool()
+    result = t.forward(_task_id)
+    print(f"Got result:\n{result}")
+    assert True
+
+@pytest.mark.skip(reason="disabled")
+@pytest.mark.parametrize("_task_id,_exp", [("7bd855d8-463d-4ed5-93ca-5fe35145f733", "")])
+def test_tool_excel_load(_task_id,_exp):
+    print(f"\nLoading Excel file for task id: {_task_id}")
+    t = ExcelLoadTool()
+    result = t.forward(_task_id)
+    print(f"Got result:\n{result}")
+    assert True
+
+@pytest.mark.skip(reason="disabled")
 @pytest.mark.parametrize("_exp", [("Introductory")])
 def test_tool_libretextx_bookshelves(_exp):
     api = MyLibreTextsAPI()