[前端+RAG] 异步上传+前端轮询自动刷新导读；PDF阅读模式合并行消除留白

langchain-chat/server/knowledge_base/file_converter.py

@@ -957,22 +957,33 @@ class FileConverter:
 
                     if text.strip():
                         any_text = True
-                        # 按行处理文本，识别标题
                         lines = text.split('\n')
+                        # 合并连续非空行为段落，空行分段，标题行独立
+                        current_para = []
                         for line in lines:
-                            line = line.strip()
-                            if not line:
+                            stripped = line.strip()
+                            if not stripped:
+                                # 空行 → 结束当前段落
+                                if current_para:
+                                    page_parts.append(f'<p>{self._escape_html("".join(current_para))}</p>')
+                                    current_para = []
                                 continue
-                            # 简单的标题检测：短行 + 无标点结尾
-                            is_heading = (len(line) < 40 and not line.endswith(('。', '，', '；', '、', '：', ',', '.', ';'))
-                                          and not line.startswith(('（', '('))
-                                          and re.match(r'^[一二三四五六七八九十\d]+[、.．]', line))
+                            # 标题检测
+                            is_heading = (len(stripped) < 30
+                                          and not stripped.endswith(('。', '，', '；', '、', '：', ',', '.', ';'))
+                                          and not stripped.startswith(('（', '('))
+                                          and re.match(r'^[一二三四五六七八九十\d]+[、.．]', stripped))
                             if is_heading:
-                                escaped = self._escape_html(line)
-                                page_parts.append(f'<h3>{escaped}</h3>')
+                                # 先输出累积的段落
+                                if current_para:
+                                    page_parts.append(f'<p>{self._escape_html("".join(current_para))}</p>')
+                                    current_para = []
+                                page_parts.append(f'<h3>{self._escape_html(stripped)}</h3>')
                             else:
-                                escaped = self._escape_html(line)
-                                page_parts.append(f'<p>{escaped}</p>')
+                                current_para.append(stripped)
+                        # 输出最后一个段落
+                        if current_para:
+                            page_parts.append(f'<p>{self._escape_html("".join(current_para))}</p>')
 
                     # 渲染表格
                     for table in tables:

langchain-chat/server/knowledge_base/kb_doc_api.py

@@ -270,6 +270,74 @@ def upload_docs(
     return BaseResponse(code=200, msg="文件上传与向量化完成", data={"failed_files": failed_files})
 
 
+def _background_generate_and_update(
+    knowledge_base_name: str,
+    file_names: List[str],
+    chunk_size: int,
+    chunk_overlap: int,
+    zh_title_enhance: bool,
+    docs: dict,
+    not_refresh_vs_cache: bool,
+):
+    """后台线程：生成 LLM 导读 + 向量化，完成后直连 MySQL 更新。"""
+    import time
+    import pymysql
+    start = time.time()
+
+    kb = KBServiceFactory.get_service_by_name(knowledge_base_name)
+
+    for filename in file_names:
+        try:
+            knowledge_file = KnowledgeFile(filename=filename, knowledge_base_name=knowledge_base_name)
+            new_loop = asyncio.new_event_loop()
+            asyncio.set_event_loop(new_loop)
+            try:
+                llm_result = new_loop.run_until_complete(knowledge_file.get_llm_result())
+            finally:
+                new_loop.close()
+
+            logger.info(f"[后台] LLM 导读生成完成: {filename}")
+
+            # 直连 MySQL 更新（用 embedding_id 匹配，因为 Java 端 embedding_id = filename）
+            try:
+                conn = pymysql.connect(**ck_mysql_config)
+                with conn.cursor() as cursor:
+                    updated = cursor.execute(
+                        "UPDATE gpt_upload_file SET article_abstract=%s, article_keywords=%s, article_paragraph=%s WHERE embedding_id=%s",
+                        (
+                            str(llm_result.get("article_abstract", "生成摘要失败")),
+                            str(llm_result.get("article_keywords", "生成关键词失败")),
+                            str(llm_result.get("article_paragraph", "生成章节速览失败")),
+                            filename
+                        )
+                    )
+                    conn.commit()
+                    logger.info(f"[后台] MySQL 更新成功: {filename}, affected rows: {updated}")
+                conn.close()
+            except Exception as db_e:
+                logger.error(f"[后台] MySQL 更新失败 {filename}: {db_e}", exc_info=True)
+        except Exception as e:
+            logger.error(f"[后台] LLM 生成失败 {filename}: {e}", exc_info=True)
+
+    # 向量化
+    try:
+        _update_docs_impl(
+            knowledge_base_name=knowledge_base_name,
+            file_names=file_names,
+            override_custom_docs=True,
+            chunk_size=chunk_size,
+            chunk_overlap=chunk_overlap,
+            zh_title_enhance=zh_title_enhance,
+            docs=docs,
+            not_refresh_vs_cache=True,
+        )
+        if kb and not not_refresh_vs_cache:
+            kb.save_vector_store()
+    except Exception as e:
+        logger.error(f"[后台] 向量化失败: {e}", exc_info=True)
+    logger.info(f"[后台] 全部完成，耗时: {time.time() - start:.2f}s")
+
+
 def upload_docs_new(
         files: List[UploadFile] = File(..., description="上传文件，支持多文件"),
         knowledge_base_name: str = Form(..., description="知识库名称", examples=["samples"]),
@@ -283,7 +351,7 @@ def upload_docs_new(
         not_refresh_vs_cache: bool = Form(False, description="暂不保存向量库（用于FAISS）"),
 ) -> BaseResponse:
     """
-    API接口：上传文件，同步生成导读（模型已优化为deepseek-v3），然后向量化
+    API接口：上传文件，快速返回（仅提取全文），LLM导读+向量化后台异步执行并直连MySQL回写
     """
     import time
     start_time = time.time()
@@ -305,64 +373,53 @@ def upload_docs_new(
     file_names = list(docs.keys())
     llm_results = {}
 
