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

package documenttext

import (
	"context"
	"log/slog"

	documenttypes "queryorchestration/internal/document/types"
	"queryorchestration/internal/serviceconfig/objectstore"

	"github.com/google/uuid"
)

func (s *Service) executeExtraction(ctx context.Context, documentId uuid.UUID) error {
	clean, err := s.cfg.GetDBQueries().GetCleanEntryByDocId(ctx, documentId)
	if err != nil {
		return err
	} else if clean.Fail.Valid {
		slog.Error("document has no clean document", "id", documentId)
		return nil
	}

	key, err := objectstore.ParseBucketKey(*clean.Key)
	if err != nil {
		return err
	}

	doc, err := documenttypes.GetFile(ctx, s.cfg, documenttypes.GetFileParams{
		Bucket:   *clean.Bucket,
		Key:      key,
		Hash:     *clean.Hash,
		Mimetype: documenttypes.ParseDBNullMimeType(clean.Mimetype),
	})
	if err != nil {
		return err
	}

	text, err := s.GetDocumentText(ctx, doc)
	if err != nil {
		return err
	}

	err = s.storeProcess(ctx, text, clean)
	if err != nil {
		return err
	}

	return nil
}