Update: I wrote up a more detailed blogpost about this specific subject on http://www.mullie.eu/geographic-searches/
--
Loop through all available towns using the Google Maps API to fetch their latitude & longitude. Save these somewhere (database). - Beware, Google will not accept an enormous amount of calls, so throttle your calls.
Then, when fetching a town, you can use code similar to the code below to grab the cities withing a certain range:
public static function getNearby($lat, $lng, $type = 'cities', $limit = 50, $distance = 50, $unit = 'km')
{
// radius of earth; @note: the earth is not perfectly spherical, but this is considered the 'mean radius'
if ($unit == 'km') $radius = 6371.009; // in kilometers
elseif ($unit == 'mi') $radius = 3958.761; // in miles
// latitude boundaries
$maxLat = (float) $lat + rad2deg($distance / $radius);
$minLat = (float) $lat - rad2deg($distance / $radius);
// longitude boundaries (longitude gets smaller when latitude increases)
$maxLng = (float) $lng + rad2deg($distance / $radius / cos(deg2rad((float) $lat)));
$minLng = (float) $lng - rad2deg($distance / $radius / cos(deg2rad((float) $lat)));
// get results ordered by distance (approx)
$nearby = (array) FrontendDB::getDB()->retrieve('SELECT *
FROM table
WHERE lat > ? AND lat < ? AND lng > ? AND lng < ?
ORDER BY ABS(lat - ?) + ABS(lng - ?) ASC
LIMIT ?;',
array($minLat, $maxLat, $minLng, $maxLng, (float) $lat, (float) $lng, (int) $limit));
return $nearby;
}
Notes about the above code:
- Own database wrapper is used, so transform to mysql_query, PDO, ...
- This will not be exact. We can't do exact spherical calculations in the DB, so we've taken the upper & lower latitude & longitude limits. This basically means that a location which is slightly further than your distance (e.g. in the far north-east, just outside of the actual radius (which actually is pretty much a circle), but still inside the max latitude & longitude (because we compare it to square limits in the database). This will just give a rough but nut 100% accurate selection of cities withing your radius.
I'll try to illustrate this:
_________________
| / |
| Y / |
| / |
|( X )|
| / |
| / |
|______\_/______|
The above circle (somewhat) is the actual radius where you want to find locations within, based upon location X. This is too hard to accomplish straight out of your DB, so what we actually fetch from the DB is the surrounding square. As you can see, it's possible that locations (like Y) fall within these max & min boundaries, though they aren't actually withing the requested radius. These can later be filtered out through PHP though.
To tackle this last issue, you could loop all results and calculate the exact distance between both your root location, and the close matches found, to calculate if they're actually within your radius. For that, you could use this code:
public static function getDistance($lat1, $lng1, $lat2, $lng2, $unit = 'km')
{
// radius of earth; @note: the earth is not perfectly spherical, but this is considered the 'mean radius'
if ($unit == 'km') $radius = 6371.009; // in kilometers
elseif ($unit == 'mi') $radius = 3958.761; // in miles
// convert degrees to radians
$lat1 = deg2rad((float) $lat1);
$lng1 = deg2rad((float) $lng1);
$lat2 = deg2rad((float) $lat2);
$lng2 = deg2rad((float) $lng2);
// great circle distance formula
return $radius * acos(sin($lat1) * sin($lat2) + cos($lat1) * cos($lat2) * cos($lng1 - $lng2));
}
This will calculate the (quasi) exact distance between location X and location Y, and then you can filter out exactly those cities that were near enough to pass the rough db-fetch, but not just near enough to actually be within your bounds.
