import os import re import gradio as gr import requests import pandas as pd import heapq from collections import Counter from io import BytesIO from youtube_transcript_api import YouTubeTranscriptApi from smolagents import tool, Tool, CodeAgent, DuckDuckGoSearchTool, HfApiModel, VisitWebpageTool, SpeechToTextTool, FinalAnswerTool from langchain_community.document_loaders import WikipediaLoader, PyPDFLoader, TextLoader from dotenv import load_dotenv import tempfile import mimetypes import logging # --- Initialize logging --- logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s') logger = logging.getLogger(__name__) # --- Load environment variables --- load_dotenv() HF_API_TOKEN = os.getenv("HF_API_TOKEN") if not HF_API_TOKEN: logger.error("HF_API_TOKEN not found in environment variables! Please set it to use the HfApiModel.") # Exit or raise an error if the token is critical for functionality # sys.exit(1) # Uncomment if you want to exit the script if token is missing # --- Utility Functions --- def extract_youtube_id(url: str) -> str: """Extract YouTube ID from various URL formats""" patterns = [ r'(?:https?:\/\/)?(?:www\.)?youtube\.com\/watch\?v=([^&]+)', r'(?:https?:\/\/)?youtu\.be\/([^?]+)', r'([a-zA-Z0-9_-]{11})' # Catches just the ID if provided directly ] for pattern in patterns: match = re.search(pattern, url) if match: return match.group(1) return "" # --- Enhanced Tools --- class WikiSearchTool(Tool): """Enhanced Wikipedia search with better formatting and error handling""" name = "wiki_search" description = "Search Wikipedia for a query. Returns up to 2 results with metadata." inputs = {"query": {"type": "string", "description": "Search term for Wikipedia"}} output_type = "string" def forward(self, query: str) -> str: try: logger.info(f"Searching Wikipedia for: {query}") docs = WikipediaLoader(query=query, load_max_docs=2).load() if not docs: logger.info(f"No Wikipedia articles found for: {query}") return "No Wikipedia articles found." formatted_results = [] for i, doc in enumerate(docs): # Limit page content length to avoid overwhelming the model, but provide enough context summary = doc.page_content[:1000] + "..." if len(doc.page_content) > 1000 else doc.page_content formatted_results.append( f"--- Wikipedia Result {i+1} ---\n" f"Title: {doc.metadata.get('title', 'N/A')}\n" f"URL: {doc.metadata.get('source', 'N/A')}\n" f"Summary: {summary}\n" ) return "\n\n".join(formatted_results) except Exception as e: logger.error(f"Wikipedia search error for '{query}': {e}") return f"Wikipedia search error: {str(e)}" class FileAnalysisTool(Tool): """Universal file analyzer for text/PDF/Excel files""" name = "file_analysis" description = "Analyze text, PDF, and Excel files. Returns extracted content." inputs = {"file_path": {"type": "string", "description": "Path to the local file"}} output_type = "string" def forward(self, file_path: str) -> str: if not os.path.exists(file_path): return f"File not found: {file_path}" try: mime_type, _ = mimetypes.guess_type(file_path) logger.info(f"Analyzing file: {file_path} with MIME type: {mime_type}") if mime_type == "application/pdf": return self._process_pdf(file_path) elif mime_type in ["application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", "application/vnd.ms-excel"]: return self._process_excel(file_path) elif mime_type and ("text" in mime_type or "csv" in mime_type): return self._process_text(file_path) else: return f"Unsupported file type for analysis: {mime_type}. Only PDF, Excel, and text/CSV files are supported." except Exception as e: logger.error(f"File analysis error for '{file_path}': {e}") return f"File analysis error: {str(e)}" def _process_pdf(self, path: str) -> str: loader = PyPDFLoader(path) docs = loader.load() content = "\n\n".join([doc.page_content for doc in docs]) # Truncate to avoid excessive token usage, provide a warning if truncated if len(content) > 8000: logger.warning(f"PDF content truncated from {len(content)} to 8000 characters for {path}") return content[:8000] + "\n... [Content truncated]" return content def _process_excel(self, path: str) -> str: df = pd.read_excel(path) # Provide a sample of the data and its basic info info = BytesIO() df.info(buf=info) info_str = info.getvalue().decode('utf-8') return (f"Excel file loaded. First 10 rows:\n{df.head(10).to_markdown()}\n\n" f"DataFrame