La interfaz está localizada; las descripciones y documentos de las Skills se traducen automáticamente con IA a su idioma. El código, los comandos, rutas y términos técnicos se mantienen en el original, y prevalece el documento del autor.
Explorar Skills
Productividad · SKILL.md

django-rest-framework-gis

django-rest-framework-gis

Guide Django REST Framework and GeoDjango projects through django-rest-framework-gis installation, GeoJSON serialization, spatial filtering, pagination, OpenAPI schema generation, and backend troubleshooting.

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentación de la Skill

GitHub

name: django-rest-framework-gis description: "Guide Django REST Framework and GeoDjango projects through django-rest-framework-gis installation, GeoJSON serialization, spatial filtering, pagination, OpenAPI schema generation, and backend troubleshooting." disable-model-invocation: true metadata: disco-role: operating license: MIT

django-rest-framework-gis

Use this repo skill when a Django REST Framework endpoint needs GeoDjango geometry as GeoJSON, GeoJSON Features or FeatureCollections, spatial query filters, GeoJSON pagination, or OpenAPI geometry schemas. It covers the public djangorestframework-gis package, not a private application model or a replacement for a spatial database.

Install and smoke-check

Install the public distribution and its declared runtime dependencies:

python -m pip install djangorestframework-gis
python -c "import rest_framework_gis; print(rest_framework_gis.get_version())"

The package requires Django, Django REST Framework, django-filter, and GeoDjango's native GEOS/GDAL/PROJ libraries for geometry operations. Add rest_framework_gis to INSTALLED_APPS after rest_framework when you want ordinary DRF ModelSerializer classes to map GeoDjango fields automatically. Use integration-testing for supported version selection, native-library checks, and spatial-database boundaries.

For a no-database JSON diagnostic, run the bundled checker from the installed skill tree:

python sub-skills/integration-testing/scripts/check_environment.py --help
python sub-skills/integration-testing/scripts/check_environment.py

Route the task

  • GeoJSON geometry or Feature I/O: read serialization. It owns GeometryField, GeometrySerializerMethodField, GeoFeatureModelSerializer, IDs, properties, bbox, SRID transforms, precision, deduplication, and null/empty behavior.
  • BBox, TMS tile, geometry, distance, or nearest/farthest queries: read spatial-filtering. It owns DRF filter backends, query parameters, view attributes, units, SRIDs, and parse errors.
  • FeatureCollection pagination or OpenAPI: read pagination-and-schema. It owns GeoJsonPagination, GeoFeatureAutoSchema, geometry nesting, bbox schemas, and DRF version-specific schema behavior.
  • Installation, GeoDjango loaders, compatibility, test setup, or database verification: read integration-testing. It separates package/GEOS/GDAL smoke from ORM and spatial-database evidence.

Cross-cutting rules

  1. Treat coordinates as GeoJSON x,y (longitude/easting, latitude/northing) and bbox values as west, south, east, north unless the application documents a different CRS.
  2. GeoFeatureModelSerializer requires Meta.geo_field; use None explicitly for a geometry-less Feature. Choose one bbox mode: auto_bbox is read-only, while bbox_geo_field is writable and mutually exclusive with auto_bbox.
  3. A GEOS/GDAL import or field serializer smoke does not prove that ORM predicates such as dwithin, contained, bboverlaps, or GeometryDistance work on the target database. Verify the actual spatial backend separately.
  4. Read references/troubleshooting.md for cross-cutting install, settings, native-library, dependency, and backend failures. Read references/repo-provenance.md before deciding whether this skill matches a changed checkout.

Scope boundary

This skill distills the package's public API and repository-tested behavior. It does not include private environment paths, the original repository as a runtime dependency, maintainer release automation, or the expensive 10,000- object performance benchmark.

Skills guardadas