Update app.py

app.py

@@ -1,4 +1,5 @@
 import os
+import re
 from typing import List, Dict, Tuple, Optional
 
 import gradio as gr
@@ -192,6 +193,72 @@ def describe_cognitive_state(state: Optional[Dict[str, int]]) -> str:
         return "mixed or uncertain cognitive state."
 
 
+# ---------- Same Question Check helpers ----------
+def _normalize_text(text: str) -> str:
+    """
+    将文本转为小写、去除标点和多余空格，用于简单相似度计算。
+    """
+    text = text.lower().strip()
+    # 去掉标点符号，只保留字母数字和空格
+    text = re.sub(r"[^\w\s]", " ", text)
+    # 合并多余空格
+    text = re.sub(r"\s+", " ", text)
+    return text
+
+
+def _jaccard_similarity(a: str, b: str) -> float:
+    tokens_a = set(a.split())
+    tokens_b = set(b.split())
+    if not tokens_a or not tokens_b:
+        return 0.0
+    return len(tokens_a & tokens_b) / len(tokens_a | tokens_b)
+
+
+def find_similar_past_question(
+    message: str,
+    history: List[Tuple[str, str]],
+    similarity_threshold: float = 0.8,
+    max_turns_to_check: int = 6,
+) -> Optional[Tuple[str, str, float]]:
+    """
+    在最近若干轮历史对话中查找与当前问题相似的既往问题。
+
+    返回:
+        (past_question, past_answer, similarity) 或 None
+    """
+    norm_msg = _normalize_text(message)
+    if not norm_msg:
+        return None
+
+    best_sim = 0.0
+    best_pair: Optional[Tuple[str, str]] = None
+    checked = 0
+
+    # 从最近一轮往前看
+    for user_q, assistant_a in reversed(history):
+        checked += 1
+        if checked > max_turns_to_check:
+            break
+
+        norm_hist_q = _normalize_text(user_q)
+        if not norm_hist_q:
+            continue
+
+        # 完全相同直接返回
+        if norm_msg == norm_hist_q:
+            return user_q, assistant_a, 1.0
+
+        sim = _jaccard_similarity(norm_msg, norm_hist_q)
+        if sim > best_sim:
+            best_sim = sim
+            best_pair = (user_q, assistant_a)
+
+    if best_pair and best_sim >= similarity_threshold:
+        return best_pair[0], best_pair[1], best_sim
+
+    return None
+
+
 # ---------- 构建 messages ----------
 def build_messages(
     user_message: str,
@@ -617,7 +684,7 @@ with gr.Blocks(title="Clare – Hanbridge AI Teaching Assistant") as demo:
         lines=8,
     )
 
-    # 主对话逻辑：更新弱项 + 认知状态 + 调用 Clare
+    # 主对话逻辑：Same Question Check + 更新弱项 + 认知状态 + 调用 Clare
     def respond(
         message,
         chat_history,
@@ -629,9 +696,37 @@ with gr.Blocks(title="Clare – Hanbridge AI Teaching Assistant") as demo:
         learning_mode_val,
         doc_type_val,
     ):
+        # 先更新弱项和认知状态
         weaknesses = update_weaknesses_from_message(message, weaknesses or [])
         cognitive_state = update_cognitive_state_from_message(message, cognitive_state)
 
+        # ---------- Same Question Check ----------
+        dup = find_similar_past_question(message, chat_history)
+        if dup is not None:
+            past_q, past_a, sim = dup
+            # 直接复用之前回答，并给一个简短提示
+            prefix_en = (
+                "I noticed this question is very similar to one you asked earlier, "
+                "so I'm showing the previous explanation again. "
+                "If there's a specific part that's still unclear, tell me and I can "
+                "re-explain it in a different way.\n\n"
+                "**Earlier answer:**\n"
+            )
+            prefix_zh = (
+                "我注意到你现在的问题和之前问过的非常相似，"
+                "所以我先把当时的回答再展示一次。"
+                "如果还有具体不清楚的地方，可以告诉我，我会换一种方式解释。\n\n"
+                "**之前的回答：**\n"
+            )
+            if language_pref_val == "中文":
+                answer = prefix_zh + past_a
+            else:
+                answer = prefix_en + past_a
+
+            new_history = chat_history + [(message, answer)]
+            return "", new_history, weaknesses, cognitive_state
+
+        # ---------- 正常调用 Clare ----------
         answer, new_history = chat_with_clare(
             message=message,
             history=chat_history,