feat: 优化后端知识库服务和文档解析

- 更新文档解析客户端
- 优化知识库服务逻辑
- 更新 protobuf 定义

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

server/internal/service/document_parser_client.go

@@ -27,6 +27,16 @@ type ParseResult struct {
 	ParserEngine  string
 }
 
+// VLMConfig VLM 模型配置
+type VLMConfig struct {
+	Enabled  bool
+	Provider string // openai, anthropic, local 等
+	Model    string
+	APIKey   string
+	BaseURL  string
+	Prompt   string
+}
+
 // NewAICoreClient 创建 AI-Core 客户端
 func NewAICoreClient(address string) (*AICoreClient, error) {
 	return &AICoreClient{address: address}, nil
@@ -56,7 +66,8 @@ func (c *AICoreClient) Close() {
 }
 
 // ParseDocument 解析文档 - 使用生成的 protobuf 代码
-func (c *AICoreClient) ParseDocument(fileURL, fileName, fileType string) (*ParseResult, error) {
+// vlmConfig 可选，如果不使用 VLM 传 nil
+func (c *AICoreClient) ParseDocument(fileURL, fileName, fileType string, vlmConfig *VLMConfig) (*ParseResult, error) {
 	if c.conn == nil {
 		if err := c.Connect(); err != nil {
 			return nil, err
@@ -72,6 +83,18 @@ func (c *AICoreClient) ParseDocument(fileURL, fileName, fileType string) (*Parse
 		FileType: fileType,
 	}
 
+	// 如果提供了 VLM 配置，添加到请求中
+	if vlmConfig != nil {
+		req.VlmConfig = &docparser.VLMConfig{
+			Enabled:  vlmConfig.Enabled,
+			Provider: vlmConfig.Provider,
+			Model:    vlmConfig.Model,
+			ApiKey:   vlmConfig.APIKey,
+			BaseUrl:  vlmConfig.BaseURL,
+			Prompt:   vlmConfig.Prompt,
+		}
+	}
+
 	ctx, cancel := context.WithTimeout(context.Background(), 120*time.Second)
 	defer cancel()
 

server/internal/service/knowledge_service.go

@@ -3,6 +3,7 @@ package service
 import (
 	"bytes"
 	"encoding/json"
+	"io"
 	"log"
 	"mime/multipart"
 	"net/http"
@@ -19,8 +20,15 @@ import (
 var knowledgeDebugLog *log.Logger
 
 func init() {
-	debugFile, _ := os.OpenFile("logs/debug.log", os.O_CREATE|os.O_WRONLY|os.O_APPEND, 0666)
-	knowledgeDebugLog = log.New(debugFile, "", log.Ldate|log.Ltime)
+	// 确保 logs 目录存在
+	os.MkdirAll("logs", 0755)
+	debugFile, err := os.OpenFile("logs/debug.log", os.O_CREATE|os.O_WRONLY|os.O_APPEND, 0666)
+	if err != nil {
+		// 如果文件打开失败，使用 discard 避免输出到控制台
+		knowledgeDebugLog = log.New(io.Discard, "", log.Ldate|log.Ltime)
+	} else {
+		knowledgeDebugLog = log.New(debugFile, "", log.Ldate|log.Ltime)
+	}
 }
 
 type KnowledgeService struct {
@@ -133,10 +141,36 @@ func (s *KnowledgeService) Update(id string, req model.UpdateKnowledgeRequest) e
 
 // Delete 删除知识库
 func (s *KnowledgeService) Delete(id string) error {
-	// 先删除关联的文档
+	// 获取知识库信息
+	kb, err := s.repo.FindByID(id)
+	if err != nil {
+		return err
+	}
+
+	// 获取知识库下所有文档
+	docs, err := s.repo.FindDocumentsByKBID(id, "")
+	if err != nil {
+		return err
+	}
+
+	// 删除每个文档的 MinIO 文件和本地 Markdown 文件
+	for _, doc := range docs {
+		// 删除 MinIO 文件
+		if doc.FileKey != "" && kb.StorageConfig.Type == "minio" {
+			s.uploadService.DeleteFileWithConfig(doc.FileKey, kb.StorageConfig)
+		}
+		// 删除本地 Markdown 文件
+		if s.markdownLocalPath != "" {
+			markdownPath := s.markdownLocalPath + "/" + doc.ID + ".md"
+			os.Remove(markdownPath)
+		}
+	}
+
+	// 删除关联的文档（数据库记录）
 	if err := s.repo.DeleteDocumentsByKBID(id); err != nil {
 		return err
 	}
+
 	return s.repo.Delete(id)
 }
 
@@ -233,7 +267,7 @@ func (s *KnowledgeService) UploadDocument(kbID string, file *multipart.FileHeade
 	go s.parseDocument(kbID, doc.ID, result.URL, kb.ParsingConfig)
 
 	// 异步调用 AI-Core gRPC 服务解析文档（获取 Markdown）
-	go s.parseDocumentWithAICore(doc.ID, result.URL, doc.Name)
+	go s.parseDocumentWithAICore(doc.ID, result.URL, doc.Name, kb.ParsingConfig)
 
 	return doc, result.URL, nil
 }
@@ -293,7 +327,7 @@ func (s *KnowledgeService) parseDocument(kbID, docID, fileURL string, config mod
 }
 
 // parseDocumentWithAICore 调用 AI-Core gRPC 服务解析文档
-func (s *KnowledgeService) parseDocumentWithAICore(docID, fileURL, fileName string) {
+func (s *KnowledgeService) parseDocumentWithAICore(docID, fileURL, fileName string, config model.ParsingConfig) {
 	if s.aiCoreClient == nil {
 		knowledgeDebugLog.Printf("[AICore] AI-Core 客户端未初始化")
 		return
@@ -301,7 +335,21 @@ func (s *KnowledgeService) parseDocumentWithAICore(docID, fileURL, fileName stri
 
 	knowledgeDebugLog.Printf("[AICore] 开始解析文档: docID=%s, fileURL=%s, fileName=%s", docID, fileURL, fileName)
 
-	result, err := s.aiCoreClient.ParseDocument(fileURL, fileName, "")
+	// 构建 VLM 配置
+	var vlmConfig *VLMConfig
+	if config.VLMEnabled {
+		vlmConfig = &VLMConfig{
+			Enabled:  config.VLMEnabled,
+			Provider: config.VLMProvider,
+			Model:    config.VLMModel,
+			APIKey:   config.VLMAPIKey,
+			BaseURL:  config.VLMBaseURL,
+			Prompt:   config.VLMPrompt,
+		}
+		knowledgeDebugLog.Printf("[AICore] VLM 配置: provider=%s, model=%s, enabled=%v", config.VLMProvider, config.VLMModel, config.VLMEnabled)
+	}
+
+	result, err := s.aiCoreClient.ParseDocument(fileURL, fileName, "", vlmConfig)
 	if err != nil {
 		knowledgeDebugLog.Printf("[AICore] 解析失败: docID=%s, err=%v", docID, err)
 		return
@@ -462,7 +510,7 @@ func (s *KnowledgeService) GetDocumentPreview(kbID, docID string, page int) (*mo
 	// 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, "")
+		result, err := s.aiCoreClient.ParseDocument(fileURL, fileName, "", nil) // Preview 不使用 VLM
 		if err != nil {
 			// 解析失败，返回文件URL
 			knowledgeDebugLog.Printf("[Preview] Parse document failed: %v", err)