I'm trying to make a mapping visualization in realtime, where I keep getting new points via websockets. The initial plotting these markers on the map seems simple, but I'm not sure what's the right way of updating a layer on Mapbox.
As of now, whenever I get a new point, I remove the old layer, create a new one and then add it on the map. The problem with this approach is that it is slow and for high number of points (>5000) it starts lagging.
// remove layer
if (this.pointsLayer != null) {
map.removeLayer(this.pointsLayer);
}
// build geoJSON
var geoJSON = { "type": "FeatureCollection", "features": [] };
geoJSON["features"] = tweets.map(function(tweet) {
return this.getGeoPoint(tweet);
}.bind(this));
// add geoJSON to layer
this.pointsLayer = L.mapbox.featureLayer(geoJSON, {
pointToLayer: function(feature, latlon) {
return L.circleMarker(latlon, {
fillColor: '#AA5042',
fillOpacity: 0.7,
radius: 3,
stroke: false
});
}
}).addTo(map);
Is there a better way?
You can create an empty GeoJSON layer by passing it a false instead of real data:
//create empty layer
this.pointsLayer = L.mapbox.featureLayer(false, {
pointToLayer: function(feature, latlon) {
return L.circleMarker(latlon, {
fillColor: '#AA5042',
fillOpacity: 0.7,
radius: 3,
stroke: false
});
}
}).addTo(map);
then use .addData to update it as new tweets come in. Something like:
// build geoJSON
var geoJSON = { "type": "FeatureCollection", "features": [] };
geoJSON["features"] = /**whatever function you use to build a single tweet's geoJSON**/
// add geoJSON to layer
this.pointsLayer.addData(geoJSON);
For a single tweet, I guess you could just create a Feature instead of a FeatureCollection, though I don't know whether that extra layer of abstraction would make any difference in terms of performance.
EDIT: Here is an example fiddle showing the .addData method at work:
http://jsfiddle.net/nathansnider/4mwrwo0t/
It does slow down noticeably if you add 10,000 points, and for 15,000 points, it's really sluggish, but I suspect that has less to do with how the points are added that the demands of rendering so many circleMarkers.
If you aren't already, you may want to try using the new Leaflet 1.0 beta, which redraws vector layers faster and is generally much more responsive with large datasets. Compare this 15,000-point example using Leaflet 0.7.5 to the same code using Leaflet 1.0.0b2. Not everything is fixed (popups take a long time to open in both), but the difference in lag time when trying to drag the map is pretty dramatic.
There's no reason to go through the intermediate step of construction a GeoJSON object just so you can add it to the map. Depending on your exact needs, you can do something like this:
tweets.forEach(function(t) {
L.marker(this.getGeoPoint(t)).addTo(map);
}, this);
You should manage the tweets object so it only contains points that are not already visible on the map, though. Deleting all the old markers, just so you can add them again, is of course going to be very slow.
I would take a look at Leaflet Realtime:
Put realtime data on a Leaflet map: live tracking GPS units, sensor data or just about anything.
https://github.com/perliedman/leaflet-realtime
Related
I'm trying to plot a supplied vector tile layer onto a map using MapBox GL JS. I've followed the documentation here but nothing apart from the basic map is being output and there are no console errors. In the browser's Network tab I can see lots of .pbf requests being returned with data so it would seem that the endpoint is passing data back, but I don't know how to determine what the problem is in plotting that data onto the map.
The code is as follows:
mapboxgl.accessToken = '[my access token]';
const map = new mapboxgl.Map({
container: 'map',
style: 'mapbox://styles/mapbox/light-v10',
zoom: 6,
center: [-0.118092, 51.509865]
});
map.once("load", () => {
map.addSource("ncn", {
type: "vector",
tiles: [
"https://tiles.arcgis.com/tiles/1ZHcUS1lwPTg4ms0/arcgis/rest/services/NCN_Vector_Tile_Package/VectorTileServer/tile/{z}/{y}/{x}.pbf"
]
});
map.addLayer({
id: "ncn-lines",
type: "line",
source: "ncn",
"source-layer": "NCN_Vector_Tile_Package",
"layout": {
"visibility": "visible"
},
'paint': {
'line-color': 'rgb(255, 0, 0)'
}
});
});
I am fairly sure that the type should be line (rather than fill) as the data is supposed to contain route lines rather than vector shapes.
I don't have access to the Arcgis server so can't see how anything is configured at that side. Can anyone suggest what might be wrong here and/or how to debug?
It looks like the value for source-layer is not correct - it should be NCN_2020. Here's a demo showing it working: https://jsbin.com/xunuhibuki/1/edit?html,output
How do you get that value? I'm not quite sure the best way, but the way I found: add ?f=html to your vector tile layer like this: https://tiles.arcgis.com/tiles/1ZHcUS1lwPTg4ms0/arcgis/rest/services/NCN_Vector_Tile_Package/VectorTileServer/?f=html then click "Styles" link at the bottom which gives you an example of how to construct your map.addLayer() commands in your mapboxgl code.