Info:\n{info_str}") def _process_text(self, path: str) -> str: with open(path, 'r', encoding='utf-8') as f: content = f.read() if len(content) > 8000: logger.warning(f"Text file content truncated from {len(content)} to 8000 characters for {path}") return content[:8000] + "\n... [Content truncated]" return content class VideoTranscriptionTool(Tool): """Enhanced YouTube transcription with multilingual support and better output""" name = "transcript_video" description = "Fetch YouTube video transcripts with optional timestamps. Supports English, French, Spanish, German." inputs = { "url": {"type": "string", "description": "YouTube URL or ID"}, "include_timestamps": {"type": "boolean", "description": "Include timestamps? (default: False)"} } output_type = "string" def forward(self, url: str, include_timestamps: bool = False) -> str: try: video_id = extract_youtube_id(url) if not video_id: return "Invalid YouTube URL or ID format. Please provide a valid YouTube URL or an 11-character video ID." logger.info(f"Attempting to transcribe video ID: {video_id}") transcript = YouTubeTranscriptApi.get_transcript( video_id, languages=['en', 'fr', 'es', 'de'] # Prioritize common languages ) if not transcript: return f"No transcript found for video ID: {video_id} in supported languages (en, fr, es, de)." if include_timestamps: formatted_transcript = "\n".join( f"[{int(seg['start']//60):02d}:{int(seg['start']%60):02d}] {seg['text']}" for seg in transcript ) else: formatted_transcript = " ".join(seg['text'] for seg in transcript) return formatted_transcript except Exception as e: logger.error(f"Transcription error for '{url}': {e}") return f"Transcription error: {str(e)}. This might be due to no available transcript or an unsupported video." class DataAnalysisTool(Tool): """Perform data analysis using pandas on structured data (CSV/Excel)""" name = "data_analysis" description = "Analyze CSV/Excel data using pandas operations. Supported operations: 'describe', 'groupby:column:aggfunc' (e.g., 'groupby:Category:mean')." inputs = { "file_path": {"type": "string", "description": "Path to the local data file (CSV or Excel)"}, "operation": {"type": "string", "description": "Pandas operation (e.g., 'describe', 'groupby:column_name:mean')"} } output_type = "string" def forward(self, file_path: str, operation: str) -> str: if not os.path.exists(file_path): return f"File not found: {file_path}" try: if file_path.endswith('.csv'): df = pd.read_csv(file_path) elif file_path.endswith('.xlsx') or file_path.endswith('.xls'): df = pd.read_excel(file_path) else: return "Unsupported file format for data analysis. Please provide a .csv or .xlsx file." logger.info(f"Performing data analysis operation '{operation}' on {file_path}") if operation == "describe": return "Descriptive Statistics:\n" + str(df.describe()) elif operation.startswith("groupby:"): parts = operation.split(":") if len(parts) == 3: _, col, agg = parts if col not in df.columns: return f"Column '{col}' not found in the DataFrame." try: result = df.groupby(col).agg(agg) return f"Groupby operation '{agg}' on column '{col}':\n" + str(result) except Exception as agg_e: return f"Error performing aggregation '{agg}' on column '{col}': {str(agg_e)}" else: return "Invalid 'groupby' operation format. Use 'groupby:column_name:agg_function'." else: return "Unsupported operation. Try: 'describe' or 'groupby:column_name:agg_function'." except Exception as e: logger.error(f"Data analysis error for '{file_path}' with operation '{operation}': {e}") return f"Data analysis error: {str(e)}. Please check file content and operation." # --- Agent Initialization --- class ResearchAgent: def __init__(self): self.model = HfApiModel( temperature=0.0, # Slightly increased temperature for more creative responses if appropriate token=HF_API_TOKEN, max_tokens=2000 ) self.tools = self._initialize_tools() self.agent = self._create_agent() def _initialize_tools(self) -> list: """Initialize all tools with enhanced capabilities""" return [ DuckDuckGoSearchTool(), WikiSearchTool(), VisitWebpageTool(), SpeechToTextTool(), # Might be less relevant for a text-based research agent but kept if needed FinalAnswerTool(), VideoTranscriptionTool(), FileAnalysisTool(), DataAnalysisTool(), self._create_excel_download_tool(), # Renamed for clarity self._create_keywords_tool() ] def _create_excel_download_tool(self): """Tool to download and parse Excel files from a specific URL""" @tool def download_and_parse_excel(task_id: str) -> dict: """ Downloads an Excel file from a predefined URL using a task_id and parses its content. Returns a dictionary with status and data (first 20 rows). """ try: url = f"https://agents-course-unit4-scoring.hf.space/files/{task_id}" logger.info(f"Attempting to download Excel from: {url}") response = requests.get(url, timeout=60) # Increased timeout for larger files response.raise_for_status() # Raise an exception for HTTP errors (4xx or 5xx) with tempfile.NamedTemporaryFile(delete=False, suffix='.xlsx') as tmp: tmp.write(response.content) temp_file_path = tmp.name df = pd.read_excel(temp_file_path) os.unlink(temp_file_path) # Clean up the temporary file logger.info(f"Successfully downloaded and parsed Excel for task_id: {task_id}") return { "task_id": task_id, "data_sample": df.head(10).to_dict(orient="records"), # Reduced to 10 for conciseness "status": "Success", "columns": df.columns.tolist(), # Added column names for context "shape": df.shape # Added shape for context } except requests.exceptions.RequestException as req_err: logger.error(f"Network or HTTP error downloading Excel for task_id '{task_id}': {req_err}") return {"status": f"Download error: {str(req_err)}"} except Exception as e: logger.error(f"Error parsing Excel for task_id '{task_id}': {e}") return {"status": f"Parsing error: {str(e)}"} return download_and_parse_excel def _create_keywords_tool(self): """Keywords extractor with TF-IDF like scoring (basic frequency for now)""" @tool def extract_keywords(text: str, top_n: int = 5) -> list: """ Extracts the most frequent keywords from a given text, excluding common stopwords. Args: text (str): The input text to extract keywords from. top_n (int): The number of top keywords to return. Returns: list: A list of the most frequent keywords. """ if not text: return [] # Use a more comprehensive list of English stopwords stopwords = set([ "a", "an", "and", "are", "as", "at", "be", "but", "by", "for", "if", "in", "into", "is", "it", "no", "not", "of", "on", "or", "such", "that", "the", "their", "then", "there", "these", "they", "this", "to", "was", "will", "with", "he", "she", "it's", "i", "we", "you", "my", "your", "our", "us", "him", "her", "his", "hers", "its", "them", "their", "what", "when", "where", "why", "how", "which", "who", "whom", "can", "could", "would", "should", "may", "might", "must", "have", "has", "had", "do", "does", "did", "am", "are", "is", "were", "been", "being", "from", "up", "down", "out", "off", "over", "under", "again", "further", "then", "once", "here", "there", "when", "where", "why", "how", "all", "any", "both", "each", "few", "more", "most", "other", "some", "such", "no", "nor", "not", "only", "own", "same", "so", "than", "too", "very", "s", "t", "can", "will", "just", "don", "should", "now" ]) words = re.findall(r'\b\w+\b', text.lower()) # Relaxed regex to capture all words filtered = [w for w in words if w not in stopwords and len(w) > 2] # Filter words less than 3 chars counter = Counter(filtered) return [word for word, _ in counter.most_common(top_n)] return extract_keywords def _create_agent(self) -> CodeAgent: """Create agent with improved system prompt""" system_prompt = """ You are an advanced, helpful, and highly analytical research assistant. Your goal is to provide accurate, comprehensive, and well-structured answers to user queries, leveraging all available tools efficiently. **Follow this robust process:** 1. **Understand the User's Need:** Carefully analyze the user's question, including any attached files or specific requests (e.g., "summarize," "analyze data," "find facts"). 2. **Formulate a Detailed Plan:** Before acting, create a clear, step-by-step plan. This plan should outline: * What information needs to be gathered. * Which tools are most appropriate for each step (e.g., `duckduckgo_search` for general web search, `wiki_search` for encyclopedic facts, `transcript_video` for YouTube, `file_analysis` or `data_analysis` for local files). * How you will combine information from different sources. * How you will verify or synthesize the findings. 