Realtime and general purpose open APIs at Digitransit

Presentation and workshop at API Days Nordic 2016, Tampere, Finland

Transcript

1. Tuukka Hastrup 2016-05-19 Realtime and general purpose open APIs at Digitransit
2. 1. APIs at HSL 2. Digitransit 3. GraphQL 4. MQTT
3. Helsinki Region and Finland • Helsinki regional transport authority (HSL.fi) • Finnish Transport Agency (FTA.fi) • Helsinki region: – 1.3 million inhabitants – 1 million boardings per day • Current HSL Journey Planner reittiopas.fi/en/ – 0.2 million users per day – Customers very satisfied: 50% NPS
4. Enabler role of government • Plan the public transport network • Provide open data and open APIs • Support app developers • Support MaaS (Mobility as a Service) startups
5. Use of HSL data Developer accounts 2009-2012 Sep 09 Data opens Sep 10 Apps4Finland Automated sign-up form Feb 11 HSL Mobile Competition 480 30 670 (789) < 5 apps +30 apps Sep 11 Apps4Finland Sep 12 Apps4Finland
6. HSL APIs dev.hsl.fi Journey planning Realtime departures Schedules Line map Disruption info Vehicle locations Biking and walking Open Data dev.hsl.fi +30 mobile apps
7. Routes Schedules Analyses Basic service Applications and services Quality monitoring Traffic planning APIs and examples Transport data: production, APIs and utilisation Open Data Novel services Responsibility of the authorities Third parties Disruptions Forecasts Vehicle locations
8. 2. Digitransit
9. 7.6.2012 Digitransit: The Concept • Integrated real-time passenger services for multi-modal trips • Open APIs first: This is not a single app, this is an application and a foundation for more: next generation, third parties, etc. • Mobile first: Small screens set the design constraints • HTML5 first: Can wrap HTML5 into native app later • Regular customers first − A commuter shouldn't need to re-choose the destination every morning and evening.
10. HTML5 mobile app • Open Data, Open API, Open Source • Leaflet • React, Relay, Fluxible • Webpack, CoffeeScript
11. Nearest routes Citybike routing
13. 3. GraphQL
15. GraphQL example (1/4) { stops { lat lon name } }
16. GraphQL example (2/4) { agency(id:"HSL") { routes { gtfsId type shortName longName }}}
17. GraphQL example (3/4) { stopsByRadius(lat:60.218, lon:24.816, radius:500){ edges { node { distance stop { gtfsId } } } } }
18. GraphQL example (4/4) { stop(id:"HSL:2111230") { name stoptimesWithoutPatterns(numberOfDepartures:20) { scheduledDeparture departureDelay trip { tripHeadsign route { type shortName } }}}}
20. Tools for GraphQL • GraphiQL • graphql-java, graphql-js, graphql-python, PostGraphQL • React & Relay, Apollo Client • A hack in Haskell: dropbox/datagraph
21. Wishlist for GraphQL tools • API management for GraphQL • Data browser
22. 4. MQTT
23. Vehicle locations
24. MQTT Example message payload (JSON): {"tsi":1431417982,"spd":5,"lat":60.17052,"long":24.94359,"dl":52 ..} MQTT topic structure: /hfp/journey/type/id/line/direction/headsign/start_time/ next_stop/geohash_level/geohash Example message topic: /hfp/journey/bus/67bf46c0/1055/1/Koskela/1105/ 1020169/4/60;24/19/74/03 Example subscriptions: /hfp/journey/# (all messages) /hfp/journey/+/+/1055/1/# (line 1055 outbound) /hfp/journey/+/+/+/+/+/+/+/60;24/19/# (geohash map rectangle)
27. Firefox: Websocket Monitor
28. MQTT in JavaScript var mqtt = require('mqtt'); var client = mqtt.connect('mqtt://mqtt.hsl.fi'); client.on('connect', function () { client.subscribe('/hfp/journey/#'); // client.publish('presence', 'Hello mqtt'); }); client.on('message', function (topic, message) { var data = JSON.parse(message); });
29. Conclusions 1. Consider how general purpose your API is 2. Consider ”Post-REST” such as GraphQL 3. Consider MQTT and topic structure for realtime
30. Thank you! @tuukkah Tuukka.Hastrup@hsl.fi Project site: digitransit.fi/en Open data at HSL: dev.hsl.fi