wechat-api/internal/logic/agent/upload_agent_data_logic.go

package agent

import (
	"context"
	"encoding/csv"
	"fmt"
	"github.com/suyuan32/simple-admin-common/msg/errormsg"
	"mime/multipart"
	agentModel "wechat-api/ent/agent"
	"wechat-api/hook/fastgpt"
	"wechat-api/internal/utils"
	"wechat-api/internal/utils/dberrorhandler"

	"wechat-api/internal/svc"
	"wechat-api/internal/types"

	"github.com/zeromicro/go-zero/core/logx"
)

type UploadAgentDataLogic struct {
	logx.Logger
	ctx    context.Context
	svcCtx *svc.ServiceContext
}

func NewUploadAgentDataLogic(ctx context.Context, svcCtx *svc.ServiceContext) *UploadAgentDataLogic {
	return &UploadAgentDataLogic{
		Logger: logx.WithContext(ctx),
		ctx:    ctx,
		svcCtx: svcCtx}
}

func (l *UploadAgentDataLogic) UploadAgentData(req *types.UploadDataReq, file multipart.File, agentId uint64) (*types.BaseDataInfo, error) {
	var count uint64 = 0

	reader := csv.NewReader(file)
	records, err := reader.ReadAll()
	if err != nil {
		return nil, err
	}

	agent, err := l.svcCtx.DB.Agent.Query().Where(agentModel.ID(agentId)).Only(l.ctx)
	if err != nil {
		return nil, dberrorhandler.DefaultEntError(l.Logger, err, req)
	}

	var params fastgpt.CreateBulkDataReq
	params.CollectionID = agent.CollectionID
	params.TrainingMode = "chunk"

	qas := make([]fastgpt.DataQuestion, 0, 100)
	for idx, record := range records {
		if idx == 0 && record[1] == "答案" {
			continue
		}
		// 空内容过滤
		if record[0] == "" || record[1] == "" {
			continue
		}

		fmt.Printf("转换前：question=%s, answer=%s \n", record[0], record[1])

		question := transCharset(record[0])

		answer := transCharset(record[1])

		fmt.Printf("转换后：question=%s, answer=%s \n", question, answer)

		qas = append(qas, fastgpt.DataQuestion{
			Q: string(question),
			A: string(answer),
		})

		length := len(qas)
		if length > 0 && length%100 == 0 {
			params.Data = qas
			//fmt.Printf("params=%+v\n", params)
			response, err := fastgpt.CreateBulkData(&params)
			if err != nil {
				l.Logger.Errorf("batch insert data to fastgpt failed. collection=%s error=%s", agent.CollectionID, err.Error())
				return nil, err
			}
			count += response.Data.InsertLen

			qas = make([]fastgpt.DataQuestion, 0, 100)
		}
	}

	if len(qas) > 0 {
		params.Data = qas
		response, err := fastgpt.CreateBulkData(&params)
		if err != nil {
			l.Logger.Errorf("batch insert data to fastgpt failed. collection=%s error=%s", agent.CollectionID, err.Error())
			return nil, err
		}
		count += response.Data.InsertLen
		qas = make([]fastgpt.DataQuestion, 0, 100)
	}

	resp := &types.BaseDataInfo{}
	resp.Code = 0
	resp.Msg = errormsg.Success
	resp.Data = fmt.Sprintf("upload %d rows", count)
	return resp, nil
}

// transCharset 自动检测编码并转换为 UTF-8
func transCharset(s string) string {
	return utils.TransCharset(s)
}