X-Agents/server/internal/service/knowledge_service.go

package service

import (
	"bytes"
	"encoding/json"
	"io"
	"log"
	"mime/multipart"
	"net/http"
	"os"
	"strings"
	"time"

	"github.com/google/uuid"
	"x-agents/server/internal/model"
	"x-agents/server/internal/repository"
)

// debugLog 专用调试日志
var knowledgeDebugLog *log.Logger

func init() {
	// 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 {
	repo             *repository.KnowledgeRepository
	modelRepo        *repository.ModelRepository
	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, markdownLocalPath string) *KnowledgeService {
	aiCoreClient, _ := NewAICoreClient(aiCoreServiceAddr)
	return &KnowledgeService{
		repo:             repo,
		modelRepo:        modelRepo,
		uploadService:    uploadService,
		pythonServiceURL: pythonServiceURL,
		aiCoreClient:     aiCoreClient,
		markdownLocalPath: markdownLocalPath,
	}
}

// Create 创建知识库
func (s *KnowledgeService) Create(req model.CreateKnowledgeRequest) (*model.KnowledgeBase, error) {
	// 验证 LLM 模型存在
	if _, err := s.modelRepo.FindByID(req.LLMModelID); err != nil {
		return nil, err
	}
	// 验证 Embedding 模型存在
	if _, err := s.modelRepo.FindByID(req.EmbeddingModelID); err != nil {
		return nil, err
	}

	kb := &model.KnowledgeBase{
		ID:               uuid.New().String(),
		Name:             req.Name,
		Description:      req.Description,
		LLMModelID:       req.LLMModelID,
		EmbeddingModelID: req.EmbeddingModelID,
		ParsingConfig:    req.ParsingConfig,
		StorageConfig:    req.StorageConfig,
		Status:           "active",
		DocumentCount:    0,
		ChunkCount:       0,
	}

	// 设置默认值
	if kb.ParsingConfig.Engine == "" {
		kb.ParsingConfig.Engine = "markitdown"
	}
	if kb.ParsingConfig.EnablePDF != false {
		kb.ParsingConfig.EnablePDF = true
	}
	if kb.ParsingConfig.Pandoc != false {
		kb.ParsingConfig.Pandoc = true
	}

	if err := s.repo.Create(kb); err != nil {
		return nil, err
	}
	return kb, nil
}

// List 获取知识库列表
func (s *KnowledgeService) List() ([]model.KnowledgeBase, error) {
	return s.repo.FindAll()
}

// GetByID 获取知识库详情
func (s *KnowledgeService) GetByID(id string) (*model.KnowledgeBase, error) {
	return s.repo.FindByID(id)
}

// Update 更新知识库
func (s *KnowledgeService) Update(id string, req model.UpdateKnowledgeRequest) error {
	updates := make(map[string]interface{})
	if req.Name != "" {
		updates["name"] = req.Name
	}
	if req.Description != "" {
		updates["description"] = req.Description
	}
	if req.LLMModelID != "" {
		// 验证模型存在
		if _, err := s.modelRepo.FindByID(req.LLMModelID); err != nil {
			return err
		}
		updates["llm_model_id"] = req.LLMModelID
	}
	if req.EmbeddingModelID != "" {
		// 验证模型存在
		if _, err := s.modelRepo.FindByID(req.EmbeddingModelID); err != nil {
			return err
		}
		updates["embedding_model_id"] = req.EmbeddingModelID
	}
	if req.ParsingConfig.Engine != "" {
		updates["parsing_config"] = req.ParsingConfig
	}
	if req.StorageConfig.Type != "" {
		updates["storage_config"] = req.StorageConfig
	}
	if req.Status != "" {
		updates["status"] = req.Status
	}

	return s.repo.Update(id, updates)
}

// Delete 删除知识库
func (s *KnowledgeService) Delete(id string) error {
	// 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)
}

// ListDocuments 获取知识库下的文档列表
func (s *KnowledgeService) ListDocuments(kbID string, status string) ([]model.KnowledgeDocument, error) {
	docs, err := s.repo.FindDocumentsByKBID(kbID, status)
	if err != nil {
		knowledgeDebugLog.Printf("[Knowledge ListDocuments] 错误: %v", err)
		return nil, err
	}
	knowledgeDebugLog.Printf("[Knowledge ListDocuments] kbID=%s, count=%d", kbID, len(docs))
	for i, doc := range docs {
		knowledgeDebugLog.Printf("[Knowledge ListDocuments] doc[%d]: id=%s, name=%s, file_url=%q", i, doc.ID, doc.Name, doc.FileURL)

