Fix knowledge map to scan upload directory when no session data is available

gemini_chatbot.py

@@ -454,6 +454,13 @@ def upload_file():
         file.save(file_path)
         logging.info(f"File saved: {file_path}")
         
+        # 检查文件是否真的已经保存
+        if not os.path.exists(file_path):
+            logging.error(f"文件保存失败，路径不存在: {file_path}")
+            return jsonify({"error": "Failed to save file", "success": False}), 500
+        else:
+            logging.info(f"文件成功保存，大小: {os.path.getsize(file_path)} 字节")
+        
         # Extract text
         text_content = None
         if file_ext.lower() == '.pdf':
@@ -483,6 +490,10 @@ def upload_file():
             summary = "Could not generate summary, but you can still ask questions about the document."
             tags = []
         
+        # 记录当前会话状态
+        logging.debug(f"当前会话ID: {session.sid if hasattr(session, 'sid') else '未知'}")
+        logging.debug(f"当前会话内容 (before update): {list(session.keys())}")
+        
         # Store session data - uploaded file context
         if 'uploaded_files_context' not in session:
             session['uploaded_files_context'] = []
@@ -509,8 +520,23 @@ def upload_file():
         
         # Ensure session changes are saved
         session.modified = True
+        
+        # 打印更新后的会话信息
+        logging.debug(f"会话更新后的内容: {list(session.keys())}")
+        logging.debug(f"uploaded_files_context 长度: {len(session.get('uploaded_files_context', []))}")
+        logging.debug(f"uploaded_files 长度: {len(session.get('uploaded_files', []))}")
+        
         logging.info(f"Stored context for {safe_filename} in session with {len(tags)} tags")
         
+        # 添加对上传目录的列表，确认文件已保存
+        try:
+            upload_dir = app.config['UPLOAD_FOLDER']
+            if os.path.exists(upload_dir):
+                files_in_dir = os.listdir(upload_dir)
+                logging.info(f"上传目录中的文件: {files_in_dir}")
+        except Exception as e:
+            logging.error(f"列出上传目录文件失败: {e}")
+        
         return jsonify({
             "success": True,
             "filename": safe_filename,
@@ -576,7 +602,7 @@ def get_knowledge_map():
     uploaded_files = session.get('uploaded_files', [])
     
     logging.debug(f"Knowledge map API - Session data keys: {session.keys()}")
-    logging.debug(f"Knowledge map API - Files found: {len(uploaded_files)}")
+    logging.debug(f"Knowledge map API - Files found in session: {len(uploaded_files)}")
     
     # 如果没有上传文件，检查context数据作为备选
     if not uploaded_files and 'uploaded_files_context' in session:
@@ -590,9 +616,77 @@ def get_knowledge_map():
             })
         logging.debug(f"Used context data as fallback, found {len(uploaded_files)} files")
     
+    # 如果会话中仍然没有文件信息，直接从上传目录扫描文件
+    if not uploaded_files:
+        logging.info("No files found in session, scanning upload directory...")
+        try:
+            upload_dir = app.config['UPLOAD_FOLDER']
+            if os.path.exists(upload_dir):
+                files_in_dir = os.listdir(upload_dir)
+                logging.info(f"Files found in upload directory: {files_in_dir}")
+                
+                # 过滤出PDF和TXT文件
+                valid_files = [f for f in files_in_dir if f.lower().endswith(('.pdf', '.txt'))]
+                
+                for filename in valid_files:
+                    # 从文件名中提取原始文件名（去除UUID前缀）
+                    original_filename = re.sub(r'^[a-f0-9\-]+_', '', filename)
+                    
+                    # 尝试从文件中提取内容并生成摘要
+                    file_path = os.path.join(upload_dir, filename)
+                    try:
+                        # 提取文本内容
+                        text_content = None
+                        if filename.lower().endswith('.pdf'):
+                            text_content = extract_text_from_pdf(file_path)
+                        elif filename.lower().endswith('.txt'):
+                            text_content = extract_text_from_txt(file_path)
+                            
+                        # 如果成功提取了文本，生成摘要和标签
+                        if text_content:
+                            try:
+                                summary, tags = get_summary_and_tags_from_gemini(text_content)
+                            except Exception as e:
+                                logging.error(f"Error generating summary for scanned file: {e}")
+                                summary = "Auto-generated summary for " + original_filename
+                                tags = ["Uncategorized"]
+                                
+                            # 将文件信息添加到上传文件列表中
+                            uploaded_files.append({
+                                'filename': original_filename,
+                                'summary': summary,
+                                'tags': tags
+                            })
+                            
+                            # 同时更新会话中的文件信息，以便后续使用
+                            if 'uploaded_files_context' not in session:
+                                session['uploaded_files_context'] = []
+                                
+                            # 检查文件是否已经在会话中
+                            file_exists = False
+                            for existing_file in session.get('uploaded_files_context', []):
+                                if existing_file.get('unique_filename') == filename:
+                                    file_exists = True
+                                    break
+                                    
+                            if not file_exists:
+                                session['uploaded_files_context'].append({
+                                    'filename': original_filename,
+                                    'unique_filename': filename,
+                                    'summary': summary,
+                                    'tags': tags
+                                })
+                            
+                            # 确保会话更改被保存
+                            session.modified = True
+                    except Exception as e:
+                        logging.error(f"Error processing scanned file {filename}: {e}")
+        except Exception as e:
+            logging.error(f"Error scanning upload directory: {e}")
+    
     # 如果还是没有上传文件，返回空数据
     if not uploaded_files:
-        logging.warning("No uploaded files found in session for knowledge map")
+        logging.warning("No uploaded files found in session or directory for knowledge map")
         return jsonify({
             "centralTopic": "Knowledge Center",
             "documents": []