Update app.py

app.py

@@ -22,30 +22,10 @@ from clare_core import (
     get_empty_input_prompt,
     summarize_conversation,
 )
-
-def respond(
-    message,
-    chat_history,
-    course_outline,
-    weaknesses,
-    cognitive_state,
-    model_name_val,
-    language_pref_val,
-    learning_mode_val,
-    doc_type_val,
-):
-    # 先根据 Auto / English / 中文 决定本轮用什么语言
-    resolved_lang = detect_language(message or "", language_pref_val)
-
-    # ---------- 空输入防护 ----------
-    if not message or not message.strip():
-        # 不更新弱项和认知状态，只给一个友好提示
-        empty_msg = get_empty_input_prompt(resolved_lang)
-        new_history = chat_history + [("", empty_msg)]
-        status_text = render_session_status(learning_mode_val, weaknesses or [], cognitive_state)
-        # 清空输入框，更新聊天记录 & 状态栏，其余状态不变
-        return "", new_history, weaknesses, cognitive_state, status_text
-        
+from rag_engine import (
+    build_rag_chunks_from_file,
+    retrieve_relevant_chunks,
+)
 
 with gr.Blocks(title="Clare – Hanbridge AI Teaching Assistant") as demo:
     gr.Markdown(
@@ -90,29 +70,39 @@ with gr.Blocks(title="Clare – Hanbridge AI Teaching Assistant") as demo:
             label="File type",
         )
 
-    # 状态：课程大纲 + 学生弱项 + 认知状态
+    # 状态：课程大纲 + 学生弱项 + 认知状态 + RAG chunks
     course_outline_state = gr.State(DEFAULT_COURSE_TOPICS)
     weakness_state = gr.State([])
     cognitive_state_state = gr.State({"confusion": 0, "mastery": 0})
+    rag_chunks_state = gr.State([])  # Session 级 RAG 向量库
 
-    # 上传 syllabus 时更新课程大纲
-    def update_outline(file, doc_type_val):
+    # 上传 syllabus 时更新课程大纲 + RAG chunks
+    def update_course_and_rag(file, doc_type_val):
+        # 更新课程大纲（保持原有逻辑）
         if file is None:
-            return DEFAULT_COURSE_TOPICS
-        if doc_type_val == "Syllabus":
-            try:
-                file_path = file.name
-                if file_path.lower().endswith(".docx"):
-                    topics = parse_syllabus_docx(file_path)
-                    return topics
-            except Exception:
-                return DEFAULT_COURSE_TOPICS
-        return DEFAULT_COURSE_TOPICS
+            topics = DEFAULT_COURSE_TOPICS
+        else:
+            if doc_type_val == "Syllabus":
+                try:
+                    file_path = file.name
+                    if file_path.lower().endswith(".docx"):
+                        topics = parse_syllabus_docx(file_path)
+                    else:
+                        topics = DEFAULT_COURSE_TOPICS
+                except Exception:
+                    topics = DEFAULT_COURSE_TOPICS
+            else:
+                topics = DEFAULT_COURSE_TOPICS
+
+        # 构建 RAG chunks
+        rag_chunks = build_rag_chunks_from_file(file, doc_type_val)
+
+        return topics, rag_chunks
 
     syllabus_file.change(
-        fn=update_outline,
+        fn=update_course_and_rag,
         inputs=[syllabus_file, doc_type],
-        outputs=[course_outline_state],
+        outputs=[course_outline_state, rag_chunks_state],
     )
 
     # 左侧聊天，右侧 Session 状态栏
@@ -157,19 +147,27 @@ with gr.Blocks(title="Clare – Hanbridge AI Teaching Assistant") as demo:
         course_outline,
         weaknesses,
         cognitive_state,
+        rag_chunks,
         model_name_val,
         language_pref_val,
         learning_mode_val,
         doc_type_val,
     ):
         # 1) 决定本轮语言（Auto / English / 中文）
-        resolved_lang = detect_language(message, language_pref_val)
+        resolved_lang = detect_language(message or "", language_pref_val)
+
+        # 2) 空输入防护
+        if not message or not message.strip():
+            empty_msg = get_empty_input_prompt(resolved_lang)
+            new_history = chat_history + [("", empty_msg)]
+            status_text = render_session_status(learning_mode_val, weaknesses or [], cognitive_state)
+            return "", new_history, weaknesses, cognitive_state, status_text
 
-        # 2) 更新弱项 & 认知状态
+        # 3) 更新弱项 & 认知状态
         weaknesses = update_weaknesses_from_message(message, weaknesses or [])
         cognitive_state = update_cognitive_state_from_message(message, cognitive_state)
 
-        # 3) Same Question Check
+        # 4) Same Question Check（session 内复用答案）
         dup = find_similar_past_question(message, chat_history)
         if dup is not None:
             past_q, past_a, sim = dup
@@ -195,7 +193,10 @@ with gr.Blocks(title="Clare – Hanbridge AI Teaching Assistant") as demo:
             status_text = render_session_status(learning_mode_val, weaknesses, cognitive_state)
             return "", new_history, weaknesses, cognitive_state, status_text
 
-        # 4) 正常调用 Clare
+        # 5) RAG：基于上传文档做检索
+        rag_context = retrieve_relevant_chunks(message, rag_chunks or [])
+
+        # 6) 正常调用 Clare（带上 RAG context）
         answer, new_history = chat_with_clare(
             message=message,
             history=chat_history,
@@ -206,6 +207,7 @@ with gr.Blocks(title="Clare – Hanbridge AI Teaching Assistant") as demo:
             course_outline=course_outline,
             weaknesses=weaknesses,
             cognitive_state=cognitive_state,
+            rag_context=rag_context,
         )
 
         status_text = render_session_status(learning_mode_val, weaknesses, cognitive_state)
@@ -219,6 +221,7 @@ with gr.Blocks(title="Clare – Hanbridge AI Teaching Assistant") as demo:
             course_outline_state,
             weakness_state,
             cognitive_state_state,
+            rag_chunks_state,
             model_name,
             language_preference,
             learning_mode,
@@ -231,12 +234,30 @@ with gr.Blocks(title="Clare – Hanbridge AI Teaching Assistant") as demo:
     def clear_all():
         empty_state = {"confusion": 0, "mastery": 0}
         status_text = render_session_status("Concept Explainer", [], empty_state)
-        return [], [], empty_state, "", "", "", status_text
+        return (
+            [],          # chatbot
+            [],          # weaknesses
+            empty_state, # cognitive_state
+            [],          # rag_chunks
+            "",          # export_box
+            "",          # quiz_box
+            "",          # summary_box
+            status_text, # session_status
+        )
 
     clear_btn.click(
         clear_all,
         None,
-        [chatbot, weakness_state, cognitive_state_state, export_box, quiz_box, summary_box, session_status],
+        [
+            chatbot,
+            weakness_state,
+            cognitive_state_state,
+            rag_chunks_state,
+            export_box,
+            quiz_box,
+            summary_box,
+            session_status,
+        ],
         queue=False,
     )