query-orchestration/internal/document/clean/create.go

// Package documentclean provides document cleaning functionality.
// Note: Text extraction pipeline has been removed. The document processing
// pipeline now ends after cleaning. See remove_texttract_and_mocks_plan.md.
package documentclean

import (
	"context"
	"fmt"

	"github.com/google/uuid"
)

// Clean processes a document by cleaning it. Previously this also triggered
// text extraction, but that feature has been removed.
func (s *Service) Clean(ctx context.Context, documentId uuid.UUID) error {
	isclean, err := s.cfg.GetDBQueries().HasDocumentCleanEntry(ctx, documentId)
	if err != nil {
		return fmt.Errorf("unable to verify if document has been cleaned: %w", err)
	}

	if !isclean {
		err = s.clean(ctx, documentId)
		if err != nil {
			return fmt.Errorf("unable to clean document: %w", err)
		}
	}

	// Note: Text extraction has been removed. Document pipeline ends here.
	return nil
}