RethinkDB supports spatial and geographic queries through geometry object support.
Geometry objects are implemented through a geographic coordinate system, with points and shapes plotted on the surface of a sphere in three-dimensional space. In addition, ReQL geometry objects can be converted to and from GeoJSON, with some limitations.
This is an overview of the system. For more details, consult the API documentation for individual geospatial commands.
Create a new table:
r.table_create('geo').run(conn)
Add a couple points:
r.table('geo').insert([ { 'id': 1, 'name': 'San Francisco', 'location': r.point(-122.423246,37.779388) }, { 'id': 2, 'name': 'San Diego', 'location': r.point(-117.220406,32.719464) } ]).run(conn)
Get the distance between the two points in San Francisco and San Diego:
r.table('geo').get(1)['location'].distance( r.table('geo').get(2)['location']).run(conn)
Add a geospatial index on the table (required for certain operations like getNearest
):
r.table('geo').index_create('location', geo=True)
Get the nearest point in the table to a specified one based on the index:
point = r.point(-122.422876,37.777128) # San Francisco r.table('geo').get_nearest(point, index='location')
Coordinates of points on the sphere’s surface are addressed by a pair of floating point numbers that denote longitude and latitude. The range of longitude is −180 through 180, which wraps around the whole of the sphere: −180 and 180 denote the same line. The range of latitude is −90 (the south pole) through 90 (the north pole).
For a more detailed explanation of this, consult the Wikipedia article on the geographic coordinate system.
Given two endpoints, a line in ReQL is the shortest path between those endpoints on the surface of the sphere, known as a geodesic. Lines can be defined with multiple points, in which case each segment of the line will be a geodesic; likewise, sides of a polygon will be geodesics. Geodesics are calculated assuming a perfect sphere.
Note that a line between the north pole and south pole (from latitude −90 to latitude 90) cannot be calculated, as all possible paths between them are the “shortest”; this may trigger an error in ReQL or it may choose an arbitrary (but technically correct) path.
Distances in ReQL are (by default) calculated assuming not a perfect sphere but an ellipsoid, using a precise and relatively fast algorithm developed by Charles Karney. The reference ellipsoid used is WGS84, the standard used for GPS. By default distances are specified in meters, but you can pass an optional argument to distance functions to specify kilometers, miles, nautical miles, and feet.
The geospatial functions are implemented through a set of new geometric object data types:
In addition, there’s a “pseudotype” called geometry which appears in documentation, to indicate that any of the geometric objects can be used with those commands.
Lines and polygons can be specified using either point objects or sequences of two-number arrays:
r.line(r.point(0,0), r.point(0,5), r.point(5,5), r.point(5,0), r.point(0,0)) r.line([0,0], [0,5], [5,5], [5,0], [0,0])
Both of those define the same square. If polygon
had been specified instead of line
they would define a filled square.
While there is a [circle] command, it approximates a circle by defining either a line or a polygon. There is no true circular data type.
To create indexes on fields containing geometry objects, you simply use the standard index_create command, setting the geo
optional argument to True
. In Python, this would be:
r.table('sites').index_create('locations', geo=True)
Just like other ReQL indexes, you can create an index using an anonymous function rather than a simple field name, as well as create multi indexes by using the multi
flag with geo
. Read the index_create API documentation for more details.
ReQL geometry objects are not GeoJSON objects, but you can convert back and forth between them with the geojson and to_geojson commands.
RethinkDB only allows conversion of GeoJSON objects which have ReQL equivalents: Point, LineString, and Polygon; MultiPoint, MultiLineString, and MultiPolygon are not supported. (You could, however, store multiple points, lines and polygons in an array and use a geospatial multi index with them.)
Only longitude/latitude coordinates are supported. GeoJSON objects that use Cartesian coordinates, specify an altitude, or specify their own coordinate reference system will be rejected.
How many dimensions are supported?
Two (latitude and longitude). Elevation is not supported.
What projections are supported?
RethinkDB supports the WGS84 World Geodetic System’s reference ellipsoid and geographic coordinate system (GCS). It does not directly support any projected coordinate system (PCS), but there are many tools available for performing such projections.
Does RethinkDB do a correct interpolation of degrees to meters along a path?
Yes. Distance calculations are done on a geodesic (either WGS84’s reference ellipsoid or a unit sphere).
Can you export to WKT or WKB?
No. However, you can export to GeoJSON and process that with other tools.
© RethinkDB contributors
Licensed under the Creative Commons Attribution-ShareAlike 3.0 Unported License.
https://rethinkdb.com/docs/geo-support/python/