Failed index of GeoPackage datasets

[mikesname]

The dataset includes records with no geometry: indexing stops on the first problematic line and the dataset remains only partially indexed, for instance:

https://geodata.ehri-project.eu/geonetwork/srv/eng/catalog.search#/metadata/62c0ed01-9c16-4ad9-964c-860388767c09

The dataset includes multipart geometries (polygons): these particular records aren’t indexed. This is true for all border datasets which we have from the Holocaust Geographies Collaborative, for instance:

https://geodata.ehri-project.eu/geonetwork/srv/eng/catalog.search#/metadata/1771b370-eb04-4e5e-a7f3-8a724fcfddad

[mikesname]

First dataset is problematic because it contains geometry of the form POINT(nan nan), which is considered invalid. Alternatives to POINT(nan nan) include NULL and POINT EMPTY

Second example also had invalid geometry due to self-intersecting polygons, possible fix:

ogr2ogr -nlt MULTIPOLYGON -makevalid OUTPUT.gpkg INPUT.gpkg

[mikesname]

Upgrading ElasticSearch from 7.x to 8.x also necessary to have geometry properly indexed.