Leverage generate_model_yaml to create LookML code

[victorurbinajz]

Describe the feature

• Hi, everyone!
• I'm using data contracts, and creating the code is not an easy task.
• In order to make the process smoother, I've been using the macro generate_model_yaml
• However, I also noticed that with some minor modifications, this macro can also be useful to create LookML code.

Describe alternatives you've considered

• This is my suggestion for the new macro:

{% macro generate_column_looker(column, model_yaml, column_desc_dict, include_data_types, parent_column_name="") %}
    {% if parent_column_name %}
        {% set column_name = parent_column_name ~ "." ~ column.name %}
    {% else %}
        {% set column_name = column.name %}
    {% endif %}

    {% do model_yaml.append('  dimension: ' ~ column_name  | lower ~ ' {') %}
    {% if include_data_types %}
        {% set version = (dbt_version.split('.')[0] ~ '.' ~  dbt_version.split('.')[1]) | float %}
        {% if version < 1.5 %}
            {% set formatted = format_column(column) %}
        {% else %}
            {% set formatted = adapter.dispatch('format_column', 'dbt')(column) %}
        {% endif %}

        ---
        {% set data_type_name = formatted['data_type'] | lower %}
        {% if data_type_name in ['varchar', 'text'] %}
            {% set data_type_name = 'string' %}
        {% endif %}
        {% if data_type_name in ['timestamp_ltz', 'timestamp','timestamp_ltz'] %}
            {% set data_type_name = 'time' %}
        {% endif %}
        {% if data_type_name in ['boolean'] %}
            {% set data_type_name = 'yesno' %}
        {% endif %}
        {% if data_type_name in ['int','bigint','float8','float'] %}
            {% set data_type_name = 'number' %}
        {% endif %}
        ---

        {% do model_yaml.append('   type: ' ~ data_type_name | lower) %}
        {% do model_yaml.append('   sql: ${TABLE}.' ~ column_name  | lower ~ ' ;;') %}
        
    {% endif %}
    {% do model_yaml.append('  }') %}
    {% do model_yaml.append('') %}

    {% if column.fields|length > 0 %}
        {% for child_column in column.fields %}
            {% set model_yaml = codegen.generate_column_yaml(child_column, model_yaml, column_desc_dict, include_data_types, parent_column_name=column_name) %}
        {% endfor %}
    {% endif %}
    {% do return(model_yaml) %}
{% endmacro %}

{% macro generate_view_code_looker(model_names=[], upstream_descriptions=False, include_data_types=True) %}

    {% set model_yaml=[] %}
    {% do model_yaml.append('\n') %}    
    
    {% if model_names is string %}
        {{ exceptions.raise_compiler_error("The `model_names` argument must always be a list, even if there is only one model.") }}
    {% else %}
        {% for model in model_names %}

            {% do model_yaml.append('view: ' ~ model ~ ' {') %}
            {% do model_yaml.append('  sql_table_name: analytics.' ~ model ~ ' ;;') %}
            {% do model_yaml.append('\n') %}
            {% set relation=ref(model) %}
            {%- set columns = adapter.get_columns_in_relation(relation) -%}
            {% set column_desc_dict =  codegen.build_dict_column_descriptions(model) if upstream_descriptions else {} %}

            {% for column in columns %}
                {% set model_yaml = generate_column_looker(column, model_yaml, column_desc_dict, include_data_types) -%}
            {% endfor %}
        {% endfor %}
    {% endif %}
    {% do model_yaml.append('\n') %}
    {% do model_yaml.append('}') %}
    
{% if execute %}

    {% set joined = model_yaml | join ('\n') -%}
    {{ log(joined, info=True) }}
    {% do return(joined) %}

{% endif %}

{% endmacro %}

• How to run this macro?

dbt run-operation generate_view_code_looker --args '{"model_names": ["model_name"]}'

Alternatively, the code can be put into a yml file

dbt run-operation generate_view_code_looker --args '{"model_names": ["model_name"]}' > sample_file.yml

Additional context

I'm using Snowflake, that's the only place where I've tested this macro.

Who will this benefit?

• Use case: this macro can be used to create the LookML basic code to create Looker views. It's a good starting point because it already adds the data type and declares all the dimensions available.

Are you interested in contributing this feature?

• For sure. I'd love to contribute.

[graciegoheen]

Hey! If you're adding contacts to your dbt models, you may want to check out the dbt-meshify package (specifically the add-contract command).

The scope of the dbt-codegen package is generating code for use within your dbt project. Since LookML code is separate from a dbt project, I'm going to close this issue for now.