Situation: I have a working site where upon entering an address, MapboxGL marks a point on the map and queries a polygon layer (queryRenderedFeatures) and displays the polygon feature containing the point.
This works; however, if I then want to geocode a second address that changes the map view, it fails the second time because map.queryRenderedFeatures returns an empty array.
var userDistrictsGeoJson;
map.on('load', function() {
//add layers from Mapbox account
addLayers(); //details left out of example, but this works.
// Listen for geocoding result
// This works the first time through, but fails if the user searchs for a second address because queryRenderedFeatures is working with a smaller set of features
geocoder.on('result', function(e) {
//need to clear geojson layer and
userDistrictsGeoJson = {
"type": "FeatureCollection",
"features": []
};
map.getSource('single-point').setData(e.result.geometry);
//project to use (pixel xy coordinates instead of lat/lon for WebGL)
var point = map.project([e.result.center[0], e.result.center[1]]);
var features = map.queryRenderedFeatures(point, { layers: ['congress-old'] });
var filter = featuresOld.reduce(function(memo, feature){
// console.log(feature.properties);
memo.push(feature.properties.GEOID);
return memo;
}, ['in', 'GEOID']);
map.setFilter('user-congress-old', filter);
var userCongressOldGeoJson = map.querySourceFeatures('congressional-districts', {
sourceLayer: 'congress_old',
filter: map.getFilter('user-congress-old')
});
userDistrictsGeoJson.features.push(userCongressOldGeoJson[0]);
var bbox = turf.bbox(userDistrictsGeoJson);
var bounds = [[bbox[0], bbox[1]], [bbox[2], bbox[3]]];
map.fitBounds(bounds, {
padding: 40
});
}); //geocoder result
}); //map load
So like I said, everything that runs on the geocodes 'result' event works the first time through, but it seems that on the second time through (user searches new address, but doesn't reload map) queryRenderedFeatures returns a smaller subset of features that doesn't include the tiles where the geocoder lands.
Any suggestions are much appreciated.
I ended up solving this by triggering the querying code once on 'moveend' event.
So now the syntax is:
geocoder.on('result', function(e){
map.once('moveend', function(){
.... rest of code
}
}
I thought I had tried this before posting the question, but seems to be working for me now.
I am exporting Google Directions routes as KML and displaying them on a Mapbox map by reading them with Omnivore and adding them to the map,
The Google KML stores each route as two Places (the start and end points) and one LineString (the route). In Mapbox I would like to show only the routes, that is to filter out the markers somehow. I'm displaying markers out of my own database and the Google markers clutter it up.
Here is my code. I change the styling of the LineStrings just to show that I can, but do not know what magic call(s) to make to not display the Points.
Thanks.
runLayer = omnivore.kml('data/xxxx.kml')
.on('ready', function() {
var llBnds = runLayer.getBounds();
map.fitBounds(llBnds);
this.eachLayer(function (layer) {
if (layer.feature.geometry.type == 'LineString') {
layer.setStyle({
color: '#4E3508',
weight: 4
});
}
if (layer.feature.geometry.type == 'Point') {
//
// Do something useful here to not display these items!!
//
}
});
})
.addTo(map);
Welcome to SO!
Many possible solutions:
Most straight forward from the code you provided, just use the removeLayer method on your runLayer Layer Group when you get a 'Point' feature.
Cleaner solution would be to filter out those features before they are even converted into Leaflet layers, through a custom GeoJSON Layer Group passed as 3rd argument of omnivore.kml, with a specified filter option:
var customLayer = L.geoJSON(null, {
filter: function(geoJsonFeature) {
// my custom filter function: do not display Point type features.
return geoJsonFeature.geometry.type !== 'Point';
}
}).addTo(map);
var runLayer = omnivore.kml('data/xxxx.kml', null, customLayer);
You can also use the style and/or onEachFeature options on customLayer to directly apply your desired style on your LineString.
I'm using Mapbox with Leaflet for drawing, editing and removing polygons etc. Every time I create a new polygon, I convert them to a GeoJSON layer and then add it to the featureGroup that I created, because I want to associate each layer with an ID property that I can use later. This is what I have:
var featureGroup = L.featureGroup().addTo(map);
var drawControl = new L.Control.Draw({
edit: {
featureGroup: featureGroup
},
draw: {
polygon: {
allowIntersection: false
},
polyline: false,
rectangle: false,
circle: false,
marker: false
}
}).addTo(map);
map.on('draw:created', addPolygon);
map.on('draw:edited', editPolygon);
map.on('draw:deleted', deletePolygon);
function deletePolygon(e) {
featureGroup.removeLayer(e.layer);
}
function editPolygon(e) {
featureGroup.eachLayer(function (layer) {
layer.eachLayer(function (layer) {
addPolygon({ layer: layer });
});
});
}
function addPolygon(e) {
var geojsonstring = e.layer.toGeoJSON();
var geojson = L.geoJson(geojsonstring);
featureGroup.addLayer(geojson);
}
When I do this, creating polygons is not a problem. But when I try to edit or delete polygons, it doesn't work properly.
