update codes

app.py

@@ -3,133 +3,57 @@ import gradio as gr
 import requests
 import inspect
 import pandas as pd
-import time
-import re
-from bs4 import BeautifulSoup
-
-from smolagents import CodeAgent, InferenceClientModel, Tool
+import random
+from smolagents import CodeAgent, InferenceClientModel, OpenAIModel, DuckDuckGoSearchTool, VisitWebpageTool
 
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
-# --- High-Performance Tool 1: 다차원 구조화 웹 검색 툴 ---
-class AdvancedSearchTool(Tool):
-    name = "web_search"
-    description = "Executes a deep web search via DuckDuckGo HTML architecture and extracts exact URLs and targeted meta-snippets."
-    inputs = {"query": {"type": "string", "description": "The precise keyword query to search for"}}
-    output_type = "string"
-
-    def forward(self, query: str) -> str:
-        try:
-            url = f"https://html.duckduckgo.com/html/?q={requests.utils.quote(query)}"
-            headers = {"User-Agent": "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36"}
-            response = requests.get(url, headers=headers, timeout=12)
-            if response.status_code != 200:
-                return f"Search Gateway Error: HTTP {response.status_code}"
-            
-            soup = BeautifulSoup(response.text, "lxml")
-            results = []
-            for i, item in enumerate(soup.select(".result__body")[:5]):
-                title_anchor = item.select_one(".result__title a")
-                snippet_div = item.select_one(".result__snippet")
-                if title_anchor and snippet_div:
-                    title = title_anchor.get_text(strip=True)
-                    link = title_anchor.get("href")
-                    # 내부 리다이렉트 URL 정제
-                    if "uddg=" in link:
-                        link = requests.utils.unquote(link.split("uddg=")[1].split("&")[0])
-                    snippet = snippet_div.get_text(strip=True)
-                    results.append(f"[{i+1}] Title: {title}\nURL: {link}\nContext: {snippet}")
-            return "\n\n".join(results) if results else "No indexing data found."
-        except Exception as e:
-            return f"Search Engine Exception: {str(e)}"
-
-# --- High-Performance Tool 2: 마크다운 변환형 웹 및 도큐먼트 파서 툴 ---
-class DeepPageVisitTool(Tool):
-    name = "visit_webpage"
-    description = "Visits a specific URL, bypasses layout boilerplate, and converts raw HTML into a dense Markdown/Table format for complex data analysis."
-    inputs = {"url": {"type": "string", "description": "The target exact URL to scrape content from"}}
-    output_type = "string"
-
-    def forward(self, url: str) -> str:
-        try:
-            headers = {"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/122.0.0.0 Safari/537.36"}
-            response = requests.get(url, headers=headers, timeout=15)
-            if response.status_code != 200:
-                return f"HTTP Access Failure: Status {response.status_code}"
-            
-            soup = BeautifulSoup(response.text, "lxml")
-            # 노이즈 태그 전량 제거
-            for element in soup(["script", "style", "nav", "footer", "header", "aside"]):
-                element.extract()
-                
-            # GAIA 핵심 지표인 '표 데이터' 보존 처리
-            for table in soup.find_all("table"):
-                markdown_table = []
-                for row in table.find_all("tr"):
-                    cells = [f" {cell.get_text(strip=True)} " for cell in row.find_all(["td", "th"])]
-                    markdown_table.append("|" + "|".join(cells) + "|")
-                if markdown_table:
-                    table.replace_with(soup.new_string("\n" + "\n".join(markdown_table) + "\n"))
-                    
-            text = soup.get_text(separator="\n")
-            text = re.sub(r'\n+', '\n', text).strip()
-            return text[:6000] # 컨텍스트 상한치 확보
-        except Exception as e:
-            return f"Page Scraping Exception: {str(e)}"
-
 # --- Basic Agent Definition ---
 class BasicAgent:
-    def __init__(self):
-        self.model = InferenceClientModel(
-            model_id="Qwen/Qwen2.5-Coder-32B-Instruct",
-            token=os.getenv("HF_TOKEN")
-        )
-        self.search_tool = AdvancedSearchTool()
-        self.visit_tool = DeepPageVisitTool()
+    def __init__(self, hf_token: str | None = None):
+        deepseek_key = os.getenv("DEEPSEEK_API_KEY")
         