		// 如果是 MinIO 内网地址，转换为代理 URL
		if doc.FileURL != "" && (strings.Contains(doc.FileURL, "10.10.10.189") || strings.Contains(doc.FileURL, ":9768")) {
			// 从 URL 中提取 fileKey（包含扩展名）
			parts := strings.Split(doc.FileURL, "/")
			if len(parts) > 0 {
				fileName := parts[len(parts)-1]
				fileKey := strings.Split(fileName, "?")[0]
				doc.FileURL = "/api/file_proxy?kb_id=" + kbID + "&key=" + fileKey
			}
		}
	}
	return docs, nil
}

// UploadDocument 上传文档到知识库
func (s *KnowledgeService) UploadDocument(kbID string, file *multipart.FileHeader) (*model.KnowledgeDocument, string, error) {
	knowledgeDebugLog.Printf("[Knowledge Upload] 开始上传文件: kbID=%s, filename=%s, size=%d", kbID, file.Filename, file.Size)

	// 验证知识库存在
	kb, err := s.repo.FindByID(kbID)
	if err != nil {
		knowledgeDebugLog.Printf("[Knowledge Upload] 错误: 知识库不存在, kbID=%s, err=%v", kbID, err)
		return nil, "", err
	}

	knowledgeDebugLog.Printf("[Knowledge Upload] 知识库配置: kbID=%s, storage_config.Type=%q, storage_config=%+v",
		kbID, kb.StorageConfig.Type, kb.StorageConfig)

	// 上传文件（根据知识库的 storage_config 选择存储方式）
	var result *UploadResponse
	if kb.StorageConfig.Type != "" {
		// 使用知识库的存储配置
		knowledgeDebugLog.Printf("[Knowledge Upload] 使用知识库存储配置: type=%s, endpoint=%s, bucket=%s",
			kb.StorageConfig.Type, kb.StorageConfig.Endpoint, kb.StorageConfig.Bucket)
		result, err = s.uploadService.UploadWithConfig(file, kb.StorageConfig)
	} else {
		// 使用全局配置
		knowledgeDebugLog.Printf("[Knowledge Upload] 使用全局存储配置")
		result, err = s.uploadService.Upload(file)
	}
	if err != nil {
		knowledgeDebugLog.Printf("[Knowledge Upload] 错误: 上传失败, err=%v", err)
		return nil, "", err
	}
	knowledgeDebugLog.Printf("[Knowledge Upload] 上传结果: success=%v, url=%s, message=%s",
		result.Success, result.URL, result.Message)
	if !result.Success {
		knowledgeDebugLog.Printf("[Knowledge Upload] 错误: 上传返回失败, message=%s", result.Message)
		return nil, "", nil
	}

	// 获取文件扩展名
	ext := getFileExt(file.Filename)

	knowledgeDebugLog.Printf("[Knowledge Upload] 准备创建文档记录: result.URL=%q, fileKey=%s, ext=%s",
		result.URL, result.FileKey, ext)

	// 创建文档记录
	doc := &model.KnowledgeDocument{
		ID:               uuid.New().String(),
		KnowledgeBaseID:  kbID,
		Name:             file.Filename,
		FileKey:          result.FileKey + ext,
		FileURL:          result.URL,
		FileSize:         file.Size,
		Status:           "parsing",
		UploadedAt:       time.Now(),
	}

	knowledgeDebugLog.Printf("[Knowledge Upload] 文档创建完成: doc.FileURL=%q", doc.FileURL)

	if err := s.repo.CreateDocument(doc); err != nil {
		return nil, "", err
	}

	// 更新知识库文档数
	s.updateDocumentCount(kbID)

	// 异步调用 Python 服务解析文档
	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, kb.ParsingConfig)

	return doc, result.URL, nil
}

// parseDocument 调用 Python 服务解析文档
func (s *KnowledgeService) parseDocument(kbID, docID, fileURL string, config model.ParsingConfig) {
	// 构建请求
	reqBody := map[string]interface{}{
		"file_url": fileURL,
		"engine":    config.Engine,
	}
	if config.Engine == "docling" && config.DoclingURL != "" {
		reqBody["docling_url"] = config.DoclingURL
	}

	body, _ := json.Marshal(reqBody)
	resp, err := http.Post(s.pythonServiceURL+"/parse", "application/json", bytes.NewBuffer(body))
	if err != nil {
		s.repo.UpdateDocument(docID, map[string]interface{}{"status": "failed"})
		return
	}
	defer resp.Body.Close()

	if resp.StatusCode != 200 {
		s.repo.UpdateDocument(docID, map[string]interface{}{"status": "failed"})
		return
	}