3. **Execute the Plan Using Tools:** Call the necessary tools, providing clear and correct arguments. If a tool fails, try to understand why and adapt your plan (e.g., try a different search query or tool). 4. **Synthesize and Verify Information:** Once you have gathered sufficient information, synthesize it into a coherent answer. Do not just list facts; explain their significance and how they relate to the original question. If there are contradictions or uncertainties, mention them. 5. **Formulate the Final Answer:** * Present your answer clearly and concisely. * Always begin your ultimate response with "FINAL ANSWER:". * If the answer is a single number, provide only the number. * If the answer is a list, provide comma-separated values. * For complex answers, use structured formats like bullet points or JSON where appropriate to enhance readability. * **Crucially, always include sources or references (e.g., URLs, Wikipedia titles, file names) where you obtained the information.** This builds trust and allows for verification. * If you used `file_analysis` or `data_analysis` tools on an uploaded file, explicitly state that you analyzed the provided file. **Important Considerations:** * **Prioritize:** If the query involves a specific file, start by analyzing that file if appropriate. * **Limitations:** If you cannot answer a question with the available tools, state that clearly. * **Conciseness:** Be as concise as possible while providing a complete and accurate answer. """ agent = CodeAgent( model=self.model, tools=self.tools, add_base_tools=True ) agent.prompt_templates["system_prompt"] = system_prompt return agent def __call__(self, question: str) -> str: logger.info(f"Received question: {question[:200]}...") # Log more of the question try: response = self.agent.run(question) logger.info(f"Response generated successfully for question: {question[:200]}") return response except Exception as e: logger.error(f"Agent execution failed for question '{question[:100]}': {str(e)}", exc_info=True) # Log full traceback return f"Error processing your request: {str(e)}. Please try again or rephrase your question." def run_and_submit_all( profile: gr.OAuthProfile | None): """ Fetches all questions, runs the BasicAgent on them, submits all answers, and displays the results. """ # --- Determine HF Space Runtime URL and Repo URL --- space_id = os.getenv("SPACE_ID") # Get the SPACE_ID for sending link to the code if profile: username= f"{profile.username}" print(f"User logged in: {username}") else: print("User not logged in.") return "Please Login to Hugging Face with the button.", None api_url = DEFAULT_API_URL questions_url = f"{api_url}/questions" submit_url = f"{api_url}/submit" # 1. Instantiate Agent ( modify this part to create your agent) try: agent = BasicAgent() except Exception as e: print(f"Error instantiating agent: {e}") return f"Error initializing agent: {e}", None # In the case of an app running as a hugging Face space, this link points toward your codebase ( usefull for others so please keep it public) agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main" print(agent_code) # 2. Fetch Questions print(f"Fetching questions from: {questions_url}") try: response = requests.get(questions_url, timeout=15) response.raise_for_status() questions_data = response.json() if not questions_data: print("Fetched questions list is empty.") return "Fetched questions list is empty or invalid format.", None print(f"Fetched {len(questions_data)} questions.") except requests.exceptions.RequestException as e: print(f"Error fetching questions: {e}") return f"Error fetching questions: {e}", None except requests.exceptions.JSONDecodeError as e: print(f"Error decoding JSON response from questions endpoint: {e}") print(f"Response text: {response.text[:500]}") return f"Error decoding server response for questions: {e}", None except Exception as e: print(f"An unexpected error occurred fetching questions: {e}") return f"An unexpected error occurred fetching questions: {e}", None # 3. Run your Agent results_log = [] answers_payload = [] print(f"Running agent on {len(questions_data)} questions...") for item in questions_data: task_id = item.get("task_id") question_text = item.get("question") 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) 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: print(f"Error running agent on task {task_id}: {e}") results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"}) if not answers_payload: print("Agent did not produce any answers to submit.") return "Agent did not produce any answers to submit.", pd.DataFrame(results_log) # 4. Prepare Submission submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload} status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..." print(status_update) # 5. Submit print(f"Submitting {len(answers_payload)} answers to: {submit_url}") try: response = requests.post(submit_url, json=submission_data, timeout=60) response.raise_for_status() result_data = response.json() final_status = ( f"Submission Successful!