Add MIH Areas data validation test

Author: alexrichey

products/pluto/models/_sources.yml

@@ -65,6 +65,8 @@ sources:
     description: Cleaned MIH option names with unique identifiers (from miharea.sql)
   - name: mih_lot_overlap
     description: Spatial overlaps between lots and MIH areas (from miharea.sql)
+  - name: mih_distinct_options
+    description: QAQC view of all distinct MIH options found in the data (from miharea.sql)
   - name: transit_zones_block_to_tz_ranked
     description: Ranked transit zone assignments by block (from transitzone.sql)
   - name: transit_zones_bbl_to_tz_ranked

products/pluto/pluto_build/sql/miharea.sql

@@ -96,6 +96,20 @@ SELECT * FROM mih_areas
 WHERE perbblgeom >= 10 OR maxpermihgeom >= 50;
 
 
+-- QAQC: Create a view of all distinct MIH options found in the data
+-- This should only contain the four valid options: Option 1, Option 2, Deep Affordability Option, Workforce Option
+-- If additional options appear, there's a source data issue
+DROP VIEW IF EXISTS mih_distinct_options CASCADE;
+CREATE VIEW mih_distinct_options AS
+WITH split_options AS (
+    SELECT DISTINCT
+        unnest(string_to_array(cleaned_option, ',')) AS option
+    FROM mih_lot_overlap
+)
+SELECT DISTINCT trim(option) AS option
+FROM split_options
+ORDER BY option;
+
 -- NOTE: GIS will likely refactor dcp_mih into this pivoted format,
 -- so much this code will likely disappear.
 --
@@ -125,7 +139,7 @@ WITH bbls_with_all_options AS (
             WHEN (all_options LIKE '%Option 3%' OR all_options LIKE '%Deep Affordability Option%') = true THEN '1'
         END AS mih_opt3,
         CASE
-            WHEN (all_options LIKE '%Deep Affordability Option%') = true THEN '1'
+            WHEN (all_options LIKE '%Workforce Option%') = true THEN '1'
         END AS mih_opt4
     FROM bbls_with_all_options
 )

products/pluto/tests/assert_only_valid_mih_options.sql

@@ -0,0 +1,36 @@
+{{
+    config(
+        tags = ['de_check', 'minor', 'major'], 
+        meta = {
+            'description': '''
+                This test checks that only the four valid MIH options appear in the source data:
+                - Option 1
+                - Option 2
+                - Deep Affordability Option
+                - Workforce Option
+                
+                Any additional options indicate a source data issue that needs to be investigated.
+            ''',
+            'next_steps': 'Contact GIS to investigate unexpected MIH option values in source data'
+        }
+    ) 
+}}
+
+WITH valid_options AS (
+    SELECT option FROM (VALUES 
+        ('Option 1'),
+        ('Option 2'),
+        ('Option 3'),
+        ('Deep Affordability Option'),
+        ('Workforce Option')
+    ) AS t(option)
+),
+
+actual_options AS (
+    SELECT * FROM {{ source('build_sources', 'mih_distinct_options') }}
+)
+
+-- Return any options that are NOT in the valid list (test fails if any rows returned)
+SELECT option
+FROM actual_options
+WHERE option NOT IN (SELECT option FROM valid_options)