upload_agent_data_logic.go 3.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125
  1. package agent
  2. import (
  3. "bytes"
  4. "context"
  5. "encoding/csv"
  6. "fmt"
  7. "github.com/suyuan32/simple-admin-common/msg/errormsg"
  8. "io"
  9. "mime/multipart"
  10. "strings"
  11. agentModel "wechat-api/ent/agent"
  12. "wechat-api/hook/fastgpt"
  13. "wechat-api/internal/utils/dberrorhandler"
  14. "wechat-api/internal/svc"
  15. "wechat-api/internal/types"
  16. "github.com/zeromicro/go-zero/core/logx"
  17. "golang.org/x/text/encoding/simplifiedchinese"
  18. "golang.org/x/text/transform"
  19. )
  20. type UploadAgentDataLogic struct {
  21. logx.Logger
  22. ctx context.Context
  23. svcCtx *svc.ServiceContext
  24. }
  25. func NewUploadAgentDataLogic(ctx context.Context, svcCtx *svc.ServiceContext) *UploadAgentDataLogic {
  26. return &UploadAgentDataLogic{
  27. Logger: logx.WithContext(ctx),
  28. ctx: ctx,
  29. svcCtx: svcCtx}
  30. }
  31. func (l *UploadAgentDataLogic) UploadAgentData(req *types.UploadDataReq, file multipart.File, agentId uint64) (*types.BaseDataInfo, error) {
  32. var count uint64 = 0
  33. reader := csv.NewReader(file)
  34. records, err := reader.ReadAll()
  35. if err != nil {
  36. return nil, err
  37. }
  38. agent, err := l.svcCtx.DB.Agent.Query().Where(agentModel.ID(agentId)).Only(l.ctx)
  39. if err != nil {
  40. return nil, dberrorhandler.DefaultEntError(l.Logger, err, req)
  41. }
  42. var params fastgpt.CreateBulkDataReq
  43. params.CollectionID = agent.CollectionID
  44. params.TrainingMode = "chunk"
  45. qas := make([]fastgpt.DataQuestion, 0, 100)
  46. for idx, record := range records {
  47. if idx == 0 && record[1] == "答案" {
  48. continue
  49. }
  50. // 空内容过滤
  51. if record[0] == "" || record[1] == "" {
  52. continue
  53. }
  54. fmt.Printf("转换前:question=%s, answer=%s \n", record[0], record[1])
  55. question := transCharset(record[0])
  56. answer := transCharset(record[1])
  57. fmt.Printf("转换后:question=%s, answer=%s \n", question, answer)
  58. qas = append(qas, fastgpt.DataQuestion{
  59. Q: string(question),
  60. A: string(answer),
  61. })
  62. length := len(qas)
  63. if length > 0 && length%100 == 0 {
  64. params.Data = qas
  65. //fmt.Printf("params=%+v\n", params)
  66. response, err := fastgpt.CreateBulkData(&params)
  67. if err != nil {
  68. l.Logger.Errorf("batch insert data to fastgpt failed. collection=%s error=%s", agent.CollectionID, err.Error())
  69. return nil, err
  70. }
  71. count += response.Data.InsertLen
  72. qas = make([]fastgpt.DataQuestion, 0, 100)
  73. }
  74. }
  75. if len(qas) > 0 {
  76. params.Data = qas
  77. response, err := fastgpt.CreateBulkData(&params)
  78. if err != nil {
  79. l.Logger.Errorf("batch insert data to fastgpt failed. collection=%s error=%s", agent.CollectionID, err.Error())
  80. return nil, err
  81. }
  82. count += response.Data.InsertLen
  83. qas = make([]fastgpt.DataQuestion, 0, 100)
  84. }
  85. resp := &types.BaseDataInfo{}
  86. resp.Code = 0
  87. resp.Msg = errormsg.Success
  88. resp.Data = fmt.Sprintf("upload %d rows", count)
  89. return resp, nil
  90. }
  91. func trim(s string) string {
  92. s = strings.TrimLeft(s, " \r\n\t")
  93. s = strings.TrimRight(s, " \r\n\t")
  94. return s
  95. }
  96. func transCharset(s string) string {
  97. s = trim(s)
  98. return s
  99. rd := transform.NewReader(bytes.NewReader([]byte(s)), simplifiedchinese.GBK.NewDecoder())
  100. bytes, err := io.ReadAll(rd)
  101. fmt.Printf("bytes=%s err=%v\n", bytes, err)
  102. return string(bytes)
  103. }