\n" f"User: {result_data.get('username')}\n" f"Overall Score: {result_data.get('score', 'N/A')}% " f"({result_data.get('correct_count', '?')}/{result_data.get('total_attempted', '?')} correct)\n" f"Message: {result_data.get('message', 'No message received.')}" ) print("Submission successful.") results_df = pd.DataFrame(results_log) return final_status, results_df except requests.exceptions.HTTPError as e: error_detail = f"Server responded with status {e.response.status_code}." try: error_json = e.response.json() error_detail += f" Detail: {error_json.get('detail', e.response.text)}" except requests.exceptions.JSONDecodeError: error_detail += f" Response: {e.response.text[:500]}" status_message = f"Submission Failed: {error_detail}" print(status_message) results_df = pd.DataFrame(results_log) return status_message, results_df except requests.exceptions.Timeout: status_message = "Submission Failed: The request timed out." print(status_message) results_df = pd.DataFrame(results_log) return status_message, results_df except requests.exceptions.RequestException as e: status_message = f"Submission Failed: Network error - {e}" print(status_message) results_df = pd.DataFrame(results_log) return status_message, results_df except Exception as e: status_message = f"An unexpected error occurred during submission: {e}" print(status_message) results_df = pd.DataFrame(results_log) return status_message, results_df # --- Build Gradio Interface using Blocks --- with gr.Blocks() as demo: gr.Markdown("# Basic Agent Evaluation Runner") gr.Markdown( """ **Instructions:** 1. Please clone this space, then modify the code to define your agent's logic, the tools, the necessary packages, etc ... 2. Log in to your Hugging Face account using the button below. This uses your HF username for submission. 3. Click 'Run Evaluation & Submit All Answers' to fetch questions, run your agent, submit answers, and see the score. --- **Disclaimers:** Once clicking on the "submit button, it can take quite some time ( this is the time for the agent to go through all the questions). This space provides a basic setup and is intentionally sub-optimal to encourage you to develop your own, more robust solution. For instance for the delay process of the submit button, a solution could be to cache the answers and submit in a seperate action or even to answer the questions in async. """ ) gr.LoginButton() run_button = gr.Button("Run Evaluation & Submit All Answers") status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False) # Removed max_rows=10 from DataFrame constructor results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True) run_button.click( fn=run_and_submit_all, outputs=[status_output, results_table] ) if __name__ == "__main__": print("\n" + "-"*30 + " App Starting " + "-"*30) # Check for SPACE_HOST and SPACE_ID at startup for information space_host_startup = os.getenv("SPACE_HOST") space_id_startup = os.getenv("SPACE_ID") # Get SPACE_ID at startup if space_host_startup: print(f"✅ SPACE_HOST found: {space_host_startup}") print(f" Runtime URL should be: https://{space_host_startup}.hf.space") else: print("ℹ️ SPACE_HOST environment variable not found (running locally?).") if space_id_startup: # Print repo URLs if SPACE_ID is found print(f"✅ SPACE_ID found: {space_id_startup}") print(f" Repo URL: https://huggingface.co/spaces/{space_id_startup}") print(f" Repo Tree URL: https://huggingface.co/spaces/{space_id_startup}/tree/main") else: print("ℹ️ SPACE_ID environment variable not found (running locally?). Repo URL cannot be determined.") print("-"*(60 + len(" App Starting ")) + "\n") print("Launching Gradio Interface for Basic Agent Evaluation...") demo.launch(debug=True, share=False)