Create expect_column_set_to_be_unique_case_insensitive.sql (#138)

* Create expect_column_set_to_be_unique_case_insensitive.sql * adjusting consistent casing test and documentation * adjusting formatting and adding integration test * Fix formatting * Formatting updates * adjusting Readme and adding test instance * Fix spelling Co-authored-by: clausherther <[email protected]>
calogica · Feb 24, 2022 · 854e653 · 854e653
1 parent 590d599
commit 854e653
Show file tree

Hide file tree

Showing 3 changed files with 58 additions and 3 deletions.
diff --git a/README.md b/README.md
@@ -88,6 +88,7 @@ To run the tests:
 - [expect_column_values_to_be_unique](#expect_column_values_to_be_unique)
 - [expect_column_values_to_be_of_type](#expect_column_values_to_be_of_type)
 - [expect_column_values_to_be_in_type_list](#expect_column_values_to_be_in_type_list)
+- [expect_column_values_to_have_consistent_casing](#expect_column_values_to_have_consistent_casing)
 
 ### Sets and ranges
 
@@ -410,6 +411,18 @@ tests:
       column_type_list: [date, datetime]
 ```
 
+### [expect_column_values_to_have_consistent_casing](macros/schema_tests/column_values_basic/expect_column_values_to_have_consistent_casing.sql)
+
+Expect a column to have consistent casing. By setting `display_inconsistent_columns` to true, the number of inconsistent values in the column will be displayed in the terminal whereas the inconsistent values themselves will be returned if the SQL compiled test is run.
+
+*Applies to:* Column
+
+```yaml
+tests:
+  - dbt_expectations.expect_column_values_to_have_consistent_casing:
+      display_inconsistent_columns: false # (Optional)
+```
+
 ### [expect_column_values_to_be_in_set](macros/schema_tests/column_values_basic/expect_column_values_to_be_in_set.sql)
 
 Expect each column value to be in a given set.

diff --git a/integration_tests/models/schema_tests/schema.yml b/integration_tests/models/schema_tests/schema.yml
@@ -274,9 +274,11 @@ models:
           - dbt_expectations.expect_column_distinct_values_to_contain_set:
               value_set: ['a','b']
               quote_values: true
-
-          - dbt_expectations.expect_column_value_lengths_to_equal   :
+          - dbt_expectations.expect_column_value_lengths_to_equal:
               value: 1
+          - dbt_expectations.expect_column_values_to_have_consistent_casing
+          - dbt_expectations.expect_column_values_to_have_consistent_casing:
+              display_inconsistent_columns: true
 
       - name: col_string_b
         tests:
@@ -328,4 +330,3 @@ models:
             datepart: day
             interval: 1
             row_condition: group_id = 4
-
diff --git a/macros/schema_tests/column_values_basic/expect_column_values_to_have_consistent_casing.sql b/macros/schema_tests/column_values_basic/expect_column_values_to_have_consistent_casing.sql
@@ -0,0 +1,41 @@
+{% test expect_column_values_to_have_consistent_casing(model, column_name, display_inconsistent_columns=False) %}
+
+with test_data as (
+
+    select
+        distinct {{ column_name }} as distinct_values
+    from
+        {{ model }}
+
+ ),
+ {% if display_inconsistent_columns %}
+ validation_errors as (
+
+    select
+        lower(distinct_values) as inconsistent_columns,
+        count(distinct_values) as set_count_case_insensitive
+    from
+        test_data
+    group by 1
+    having
+        count(distinct_values) > 1
+
+ )
+ select * from validation_errors
+ {% else %}
+ validation_errors as (
+
+    select
+        count(1) as set_count,
+        count(distinct lower(distinct_values)) as set_count_case_insensitive
+    from
+        test_data
+
+ )
+ select *
+ from
+    validation_errors
+ where
+    set_count != set_count_case_insensitive
+ {% endif %}
+ {%- endtest -%}