Skip to content

Children elements are not created on transkribus import

Refs https://redmine.teklia.com/issues/3390

I'm trying to import a transkribus archive (available as /data/humu/export_job_4461689.zip on xenarque.vpn), using the import from this project using this command:

docker run  \
	-e ARKINDEX_API_URL=https://demo.arkindex.org/api/v1 \
	-e ARKINDEX_API_TOKEN=<redacted> \
	-e ARKINDEX_WORKER_RUN_ID=e2a255c8-bf14-4014-82fe-7b6121e36acd \
	-v /data/humu/export_job_4461689.zip:/export.zip \
	registry.gitlab.com/teklia/arkindex/tasks \
	python -m arkindex_tasks.import_transkribus \
	--archive-path=/export.zip \
	--corpus c0e861c2-9ff3-4047-a146-950e710b6c22

Unfortunately the children elements are not created:

[ERROR] Failed importing transcript 1328264/Copy_of_HTR_Validation_Set_'NorHand_pylaia'/page/no-nb_digimanus_147901_0001.xml: 400 - {'elements': {'non_field_errors': ['This list may not be empty.']}}

I also see some warnings about polygons out of bounds, this seems related as it happens right before the failure:

[INFO] Importing transcript 1328264/Copy_of_HTR_Validation_Set_'NorHand_pylaia'/page/no-nb_digimanus_147909_0002.xml
[WARNING] Ignoring region r1 (this polygon is entirely outside the image's bounds)

The export is only 2 days old, I hope transkribus did not change something in there...