Update app.py

app.py

@@ -2,12 +2,18 @@ import os
 from dotenv import load_dotenv
 import gradio as gr
 from openai import OpenAI
+from datetime import datetime
 
 # 載入環境變數
 load_dotenv()
 client = OpenAI(api_key=os.getenv('OPENAI_API_KEY'))
 
-def grade_essay(essay_content, additional_prompt=""):
+def evaluate_essay(message, history, additional_prompt):
+    if not message.strip():
+        return history + [
+            {"role": "assistant", "content": "請輸入作文內容進行評分。"}
+        ]
+    
     criteria = {
         '題旨發揮': {'weight': 0.4, 'max_score': 40},
         '資料掌握': {'weight': 0.2, 'max_score': 20},
@@ -21,54 +27,29 @@ def grade_essay(essay_content, additional_prompt=""):
         'C+': 65, 'C': 60, 'C-': 55,
         '0': 0
     }
-
-    def get_evaluation(prompt):
-        response = client.chat.completions.create(
-            model="gpt-3.5-turbo",
-            messages=[
-                {"role": "system", "content": "你是一位資深的國文作文評閱委員，請依據提供的評分規準進行評分。"},
-                {"role": "user", "content": prompt}
-            ],
-            temperature=0.7
-        )
-        return response.choices[0].message.content
-
-    def parse_response(response):
-        lines = response.lower().split('\n')
-        grade = '0'
-        comment = ""
-        
-        for line in lines:
-            if '等第：' in line or '等第:' in line:
-                grade_text = line.split('：')[-1].strip().upper()
-                if grade_text in grade_scores:
-                    grade = grade_text
-            elif '評語：' in line or '評語:' in line:
-                comment = line.split('：')[-1].strip()
-        
-        return grade, comment
-
+    
+    history.append({"role": "user", "content": f"要評分的作文內容：\n\n{message}"})
+    
     try:
         scores = {}
-        feedback = {}
-        result_text = []
+        feedback = []
+        total_score = 0
         
-        # 添加額外提示詞到作文內容
-        if additional_prompt:
-            evaluation_content = f"{essay_content}\n\n額外評分提示：{additional_prompt}"
-        else:
-            evaluation_content = essay_content
+        # 評分前的提示訊息
+        history.append({
+            "role": "assistant", 
+            "content": "正在進行評分分析...",
+            "metadata": {"title": "🔍 開始評分"}
+        })
         
+        # 評估每個面向
         for criterion, details in criteria.items():
             prompt = f"""評估以下作文的{criterion}（權重{details['weight']*100}%）：
 
-評分標準：
-A等（A+、A、A-）：優秀表現
-B等（B+、B、B-）：中等表現
-C等（C+、C、C-）：待加強表現
-
 作文內容：
-{evaluation_content}
+{message}
+
+{additional_prompt if additional_prompt else ''}
 
 請依三等九級制（A+、A、A-、B+、B、B-、C+、C、C-）評分，並提供簡短評語。
 如果是缺考、未作答、完全文不對題或作答內容完全照抄試題，請給予0分。
@@ -76,67 +57,102 @@ C等（C+、C、C-）：待加強表現
 請按以下格式回覆：
 等第：
 評語："""
+
+            response = client.chat.completions.create(
+                model="gpt-3.5-turbo",
+                messages=[
+                    {"role": "system", "content": "你是一位資深的國文作文評閱委員，請依據提供的評分規準進行評分。"},
+                    {"role": "user", "content": prompt}
+                ],
+                temperature=0.7
+            )
             
-            response = get_evaluation(prompt)
-            grade, comment = parse_response(response)
+            result = response.choices[0].message.content
+            lines = result.lower().split('\n')
+            grade = '0'
+            comment = ""
             
-            raw_score = grade_scores[grade]
-            weighted_score = (raw_score / 100) * details['max_score']
+            for line in lines:
+                if '等第：' in line or '等第:' in line:
+                    grade_text = line.split('：')[-1].strip().upper()
+                    if grade_text in grade_scores:
+                        grade = grade_text
+                elif '評語：' in line or '評語:' in line:
+                    comment = line.split('：')[-1].strip()
             
