feat: 完善后端知识库服务和配置

- 优化 AI-Core 客户端调用
- 添加更多知识库配置选项
- 完善文档解析逻辑

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

server/internal/service/document_parser_client.go

@@ -7,6 +7,8 @@ import (
 
 	"google.golang.org/grpc"
 	"google.golang.org/grpc/credentials/insecure"
+
+	docparser "x-agents/server"
 )
 
 // AICoreClient AI-Core 文档解析服务客户端
@@ -53,7 +55,7 @@ func (c *AICoreClient) Close() {
 	}
 }
 
-// ParseDocument 解析文档
+// ParseDocument 解析文档 - 使用生成的 protobuf 代码
 func (c *AICoreClient) ParseDocument(fileURL, fileName, fileType string) (*ParseResult, error) {
 	if c.conn == nil {
 		if err := c.Connect(); err != nil {
@@ -61,17 +63,16 @@ func (c *AICoreClient) ParseDocument(fileURL, fileName, fileType string) (*Parse
 		}
 	}
 
-	// 使用 gRPC raw bytes 调用
-	// 由于没有生成 protobuf 代码，使用 raw bytes 方式调用
-	client := NewDocumentParserClient(c.conn)
+	// 使用生成的 protobuf 客户端
+	client := docparser.NewDocumentParserClient(c.conn)
 
-	req := &ParseRequest{
+	req := &docparser.ParseRequest{
 		FileUrl:  fileURL,
 		FileName: fileName,
 		FileType: fileType,
 	}
 
-	ctx, cancel := context.WithTimeout(context.Background(), 60*time.Second)
+	ctx, cancel := context.WithTimeout(context.Background(), 120*time.Second)
 	defer cancel()
 
 	resp, err := client.ParseDocument(ctx, req)
@@ -80,53 +81,11 @@ func (c *AICoreClient) ParseDocument(fileURL, fileName, fileType string) (*Parse
 	}
 
 	return &ParseResult{
-		Success:       resp.Success,
-		Content:       resp.Content,
-		Message:       resp.Message,
-		ContentLength: resp.ContentLength,
-		FileType:      resp.FileType,
-		ParserEngine:  resp.ParserEngine,
+		Success:       resp.GetSuccess(),
+		Content:       resp.GetContent(),
+		Message:       resp.GetMessage(),
+		ContentLength: resp.GetContentLength(),
+		FileType:      resp.GetFileType(),
+		ParserEngine:  resp.GetParserEngine(),
 	}, nil
 }
-
-// 以下是手动定义的 protobuf messages（与 proto 文件一致）
-// 不需要生成 .pb.go 文件，直接手动定义
-
-type ParseRequest struct {
-	FileUrl       string
-	FileName     string
-	FileType     string
-	ParserEngine string
-}
-
-type ParseResponse struct {
-	Success       bool
-	Content       string
-	Message       string
-	ContentLength int32
-	FileType      string
-	ParserEngine  string
-}
-
-// DocumentParserClient gRPC 客户端接口（手动实现）
-type DocumentParserClient interface {
-	ParseDocument(ctx context.Context, in *ParseRequest, opts ...grpc.CallOption) (*ParseResponse, error)
-}
-
-type documentParserClient struct {
-	cc grpc.ClientConnInterface
-}
-
-// NewDocumentParserClient 创建 DocumentParser 客户端
-func NewDocumentParserClient(cc grpc.ClientConnInterface) DocumentParserClient {
-	return &documentParserClient{cc: cc}
-}
-
-func (c *documentParserClient) ParseDocument(ctx context.Context, in *ParseRequest, opts ...grpc.CallOption) (*ParseResponse, error) {
-	out := new(ParseResponse)
-	err := c.cc.Invoke(ctx, "/docparser.DocumentParser/ParseDocument", in, out, opts...)
-	if err != nil {
-		return nil, err
-	}
-	return out, nil
-}

server/internal/service/knowledge_service.go

@@ -29,9 +29,10 @@ type KnowledgeService struct {
 	uploadService    *UploadService
 	pythonServiceURL string
 	aiCoreClient     *AICoreClient
+	markdownLocalPath string // Markdown 本地存储路径
 }
 
-func NewKnowledgeService(repo *repository.KnowledgeRepository, modelRepo *repository.ModelRepository, uploadService *UploadService, pythonServiceURL, aiCoreServiceAddr string) *KnowledgeService {
+func NewKnowledgeService(repo *repository.KnowledgeRepository, modelRepo *repository.ModelRepository, uploadService *UploadService, pythonServiceURL, aiCoreServiceAddr, markdownLocalPath string) *KnowledgeService {
 	aiCoreClient, _ := NewAICoreClient(aiCoreServiceAddr)
 	return &KnowledgeService{
 		repo:             repo,
@@ -39,6 +40,7 @@ func NewKnowledgeService(repo *repository.KnowledgeRepository, modelRepo *reposi
 		uploadService:    uploadService,
 		pythonServiceURL: pythonServiceURL,
 		aiCoreClient:     aiCoreClient,
+		markdownLocalPath: markdownLocalPath,
 	}
 }
 
