Use GeoJSON format

Index complex geometries like polygons and multi-polygons using the GeoJSON standard format.

GeoJSON is a standardized format for encoding geographic data structures. Meilisearch supports GeoJSON through the _geojson field, allowing you to index complex shapes like polygons and multi-polygons in addition to simple point coordinates.

Use GeoJSON when your documents represent areas (neighborhoods, properties, delivery zones) rather than single points.

The _geojson field

To use GeoJSON, add a _geojson field to your documents. The value must follow the GeoJSON specification.

Point geometry

For simple point locations, you can use either the _geo field or a GeoJSON Point:

{
  "id": 1,
  "name": "Nàpiz' Milano",
  "address": "Viale Vittorio Veneto, 30, 20124, Milan, Italy",
  "_geojson": {
    "type": "Feature",
    "geometry": {
      "type": "Point",
      "coordinates": [9.1967508, 45.4777599]
    }
  }
}

GeoJSON uses longitude first, latitude second ([lng, lat]). This is the opposite order from the _geo field, which uses lat and lng as named keys.

Polygon geometry

Use a Polygon to represent an area like a neighborhood, a property boundary, or a delivery zone:

{
  "id": 10,
  "name": "Quartiere Brera",
  "type": "neighborhood",
  "_geojson": {
    "type": "Feature",
    "geometry": {
      "type": "Polygon",
      "coordinates": [[
        [9.1850, 45.4730],
        [9.1920, 45.4730],
        [9.1920, 45.4780],
        [9.1850, 45.4780],
        [9.1850, 45.4730]
      ]]
    }
  }
}

In GeoJSON Polygon format, the coordinates array contains one or more linear rings. The first ring defines the outer boundary, and the last coordinate must repeat the first to close the ring.

Meilisearch does not support polygons with holes. If your polygon includes an inner ring (a hole), Meilisearch ignores the hole and treats the polygon as a solid shape.

MultiPolygon geometry

Use a MultiPolygon when a single document covers multiple separate areas:

{
  "id": 20,
  "name": "Downtown delivery zone",
  "type": "delivery_area",
  "_geojson": {
    "type": "Feature",
    "geometry": {
      "type": "MultiPolygon",
      "coordinates": [
        [[
          [9.1800, 45.4600],
          [9.1900, 45.4600],
          [9.1900, 45.4700],
          [9.1800, 45.4700],
          [9.1800, 45.4600]
        ]],
        [[
          [9.2000, 45.4650],
          [9.2100, 45.4650],
          [9.2100, 45.4750],
          [9.2000, 45.4750],
          [9.2000, 45.4650]
        ]]
      ]
    }
  }
}

Filtering and sorting with GeoJSON documents

Filtering works the same way with GeoJSON documents as with _geo documents. Add _geo to filterableAttributes, then use _geoRadius, _geoBoundingBox, or _geoPolygon in your search queries.

curl \
  -X PUT 'MEILISEARCH_URL/indexes/neighborhoods/settings/filterable-attributes' \
  -H 'Content-type:application/json' \
  --data-binary '["_geo"]'

Then search as usual:

curl \
  -X POST 'MEILISEARCH_URL/indexes/neighborhoods/search' \
  -H 'Content-type:application/json' \
  --data-binary '{
    "filter": "_geoRadius(45.4700, 9.1880, 1000)"
  }'

When a document has a _geojson Polygon or MultiPolygon, Meilisearch checks whether the filter area intersects with the document’s geometry.

Sorting with _geoPoint only works with the _geo field. It is not possible to sort documents based on _geojson data.

Using _geo and _geojson together

If your application needs both distance-based sorting and polygon-based filtering, add both fields to your documents:

{
  "id": 10,
  "name": "Quartiere Brera",
  "type": "neighborhood",
  "_geo": {
    "lat": 45.4755,
    "lng": 9.1885
  },
  "_geojson": {
    "type": "Feature",
    "geometry": {
      "type": "Polygon",
      "coordinates": [[
        [9.1850, 45.4730],
        [9.1920, 45.4730],
        [9.1920, 45.4780],
        [9.1850, 45.4780],
        [9.1850, 45.4730]
      ]]
    }
  }
}

When a document contains both fields, Meilisearch:

• Uses _geo for sorting with _geoPoint
• Uses _geojson for filtering with _geoPolygon
• Matches both _geo and _geojson values when filtering with _geoRadius and _geoBoundingBox

Limitations

• Transmeridian shapes are not supported. If your shape crosses the 180th meridian, split it into two separate shapes grouped as a MultiPolygon or MultiLine.
• Polygons with holes are not supported. Meilisearch ignores inner rings and treats polygons as solid shapes.
• CSV files do not support _geojson. Use JSON or NDJSON format for documents with GeoJSON data.

Geo search overview

Learn about all geo search capabilities, including _geo and _geojson.

GeoJSON specification

Official GeoJSON format documentation.