Update agent.py

agent.py

@@ -4,11 +4,10 @@ import requests
 import tempfile
 import pandas as pd
 from openai import OpenAI
-
 try:
     from duckduckgo_search import DDGS
 except ImportError:
-    DDGS = None  # In case of install problems
+    DDGS = None
 
 PROMPT = (
     "You are a general AI assistant. I will ask you a question. "
@@ -51,7 +50,6 @@ class BasicAgent:
                 f.flush()
                 excel_path = f.name
             df = pd.read_excel(excel_path)
-            # Try to sum 'Sales' where 'Type' == 'food'
             if "Type" in df.columns and "Sales" in df.columns:
                 total = df[df["Type"].str.lower() == "food"]["Sales"].sum()
                 return f"{round(total, 2)}"
@@ -60,40 +58,6 @@ class BasicAgent:
         except Exception as e:
             return ""
 
-    def transcribe_audio(self, file_url: str) -> str:
-        import openai
-        openai.api_key = os.getenv("OPENAI_API_KEY")
-        try:
-            r = requests.get(file_url, timeout=20)
-            r.raise_for_status()
-            # Guess extension from url or response
-            ext = ".mp3"
-            if file_url.endswith(".wav"):
-                ext = ".wav"
-            with tempfile.NamedTemporaryFile(suffix=ext, delete=False) as f:
-                f.write(r.content)
-                f.flush()
-                audio_path = f.name
-            transcript = openai.Audio.transcribe("whisper-1", open(audio_path, "rb"))
-            return transcript.get("text", "")
-        except Exception as e:
-            return ""
-
-    def execute_python(self, file_url: str) -> str:
-        try:
-            r = requests.get(file_url, timeout=20)
-            r.raise_for_status()
-            code = r.content.decode("utf-8")
-            import io, contextlib
-            buf = io.StringIO()
-            with contextlib.redirect_stdout(buf):
-                exec(code, {})
-            output = buf.getvalue().strip().split('\n')[-1]
-            numbers = re.findall(r'[-+]?\d*\.\d+|\d+', output)
-            return numbers[-1] if numbers else output
-        except Exception as e:
-            return ""
-
     def fetch_file_url(self, task_id):
         DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
         try:
@@ -105,34 +69,55 @@ class BasicAgent:
             pass
         return None
 
+    def solve_chess_image(self, image_url: str) -> str:
+        """Uses GPT-4o Vision to answer a chess image question (requires image URL)."""
+        prompt = (
+            "You are a chess engine. Only answer with the best move for Black in algebraic notation (e.g., Qd1#). "
+            "Do not explain your reasoning, do not include any commentary, only the move."
+        )
+        try:
+            response = self.llm.chat.completions.create(
+                model="gpt-4o",
+                messages=[
+                    {"role": "system", "content": prompt},
+                    {
+                        "role": "user",
+                        "content": [
+                            {"type": "text", "text": prompt},
+                            {"type": "image_url", "image_url": {"url": image_url}},
+                        ],
+                    }
+                ],
+                max_tokens=32,
+                temperature=0.0,
+            )
+            result = response.choices[0].message.content.strip()
+            # Remove commentary, keep only first move if any
+            move = re.findall(r"\b([KQRNB]?[a-h]?[1-8]?x?[a-h][1-8](?:=[QRNB])?#?)\b", result)
+            if move:
+                return move[0]
+            return result
+        except Exception as e:
+            return ""
+
     def __call__(self, question: str, task_id: str = None) -> str:
         file_url = self.fetch_file_url(task_id) if task_id else None
         file_result = None
+
+        # --- Chess image detection (heuristic: "chess", "move", "image", or "position") ---
+        if file_url and ("chess" in question.lower() or "move" in question.lower() or "image" in question.lower() or "position" in question.lower()):
+            chess_result = self.solve_chess_image(file_url)
+            if chess_result and len(chess_result) < 10:  # likely algebraic notation
+                return chess_result
+
+        # --- Excel heuristic (only try for likely Excel file) ---
         ext = file_url.split('.')[-1].lower() if file_url else ""
-        # --- Try all known file tools by extension ---
-        if file_url:
-            # Excel (any file: always try)
-            if ext in ["xlsx", "xls"] or "excel" in question.lower() or "spreadsheet" in question.lower():
-                file_result = self.excel_tool(file_url)
-                if file_result and re.match(r'^\d+(\.\d+)?$', file_result):
-                    return file_result
-            # Audio
-            elif ext in ["mp3", "wav"] or "audio" in question.lower() or "transcribe" in question.lower():
-                file_result = self.transcribe_audio(file_url)
-                if file_result and file_result.strip():
-                    return file_result
-            # Python code
-            elif ext == "py":
-                file_result = self.execute_python(file_url)
-                if file_result and file_result.strip():
-                    return file_result
-            # Fallback: try Excel anyway
-            if not file_result:
-                file_result = self.excel_tool(file_url)
-                if file_result and re.match(r'^\d+(\.\d+)?$', file_result):
-                    return file_result
+        if file_url and (ext in ["xlsx", "xls"] or "excel" in question.lower() or "spreadsheet" in question.lower()):
+            file_result = self.excel_tool(file_url)
+            if file_result and re.match(r'^\d+(\.\d+)?$', file_result):
+                return file_result
 
-        # --- Web search and LLM as before ---
+        # --- Web search + LLM as before ---
         search_snippet = self.web_search(question)
         prompt = PROMPT + f"\n\nWeb search results:\n{search_snippet}\n\nQuestion: {question}"
         response = self.llm.chat.completions.create(