When I try to edit a polygon, it tells me "TypeError: i.Editing is undefined". It doesn't allow me to cancel editing as well.
When I try to delete a polygon, it is not displayed in the map anymore, but it is still not removed from the featureGroup.
What am I doing wrong here?
Edit: The way I'm currently doing this is the same way that ghybs has suggested. But the problem is, once all the edits are done, the polygons are saved to a database (I convert it to a WKT string to save in SQLServer). And when the page is loaded the next time, the polygons are loaded back from the database, and the user can edit or delete them and save it back to the database.
As it is right now, when the user makes the edit and saves them again, it only creates duplicate polygons. and I don't know of any way to connect the edited polygons to the ones from the database.
So I thought if I could convert them to GeoJSON and assign an ID property to each layer (something like ID=0 if it is a new layer, and the corresponding polygonID from the database if it is loaded from the database). So that when they are saved again, I can check this ID and based on that, I can either update the available polygon, or create a new polygon in the database.
Is there a better way of doing this?
Not sure exactly why in addPolygon you go through a GeoJSON object that you convert back into a Leaflet layer group through L.geoJson?
You could have directly added the created layer, as in Leaflet.draw "draw:created" example:
function addPolygon(e) {
var layer = e.layer;
var feature = layer.feature = layer.feature || {}; // Initialize layer.feature
// use the feature.id: http://geojson.org/geojson-spec.html#feature-objects
feature.id = 0; // you can change it with your DB id once created.
featureGroup.addLayer(layer);
// record into DB, assuming it returns a Promise / Deferred.
recordToDb(layer.toGeoJSON()).done(function (result) {
feature.id = result.id; // Update the layer id.
});
}
As for the reason for you error, it is due to the fact that you add a (GeoJSON) Layer Group to your featureGroup, which Leaflet.draw plugin does not know how to handle. You must add only "non group" layers.
See also: https://gis.stackexchange.com/questions/203540/how-to-edit-an-existing-layer-using-leaflet
I have two geoJson layers being loaded - both layers are the same data for testing purposes, but being drawn from two different json files. When I turn the layers on and off in the layer controller, the draw order of the layers change.
Any ideas why this is happening?
I have put my code into a JSFiddle: http://jsfiddle.net/lprashad/ph5y9/10/ and the JS is below:
//styling for watersheds_copy
var Orange = {
"color": "#ff7800",
"weight": 5,
"opacity": 0.65
};
var Water_Orange = L.geoJson(watersheds_copy, {
style: Orange
});
Water_Orange.addData(watersheds_copy);
//these are blue
var Water_blue = L.geoJson(watersheds, {});
Water_blue.addData(watersheds);
//This sets the inital order - last in layer list being on top. Except minimal - tile layer is always on bottom
var map = L.map('map', {
center: [41.609, -74.028],
zoom: 8,
layers: [minimal, Water_Orange, Water_blue]
});
var baseLayers = {
"Minimal": minimal,
"Night View": midnight
};
//This controls the order in the layer switcher. This does not change draw order
var overlays = {
"Water_Orange": Water_Orange,
"Water_blue": Water_blue
};
L.control.layers(baseLayers, overlays).addTo(map);
LP
While searching I happened upon this site that shows some of the Leaflet code:
http://ruby-doc.org/gems/docs/l/leaflet-js-0.7.0.3/lib/leaflet/src/control/Control_Layers_js.html
In it I found this condition for the application of autoZIndex:
if (this.options.autoZIndex && layer.setZIndex) {
this._lastZIndex++;
layer.setZIndex(this._lastZIndex);
}
TileLayer is the only layer type that has a setZIndex function, so apparently autoZIndex only works there.
I'm not sure which annoys me more. This incredible limitation or the fact that Leafet documentation doesn't point it out.
At least on 0.7.2, I had to use bringToFront in the callback of map.on('overlayadd'). autoZIndex: false did not work in my case neither. A comment on this issue may explain the reason.
It's not specific to L.GeoJson layers. As far as I can tell, it's true of all Leaflet layers with layer control. The last layer turned on is simply on top. I don't think this is a bug either. It's predictable behavior which I use and depend on when I'm designing maps with layer control...