Merge branch 'feat/add-more-tools'

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.mp3 filter=lfs diff=lfs merge=lfs -text
+*.xlsx filter=lfs diff=lfs merge=lfs -text

ShrewdAgent.py

@@ -13,7 +13,12 @@ from langgraph.pregel import PregelProtocol
 from loguru import logger
 from pydantic import SecretStr
 
+from tools.excel_to_text import excel_to_text
+from tools.execute_python_code_from_file import execute_python_code_from_file
+from tools.maths import add_integers
 from tools.produce_classifier import produce_classifier
+from tools.sort_words_alphabetically import sort_words_alphabetically
+from tools.transcribe_audio import transcribe_audio
 from tools.web_page_information_extractor import web_page_information_extractor
 from tools.wikipedia_search import wikipedia_search
 from tools.youtube_transcript import youtube_transcript
@@ -44,6 +49,11 @@ class ShrewdAgent:
             web_page_information_extractor,
             youtube_transcript,
             produce_classifier,
+            sort_words_alphabetically,
+            excel_to_text,
+            execute_python_code_from_file,
+            add_integers,
+            transcribe_audio,
         ]
         self.llm = ChatOpenAI(
             model="gpt-4o-mini",
@@ -60,7 +70,7 @@ class ShrewdAgent:
         logger.info("ShrewdAgent initialized.")
 
     def __call__(self, question: str) -> str:
-        logger.info(f"Agent received question: {question}")
+        logger.info(f"Agent received question:\n{question}")
         accumulated_response = []
         try:
             for chunk in self.agent.stream(
@@ -72,10 +82,10 @@ class ShrewdAgent:
             ):
                 assistant = chunk.get("assistant")
                 if assistant:
-                    logger.debug(f"{assistant.get('messages')[0].pretty_repr()}")
+                    logger.debug(f"\n{assistant.get('messages')[0].pretty_repr()}")
                 tools = chunk.get("tools")
                 if tools:
-                    logger.debug(f"{tools.get('messages')[0].pretty_repr()}")
+                    logger.debug(f"\n{tools.get('messages')[0].pretty_repr()}")
                 accumulated_response.append(chunk)
 
         except GraphRecursionError as e:

app.py

@@ -78,11 +78,13 @@ 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 not task_id or question_text is None:
             print(f"Skipping item with missing task_id or question: {item}")
             continue
         try:
-            submitted_answer = agent(question_text)
+            question_with_attachment = compute_question_with_attachment(question_text, task_id, file_name)
+            submitted_answer = agent(question_with_attachment)
             answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
             results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
         except Exception as e:
@@ -173,6 +175,14 @@ with gr.Blocks() as demo:
         outputs=[status_output, results_table]
     )
 
+
+def compute_question_with_attachment(question: str, task_id: str, file_name: str) -> str:
+    if file_name:
+        return f"{question}\n\nAttached file: https://agents-course-unit4-scoring.hf.space/files/{task_id}"
+    else:
+        return question
+
+
 if __name__ == "__main__":
     print("\n" + "-"*30 + " App Starting " + "-"*30)
     # Check for SPACE_HOST and SPACE_ID at startup for information

data/7bd855d8-463d-4ed5-93ca-5fe35145f733.xlsx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:196b64ea8f72841cb0c4fc972ca82a28e74de926c402bb043305e12e05e012b7
+size 5285

data/99c9cc74-fdc8-46c6-8f8d-3ce2d3bfeea3.mp3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b218c951c1f888f0bbe6f46c080f57afc7c9348fffc7ba4da35749ff1e2ac40f
+size 179304

data/f918266a-b3e0-4914-865d-4faa564f1aef.py

@@ -0,0 +1,36 @@
+import time
+from random import randint
+
+
+class UhOh(Exception):
+    pass
+
+class Hmm:
+    def __init__(self):
+        self.value = randint(-100, 100)
+
+    def Yeah(self):
+        if self.value == 0:
+            return True
+        else:
+            raise UhOh()
+
+def Okay():
+    while True:
+        yield Hmm()
+
+def keep_trying(go, first_try=True):
+    maybe = next(go)
+    try:
+        if maybe.Yeah():
+            return maybe.value
+    except UhOh:
+        if first_try:
+            print("Working...")
+            print("Please wait patiently...")
+        time.sleep(0.1)
+        return keep_trying(go, first_try=False)
+
+if __name__ == "__main__":
+    go = Okay()
+    print(f"{keep_trying(go)}")

requirements.txt

@@ -11,4 +11,6 @@ beautifulsoup4~=4.13.4
 readability-lxml~=0.8.4.1
 youtube-transcript-api~=1.0.3
 wikipedia~=1.4.0
-langchain_tavily~=0.1.6
+langchain_tavily~=0.1.6
+rizaio~=0.11.0
+openai-whisper~=20240930

tools/excel_to_text.py

@@ -0,0 +1,32 @@
+import pandas as pd
+from langchain_core.tools import tool
+from loguru import logger
+
+from tools.load_file import load_file
+
+
+@tool("excel_to_text_tool", parse_docstring=True)
+def excel_to_text(path_or_url: str) -> str:
+    """
+    Read an Excel file and return a text representation of the table data.
+
+    Args:
+        path_or_url (str): Either a local file path to an Excel file or a URL pointing to an Excel file.
+
+    Returns:
+        str: Text representation of the table data
+    """
+    logger.info(f"use excel_to_text with param: {path_or_url}")
+    try:
+        excel = load_file(path_or_url)
+        df = pd.read_excel(excel)
+        return df.to_markdown(index=False)
+
+    except Exception as e:
+        return f"Error reading Excel file: {str(e)}"
+
+
+if __name__ == "__main__":
+    print(excel_to_text.invoke("../data/7bd855d8-463d-4ed5-93ca-5fe35145f733.xlsx"))
+    print(
+        excel_to_text.invoke("https://agents-course-unit4-scoring.hf.space/files/7bd855d8-463d-4ed5-93ca-5fe35145f733"))

