Deploying OSMR

I'm working on a routing service, and been playing around with different projects based off of the OpenStreetMap project.

Here's what I went through to deploy the OpenSourceRoutingMachine project locally.

1
  docker volume create map_data

We'll startup a container that has that volume mounted it in, and then download the data from the internet and process it as needed.

Prepare the data

1
  docker run -it --rm -v map_data:/data ghcr.io/project-osrm/osrm-backend bash

https://download.geofabrik.de/ has a lot of files available that we can use to import, lets find an .osm.pbf file, say something at https://download.geofabrik.de/north-america.html

Inside of the container:

1
2
3
4
5
  FILE=north-america
  # FILE=connecticut
  apt-get update && apt-get install -y wget 
  cd /data
  wget https://download.geofabrik.de/${FILE}-latest.osm.pbf

That file is 14G fyi, and takes about 15m just to download on my connection!

Then extract:

1
2
3
  osrm-extract -p /opt/car.lua /data/${FILE}-latest.osm.pbf || echo "osrm-extract failed"
  osrm-partition /data/${FILE}-latest.osrm || echo "osrm-partition failed"
  osrm-customize /data/${FILE}-latest.osrm || echo "osrm-customize failed"

And you can quit out of that container.

Run the server

1
2
3
4
5
6
  docker run -it \
         -p 5000:5000 \
         -v map_data:/data \
         ghcr.io/project-osrm/osrm-backend osrm-routed \
         --algorithm mld \
         /data/north-america-latest.osrm

Once that gets going, we can hit it locally and pass the results through jq:

1
2
  curl "http://127.0.0.1:5000/route/v1/driving/-73.364755,41.838942;-73.409656,41.752237?steps=true" \
      | jq -r '.routes | .[] | .legs | .[] | .steps | .[] | .name'

Popple Swamp Road
River Road South
Kent Road South
Warren Hill Road
Cornwall Road
Kent Road
Segar Mountain Road
Kenico Road
Gorham Road
Old Homestead Lane
Stonewall Lane

CLI Testing

There's also a library that makes better words out of the responses.

1
  npm i osrm-text-instructions

Quick formatting demo:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
  var version = 'v5';
  var osrmTextInstructions = require('osrm-text-instructions')(version);

  fetch(
      "http://127.0.0.1:5000/route/v1/driving/-73.364755,41.838942;-73.409656,41.752237?steps=true" )
      .then( (response) => response.json() )
      .then( (response) => {
          response.routes[0].legs.forEach(function(leg) {
              leg.steps.forEach(function(step) {
                  instruction = osrmTextInstructions.compile('en', step)
                  console.log( instruction )
              });
          })
      })

Which gives us:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
  Head southwest on Popple Swamp Road
  Turn right onto River Road South
  Go straight onto Kent Road South (US 7)
  Make a slight left onto Warren Hill Road (CT 45)
  Continue onto Cornwall Road (CT 45)
  Turn right onto Kent Road (CT 341)
  Continue onto Segar Mountain Road (CT 341)
  Turn right onto Kenico Road
  Continue slightly left onto Gorham Road
  Turn right onto Old Homestead Lane
  Continue left onto Stonewall Lane
  Turn right
  You have arrived at your destination, on the right

Starting a front end

It does come with a front end, but I think Leaflet is more promising as a start.

1
2
  docker run -p 9966:9966 osrm/osrm-frontend
  xdg-open 'http://127.0.0.1:9966'

Deploying OSMR

runs pretty fast locally

Contents

Prepare the data

Run the server

CLI Testing

Starting a front end

References

Port 5000 already open on OSX

airplay receiver

Wait for the download to finish with puppeteer

Page.downloadProcess