Implementing Schema Generation

Cube.js Schema is Javascript code, which means the full power of this language can be used to configure your schema definitions. In this guide we generate several measure definitions based on an array of strings.

One example, based on a real world scenario, is when you have a single events table containing an event_type and user_id column. Based on this table you want to create a separate user count measure for each event.

It can be done as simple as

const events = ['app_engagement', 'login', 'purchase'];

cube(`Events`, {
  sql: `select * from events`,

  measures: Object.assign(
    {
      count: {
        type: `count`,
      },
    },
    events
      .map((e) => ({
        [`${e}_userCount`]: {
          type: `countDistinct`,
          sql: `user_id`,
          filters: [
            {
              sql: `${CUBE}.event_type = '${e}'`,
            },
          ],
        },
      }))
      .reduce((a, b) => Object.assign(a, b))
  ),
});

In this case we use standard Javascript functions Object.assign, Array.map and Array.reduce to add user count measure definitions based on events array. This approach allows you to maintain list of events in very concise manner without boilerplate code. This configuration can be reused using export / import feature.

Please refer to asyncModule documentation to learn how to use databases and other data sources for schema generation.