Command Line Interface

There are a few cli methods Zenlytic's metrics layer provides using its cli ml.

The first is the debug command. Once you finish git setup and connection setup, simply run the following command from the terminal.

$ ml debug

Initialize

To initialize your project run this command.

$ ml init

This will create folders for models and views in addition to creating a file called zenlytic_project.yml that will tell Zenlytic which dbt profile to look for, and where to find the data model itself.

An example of that file will look like this

name: my_project_name # This can be anything you wantmode: metricflow (only necessary when using dbt Metricflow)profile: my_dbt_profile # This references the name in your dbt profiles.yml file
model-paths: ['models'] # This is the folder where your models are (optional with dbt Metricflow)view-paths: ['views'] # This is the folder where your views are (unnecessary with dbt Metricflow)dashboard-paths: ['dashboards'] # This is the folder where your dashboards are

Seeding

One of the most helpful if you're just starting out is the seed method. If you haven't started writing your data model or if you've recently added a new database, it's a pain to go in and manually set up the views and columns. seed allows you to specify a database and optionally a schema, and it will programmatically read the tables in that database or schema and generate the views and fields to give you a nice starting place.

Here are some examples (the $ indicates these examples are taking place in the terminal):

If you want to generate initial views from only tables in the prod_zendesk schema inside the analytics database

$ ml seed --database analytics --schema prod_zendesk

If you want to generate initial views from all tables inside the analytics database

$ ml seed --database analytics 

If you want to generate initial views from the zendesk_metrics table

$ ml seed --database analytics --schema prod_zendesk --table zendesk_metrics

Validation

Another helpful feature is the validate feature. It allows you to statically check if you have any broken references or invalid types in your project before deploying it anywhere or using it top query live data. It is run with the following command.

$ ml validate

Two additional helper commands are the list and show commands. The list command lists the names of objects from the type you pass, and the show command shows printable attributes of the object with the name you pass. Here are some examples.

This lists the names of all the models in your project

$ ml list

This command lists all the metrics (fields of type measure) in the view transactions

$ ml list --view transactions metrics

This command shows the attributes of the field total_revenue

$ ml show --type field total_revenue

This command shows the attributes of the view transaction_lines

$ ml show --type view transaction_lines

Have any other ideas for good cli tools to help with development? Open a GitHub issue and let us know!