-        # 가드레일 강화를 위해 verbosity_level을 높이고 복합 연산 지시 프롬프트 템플릿 조정
-        self.agent = CodeAgent(
-            tools=[self.search_tool, self.visit_tool],
-            model=self.model,
-            max_steps=12,
-            verbosity_level=2
+        if deepseek_key:
+            model = OpenAIModel(
+                model_id="deepseek-chat",
+                api_base="https://api.deepseek.com",
+                api_key=deepseek_key,
+            )
+        else:
+            model = InferenceClientModel(
+                model_id="Qwen/Qwen2.5-Coder-32B-Instruct",
+                token=hf_token
+            )
+                
+        search_tool = DuckDuckGoSearchTool()
+        visit_tool = VisitWebpageTool()
+                
+        self.alfred = CodeAgent(
+            tools=[search_tool, visit_tool], 
+            model=model,
+            add_base_tools=True,  
+            planning_interval=3   
         )
-        print("BasicAgent: Guardrail & Self-Correction Engine Loaded.")
 
     def __call__(self, question: str) -> str:
-        print(f"Agent received question (first 50 chars): {question[:50]}...")
         try:
-            # 에이전트가 단발성 판단을 내리지 않고 명확한 실행 계획(Execution Plan)을 세우도록 강제하는 엔지니어링 프롬프트
-            structured_prompt = (
-                f"You are an expert AI agent solving a GAIA task.\n"
-                f"Task: {question}\n\n"
-                f"Strict Protocol:\n"
-                f"1. Plan: Break down the research and computation into clear sub-tasks.\n"
-                f"2. Action: Use your code interpreter or tools to gather and verify facts.\n"
-                f"3. Self-Correction: If any code execution fails with a Traceback, analyze the error, rewrite the script, and run it again.\n"
-                f"4. Output: Extract the absolute raw answer value (e.g., specific number, name, date) without any markdown formatting wrappers or conversational text. Present this on the very last line."
-            )
-            
-            result = self.agent.run(structured_prompt)
+            result = self.alfred.run(question)
             if result is None:
                 return "unknown"
-                
-            # 정답 유실 방지를 위한 최종 타겟 파싱 가드레일 처리
-            final_output = str(result).strip()
-            if "\n" in final_output:
-                final_output = final_output.split("\n")[-1].replace("Final Answer:", "").strip()
-            return final_output
-            
+            return str(result).strip()
         except Exception as e:
-            print(f"Critical System Failure during agent execution: {e}")
+            print(f"Error during agent runtime execution: {e}")
             return "unknown"
 
-def run_and_submit_all( profile: gr.OAuthProfile | None):
+def run_and_submit_all(profile: gr.OAuthProfile | None, oauth_token: gr.OAuthToken | None = None):
+    """
+    Fetches all questions, runs the BasicAgent on them, submits all answers,
+    and displays the results.
+    """
     space_id = os.getenv("SPACE_ID")
-
     if profile:
-        username= f"{profile.username}"
+        username = f"{profile.username}"
         print(f"User logged in: {username}")
     else:
         print("User not logged in.")
@@ -140,10 +64,12 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
     submit_url = f"{api_url}/submit"
 
     try:
-        agent = BasicAgent()
+        hf_token = oauth_token.token if oauth_token else os.getenv("HF_TOKEN")
+        agent = BasicAgent(hf_token=hf_token)
     except Exception as e:
         print(f"Error instantiating agent: {e}")
         return f"Error initializing agent: {e}", None
+
     agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
     print(agent_code)
 
@@ -211,7 +137,7 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
         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)}"
+            error_detail += f" Detail: {error_json.get('detail', error_json.get('message', e.response.text))}"
         except requests.exceptions.JSONDecodeError:
             error_detail += f" Response: {e.response.text[:500]}"
         status_message = f"Submission Failed: {error_detail}"
@@ -239,21 +165,26 @@ with gr.Blocks() as demo:
     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.
+        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.
         """
     )
-
     gr.LoginButton()
     run_button = gr.Button("Run Evaluation & Submit All Answers")
     status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
     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__":
+    space_host_startup = os.getenv("SPACE_HOST")
+    space_id_startup = os.getenv("SPACE_ID")
+    if space_host_startup:
+        print(f"✅ SPACE_HOST found: {space_host_startup}")
+    if space_id_startup:
+        print(f"✅ SPACE_ID found: {space_id_startup}")
     demo.launch(debug=True, share=False)

requirements.txt

@@ -1,7 +1,7 @@
-smolagents>=1.0.0
-beautifulsoup4>=4.12.0
-lxml>=5.1.0
-pandas>=2.0.0
-openpyxl>=3.1.0
-requests>=2.31.0
+smolagents>=1.1.0
+duckduckgo-search>=6.2.1
+openai
+markdownify
+pandas
+requests
 gradio