feat: rag初始版

service/dataset.go

@@ -0,0 +1,87 @@
+package service
+
+import (
+	"context"
+	"rag/dao"
+	"rag/model/dto"
+
+	"github.com/gogf/gf/v2/util/gconv"
+)
+
+var Dataset = new(datasetService)
+
+type datasetService struct{}
+
+// Create 创建数据集
+func (s *datasetService) Create(ctx context.Context, req *dto.CreateDatasetReq) (res *dto.CreateDatasetRes, err error) {
+	id, err := dao.Dataset.Insert(ctx, req)
+	if err != nil {
+		return
+	}
+	return &dto.CreateDatasetRes{Id: id}, nil
+}
+
+// Update 更新数据集
+func (s *datasetService) Update(ctx context.Context, req *dto.UpdateDatasetReq) (err error) {
+	_, err = dao.Dataset.Update(ctx, req)
+	return
+}
+
+// Delete 删除数据集
+func (s *datasetService) Delete(ctx context.Context, req *dto.DeleteDatasetReq) (err error) {
+	_, err = dao.Dataset.Delete(ctx, req)
+	return
+}
+
+// List 数据集列表
+func (s *datasetService) List(ctx context.Context, req *dto.ListDatasetReq) (res *dto.ListDatasetRes, err error) {
+	list, total, err := dao.Dataset.List(ctx, req)
+	if err != nil {
+		return
+	}
+	res = &dto.ListDatasetRes{
+		Total: total,
+	}
+	err = gconv.Struct(list, &res.List)
+	return
+}
+
+//// Search 搜索(示例,实际需要调用向量库)
+//func (s *datasetService) Search(ctx context.Context, req *dto.SearchReq) (res *dto.SearchRes, err error) {
+//	// 1. 获取数据集信息
+//	kb, err := dao.Dataset.GetByID(ctx, req)
+//	if err != nil {
+//		return nil, err
+//	}
+//
+//	// 2. 获取文件块
+//	chunks, err := dao.Chunk.FindChunksByKBIDWithLimit(ctx, req.KBID, 0, req.TopK)
+//	if err != nil {
+//		return nil, err
+//	}
+//
+//	// 3. TODO: 使用向量检索(需要集成向量库)
+//	// 暂时使用简单的关键词匹配
+//	results := make([]dto.SearchResult, 0)
+//	for _, chunk := range chunks {
+//		results = append(results, dto.SearchResult{
+//			Content:    chunk.Content,
+//			Score:      0.8, // TODO: 计算实际向量相似度
+//			DocumentID: chunk.DocumentID,
+//			ChunkIndex: chunk.Index,
+//		})
+//	}
+//
+//	g.Log().Infof(ctx, "数据集[%s]搜索完成,查询:%s,结果数:%d", kb.Name, req.Query, len(results))
+//
+//	return &dto.SearchRes{Results: results}, nil
+//}
+//
+//// formatChunks 格式化文件块为上下文
+//func (s *datasetService) formatChunks(chunks []*entity.DocumentChunk) string {
+//	var sb strings.Builder
+//	for i, chunk := range chunks {
+//		sb.WriteString(fmt.Sprintf("[%d] %s\n\n", i+1, chunk.Content))
+//	}
+//	return sb.String()
+//}