Published on

Using the API with Leaflet (part 2)


In this post I put a geocoder on the map, so that users can find street level crimes by postcode or place name. I also use a clustering method to aggregate the crime data to a single point at higher zoom levels.

My previous post looked at querying the API to get police neighbourhood boundaries displayed in a leaflet map. By adding a geocoder, users can ENTER postcodes, street names or place names to look up crime data, which helps if they aren't sure what police neighbourhood or police force they live in.

In this post I'm using the mapbox geocoder API, but others are available such as the Google Geocoding API and the ArcGIS Geocoding API. With the free MapBox geocoding service, any results must be displayed on a MapBox map, so if implementing elsewhere just be sure to check what map you're actually displaying your results on. If you're paying for enterprise level geocoding you can use the results anywhere.

To start, add the following js and css files to the head of the html file. These allow access to the MarkerCluster plugin.

Then create a geocoder object using the mapbox.places geocoder index ID and add it to the map.

//create a geocoder
var geocoderControl = L.mapbox.geocoderControl('mapbox.places', {
  autocomplete: true,

//add it to the map

Using the mapbox.places ID allows you to make 1 geocode per request and to cache any results for up to 30 days.

Next, add an event listener to the geocoderControl which is fired when a user selects one of the autocomplete options listed in the geocoder.

geocoderControl.on('select', function(res) {

//get the lat lon
var latlng = (;

//construct the url to call the neighbourhood boundary
var url = ""
	+ "q=" + latlng[1] + "," + latlng[0];

$.getJSON(url, function(data) {
	if ( data ) {
		//add a new marker at the lat lon of postcode centroid
		var marker = L.marker([latlng[1], latlng[0]]);

		loadBoundary(data.force, data.neighbourhood);

This returns a GeoJSON object from which the latitude and longitude of the center can be found using ( This is then used in a jQuery call to the locate-neighbourhood method.

The jQuery.getJSON method allows for a callback function to be put in place. This is used to check the return contains data, to add a standard leaflet marker to the map at the correct latitude and longitude, and finally to call the loadBoundary function from part 1 of this post with the force name and neighbourhood name as parameters.

To make it easier to load neighbourhood boundaries, some of the code in the neighbourhoodChanged function from pat 1 is stripped out and becomes its own function, which is now called from both the neighbourhoodChanged function and from the callback function in the geocoder event listener.

var loadBoundary = function (forceID, neighbourhoodID) {
  var latlng = []
  var url = '' + forceID + '/' + neighbourhoodID + '/boundary'

  $.getJSON(url, function (data) {
    //create an array of boundary lat lon pairs
    $.each(data, function (i, item) {
      latlng.push(new L.LatLng(data[i].latitude, data[i].longitude))

    //if a layer is already present, remove it
    if (areaLayer) {

    //create a new polygon object using the latlng array
    areaLayer = new L.Polygon(latlng, {
      clickable: true,
      weight: 3,
      opacity: 0.4,
      fillOpacity: 0.1,

    //redraw the map to the bounds of the new polygon
    //add the polygon to the map


At the end of this function, a new function called getStreetCrimes is called, which takes the neighbourhood polygon as a custom area and an optional date parameter in YYYY-MM format and uses to POST the custom area and date data to the API. In this instance POST is used over GET due to GET having a URL character limit of 4,094 , which can easily be exceeded when using latitude longitude pairs for a neighbourhood boundary.

A callback function is implemented, so that a successful query fires the addCrimeLayer function, which is again a new function.

var getStreetCrimes = function (polygon, date) {
  var data = ''

  var url = ''

  $.each(polygon, function (i, item) {
    data += polygon[i].lat + ',' + polygon[i].lng + ':'

  if (typeof date !== 'undefined') {
    data += '&date=' + date
  } else {
    data = data.substring(0, data.length - 1)

  data = 'poly=' + data + '&date=2015-03'

  $.post(url, data, function (crimes) {

The addCrimeLayer works on a Leaflet MarkerClusterGroup, defined towards the start of the javascript.

//create crime cluster marker
var crime_markers = new L.MarkerClusterGroup()

It first clears any existing layers, then for each object in the crimes array, defined the objects category and the street name to which the crime was attributed before creating a new marker for that object, adding this as a layer to the crime_markers MarkerClusterGroup and binding a popup to it with the street name and crime category.

var addCrimeLayer = function (crimes) {

  for (var i = 0; i < crimes.length; i++) {
    var a = crimes[i]
    var content = '<p>Category:	' + a.category + '<br /> Location:	' +

    var crime_marker = L.marker(new L.LatLng(a.location.latitude, a.location.longitude), {
      icon: L.mapbox.marker.icon({
        'marker-color': '9C9E99',



Whilst not perfect (e.g. force name and neighbourhood name don't update when the geocoder is used) this is a good example of how easy it is to use Leaflet to display geographic data, even when there are high densities of points. Also, it's great to see the site putting all of this data out there as Open Data under an Open Government License.