Add expect_column_distinct_count_to_equal_other_table (#50)

README.md

@@ -539,6 +539,40 @@ tests:
       value_set: ['a','b','c']
 ```
 
+
+#### [expect_column_distinct_count_to_equal_other_table](macros/schema_tests/aggregate_functions/expect_column_distinct_count_to_equal_other_table.sql)
+
+Expect the number of distinct column values to be equal to number of distinct values in another model.
+
+*Applies to:* Model, Column, Seed, Source
+
+This can be applied to a model:
+
+```yaml
+models: # or seeds:
+  - name: my_model_1
+    tests:
+      - dbt_expectations.expect_column_distinct_count_to_equal_other_table:
+          column_name: col_1
+          compare_model: ref("my_model_2")
+          compare_column_name: col_2
+```
+
+or at the column level:
+
+```yaml
+models: # or seeds:
+  - name: my_model_1
+    columns:
+      - name: col_1
+        tests:
+          - dbt_expectations.expect_column_distinct_count_to_equal_other_table:
+              compare_model: ref("my_model_2")
+              compare_column_name: col_2
+```
+
+If `compare_model` or `compare_column_name` are no specified, `model` and `column_name` are substituted. So, one could compare distinct counts of two different columns in the same model, or identically named columns in separate models etc.
+
 #### [expect_column_mean_to_be_between](macros/schema_tests/aggregate_functions/expect_column_mean_to_be_between.sql)
 
 Expect the column mean to be between a min_value value and a max_value value (inclusive).

integration_tests/models/schema_tests/schema.yml

@@ -26,6 +26,13 @@ models:
     tests:
         - dbt_expectations.expect_table_columns_to_match_ordered_list:
             column_list: ["date_day", "row_value", "row_value_log"]
+        - dbt_expectations.expect_column_distinct_count_to_equal_other_table:
+            column_name: date_day
+            compare_model: ref("timeseries_data_extended")
+            compare_column_name: date_day
+        - dbt_expectations.expect_column_distinct_count_to_equal_other_table:
+            column_name: date_day
+
     columns:
       - name: date_day
         tests:
@@ -38,6 +45,9 @@ models:
               column_type_list: [date, "{{ dbt_expectations.type_datetime() | trim }}"]
           - dbt_expectations.expect_column_values_to_be_increasing:
               sort_column: date_day
+          - dbt_expectations.expect_column_distinct_count_to_equal_other_table:
+              compare_model: ref("timeseries_data_extended")
+
 
       - name: row_value
         tests:

macros/schema_tests/aggregate_functions/expect_column_distinct_count_to_equal_other_table.sql

@@ -0,0 +1,27 @@
+{% macro test_expect_column_distinct_count_to_equal_other_table(model,
+                                                                compare_model,
+                                                                column_name,
+                                                                compare_column_name,
+                                                                row_condition=None,
+                                                                compare_row_condition=None
+                                                                ) %}
+{%- set expression -%}
+count(distinct {{ column_name }})
+{%- endset -%}
+{%- set compare_expression -%}
+{%- if compare_column_name -%}
+count(distinct {{ compare_column_name }})
+{%- else -%}
+{{ expression }}
+{%- endif -%}
+{%- endset -%}
+{{ dbt_expectations.test_equal_expression(
+    model,
+    expression=expression,
+    compare_model=compare_model,
+    compare_expression=compare_expression,
+    row_condition=row_condition,
+    compare_row_condition=compare_row_condition,
+    return_difference=True
+) }}
+{%- endmacro -%}