@@ -307,6 +309,13 @@ func (s *KnowledgeService) parseDocumentWithAICore(docID, fileURL, fileName stri
 
 	if result.Success && result.Content != "" {
 		knowledgeDebugLog.Printf("[AICore] 解析成功: docID=%s, contentLength=%d", docID, len(result.Content))
+
+		// 保存到本地文件
+		markdownPath := s.saveMarkdownToFile(docID, fileName, result.Content)
+		if markdownPath != "" {
+			knowledgeDebugLog.Printf("[AICore] Markdown 保存到本地: docID=%s, path=%s", docID, markdownPath)
+		}
+
 		// 更新文档的 Content 字段
 		s.repo.UpdateDocument(docID, map[string]interface{}{
 			"content": result.Content,
@@ -316,6 +325,31 @@ func (s *KnowledgeService) parseDocumentWithAICore(docID, fileURL, fileName stri
 	}
 }
 
+// saveMarkdownToFile 保存 Markdown 内容到本地文件
+func (s *KnowledgeService) saveMarkdownToFile(docID, fileName, content string) string {
+	if s.markdownLocalPath == "" {
+		s.markdownLocalPath = "resource/markdown"
+	}
+
+	// 创建目录
+	if err := os.MkdirAll(s.markdownLocalPath, 0755); err != nil {
+		knowledgeDebugLog.Printf("[AICore] 创建目录失败: path=%s, err=%v", s.markdownLocalPath, err)
+		return ""
+	}
+
+	// 生成文件名（用 docID + .md）
+	markdownFileName := docID + ".md"
+	markdownPath := s.markdownLocalPath + "/" + markdownFileName
+
+	// 写入文件
+	if err := os.WriteFile(markdownPath, []byte(content), 0644); err != nil {
+		knowledgeDebugLog.Printf("[AICore] 保存 Markdown 失败: path=%s, err=%v", markdownPath, err)
+		return ""
+	}
+
+	return markdownPath
+}
+
 // DeleteDocument 删除文档
 func (s *KnowledgeService) DeleteDocument(kbID, docID string) error {
 	// 验证文档存在
@@ -400,24 +434,66 @@ func (s *KnowledgeService) GetDocumentPreview(kbID, docID string, page int) (*mo
 		return nil, nil
 	}
 
-	// 如果已解析，返回解析内容；否则返回文件 URL
-	if doc.Status == "parsed" {
-		// TODO: 从存储中读取解析内容（可以存到数据库或文件）
-		// 暂时返回文件 URL
-		fileURL, _ := s.uploadService.GetFileURL(doc.FileKey)
+	// 获取文件URL
+	fileURL, _ := s.uploadService.GetFileURL(doc.FileKey)
+
+	// 根据文件类型决定预览方式
+	fileName := doc.Name
+	isPDF := strings.HasSuffix(strings.ToLower(fileName), ".pdf")
+	isOffice := false
+	officeExts := []string{".csv", ".xlsx", ".xls", ".docx", ".doc", ".pptx", ".ppt", ".txt", ".md"}
+	for _, ext := range officeExts {
+		if strings.HasSuffix(strings.ToLower(fileName), ext) {
+			isOffice = true
+			break
+		}
+	}
+
+	// PDF文件返回文件URL
+	if isPDF {
 		return &model.DocumentPreviewResponse{
 			TotalPages:  1,
 			CurrentPage: page,
 			Content:     fileURL,
+			ContentType: "url",
 		}, nil
 	}
 
-	// 未解析，返回文件 URL
-	fileURL, _ := s.uploadService.GetFileURL(doc.FileKey)
+	// Office文件调用解析服务转换为HTML
+	if isOffice && s.aiCoreClient != nil {
+		knowledgeDebugLog.Printf("[Preview] Parsing office file: %s, URL: %s", fileName, fileURL)
+		result, err := s.aiCoreClient.ParseDocument(fileURL, fileName, "")
+		if err != nil {
+			// 解析失败，返回文件URL
+			knowledgeDebugLog.Printf("[Preview] Parse document failed: %v", err)
+			return &model.DocumentPreviewResponse{
+				TotalPages:  1,
+				CurrentPage: page,
+				Content:     fileURL,
+				ContentType: "url",
+			}, nil
+		}
+
+		knowledgeDebugLog.Printf("[Preview] Parse result: success=%v, content_length=%d", result.Success, len(result.Content))
+
+		// 返回HTML内容
+		if result.Success && result.Content != "" {
+			knowledgeDebugLog.Printf("[Preview] Returning HTML content, length: %d", len(result.Content))
+			return &model.DocumentPreviewResponse{
+				TotalPages:  1,
+				CurrentPage: page,
+				Content:     result.Content,
+				ContentType: "html",
+			}, nil
+		}
+	}
+
+	// 其他情况返回文件URL
 	return &model.DocumentPreviewResponse{
 		TotalPages:  1,
 		CurrentPage: page,
 		Content:     fileURL,
+		ContentType: "url",
 	}, nil
 }