Elasticsearch rejects geometries

[rutgervanwilligen]

See https://gist.github.com/rutgervanwilligen/5bc20ee44e839d7ca4a2 for a list of rejected geometries, including the reasons of rejection and a sample geometry.

• [x] Gemeentegeschiedenis: Find cause
  • Found! Several wrongly formatted geometries. Source has provided new geometries.
• [x] Gemeentegeschiedenis: Import new geometries and check if they're fixed.
• [x] Atlas-verstedelijking: Find cause
  • Found: Multipolygons with self-intersecting points
• [ ] Atlas-verstedelijking: Fix polygons ( @reinvantveer is working on it) and import
• [x] Gewesten: Find cause
  • Found: Self intersecting polygons
• [ ] Gewesten: Fix and reimport

[rutgervanwilligen]

Gemeentegeschiedenis/106 contains a syntactically correct GeoJSON geometry, but is not valid.

http://bl.ocks.org/d/5334d09b72d09c940730

The geometry is formatted as a single polygon, but this contains two islands. Syntactically speaking, the second 'island' in the polygon array is interpreted as a hole in the first, instead of a second island.

[reinvantveer]

Should geometries be OGC-simple feature specification valid? This could ve checked during conversion if checking at the source is the preferred solution.

Verzonden vanaf Samsung Mobile

-------- Oorspronkelijk bericht -------- Van: Rutger van Willigen notifications@github.com Datum:11-03-2015 14:54 (GMT+01:00) Aan: histograph/core core@noreply.github.com Cc: Onderwerp: Re: [core] Elasticsearch rejects geometries (#19)

Gemeentegeschiedenis/106 contains a syntactically correct GeoJSON geometry, but is not valid.

http://bl.ocks.org/d/5334d09b72d09c940730

The geometry is formatted as a single polygon, but this contains two islands. Syntactically speaking, the second 'island' in the polygon array is interpreted as a hole in the first, instead of a second.

— Reply to this email directly or view it on GitHubhttps://github.com/histograph/core/issues/19#issuecomment-78266688.

[reinvantveer]

So often, the geometry type should be multipolygon, not polygon?

Verzonden vanaf Samsung Mobile

-------- Oorspronkelijk bericht -------- Van: Rutger van Willigen notifications@github.com Datum:11-03-2015 12:58 (GMT+01:00) Aan: histograph/core core@noreply.github.com Cc: Onderwerp: [core] Elasticsearch rejects geometries (#19)

See https://gist.github.com/rutgervanwilligen/5bc20ee44e839d7ca4a2 for a list of rejected geometries, including the reasons of rejection and a sample geometry.

— Reply to this email directly or view it on GitHubhttps://github.com/histograph/core/issues/19.

[rutgervanwilligen]

All of the Gemeentegeschiedenis geometries are now accepted.

[reinvantveer]

:+1:

[mmmenno]

:-)

[reinvantveer]

PostGIS valid geometries are sometimes rejected by Elasticsearch shape indexer

See also: this link

Elasticsearch has (custom?) geometry checking procedure using Java Topology Suite and Spatial4j:

import com.spatial4j.core.shape.Shape;
import com.vividsolutions.jts.geom.*;

From here

Node conversion scripts need similar checking for the geometries to be inserted I suppose.

[reinvantveer]

Perhaps use this

[wires]

kapotte geometrieren op andere queue zetten?

[jobspierings]

@bertspaan zegt:

heeft te maken met strengheid van Lucene. Meeste gefixed - nieuwe issues aanmaken als tegenkomen.