This is the UI of the Reittiloki application. The project also includes a server-side component, which can be found at https://github.com/HSLdevcom/transitlog-server.
Reittiloki is an application for viewing recorded historical high-frequency positioning (HFP) data from all of HSL's vehicles in the larger Helsinki region. This includes buses, trams, metro, trains and ferries. It is used to inspect and improve the quality of the capitol region public transport.
All vehicles broadcast their current position (plus other information) over MQTT once per second. This data is stored by the Transitlog project and can be queried for up to 4 months. We are working on making older data available in Reittiloki. This data is combined with and compared against planned schedules and routes to produce the data which is visualized on a map and queryable with Transitlog UI.
Reittiloki is a specialist application that requires a certain level of knowledge of the concepts and data that the application deals with. The user must also know what their objective in using the application is and what they are looking for. While some effort has been made to make the UI intuitive to use, the data-density of the application is unavoidable. The goal of the UI is that you need to learn how things are done, but only once.
Transitlog UI is a desktop-only app and it is not fun to use on mobile. Please use Chrome-based or Firefox-based browsers with Transitlog UI.
These are some challenges that we've solved in Transitlog UI.
Each day in Reittiloki is actually around 28 hours long. We call this 24h+ time
. This is because traffic operation does not stop at midnight, but continues into the night until about 4:30 am. The night traffic is still counted as belonging to the previous day in all source data that Transitlog UI uses. After 4:30 there is usually in a brief pause with the traffic picking up again for the current day at around 5 am.
Transitlog UI needs a way to translate between 24h+ times and normal 24h times. Transitlog must also always show all times as Finnish times, so we cannot rely on the native Date object since it will show the current time of the location where the user is. HSL employees will sometimes access Reittiloki from outside Finland which would be a problem.
To solve this, time is represented as seconds from midnight. This is most evident in the time slider above the map. Sliding it will decrease or increase the seconds since midnight. A handful of helper functions can be used to translate the seconds into time strings, either for display or for querying from the server. Time displays will display the 24h+ time (eg. "28:42:00"). For querying, the constant departure time is usually used along with the date.
The app displays a very large amount of data, which is mostly solved server-side. The UI uses a variety of caching methods depending on what the situation calls for.
For example, when you have a departure selected, a marker representing the vehicle will be displayed on the map and it moves according to where the vehicle as at the currently selected time. When dragging the slider, all the HFP data points need to be traversed frequently to get the exact HFP item which corresponds to the current time. The solution is to index each HFP item in a Map by the timestamp, and then the current time can be used to retrieve the HFP item from the Map. The collection of HFP data points are thus only traversed once.
This is a Create React App project. We use modern React with hooks.
State is handled with MobX and Mobx-App.
GraphQL is used for communicating with the server and it is managed by Apollo. Graphql-codegen is used to integrate the GraphQL schema with the UI.
There is no router, but we encode all the app's state into the URL on all state changes and use HTML5 history to manipulate the URL without page transitions. This enables very deep linking and state sharing.
The map is powered by Leaflet and React Leaflet.
Styling is handled by styled-components.
Docker is used to package the app and run it in production.
Testing is mainly performed with Cypress.io, but we also have a few Jest unit and integration tests.
The application is deployed on Azure and runs in a Docker Swarm.
Transitlog UI mainly displays data from the Transitlog Server, but there are also a few other integrations. The dependencies are:
Run yarn
once to install dependencies.
In the project directory, you can run:
yarn start
Runs the app in the development mode.
Open http://localhost:3000 to view it in the browser.
The page will reload if you make edits.
yarn run build
Builds the app for production to the build
folder. It correctly bundles React in production mode and optimizes the build for the best performance.
yarn run codegen
If you update the schema in transitlog-server, be sure to tun Codegen to bring the updated schema to the UI.
./deploy-env.sh
A custom build script that builds a Docker image for a specific environment. When asked, press the number for the environment you want to build for.
Before running the script, ensure that you are logged in to Docker hub through Docker as the script will try to push the image.
./deploy-all.sh
Builds and tags Docker images for all environments.
Each environment has its own set of environment variables. Since this is a frontend app, no secrets should be included in any env files.
When building for an environment, the corresponding .env file will be included as .env.production
, which CRA uses as the de facto env file of the build. The env file that is included in the build is determined by the Docker BUILD_END
ARG, which should be either dev
, stage
, production
, or match any other of the available env file endings. Just use the deploy scripts and it will work.
