feat: Add reprocess command

internal/service/dataset.go

@@ -211,18 +211,40 @@ func (s *DatasetService) Upload(ctx context.Context, in UploadInput) (domain.Dat
 
 	// Kick off the appropriate async job per file type. If enqueueing fails the
 	// row exists, so record the failure rather than leaving it stuck.
-	var enqueueErr error
-	switch in.FileType {
-	case domain.FileTypeVectorWithKato:
-		enqueueErr = s.jobs.EnqueueParse(ctx, dataset.ID)
-	case domain.FileTypeVector:
-		enqueueErr = s.jobs.EnqueueProperties(ctx, dataset.ID)
-	case domain.FileTypeRaster:
-		enqueueErr = s.jobs.EnqueueConvert(ctx, dataset.ID)
+	if err := s.enqueueProcessing(ctx, dataset); err != nil {
+		_ = s.repo.MarkParseFailed(ctx, dataset.ID, "failed to enqueue processing: "+err.Error())
+		return domain.Dataset{}, fmt.Errorf("enqueue processing: %w", err)
 	}
-	if enqueueErr != nil {
-		_ = s.repo.MarkParseFailed(ctx, dataset.ID, "failed to enqueue processing: "+enqueueErr.Error())
-		return domain.Dataset{}, fmt.Errorf("enqueue processing: %w", enqueueErr)
+	return dataset, nil
+}
+
+// enqueueProcessing schedules the appropriate async job for a dataset based on
+// its file type: vector_with_kato is parsed for column selection, plain vector
+// has its attribute table extracted into properties, and raster is converted to
+// a COG.
+func (s *DatasetService) enqueueProcessing(ctx context.Context, d domain.Dataset) error {
+	switch d.FileType {
+	case domain.FileTypeVectorWithKato:
+		return s.jobs.EnqueueParse(ctx, d.ID)
+	case domain.FileTypeVector:
+		return s.jobs.EnqueueProperties(ctx, d.ID)
+	case domain.FileTypeRaster:
+		return s.jobs.EnqueueConvert(ctx, d.ID)
+	default:
+		return fmt.Errorf("%w: unknown file_type %q", domain.ErrValidation, d.FileType)
+	}
+}
+
+// Reprocess re-enqueues the processing job for an existing dataset, restarting
+// its asynchronous pipeline from the step appropriate to its file type. Useful
+// for retrying after a transient failure or a worker that was behind the schema.
+func (s *DatasetService) Reprocess(ctx context.Context, id uuid.UUID) (domain.Dataset, error) {
+	dataset, err := s.repo.GetByID(ctx, id)
+	if err != nil {
+		return domain.Dataset{}, err
+	}
+	if err := s.enqueueProcessing(ctx, dataset); err != nil {
+		return domain.Dataset{}, fmt.Errorf("enqueue processing: %w", err)
 	}
 	return dataset, nil
 }