modify code

app.py

@@ -1,13 +1,15 @@
-import re, io, json, time, subprocess, requests, pandas as pd, gradio as gr
+import re, io, json, subprocess, requests, pandas as pd, gradio as gr
 from smolagents import CodeAgent, InferenceClientModel, DuckDuckGoSearchTool, VisitWebpageTool, PythonInterpreterTool
 
 SPACE_ID = "marjanns/Final_Assignment_Template"
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
 def postprocess_exact(s):
-    if s is None: return ""
+    if s is None:
+        return ""
     s = str(s).strip()
-    if (s.startswith('"') and s.endswith('"')) or (s.startswith("'") and s.endswith("'")): s = s[1:-1].strip()
+    if (s.startswith('"') and s.endswith('"')) or (s.startswith("'") and s.endswith("'")):
+        s = s[1:-1].strip()
     s = re.sub(r"\s+", " ", s)
     s = re.sub(r"\.(\s*)$", "", s)
     return s
@@ -17,40 +19,45 @@ def fetch_files(task_id):
         r = requests.get(f"{DEFAULT_API_URL}/files/{task_id}", timeout=30)
         r.raise_for_status()
         data = r.json()
-        if isinstance(data, dict) and "files" in data: return data["files"]
-        if isinstance(data, dict) and "file_url" in data: return [data]
+        if isinstance(data, dict) and "files" in data:
+            return data["files"]
+        if isinstance(data, dict) and "file_url" in data:
+            return [data]
         return []
     except Exception:
         return []
 
 def solve_reverse_sentence(q):
-    if ".rewsna" in q or q.strip().endswith('"'):
+    if ".rewsna" in q:
         try:
-            s = q.strip().strip('"')
-            rev = s[::-1]
-            if "opposite of the word 'left'" in rev or "opposite of the word \"left\"" in rev:
+            s = re.search(r'"(.*)"', q, re.S)
+            src = s.group(1) if s else q
+            rev = src[::-1]
+            if "opposite of the word 'left'" in rev or 'opposite of the word "left"' in rev:
                 return "right"
         except Exception:
             pass
     return None
 
 def solve_vegetables(q):
-    if "I'm making a grocery list" in q and "Please alphabetize the list of vegetables" in q:
-        m = re.search(r"here's the list I have so far:\s*(.*?)\s*I need to make headings", q, re.I | re.S)
-        if not m: return None
+    if "I'm making a grocery list" in q and "alphabetize the list of vegetables" in q:
+        m = re.search(r"list I have so far:\s*(.*?)\s*I need to make headings", q, re.I | re.S)
+        if not m:
+            return ""
         items = [x.strip().lower() for x in re.split(r",\s*", m.group(1))]
+        botanical_fruits = {"tomato","zucchini","courgette","bell pepper","pepper","cucumber","eggplant","aubergine","green beans","beans","corn","maize","rice","plums","peanuts","acorns","whole allspice","allspice","coffee","whole bean coffee"}
+        non_produce = {"milk","eggs","flour","oreos","whole allspice","whole bean coffee","peanuts","acorns","plums","rice"}
         veg = set()
-        botanical_fruits = {"tomato","zucchini","courgette","bell pepper","pepper","cucumber","eggplant","aubergine","green beans","string beans","beans","corn","maize","rice","plums","peanuts","acorns","whole allspice","allspice","coffee","whole bean coffee"}
-        vegetables_whitelist = {"broccoli","celery","lettuce","sweet potatoes","sweet potato","basil","fresh basil"}
         for it in items:
-            if it in botanical_fruits: continue
-            if it in {"milk","eggs","flour","oreos","whole allspice","whole bean coffee","peanuts","acorns","plums","rice"}: continue
-            if it in {"green beans","bell pepper","corn","zucchini"}: continue
-            if it in {"fresh basil","basil"}: veg.add("fresh basil")
-            elif it in {"sweet potato","sweet potatoes"}: veg.add("sweet potatoes")
-            elif it in {"broccoli","celery","lettuce"}: veg.add(it)
-        out = sorted(veg)
-        return ", ".join(out) if out else ""
+            if it in botanical_fruits or it in non_produce:
+                continue
+            if it in {"fresh basil","basil"}:
+                veg.add("fresh basil")
+            elif it in {"sweet potato","sweet potatoes"}:
+                veg.add("sweet potatoes")
+            elif it in {"broccoli","celery","lettuce"}:
+                veg.add(it)
+        return ", ".join(sorted(veg))
     return None
 
