From a45a1d012b40f678a659d417eed50f76fcd4021d Mon Sep 17 00:00:00 2001
From: William Ridgeway <10358980+wrridgeway@users.noreply.github.com>
Date: Tue, 21 May 2024 17:09:07 -0500
Subject: [PATCH] Add `expression_is_false` dbt test macro (#464)

* Add new dbt test

* Use new test for nonlivable chars test

* Update dbt generic test readme
---
 .../schema/default.vw_pin_condo_char.yml      |  6 ++---
 dbt/tests/generic/README.md                   | 18 +++++++++++--
 .../generic/test_expression_is_false.sql      | 27 +++++++++++++++++++
 3 files changed, 46 insertions(+), 5 deletions(-)
 create mode 100644 dbt/tests/generic/test_expression_is_false.sql

diff --git a/dbt/models/default/schema/default.vw_pin_condo_char.yml b/dbt/models/default/schema/default.vw_pin_condo_char.yml
index 4a455db34..36be5e2be 100644
--- a/dbt/models/default/schema/default.vw_pin_condo_char.yml
+++ b/dbt/models/default/schema/default.vw_pin_condo_char.yml
@@ -98,11 +98,11 @@ models:
       - not_null:
           name: default_vw_pin_condo_char_bldg_is_mixed_use_not_null
           column_name: bldg_is_mixed_use
-      - expression_is_true:
+      - expression_is_false:
           name: default_vw_pin_condo_char_nonlivable_no_chars
           expression: |
-            (NOT is_common_area AND NOT is_parking_space)
-            OR (char_bedrooms IS NULL AND char_full_baths IS NULL)
+            (is_common_area OR is_parking_space)
+            AND (char_bedrooms IS NOT NULL OR char_full_baths IS NOT NULL)
           config:
             error_if: ">48080" # as of 2024-05-21
       # TODO: Non-liveable unit heuristics
diff --git a/dbt/tests/generic/README.md b/dbt/tests/generic/README.md
index c3bf4bf15..6a7f8dfb2 100644
--- a/dbt/tests/generic/README.md
+++ b/dbt/tests/generic/README.md
@@ -12,6 +12,7 @@ to define our test suite.
 - [`test_column_length`](#test_column_length)
 - [`test_columns_match`](#test_columns_match)
 - [`test_count_is_consistent`](#test_count_is_consistent)
+- [`test_expression_is_false`](#test_expression_is_false)
 - [`test_expression_is_true`](#test_expression_is_true)
 - [`test_is_null`](#test_is_null)
 - [`test_no_extra_whitespace`](#test_no_extra_whitespace)
@@ -110,6 +111,19 @@ the grouping column and a column called `count` with the count of rows for that
 
 * `group_column` (required string): The column to use for grouping.
 
+### `test_expression_is_false`
+
+Asserts that a valid SQL expression is false for all rows. In other words, filters for
+rows where a given `expression` is true. Often useful for idiosyncratic comparisons
+across columns that are not easily generalized into generic tests.
+
+**Parameters**:
+
+* `expression` (required string): A valid SQL expression to apply to the column or table.
+* `additional_select_columns` (optional list of strings): Additional columns to select for
+  failure output. The column the test is defined on will always be selected regardless
+  of this value.
+
 ### `test_expression_is_true`
 
 Asserts that a valid SQL expression is true for all rows. In other words, filters for
@@ -221,7 +235,7 @@ using `parid` and `taxyr`; as a result, it filters out mixed-use parcels as well
     `taxyr`.
   * `alias` (optional string): The name of the column to use for output. Necessary because
     aggregation functions as represented by `agg_func` require aliases in SQL. Defaults to
-    `<agg_func>_<column_name>`. 
+    `<agg_func>_<column_name>`.
 
 ### `test_row_count`
 
@@ -275,7 +289,7 @@ the test.
     columns specified in the `group_by` parameter.
   * `alias` (optional string): The name of the column to use for output. Necessary because
     aggregation functions as represented by `agg_func` require aliases in SQL. Defaults to
-    `<agg_func>_<column_name>`. 
+    `<agg_func>_<column_name>`.
 
 ### `test_sequential_values`
 
diff --git a/dbt/tests/generic/test_expression_is_false.sql b/dbt/tests/generic/test_expression_is_false.sql
new file mode 100644
index 000000000..858d0c65d
--- /dev/null
+++ b/dbt/tests/generic/test_expression_is_false.sql
@@ -0,0 +1,27 @@
+-- Filter for rows where a given `expression` is true.
+--
+-- Adapted from our own `expression_is_true`. Supports an optional
+-- `additional_select_columns` option representing an array of columns to select
+-- for failing rows in addition to `column_name`. If no
+-- `additional_select_columns` array is provided, defaults to selecting the
+-- column represented by `column_name`; if `column_name` is also missing, falls
+-- back to selecting 1 for failing rows.
+{% test expression_is_false(
+    model, column_name, expression, additional_select_columns=[]
+) %}
+    {%- set select_columns_csv = format_additional_select_columns(
+        additional_select_columns
+    ) -%}
+    {%- if column_name -%}
+        {%- set columns_csv = column_name -%}
+        {%- if select_columns_csv -%}
+            {%- set columns_csv = columns_csv ~ ", " ~ select_columns_csv -%}
+        {%- endif -%}
+    {%- elif select_columns_csv -%} {%- set columns_csv = select_columns_csv -%}
+    {%- else -%} {%- set columns_csv = "1 AS fail" -%}
+    {%- endif -%}
+
+    select {{ columns_csv }}
+    from {{ model }}
+    where ({{ column_name }} {{ expression }})
+{% endtest %}