tools/execute_python_code_from_file.py

@@ -0,0 +1,38 @@
+from langchain_core.tools import tool, ToolException
+from loguru import logger
+from rizaio import Riza
+
+from tools.load_file import load_file
+
+
+@tool("execute_python_code_from_file_tool", parse_docstring=True)
+def execute_python_code_from_file(path_or_url: str) -> str:
+    """
+    Executes a given Python code snippet.
+
+    Args:
+        path_or_url (str): A string containing the Python code to execute.
+
+    Returns:
+        str: The standard output resulting from the execution of the code.
+    """
+    logger.info(f"execute_python_code_from_file with param: {path_or_url}")
+    code = load_file(path_or_url).read().decode("utf-8")
+    logger.debug(f"Python code:\n{code}")
+    client = Riza()
+    output = client.command.exec(
+        language="python",
+        code=code,
+    )
+
+    if output.exit_code > 0:
+        raise ToolException(
+            f"Riza code execution returned a non-zero exit code. "
+            f"The output captured from stderr was:\n{output.stderr}"
+        )
+    return output.stdout
+
+
+if __name__ == "__main__":
+    print(execute_python_code_from_file.invoke("../data/f918266a-b3e0-4914-865d-4faa564f1aef.py"))
+    print(execute_python_code_from_file.invoke("https://agents-course-unit4-scoring.hf.space/files/f918266a-b3e0-4914-865d-4faa564f1aef"))

tools/load_file.py

@@ -0,0 +1,22 @@
+import io
+from urllib.parse import urlparse
+
+import requests
+
+
+def load_file(path_or_url: str) -> io.BytesIO:
+    if _is_url(path_or_url):
+        response = requests.get(path_or_url)
+        response.raise_for_status()
+        return io.BytesIO(response.content)
+    else:
+        with open(path_or_url, 'rb') as f:
+            return io.BytesIO(f.read())
+
+
+def _is_url(path_or_url):
+    try:
+        result = urlparse(path_or_url)
+        return result.scheme in ('http', 'https')
+    except ValueError:
+        return False

tools/maths.py

@@ -0,0 +1,33 @@
+from typing import List
+
+from langchain_core.tools import tool
+from loguru import logger
+
+
+@tool("add_integers_tool", parse_docstring=True)
+def add_integers(numbers: List[int]) -> int:
+    """
+    Add a list of integers together and return their sum.
+
+    This tool takes a list of integers and calculates their sum. It's useful for
+    performing basic arithmetic operations on multiple numbers at once.
+
+    Args:
+        numbers (List[int]): A list of integers to be summed.
+
+    Returns:
+        int: The sum of all integers in the input list.
+    """
+    logger.info(f"use add_integers with param: {numbers}")
+    if not numbers:
+        raise ValueError("Input list cannot be empty")
+
+    try:
+        return sum(numbers)
+    except TypeError as e:
+        raise TypeError("All elements in the list must be integers") from e
+
+
+if __name__ == "__main__":
+    result = add_integers.invoke({"numbers": [10577, 10037, 10107, 9888, 10089, 10061, 10089, 9917, 10031]})
+    print(result)

tools/sort_words_alphabetically.py

@@ -0,0 +1,26 @@
+from typing import List
+
+from langchain_core.tools import tool
+from loguru import logger
+
+
+@tool("sort_words_alphabetically_tool", parse_docstring=True)
+def sort_words_alphabetically(words: List[str]) -> str:
+    """Sort a list of words alphabetically and join them with commas.
+
+    Args:
+        words (List[str]): List of strings to be sorted
+
+    Returns:
+        str: A string containing all sorted words joined with commas
+    """
+    logger.info(f"use sort_words_alphabetically with param: {words}")
+    sorted_words = sorted(words)
+
+    result = ", ".join(sorted_words)
+
+    return result
+
+
+if __name__ == "__main__":
+    print(sort_words_alphabetically.invoke({"words": ["broccoli", "celery", "sweet potatoes", "lettuce"]}))

tools/transcribe_audio.py

@@ -0,0 +1,42 @@
+import tempfile
+
+import whisper
+from langchain_core.tools import tool
+from loguru import logger
+
+from tools.load_file import load_file
+
+
+@tool("transcribe_audio_tool", parse_docstring=True)
+def transcribe_audio(file_name: str) -> str:
+    """
+    Convert speech from an audio file or URL to text.
+
+    Args:
+        file_name (str): Either a local file path to an audio file or a URL pointing to an audio file.
+
+
+    Returns:
+        str: The transcribed text from the audio file
+    """
+    logger.info(f"use transcribe_audio_tool with param: {file_name}")
+    model_name = "base"
+
+    try:
+        audio_bytes = load_file(file_name)
+        with tempfile.NamedTemporaryFile() as tmp:
+            tmp.write(audio_bytes.getvalue())
+            tmp.flush()
+            model = whisper.load_model(model_name)
+            result = model.transcribe(tmp.name, fp16=False)
+            return result["text"]
+
+    except Exception as e:
+        print(f"Error transcribing audio: {str(e)}")
+        return ""
+
+
+if __name__ == "__main__":
+    print(transcribe_audio.invoke("../data/99c9cc74-fdc8-46c6-8f8d-3ce2d3bfeea3.mp3"))
+    print(transcribe_audio.invoke(
+        "https://agents-course-unit4-scoring.hf.space/files/99c9cc74-fdc8-46c6-8f8d-3ce2d3bfeea3"))