package documenttext import ( "context" "errors" "time" "queryorchestration/internal/database/repository" "queryorchestration/internal/serviceconfig/build" "queryorchestration/internal/serviceconfig/objectstore" "github.com/aws/aws-sdk-go-v2/service/textract" "github.com/aws/aws-sdk-go-v2/service/textract/types" "github.com/google/uuid" ) func (s *Service) triggerExtract(ctx context.Context, clean *repository.Currentcleanentry) error { version := build.GetVersionUnixTimestamp() return s.cfg.ExecuteDBTransaction(ctx, func(ctx context.Context, q *repository.Queries) error { triggerId, err := s.cfg.GetDBQueries().AddDocumentTextTrigger(ctx, &repository.AddDocumentTextTriggerParams{ Cleanid: clean.ID, Version: version, }) if err != nil { return err } jobTag := triggerId.String() filename, err := s.getOutputFilename(triggerId) if err != nil { return err } key := objectstore.BucketKey{ CreatedAt: time.Now().UTC(), Location: objectstore.TextTextract, ClientID: clean.Clientid, Filename: filename, } keyStr := key.String() res, err := s.cfg.GetTextractClient().StartDocumentTextDetection(ctx, &textract.StartDocumentTextDetectionInput{ DocumentLocation: &types.DocumentLocation{ S3Object: &types.S3Object{ Bucket: clean.Bucket, Name: clean.Key, }, }, JobTag: &jobTag, OutputConfig: &types.OutputConfig{ S3Bucket: clean.Bucket, S3Prefix: &keyStr, }, }) if err != nil { return err } else if res.JobId == nil { return errors.New("No job id returned from textract") } err = s.cfg.GetDBQueries().AddDocumentTextTriggerJobId(ctx, &repository.AddDocumentTextTriggerJobIdParams{ ID: triggerId, Textractjobid: res.JobId, }) if err != nil { return err } return nil }) } func (s *Service) getOutputFilename(triggerId uuid.UUID) (string, error) { if triggerId == uuid.Nil { return "", errors.New("trigger id required") } return triggerId.String(), nil } func (s *Service) getTriggerIdFromKey(key objectstore.BucketKey) (uuid.UUID, error) { triggerId, err := uuid.Parse(key.Filename) if err != nil { return uuid.Nil, err } return triggerId, nil }