Feature/surrogate key trim

integration_tests/data/sql/data_generate_surrogate_key_trim.csv

@@ -0,0 +1,5 @@
+column_1,column_2,column_3
+a,b,c
+  a  ,  b  ,  c  
+a  ,b,  c
+  a,  b,  c

integration_tests/models/sql/schema.yml

@@ -203,6 +203,11 @@ models:
       - assert_equal:
           actual: actual_all_columns_list
           expected: expected_all_columns
+
+  - name: test_generate_surrogate_key_trim
+    description: "Test that trim parameter correctly removes leading/trailing whitespace before hashing"
+    # Note: Aggregate tests are in tests/sql/test_generate_surrogate_key_trim_distinct_count.sql
+    # because expression_is_true cannot use aggregate functions in WHERE clause
 
   - name: test_union
     data_tests:

integration_tests/models/sql/test_generate_surrogate_key_trim.sql

@@ -0,0 +1,18 @@
+with data as (
+
+    select * from {{ ref('data_generate_surrogate_key_trim') }}
+
+)
+
+select
+    -- Test without trim (default behavior) - should produce different hashes for whitespace variations
+    {{ dbt_utils.generate_surrogate_key(['column_1', 'column_2', 'column_3']) }} as key_no_trim,
+
+    -- Test with trim enabled - should produce identical hashes regardless of whitespace
+    {{ dbt_utils.generate_surrogate_key(['column_1', 'column_2', 'column_3'], trim=true) }} as key_with_trim,
+
+    column_1,
+    column_2,
+    column_3
+
+from data

integration_tests/profiles.yml

@@ -45,4 +45,4 @@ integration_tests:
       database: "{{ env_var('SNOWFLAKE_DATABASE') }}"
       warehouse: "{{ env_var('SNOWFLAKE_WAREHOUSE') }}"
       schema: "{{ env_var('SNOWFLAKE_SCHEMA') }}"
-      threads: 10
+      threads: 10 

integration_tests/setup_env.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+export POSTGRES_HOST=localhost
+export POSTGRES_USER=dummy
+export POSTGRES_PASS=dummy
+export POSTGRES_PORT=5432
+export POSTGRES_DATABASE=dummy
+echo "Environment variables set for dbt testing"

integration_tests/tests/sql/test_generate_surrogate_key_trim_distinct_count.sql

@@ -0,0 +1,15 @@
+-- Test that the trim parameter correctly affects the number of distinct surrogate keys
+-- key_with_trim should have 1 distinct value (all whitespace trimmed = identical inputs)
+-- key_no_trim should have 4 distinct values (whitespace preserved = different inputs)
+
+with counts as (
+    select
+        count(distinct key_with_trim) as distinct_count_with_trim,
+        count(distinct key_no_trim) as distinct_count_no_trim
+    from {{ ref('test_generate_surrogate_key_trim') }}
+)
+
+select *
+from counts
+where distinct_count_with_trim != 1
+   or distinct_count_no_trim != 4

macros/sql/generate_surrogate_key.sql

@@ -1,8 +1,8 @@
-{%- macro generate_surrogate_key(field_list) -%}
-    {{ return(adapter.dispatch('generate_surrogate_key', 'dbt_utils')(field_list)) }}
+{%- macro generate_surrogate_key(field_list, trim=false) -%}
+    {{ return(adapter.dispatch('generate_surrogate_key', 'dbt_utils')(field_list, trim)) }}
 {% endmacro %}
 
-{%- macro default__generate_surrogate_key(field_list) -%}
+{%- macro default__generate_surrogate_key(field_list, trim=false) -%}
 
 {%- if var('surrogate_key_treat_nulls_as_empty_strings', False) -%}
     {%- set default_null_value = "" -%}
@@ -14,9 +14,15 @@
 
 {%- for field in field_list -%}
 
-    {%- do fields.append(
-        "coalesce(cast(" ~ field ~ " as " ~ dbt.type_string() ~ "), '" ~ default_null_value  ~"')"
-    ) -%}
+    {%- if trim -%}
+        {%- do fields.append(
+            "coalesce(trim(cast(" ~ field ~ " as " ~ dbt.type_string() ~ ")), '" ~ default_null_value  ~"')"
+        ) -%}
+    {%- else -%}
+        {%- do fields.append(
+            "coalesce(cast(" ~ field ~ " as " ~ dbt.type_string() ~ "), '" ~ default_null_value  ~"')"
+        ) -%}
+    {%- endif -%}
 
     {%- if not loop.last %}
         {%- do fields.append("'-'") -%}
@@ -26,4 +32,4 @@
 
 {{ dbt.hash(dbt.concat(fields)) }}
 
-{%- endmacro -%}
+{%- endmacro -%}