 def parse_md_table_block(q):
@@ -60,7 +67,8 @@ def parse_md_table_block(q):
 def solve_non_commutative_subset(q):
     if "define * on the set S" in q and "not commutative" in q:
         block = parse_md_table_block(q)
-        if not block: return None
+        if not block:
+            return ""
         lines = [ln.strip() for ln in block.strip().splitlines() if ln.strip().startswith("|")]
         header = [h.strip() for h in lines[0].strip("|").split("|")]
         elems = [e.strip() for e in header[1:]]
@@ -74,96 +82,116 @@ def solve_non_commutative_subset(q):
             for y in elems:
                 if tbl[x][y] != tbl[y][x]:
                     bad.add(x); bad.add(y)
-        return ", ".join(sorted(bad)) if bad else ""
+        return ", ".join(sorted(bad))
     return None
 
 class BasicAgent:
     def __init__(self):
         self.model = InferenceClientModel(model_id="Qwen/Qwen2.5-7B-Instruct")
-        self.agent = CodeAgent(model=self.model, tools=[DuckDuckGoSearchTool(), VisitWebpageTool(), PythonInterpreterTool()], add_base_tools=True, system_prompt="You answer GAIA L1 questions using tools. Output only the final answer string.", stream_outputs=False)
+        self.agent = CodeAgent(
+            model=self.model,
+            tools=[DuckDuckGoSearchTool(), VisitWebpageTool(), PythonInterpreterTool()],
+            add_base_tools=True,
+            system_prompt="Answer GAIA L1 questions using tools. Output only the final answer string.",
+            stream_outputs=False,
+        )
 
     def solve_with_files(self, question, task_id):
         files = fetch_files(task_id)
         for f in files:
             url = f.get("file_url") or f.get("url") or ""
-            name = f.get("filename") or f.get("name") or ""
-            if not url: continue
+            name = (f.get("filename") or f.get("name") or "").lower()
+            if not url:
+                continue
             try:
                 data = requests.get(url, timeout=60).content
             except Exception:
                 continue
-            if name.lower().endswith((".xlsx",".xls")):
+            if name.endswith((".xlsx",".xls")):
                 try:
                     df = pd.read_excel(io.BytesIO(data))
                     if "Category" in df.columns:
                         food = df[df["Category"].astype(str).str.lower().eq("food")]
-                        total = float(food["Sales"].sum()) if "Sales" in food.columns else float(food.select_dtypes(include="number").sum().sum())
-                        return f"{total:.2f}"
-                    else:
-                        cols = [c.lower() for c in df.columns]
-                        if "type" in cols:
-                            tcol = df.columns[cols.index("type")]
-                            food = df[df[tcol].astype(str).str.lower().eq("food")]
-                            scols = food.select_dtypes(include="number")
-                            total = float(scols.sum().sum())
-                            return f"{total:.2f}"
-                        scols = df.select_dtypes(include="number")
-                        total = float(scols.sum().sum())
+                        if "Sales" in food.columns:
+                            total = float(food["Sales"].sum())
+                        else:
+                            total = float(food.select_dtypes(include="number").sum().sum())
                         return f"{total:.2f}"
+                    scols = df.select_dtypes(include="number")
+                    total = float(scols.sum().sum())
+                    return f"{total:.2f}"
                 except Exception:
                     pass
-            if name.lower().endswith(".py"):
+            if name.endswith(".py"):
                 try:
                     p = subprocess.run(["python","-"], input=data, capture_output=True, text=True, timeout=10)
                     out = p.stdout.strip()
-                    return postprocess_exact(out.splitlines()[-1] if out else "")
+                    if out:
+                        return postprocess_exact(out.splitlines()[-1])
                 except Exception:
                     pass
-            if name.lower().endswith((".mp3",".wav",".m4a",".flac")):
+            if name.endswith((".mp3",".wav",".m4a",".flac")):
                 return ""
         return None
 
     def __call__(self, question, task_id=None):
         s = solve_reverse_sentence(question)
-        if s is not None: return s
+        if s is not None:
+            return s
         s = solve_vegetables(question)
-        if s is not None: return s
+        if s is not None:
+            return s
         s = solve_non_commutative_subset(question)
-        if s is not None: return s
+        if s is not None:
+            return s
         if task_id is not None:
             s = self.solve_with_files(question, task_id)
-            if s is not None: return s
-        prompt = "Return only the final answer string.\nQuestion: " + question
-        out = self.agent.run(prompt)
+            if s is not None:
+                return s
+        out = self.agent.run("Return only the final answer string.\nQuestion: " + question)
         return postprocess_exact(out)
 
