Home | Tasks | Build/Deployment

RFCx Core API

Introduction

The reposititory consists of several services sharing a common base: Core API, Core Tasks, Noncore API (Guardian-specific), MQTT (Guardian-specific). All the services require a Timescale database (based on Postgres). Some services also dependend on AWS (S3 and SNS/SQS), Auth0, Firebase, Mailchimp/Mandrill, Stripe and Classy.

Quick start (Core API)

Requirements:

• Node 20.9.0 (can be installed via nvm module and .nvmrc file)
• yarn
• Docker
• psql

1. Install dependencies yarn

2. Copy config file cp common/config/env_vars.js.sample common/config/env_vars.js

3. Start the db, run migrations and seeds yarn serve-db:core

4. Run the Core API in local dev mode yarn dev:core

5. Navigate to the API docs http://localhost:8080/docs/

6. Start the test db yarn serve-db:test

7. Make sure your tests are passing yarn test

Project configuration for local development

Node version

Required Nodejs version is specified in .nvmrc file. You can use nvm to install/select the correct node version.

If required node version is not yet installed on your machine:

nvm install

If required node version is installed on your machine:

nvm use

Env variables

Clone and copy ./common/config/env_vars.js.sample into ./common/config/env_vars.js and fill it with required vars.

Run all the required support services

Install dependencies:

yarn

We can use docker-compose to run a local Timescale database and other services. You can use docker compose up -d or docker compose down, or simply:

yarn serve-db:core

Or for noncore:

yarn serve-db:noncore

For some users that have a local database driver installed on port 5432, the command up top might not work because it's trying to access your local database. Please change your local database port before trying to run the command again.

Create the database tables

To run sync and migrations

For MySQL (noncore):

yarn migrate:noncore

For TimescaleDB (core):

yarn migrate:core

Seed the database

Requires psql (on Mac with brew, use brew install libpq and follow steps to add to your path).

If you are using a local dev environment with docker compose then:

./common/_cli/seed.sh

Otherwise, specify your host/user/pass/etc as arguments:

./common/_cli/seed.sh USERNAME PASSWORD HOSTNAME PORT DATABASENAME

For some users that are on Windows, your seed command might fail because of Windows's line endings (CRLF). So you have to convert the line endings from CRLF to LF first to be able to run the command. You can fire up WSL and use tr

tr -d "\r" < file

Running the API locally

HTTP API

To run the core API with auto-reload:

yarn dev:core

Similarly, for the non-core API (only v1/v2 endpoints, without core):

yarn dev:noncore

Without live reload (how it runs in production):

yarn start:core
yarn start:noncore

MQTT API

To start the MQTT client and API, open another terminal then:

yarn start:mqtt

Testing

This project uses Jest for testing and ESLint for linting.

Linting

Run ESLint:

yarn lint

Fix all fixable errors:

yarn lint-fix

Recommend developers to setup ESLint in their IDE. For VS Code, settings are included in the workspace settings already and lint/format automatically when the the ESLint extension is installed.

Unit tests

Unit tests don't require a database. They typically test the logic of a function or module on its own. Mark a test as a unit test by naming it *.unit.test.js.

In general, unit tests should be saved alongside the code files that they apply to (e.g. for a module xyz.js, if there are tests then they will be stored in xyz.unit.test.js).

Run all unit tests:

yarn test:unit

Run a specific test suite (matching a pattern, e.g. any file containing array in the file path):

yarn test:unit array

When targeting a specific test suite, to run a single test case you can temporarily add .only to the case:

test.only('can camelize', () => { ... })

Integration tests

The goal of integration tests is to test a set of components (functions, modules) working together. Integration tests often require a database (but it is not a requirement).

To run all the integration tests, you will need to have a test database running. It's recommended to use a separate database from your development database because data will be dropped from the tables when running tests.

To start a test database (requires Docker to be running):

yarn serve-db:test

Run all the integration tests:

yarn test:int

Similar to unit tests, you can run individual test suites (matching a pattern):

yarn test:int classifier-jobs/dequeue

If you want to run unit and integration tests together then:

yarn test