	// 解析响应
	var result map[string]interface{}
	if err := json.NewDecoder(resp.Body).Decode(&result); err != nil {
		s.repo.UpdateDocument(docID, map[string]interface{}{"status": "failed"})
		return
	}

	if success, ok := result["success"].(bool); ok && success {
		// 解析成功，更新状态
		chunks := []string{}
		if c, ok := result["chunks"].([]interface{}); ok {
			for _, chunk := range c {
				if c, ok := chunk.(string); ok {
					chunks = append(chunks, c)
				}
			}
		}

		s.repo.UpdateDocument(docID, map[string]interface{}{
			"status":      "parsed",
			"chunk_count": len(chunks),
		})

		// 更新知识库的 chunk_count
		s.updateChunkCount(kbID)
	} else {
		s.repo.UpdateDocument(docID, map[string]interface{}{"status": "failed"})
	}
}

// parseDocumentWithAICore 调用 AI-Core gRPC 服务解析文档
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
	}

	knowledgeDebugLog.Printf("[AICore] 开始解析文档: docID=%s, fileURL=%s, fileName=%s", docID, 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
	}

	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,
		})
	} else {
		knowledgeDebugLog.Printf("[AICore] 解析返回失败: docID=%s, message=%s", docID, result.Message)
	}
}

// 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
	doc, err := s.repo.FindDocumentByID(docID)
	if err != nil {
		return err
	}
	if doc.KnowledgeBaseID != kbID {
		return nil
	}

	// 获取知识库配置
	kb, err := s.repo.FindByID(kbID)
	if err != nil {
		return err
	}

	// 删除文件
	if doc.FileKey != "" {
		knowledgeDebugLog.Printf("[Knowledge DeleteDocument] 删除文件: kbID=%s, docID=%s, fileKey=%s, storageType=%s",
			kbID, docID, doc.FileKey, kb.StorageConfig.Type)
		if kb.StorageConfig.Type != "" {
			// 使用知识库的存储配置删除
			s.uploadService.DeleteFileWithConfig(doc.FileKey, kb.StorageConfig)
		} else {
			// 使用全局配置删除
			s.uploadService.DeleteFile(doc.FileKey)
		}
	}

	// 删除文档记录
	if err := s.repo.DeleteDocument(docID); err != nil {
		return err
	}

	// 更新知识库文档数
	s.updateDocumentCount(kbID)

	return nil
}

// ReparseDocument 重新解析文档
func (s *KnowledgeService) ReparseDocument(kbID, docID string) error {
	// 验证文档存在
	doc, err := s.repo.FindDocumentByID(docID)
	if err != nil {
		return err
	}
	if doc.KnowledgeBaseID != kbID {
		return nil
	}

	// 获取知识库配置
	kb, err := s.repo.FindByID(kbID)
	if err != nil {
		return err
	}

	// 获取文件 URL
	fileURL, err := s.uploadService.GetFileURL(doc.FileKey)
	if err != nil {
		return err
	}

	// 重置状态为 parsing
	s.repo.UpdateDocument(docID, map[string]interface{}{"status": "parsing"})

	// 异步重新解析
	go s.parseDocument(kbID, docID, fileURL, kb.ParsingConfig)

	return nil
}

// GetDocumentPreview 获取文档预览
func (s *KnowledgeService) GetDocumentPreview(kbID, docID string, page int) (*model.DocumentPreviewResponse, error) {
	// 验证文档存在
	doc, err := s.repo.FindDocumentByID(docID)
	if err != nil {
		return nil, err
	}
	if doc.KnowledgeBaseID != kbID {
		return nil, nil
	}

	// 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
	}

	// 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, "", 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)
			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
}

// updateDocumentCount 更新知识库文档数
func (s *KnowledgeService) updateDocumentCount(kbID string) {
	count, _ := s.repo.CountDocumentsByKBID(kbID)
	s.repo.Update(kbID, map[string]interface{}{"document_count": int(count)})
}

// updateChunkCount 更新知识库 chunk 数
func (s *KnowledgeService) updateChunkCount(kbID string) {
	docs, _ := s.repo.FindDocumentsByKBID(kbID, "parsed")
	totalChunks := 0
	for _, doc := range docs {
		totalChunks += doc.ChunkCount
	}
	s.repo.Update(kbID, map[string]interface{}{"chunk_count": totalChunks})
}

func getFileExt(filename string) string {
	exts := []string{".pdf", ".docx", ".xlsx", ".pptx", ".txt", ".md", ".html"}
	for _, ext := range exts {
		if len(filename) >= len(ext) && filename[len(filename)-len(ext):] == ext {
			return ext
		}
	}
	if len(filename) > 4 {
		return filename[len(filename)-4:]
	}
	return ""
}