Can you give an example on how to find nearby points within 1km radius ?
WP-GeoMeta uses MySQL’s built-in spatial functions, which doesn’t include ST_DWithin at this time.
So finding points within a radius is a two step process.
1) First, we buffer the center point by 1km to get a polygon representing our area of interest.
2) Second, we intersect our buffer with the points data.
Now, with code:
/* Get the center point at GeoJSON. If you're using WP-GeoMeta,
* you're probably storing your geometry as GeoJSON anyways.
* lat/lng is also supported, but then you'll need to construct
* GeoJSON manually
*/
$center_point = get_post_meta(15, 'geom');
/*
* MySQL supports ST_Buffer, but the buffer units must be the same as the
* units for your projection. WP-GeoMeta includes custom SQL functions
* for buffering in meters or miles: https://github.com/BrilliantPlugins/wp-geometa-lib/blob/master/sql/buffer_point.sql
*
* Buffer the center point by 1000 meters, using 8 points per quarter circle in
* the output polygon. Note: This follows the same syntax as PostGIS's ST_Buffer
*/
$buffered_shape = WP_GeoUtil::wp_buffer_point_m($center_point,1000,8);
/*
* Now run WP_Query with two meta_query params. inbuffer filters for posts within our buffer. distance is calculated so taht we can order by distance from $center_point
*/
$q = new WP_Query( array(
'orderby' => 'distance',
'order' => 'ASC',
'meta_query' => array(
'inbuffer' => array(
'key' => 'geom',
'compare' => 'ST_INTERSECTS',
'value' => $buffered_shape
),
'distance' => array(
'key' => 'geom',
'compare' => 'ST_DISTANCE',
'value' => $center_point,
'type' => 'DECIMAL(10,7)'
)
)
));
while($q->have_posts() ) {
$q->the_post();
print "\t* " . get_the_title() . "\n";
}
Does your plugin use extra tables other than wp meta ?
Yes, but it should be transparent and all original data is still stored in the regular meta tables.
WP-GeoMeta creates wp_postmeta_geo, wp_usermeta_geo, wp_termmeta_geo and wp_commentmeta_geo. These tables have a meta value column of type GEOMETRYCOLLECTION.
When you or a plugin save postmeta (or other types of meta), WP-GeoMeta checks the meta value is GeoJSON, and if so, saves a copy of the data in MySQL’s geometry format.
Later when you query postmeta (or another type of meta), WP-GeoMeta detects if you’re using a spatial function, and if so, it alters the query to fetch that data from the appropriate *meta_geo table.
Because the original/source data is still stored in postmeta, WP-GeoMeta shouldn’t affect any other plugins or where data is stored. It also means that if something goes wrong, the *meta_geo tables can be rebuilt from the original data in the postmeta table.
Any benchmarks for millions of points ?
I’m afraid not. I suspect that querying for points within 1km of a central point will be faster with WP-GeoMeta than with a custom haversine formula on the postmeta table.
WP-GeoMeta takes advantage of MySQL’s spatial indexes, which should make the ST_INTERSECTS query much faster than alternatives. However, I haven’t run any benchmarks so I can’t be sure.
