Support sql as a unit test `format`

  - name: reworked_compare_struct
    model: unit_reworked_compare_struct
    given:
      - input: ref('unit_test_struct_model_a')
        format: sql
        rows: |
          select 1 as id, 'John Doe' as col1, object_construct('street', '123 Main St', 'city', 'Anytown', 'state', 'CA') as col2
      - input: ref('unit_test_struct_model_b')
        format: sql
        rows: |
          select 1 as id, 'John Doe' as col1, object_construct('street', '123 Main St', 'city', 'Anytown', 'state', 'CA') as col2
    expect:
      rows:
        - {"id": 1, "dbt_audit_row_status": "identical", "dbt_audit_num_rows_in_status": 1}
    overrides:
      vars:
        reworked_compare__columns: ['id', 'col1', 'col2']
        reworked_compare__event_time:
        reworked_compare__primary_key_columns: ['id']

sql format has rows as a string, and also supports defining a fixture. Basically the same as csv

dbt-labs / dbt-jsonschema

Support sql as a unit test `format` #140