The app is deployed to a Docker swarm running on Azure. The deployment itself is managed by Gitlab pipelines, one repo for each environment. This repo also contains the service configuration for the app in the swarm, as well as the nginx configuration.
First, build and push a new image for the environment you want to update, then simply run the pipeline for the corresponding environment:
Dev: https://gitlab.hsl.fi/transitlog/transitlog-app-dev-deploy
Stage: https://gitlab.hsl.fi/transitlog/transitlog-app-stage-deploy
Prod: https://gitlab.hsl.fi/transitlog/transitlog-app-prod-deploy
Each branch has a corresponding Github actions workflow that triggers on certain events. The master branch deploys to dev, which is triggered by a merged PR. The staging and production branches deploy to their respective environments on all pushes.
The github actions workflow builds, pushes and deploys the app to Azure. So when using Github actions, you do not need to use the ./deploy-env
scripts or manually trigger the pipeline.
After the update is deployed, the actions workflow starts the Cypress E2E testsuite.
All Github actions workflow statuses are posted to the #transitlog-deployment
channel in the HSLdevcom Slack.
A Cypress testsuite is also triggered to run every day on a schedule. The result of this is also posted to the #transitlog-deployment
channel.
To develop and deploy a feature or bugfix, follow these steps:
./deploy-env.sh
and select 2 (for development) when asked by the script. This will build and push an image for the dev environment. Then go to the deployment repo in Gitlab and run the pipeline.master
branch into the staging
branch. A github actions workflow will start and update the staging environment.staging
branch into the production
branch.You can always also circumvent the Github workflow by building the images with the deploy.sh
script and triggering the pipeline in Gitlab manually.
The following section outlines the architecture and how the application should be developed. If you are unfamiliar with any of the main libraries listed in the Tech Stack section, please familiarise yourself with those as how they work is not in the scope of this documentation. This includes Create React App and React apps in general.
The source files are divided into directories based on what they do. The main directories are:
The container directory for all source files. The root contains central bootstrap files.
Contains all visible view components. There may be React components elsewhere too, but most of the visible stuff is in here. The root level contains common components, and the components belonging to the main sections of the app are divided into subdirectories.
The authentication interface is contained here in the auth directory.
This directory contains all helper functions. These are primarily shorter functions used in multiple places throughout the app.
Custom React Hooks are separated from the helper functions into this directory.
The HSL app icon library is contained here.
This directory contains the translations for all text in the UI. The texts are subdivided into alerts, help and ui. Alerts contains only translations of the alert and cancellations categories and types, help contains tooltip texts and UI contains shorter UI strings.
Contains most of the GraphQL queries used throughout the app as higher-order components which can be used to wrap the component you want to provide data for. Note that some components use locally-defined queries too.
Contains the state of the app, divided into different stores based on the area of the state they handle. It also contains store update actions as well as the UrlManager which is essentially a store.
Contains the Cypress test suite.
The following modules are central to making Transitlog UI tick.
This is the main component that ties everything together. It renders the main elements of the UI; the map, the sidepanel and the filter bar. It also fetches some commonly used data and distributes it through the components.
This is also where the useAuth
hook is used. It plucks the HSL ID authentication code from the URL upon return to the app and sends it to the server for to authenticate the user. When no code, it fetches the user info and updates the app state to show the logged-in user.
This is where Apollo GraphQL is set up.
This is where all CSS variables are defined.
This is where app configuration is located. Some values come from the environment variables.
This file ties all stores together and initializes them with Mobx-App. It then provides the store context which wraps the whole app. If you add a store, be sure to add the store to this list.
This is where Leaflet is configured and mounted. The actual map components for Transitlog are mounted in the MapContent component.
This is where all Transitlog custom map components are mounted. All Leaflet setup is done in the Map component. This component is largely where the "what is visible when" logic for the map is located.
This is the filterbar on the top of the page, containing search fields and date/time fields.
This is the sidepanel component which renders the sidebar. It displays various lists, mostly departures, by route, stop, terminal, vehicle, area or week, depending on what selections are made in the app. It is also the parent of the Journey Details sidebar which appears when a journey is selected.
This is the Journey Panel component which appears when a journey is selected. it displays detailed information about the journey.
Substitute the configFile with the environment you want to test. Leave the whole --env
option out to test the local version.
yarn run cypress run --env configFile=production
The configFile value works the same as above.
yarn run cypress open --env configFile=production