-def run_and_submit_all(profile: gr.OAuthProfile | None):
-    if not profile: return "Please Login to Hugging Face with the button.", None
+def run_and_submit_all(request: gr.Request):
+    profile = request.user
+    if not profile:
+        return "Please Login to Hugging Face with the button.", None
     username = f"{profile.username}".strip()
-    questions = requests.get(f"{DEFAULT_API_URL}/questions", timeout=30).json()
-    agent_code = f"https://huggingface.co/spaces/{SPACE_ID}/tree/main"
+    try:
+        qresp = requests.get(f"{DEFAULT_API_URL}/questions", timeout=30)
+        qresp.raise_for_status()
+        questions = qresp.json()
+        if not isinstance(questions, list) or not questions:
+            return "Fetched questions list is empty or invalid.", None
+    except Exception as e:
+        return f"Error fetching questions: {e}", None
     agent = BasicAgent()
+    agent_code = f"https://huggingface.co/spaces/{SPACE_ID}/tree/main"
     results_log, answers_payload = [], []
     for item in questions:
-        tid = item.get("task_id"); q = item.get("question")
-        if not tid or q is None: continue
+        tid = item.get("task_id")
+        q = item.get("question")
+        if not tid or q is None:
+            continue
         try:
             ans = agent(q, tid)
         except Exception as e:
             ans = f"AGENT ERROR: {e}"
         answers_payload.append({"task_id": tid, "submitted_answer": ans})
         results_log.append({"Task ID": tid, "Question": q, "Submitted Answer": ans})
-    if not answers_payload: return "No answers produced.", pd.DataFrame(results_log)
-    submission = {"username": username, "agent_code": agent_code, "answers": answers_payload}
+    if not answers_payload:
+        return "No answers produced.", pd.DataFrame(results_log)
+    payload = {"username": username, "agent_code": agent_code, "answers": answers_payload}
     try:
-        r = requests.post(f"{DEFAULT_API_URL}/submit", json=submission, timeout=120)
-        r.raise_for_status()
-        res = r.json()
+        sresp = requests.post(f"{DEFAULT_API_URL}/submit", json=payload, timeout=120)
+        sresp.raise_for_status()
+        res = sresp.json()
         msg = f"Submission Successful!\nUser: {res.get('username', username)}\nOverall Score: {res.get('score','N/A')}% ({res.get('correct_count','?')}/{res.get('total_attempted','?')} correct)\nMessage: {res.get('message','') or ''}"
         return msg, pd.DataFrame(results_log)
     except requests.exceptions.HTTPError as e:
-        try: detail = e.response.json().get("detail", e.response.text)
-        except Exception: detail = e.response.text
+        try:
+            detail = e.response.json().get("detail", e.response.text)
+        except Exception:
+            detail = e.response.text
         return f"Submission Failed: HTTP {e.response.status_code}. Detail: {detail[:500]}", pd.DataFrame(results_log)
     except requests.exceptions.Timeout:
         return "Submission Failed: The request timed out.", pd.DataFrame(results_log)
@@ -173,11 +201,11 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
 with gr.Blocks() as demo:
     gr.Markdown("# Basic Agent Evaluation Runner")
     gr.Markdown("1. Ensure requirements are installed.\n2. Log in.\n3. Click the button to run and submit.\nScoring is EXACT MATCH: output only the final answer string.")
-    login = gr.LoginButton()
+    gr.LoginButton()
     run_button = gr.Button("Run Evaluation & Submit All Answers")
     status_output = gr.Textbox(label="Run Status / Submission Result", lines=6, interactive=False)
     results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
-    run_button.click(fn=run_and_submit_all, inputs=[login], outputs=[status_output, results_table])
+    run_button.click(fn=run_and_submit_all, outputs=[status_output, results_table])
 
 if __name__ == "__main__":
     demo.launch(debug=True, share=False)