- Ruby v3.1
- Bundler v2.3.3
- PostgreSQL v14 with postgis extension
Development dependency: redis-server
These are the steps to run the project locally:
The root directory of the backend application is in heco-invest/backend.
You will need to create an .env file in the project root with values of environment variables required by the application. See ENV VARS documented in ENV_VARS.md
For documentation of server deployment, please refer to the infrastructure documentation.
In the root directory run bundle install.
You can use docker-compose.yml to run postgres service.
docker-compose up -d
bin/rails db:setup to setup the database
bin/dev and access the project on http://localhost:4000
If you want to debug the Rails app, running it through foreman could be not the best idea. In that case you can run css and js bundling
using foreman bin/watch and the server in the standard way bin/rails server in a separate terminal tab.
To get rid of basic auth, remove the username / password env vars from .env.
See the generated api docs (described below) for available API endpoints. The backoffice is accessed at /backoffice. All route paths are prefixed with the value of the RAILS_RELATIVE_URL_ROOT env var. For example, if RAILS_RELATIVE_URL_ROOT=backend, then all the API endpoints and backoffice urls will be prefixed e.g. /backend/backoffice.
We use cloudtasker gem to schedule ad-hoc background jobs. In development, you need redis-server running and a cloudtasker process.
bundle exec rspec
SWAGGER_DRY_RUN=0 rake rswag:specs:swaggerize
Documentation can be found at /api-docs.
On the first run, the match_snapshot matcher will always return success and it will store a snapshot file. On the next runs, it will compare the response with the file content.
If you need to replace snapshots, run the specs with:
REPLACE_SNAPSHOTS=true bundle exec rspec
If you only need to add, remove or replace data without replacing the whole snapshot:
CONSERVATIVE_UPDATE_SNAPSHOTS=true bundle exec rspec
bin/rails standard
To fix linter issues
bin/rails standard:fix
We use Transifex to provide the best UX for translators. The system is set up to use Zulu (zu) as the main language. This is a trick to enable translators to also update English translations (read more here).
Static content that needs to be translated has to be added only to config/locales/zu.yml file. To push new translations to Transifex use tx push -s.
To download already translated content bin/rails transifex:pull
We use the Google translate API to provide user-generated content in the languages of the platform. This runs as a background job. As with any external service this may sometimes be unavailable. In case of missing translations, the first place to look are the logs on the jobs instance and the error reporting service.
We use ARIES to fetch project demand scores for a given geometry and use those to calculate impact scores. This runs as a background job. As with any external service, this may sometimes be unavailable, and so there are a limited number of retries. In case of missing impact scores, the first place to look are the logs on the jobs instance and the error reporting service. Typical reasons for such situations are:
- ARIES is down
- The geometry contains some artefacts that ARIES is not able to handle
Proposed steps which should be taken in such case are:
- analyse logs and decide which part of communication with ARIES failed and what kind of error is raised (e.g. timeout, 500, etc.)
- if the error is not related to the geometry, but to the service itself, double check that ARIES is up and running and try to re-run the job manually
- if the error is related to the project geometry, try to fix it and re-run the job manually
More information about the integration can be found here.