Tranforming array<array<double>> to ST_MultiPolygon

[Andarius]

I'm trying to load a valid Geojson file using SPARK SQL and then transforming the coordinates to ST_MultiPolygon.

The jars I use are:

esri-geometry-api-2.0.0.jar
spatial-sdk-json-1.2.0.jar
spatial-sdk-hive-1.2.0.jar

The schema of the input data is:

 |-- geometry: struct (nullable = true)
 |    |-- coordinates: array (nullable = true)
 |    |    |-- element: array (containsNull = true)
 |    |    |    |-- element: array (containsNull = true)
 |    |    |    |    |-- element: array (containsNull = true)
 |    |    |    |    |    |-- element: double (containsNull = true)
 |    |-- type: string (nullable = true)
 |-- properties: struct (nullable = true)
 |-- type: string (nullable = true)

The process fails when I try the following:

SELECT
    st_multipolygon(geometry.coordinates[0][0]) as geo
FROM test

I get:

No handler for Hive UDF 'com.esri.hadoop.hive.ST_MultiPolygon': org.apache.hadoop.hive.ql.exec.NoMatchingMethodException: No matching method for class com.esri.hadoop.hive.ST_MultiPolygon with (array<array<double>>). Possible choices: _FUNC_(array<array<double>>) _FUNC_(string) ;

So I pass an array<array> and it expects an array<array>.

It works if I try:

SELECT
    st_multipolygon(geometry.coordinates[0][0][0], geometry.coordinates[0][0][1], ...) as geo
FROM test

Any idea of what's wrong (or what I'm doing wrong ?)

[randallwhitman]

Spatial-Framework documentation says ST_MultiPolygon(array(x1, y1, x2, y2, ... ), array(x1, y1, x2, y2, ... ), ... ), so it looks like st_multipolygon(geometry.coordinates[0][0][0], geometry.coordinates[0][0][1], ...) passes array<double> as expected and works correctly.

Maybe the only oddity is - is Spark-SQL presenting a botched error message?

[randallwhitman]

Cross-reference #36, #126.

[randallwhitman]

Try ST_GeomFromGeoJson.