+    # 保存文件 + 提取全文（快速，不调 LLM）
     for result in _save_files_in_thread(files, knowledge_base_name=knowledge_base_name, override=override):
         filename = result["data"]["file_name"]
         if result["code"] != 200:
             failed_files[filename] = result["msg"]
-
         if filename not in file_names:
             file_names.append(filename)
 
         try:
             knowledge_file = KnowledgeFile(filename=filename, knowledge_base_name=knowledge_base_name)
-            import concurrent.futures
-            def run_async_in_thread():
-                new_loop = asyncio.new_event_loop()
-                asyncio.set_event_loop(new_loop)
-                try:
-                    return new_loop.run_until_complete(knowledge_file.get_llm_result())
-                finally:
-                    new_loop.close()
-
-            with concurrent.futures.ThreadPoolExecutor(max_workers=1) as executor:
-                future = executor.submit(run_async_in_thread)
-                llm_result = future.result()
+            full_text_data = knowledge_file.get_full_text()
+            import json as _json
+            try:
+                full_text = _json.loads(full_text_data).get("full_text", "")
+            except:
+                full_text = ""
             llm_results[filename] = {
-                "full_text": llm_result.get("full_text", "获取全文失败"),
-                "article_abstract": llm_result.get("article_abstract", "生成摘要失败"),
-                "article_keywords": llm_result.get("article_keywords", "生成关键词失败"),
-                "article_paragraph": llm_result.get("article_paragraph", "生成章节速览失败")
+                "full_text": full_text,
+                "article_abstract": "导读生成中，请稍后刷新...",
+                "article_keywords": "导读生成中，请稍后刷新...",
+                "article_paragraph": "导读生成中，请稍后刷新..."
             }
         except Exception as e:
-            logger.error(f"生成LLM结果时出错：{e}", exc_info=e if log_verbose else None)
+            logger.error(f"提取全文失败 {filename}: {e}")
             llm_results[filename] = {
-                "article_abstract": "生成摘要失败",
-                "article_keywords": "生成关键词失败",
-                "article_paragraph": "生成章节速览失败"
+                "full_text": "",
+                "article_abstract": "导读生成中，请稍后刷新...",
+                "article_keywords": "导读生成中，请稍后刷新...",
+                "article_paragraph": "导读生成中，请稍后刷新..."
             }
 
-    if to_vector_store:
-        update_st = time.time()
-        result = _update_docs_impl(
-            knowledge_base_name=knowledge_base_name,
-            file_names=file_names,
-            override_custom_docs=True,
-            chunk_size=chunk_size,
-            chunk_overlap=chunk_overlap,
-            zh_title_enhance=zh_title_enhance,
-            docs=docs,
-            not_refresh_vs_cache=True,
-        )
-        failed_files.update(result.data["failed_files"])
-        if not not_refresh_vs_cache:
-            kb.save_vector_store()
-        logger.info(f'向量化用时：{time.time() - update_st}')
-    logger.info(f"总执行时间: {time.time() - start_time:.2f}s")
-    return BaseResponse(code=200, msg="文件上传与向量化完成", data={
+    # 后台线程：LLM 导读 + 向量化 + MySQL 回写
+    import threading
+    threading.Thread(
+        target=_background_generate_and_update,
+        args=(knowledge_base_name, file_names, chunk_size, chunk_overlap,
+              zh_title_enhance, docs, not_refresh_vs_cache),
+        daemon=True
+    ).start()
+
+    logger.info(f"上传+全文提取: {time.time() - start_time:.2f}s，后台生成中")
+    return BaseResponse(code=200, msg="文件上传完成", data={
         "failed_files": failed_files,
         "llm_results": llm_results
     })
 
+
 def delete_docs(
         knowledge_base_name: str = Body(..., examples=["samples"]),
         file_names: List[str] = Body(..., examples=[["file_name.md", "test.txt"]]),