Importación de GeoJSON con un array de objetos profundamente anidado

Pregunta

¿Cómo importo GeoJSON con un array anidado de objetos?

Respuesta

Para este tutorial, usaremos datos abiertos disponibles públicamente aquí. Puede encontrar una copia aquí.

Descargue los datos en formato GeoJSON y cambie el nombre del archivo a geojson.json.
Comprenda la estructura.

DESCRIBE TABLE file('geojson.json', 'JSON')
┌

Cree una tabla para almacenar las filas de GeoJSON.

Aquí, el requisito es generar una fila para cada object del features array. El tipo de dato inferido para el campo geometry sugiere que corresponde al tipo de dato MultiPolygon de ClickHouse.

create table geojson 
(
    type String,
    name String,
    crsType String,
    crsName String,
    featureType String,
    id Int64,
    inspiredId String,
    natCode String,
    nameUnit String,
    codNut1 String,
    codNut2 String,
    codNut3 String,
    codigoIne String,
    shapeLength Float64,
    shapeArea Float64,
    geometryType String,
    geometry MultiPolygon
)
engine = MergeTree
order by id;

Prepare los datos.

El objetivo principal de la consulta es verificar que obtenemos una fila por cada objeto del array features.

SELECT
    type AS type,
    name AS name,
    crs.type AS crsType,
    crs.properties.name AS crsName,
    features.type AS featureType,
    features.properties.FID AS id,
    features.properties.INSPIREID AS inspiredId,
    features.properties.NATCODE AS natCode,
    features.properties.NAMEUNIT AS nameUnit,
    features.properties.CODNUT1 AS codNut1,
    features.properties.CODNUT2 AS codNut2,
    features.properties.CODNUT3 AS codNut3,
    features.properties.CODIGOINE AS codigoIne,
    features.properties.SHAPE_Length AS shapeLength,
    features.properties.SHAPE_Area AS shapeArea,
    features.geometry.type AS geometryType
    --,features.geometry.coordinates
FROM file('municipios_ign.geojson', 'JSON')
ARRAY JOIN features
LIMIT 5

┌

Inserte los datos.

INSERT INTO geojson
SELECT
    type AS type,
    name AS name,
    crs.type AS crsType,
    crs.properties.name AS crsName,
    features.type AS featureType,
    features.properties.FID AS id,
    features.properties.INSPIREID AS inspiredId,
    features.properties.NATCODE AS natCode,
    features.properties.NAMEUNIT AS nameUnit,
    features.properties.CODNUT1 AS codNut1,
    features.properties.CODNUT2 AS codNut2,
    features.properties.CODNUT3 AS codNut3,
    features.properties.CODIGOINE AS codigoIne,
    features.properties.SHAPE_Length AS shapeLength,
    features.properties.SHAPE_Area AS shapeArea,
    features.geometry.type AS geometryType,
    features.geometry.coordinates as geometry
FROM file('municipios_ign.geojson', 'JSON')
ARRAY JOIN features

Aquí aparece el siguiente error:

Code: 53. DB::Exception: Received from localhost:9000. DB::Exception: ARRAY JOIN requires array or map argument. (TYPE_MISMATCH)
Received exception from server (version 24.1.2):

Esto se debe al análisis de features.geometry.coordinates.

Comprobemos su tipo de dato.

SELECT DISTINCT toTypeName(features.geometry.coordinates) AS geometry
FROM file('municipios_ign.geojson', 'JSON')
ARRAY JOIN features

┌

Se puede solucionar convirtiendo multipolygon.properties.coordinates a Array(Array(Array(Tuple(Float64,Float64)))). Para ello, podemos usar la función arrayMap(func,arr1,…).

SELECT distinct
    toTypeName(
        arrayMap(features.geometry.coordinates->
                    arrayMap(features.geometry.coordinates-> 
                            arrayMap(features.geometry.coordinates-> (features.geometry.coordinates[1],features.geometry.coordinates[2]) 
                    ,features.geometry.coordinates),
                features.geometry.coordinates),
        features.geometry.coordinates)
    ) as toTypeName
FROM file('municipios_ign.geojson', 'JSON')
ARRAY JOIN features;

┌

Inserte los datos.

INSERT INTO geojson
SELECT
    type as type,
    name as name,
    crs.type as crsType,
    crs.properties.name as crsName,
    features.type as featureType,
    features.properties.FID id,
    features.properties.INSPIREID inspiredId,
    features.properties.NATCODE natCode,
    features.properties.NAMEUNIT nameUnit,
    features.properties.CODNUT1 codNut1,
    features.properties.CODNUT2 codNut2,
    features.properties.CODNUT3 codNut3,
    features.properties.CODIGOINE codigoIne,
    features.properties.SHAPE_Length shapeLength,
    features.properties.SHAPE_Area shapeArea,
    features.geometry.type geometryType,
    arrayMap(features.geometry.coordinates->
                arrayMap(features.geometry.coordinates-> 
                        arrayMap(features.geometry.coordinates-> (features.geometry.coordinates[1],features.geometry.coordinates[2]),features.geometry.coordinates)
                ,features.geometry.coordinates)
    ,features.geometry.coordinates) geometry
FROM file('municipios_ign.geojson', 'JSON')
ARRAY JOIN features;

SELECT count()
FROM geojson

┌

Conclusión

Trabajar con JSON puede ser una tarea compleja. En este tutorial se abordó un caso en el que un array de objetos anidados podía dificultar aún más esta tarea. Para cualquier otro requisito relacionado con JSON, consulta nuestra documentación.