query-orchestration/internal/document/text/extract.go

package documenttext

import (
	"context"
	"queryorchestration/internal/database"
	"queryorchestration/internal/database/repository"
	"queryorchestration/internal/document"

	"github.com/google/uuid"
)

type ExtractionParams struct {
	ID       uuid.UUID
	Location document.Location
}

func (s *Service) executeExtraction(params *ExtractionParams) (*document.Location, error) {
	// TODO - various extraction tasks
	return &params.Location, nil
}

func (s *Service) extract(ctx context.Context, id uuid.UUID) error {
	docId := database.MustToDBUUID(id)

	entry, err := s.cfg.GetDBQueries().GetDocumentCleanEntry(ctx, docId)
	if err != nil {
		return err
	}

	outLocation, err := s.executeExtraction(&ExtractionParams{
		ID: id,
		Location: document.Location{
			Bucket: entry.Bucket,
			Key:    entry.Key,
		},
	})
	if err != nil {
		return err
	}

	version := s.svc.Version.GetVersion()

	err = s.cfg.GetDBQueries().AddDocumentTextEntry(ctx, &repository.AddDocumentTextEntryParams{
		Version:      version,
		Bucket:       outLocation.Bucket,
		Key:          outLocation.Key,
		Cleanentryid: entry.ID,
	})
	if err != nil {
		return err
	}

	return nil
}