-            result_text.append(f"【{criterion}】")
-            result_text.append(f"等第: {grade}")
-            result_text.append(f"加權分數: {weighted_score:.1f}/{details['max_score']}")
-            result_text.append(f"評語: {comment}")
-            result_text.append("")
+            weighted_score = (grade_scores[grade] / 100) * details['max_score']
+            total_score += weighted_score
             
-            scores[criterion] = weighted_score
+            feedback.append(f"**{criterion}**  \n")
+            feedback.append(f"等第：{grade}  \n")
+            feedback.append(f"得分：{weighted_score:.1f}/{details['max_score']}  \n")
+            feedback.append(f"評語：{comment}  \n\n")
         
-        total_score = sum(scores.values())
-        result_text.append(f"總分: {total_score:.1f}/100")
-        
-        return "\n".join(result_text)
+        # 添加評分結果
+        history.append({
+            "role": "assistant",
+            "content": "".join(feedback) + f"**總分：{total_score:.1f}/100**",
+            "metadata": {"title": "📝 評分結果"}
+        })
         
     except Exception as e:
-        return f"評分過程發生錯誤: {str(e)}"
+        history.append({
+            "role": "assistant",
+            "content": f"評分過程發生錯誤：{str(e)}",
+            "metadata": {"title": "❌ 錯誤"}
+        })
+    
+    return history
 
-# 創建 Gradio 介面
 def create_interface():
     with gr.Blocks(title="國文作文自動評分系統") as iface:
         gr.Markdown("""
         # 國文作文自動評分系統
         ## 評分標準說明
-        本系統依據以下四個面向進行評分：
+        本系統採用以下四個面向進行評分：
         - 題旨發揮 (40%)
         - 資料掌握 (20%)
         - 結構安排 (20%)
         - 字句運用 (20%)
         
-        採用三等九級制（A+、A、A-、B+、B、B-、C+、C、C-）進行評分。
+        評分採用三等九級制（A+、A、A-、B+、B、B-、C+、C、C-）
         """)
         
-        with gr.Column():
-            essay_input = gr.Textbox(
-                label="作文內容",
-                placeholder="請在此輸入作文內容...",
-                lines=10
-            )
-            prompt_input = gr.Textbox(
-                label="額外評分提示（選填）",
-                placeholder="可輸入額外的評分提示或要求...",
-                lines=2
-            )
-            grade_button = gr.Button("開始評分", variant="primary")
-            output = gr.Textbox(label="評分結果", lines=15)
+        with gr.Row():
+            with gr.Column(scale=4):
+                chatbot = gr.Chatbot(
+                    value=[],
+                    height=500,
+                    bubble_full_width=False,
+                    avatar_images=("user.png", "assistant.png"),
+                    show_copy_button=True,
+                    layout="bubble",
+                    render_markdown=True
+                )
+                msg = gr.Textbox(
+                    label="請輸入作文內容",
+                    placeholder="在此輸入作文...",
+                    lines=3
+                )
+                additional_prompt = gr.Textbox(
+                    label="額外評分提示（選填）",
+                    placeholder="可輸入額外的評分要求或提示...",
+                    lines=2
+                )
+        
+        with gr.Row():
+            submit = gr.Button("開始評分", variant="primary")
+            clear = gr.Button("清除對話")
             
-        grade_button.click(
-            fn=grade_essay,
-            inputs=[essay_input, prompt_input],
-            outputs=output
+        submit.click(
+            evaluate_essay,
+            inputs=[msg, chatbot, additional_prompt],
+            outputs=chatbot
         )
-    
+        clear.click(lambda: [], outputs=chatbot)
+        
     return iface
 
-# 啟動應用
 if __name__ == "__main__":
     demo = create_interface()
     demo.launch(share=True)