aggiunti tool: wikipedia_search_3, download_taskid_file

agent.py

@@ -13,7 +13,7 @@ from langgraph.graph import StateGraph, END
 from langchain_google_genai import ChatGoogleGenerativeAI
 
 # Local imports
-from tools import get_search_tool, get_wikipedia_tool, wikipedia_search, execute_python_code_from_file
+from tools import get_search_tool, get_wikipedia_tool, wikipedia_search, wikipedia_search_3, execute_python_code_from_file, download_taskid_file
 
 # Nota: per i test in locale si usa il .env
 #       su HuggingFace invece si usano le variabili definite in Settings/"Variables and secrets"
@@ -25,13 +25,23 @@ GEMINI_TEMPERATURE = float(os.environ.get("GEMINI_TEMPERATURE"))
 
 TOOLS_CALL_DELAY = 1.5
 
-GENERAL_AGENT_INSTRUCTIONS = """You are a general AI assistant. I will ask you a question.
+GENERAL_AGENT_INSTRUCTIONS = """You are a helpful assistant tasked with answering questions using a set of tools.
+Now, I will ask you a question. Report your thoughts, and provide your answer.
 Your 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.
 Provide only the answer, without notes, explanations or comments."""
 
+# GENERAL_AGENT_INSTRUCTIONS = """
+# You are a general AI assistant. Your purpose is to answer questions and complete tasks accurately and concisely.
+# You have access to various tools to help you gather information and perform actions.
+# Always prioritize using your tools to find factual information if a question requires it.
+# If a question can be answered directly from your knowledge, do so.
+# If you use a tool, provide only the direct result or answer based on the tool's output.
+# Do not include any conversational filler, explanations of your thought process, or pleasantries unless specifically asked.
+# """
+
 #
 # Inizializza il modello e gli associa i tool
 #
@@ -47,7 +57,7 @@ chat = ChatGoogleGenerativeAI(
 search_tool = get_search_tool()
 #wikipedia_tool = get_wikipedia_tool()
 
-tools = [search_tool, wikipedia_search, execute_python_code_from_file]
+tools = [search_tool, wikipedia_search_3, execute_python_code_from_file, download_taskid_file]
 
 # Bind tools to the model
 chat_with_tools = chat.bind_tools(tools)

app.py

@@ -85,6 +85,12 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
         if not task_id or question_text is None:
             print(f"Skipping item with missing task_id or question: {item}")
             continue
+
+        # Se c'e' un file associato, agginge alla question file e task_id e suggerisce il tool da utilizzare per scaricarlo
+        file_name = item.get("file_name")
+        if file_name is not None:
+            question_text += " file_name: {}, task:id: {} (use tool 'download_taskid_file' to get it)".format(file_name, task_id)
+
         try:
             # submitted_answer = agent(question_text)
             messages = [HumanMessage(content = question_text)]

tools.py

@@ -1,9 +1,11 @@
 import os, sys
 from dotenv import load_dotenv
+import requests
 
 from langchain_community.tools import DuckDuckGoSearchRun
 from langchain_community.utilities import WikipediaAPIWrapper
 from langchain_community.tools import WikipediaQueryRun
+from langchain_community.document_loaders import WikipediaLoader
 import wikipedia
 from langchain_core.tools import tool
 from langchain.tools import Tool
@@ -12,6 +14,8 @@ from langchain.tools import Tool
 import subprocess
 
 
+DATASET_API_URL = 'https://agents-course-unit4-scoring.hf.space'
+
 load_dotenv()
 WIKIPEDIA_TOP_K_RESULTS = int(os.environ.get("WIKIPEDIA_TOP_K_RESULTS"))
 WIKIPEDIA_DOC_CONTENT_CHARS_MAX = int(os.environ.get("WIKIPEDIA_DOC_CONTENT_CHARS_MAX"))
@@ -24,6 +28,8 @@ def get_search_tool():
     return search_tool
 
 
+# Wikipedia tool 1: usa WikipediaQueryRun dal package 'langchain_community.tools'
+# problema: sembra ottenere solo i summary
 def get_wikipedia_tool():
 
     #print("WIKIPEDIA_TOP_K_RESULTS:{}, WIKIPEDIA_DOC_CONTENT_CHARS_MAX:{}".format(WIKIPEDIA_TOP_K_RESULTS, WIKIPEDIA_DOC_CONTENT_CHARS_MAX))
@@ -37,6 +43,7 @@ def get_wikipedia_tool():
     return wikipedia_tool
 
 
+# Wikipedia tool 2: utilizza direttamente il package 'wikipedia'
 @tool
 def wikipedia_search(query: str) -> str:
     """
@@ -64,6 +71,25 @@ def wikipedia_search(query: str) -> str:
         return f"Error occurred: {str(e)}"
 
 
+# Wikipedia tool 3: utilizza WikipediaLoader dla package 'langchain_community.document_loaders' 
+@tool
+def wikipedia_search_3(query: str) -> str:
+    """
+    Search Wikipedia and return the full content of the most relevant articles.
+    Args:
+        query: The search query.
+    """
+    search_docs = WikipediaLoader(query=query,
+                                  load_max_docs=WIKIPEDIA_TOP_K_RESULTS,
+                                  doc_content_chars_max=WIKIPEDIA_DOC_CONTENT_CHARS_MAX,
+                                  load_all_available_meta=True).load()
+    formatted_search_docs = "\n\n---\n\n".join(
+        [
+            f'<Document source="{doc.metadata["source"]}" page="{doc.metadata.get("page", "")}"/>\n{doc.page_content}\n</Document>'
+            for doc in search_docs
+        ])
+    return {"wiki_results": formatted_search_docs}
+
 
 
 @tool
@@ -108,3 +134,18 @@ def execute_python_code_from_file(file_path: str) -> str:
 
     except Exception as e:
         return f"An unexpected error occurred during code execution: {e}"
+
+
+@tool
+def download_taskid_file(task_id: str, file_name: str) -> str:
+    """
+    Downloads the file associated with the given task_id (if any). Returns the absolute path of the file
+    """
+    try:
+        response = requests.get(f"{DATASET_API_URL}/files/{task_id}", timeout=20)
+        response.raise_for_status()
+        with open(file_name, 'wb') as file:
+            file.write(response.content)
+        return os.path.abspath(file_name)
+    except Exception as e:
+        return "Error occurred: {}".format(e)