Fix: handle whitespace trimming in dbt test-to-macro jinja normalization (#5232)

Author: georgesittas

examples/sushi_dbt/models/schema.yml

@@ -21,6 +21,8 @@ models:
         tests:
           - less_than_amount:
               amount: 1000
+          - greater_than_amount:
+              amount: 0
       - name: ds
         description: Date
   - name: top_waiters

examples/sushi_dbt/tests/generic/greater_than_amount.sql

@@ -0,0 +1,5 @@
+{%- test greater_than_amount(model, column_name, amount) -%}
+	select *
+	from {{ model }}
+	where {{ column_name }} < {{ amount }}
+{%- endtest -%}

sqlmesh/dbt/manifest.py

@@ -673,12 +673,17 @@ def _node_base_config(node: ManifestNode) -> t.Dict[str, t.Any]:
 
 
 def _convert_jinja_test_to_macro(test_jinja: str) -> str:
-    TEST_TAG_REGEX = r"\s*{%\s*test\s+"
-    ENDTEST_REGEX = r"{%\s*endtest\s*%}"
+    TEST_TAG_REGEX = r"\s*{%-?\s*test\s+"
+    ENDTEST_REGEX = r"{%-?\s*endtest\s*-?%}"
+
     match = re.match(TEST_TAG_REGEX, test_jinja)
     if not match:
         # already a macro
         return test_jinja
 
-    macro = "{% macro test_" + test_jinja[match.span()[-1] :]
-    return re.sub(ENDTEST_REGEX, "{% endmacro %}", macro)
+    test_tag = test_jinja[: match.span()[-1]]
+
+    macro_tag = re.sub(r"({%-?\s*)test\s+", r"\1macro test_", test_tag)
+    macro = macro_tag + test_jinja[match.span()[-1] :]
+
+    return re.sub(ENDTEST_REGEX, lambda m: m.group(0).replace("endtest", "endmacro"), macro)

tests/dbt/test_manifest.py

@@ -7,7 +7,7 @@
 from sqlmesh.core.config import ModelDefaultsConfig
 from sqlmesh.dbt.basemodel import Dependencies
 from sqlmesh.dbt.context import DbtContext
-from sqlmesh.dbt.manifest import ManifestHelper
+from sqlmesh.dbt.manifest import ManifestHelper, _convert_jinja_test_to_macro
 from sqlmesh.dbt.profile import Profile
 from sqlmesh.dbt.builtin import Api, _relation_info_to_relation
 from sqlmesh.dbt.util import DBT_VERSION
@@ -257,3 +257,45 @@ def test_top_level_dbt_adapter_macros():
     customers_macros = helper.macros("customers")
     assert not customers_macros["default__current_engine"].info.is_top_level
     assert not customers_macros["duckdb__current_engine"].info.is_top_level
+
+
+def test_convert_jinja_test_to_macro():
+    # Test block with whitespace trimming
+    test_input = """{%- test assert_positive(model, column_name) -%}
+    select * from {{ model }} where {{ column_name }} <= 0
+{%- endtest -%}"""
+
+    expected_output = """{%- macro test_assert_positive(model, column_name) -%}
+    select * from {{ model }} where {{ column_name }} <= 0
+{%- endmacro -%}"""
+
+    assert _convert_jinja_test_to_macro(test_input) == expected_output
+
+    # Test block without whitespace trimming
+    test_input_no_ws = """{% test assert_positive(model, column_name) %}
+    select * from {{ model }} where {{ column_name }} <= 0
+{% endtest %}"""
+
+    expected_output_no_ws = """{% macro test_assert_positive(model, column_name) %}
+    select * from {{ model }} where {{ column_name }} <= 0
+{% endmacro %}"""
+
+    assert _convert_jinja_test_to_macro(test_input_no_ws) == expected_output_no_ws
+
+    # Test block with mixed whitespace trimming
+    test_input_mixed = """{%- test complex_test(model, column_name='id') %}
+    select count(*) from {{ model }} where {{ column_name }} is null
+{% endtest -%}"""
+
+    expected_output_mixed = """{%- macro test_complex_test(model, column_name='id') %}
+    select count(*) from {{ model }} where {{ column_name }} is null
+{% endmacro -%}"""
+
+    assert _convert_jinja_test_to_macro(test_input_mixed) == expected_output_mixed
+
+    # Test already converted macro (should return unchanged)
+    macro_input = """{%- macro test_already_converted(model) -%}
+    select * from {{ model }}
+{%- endmacro -%}"""
+
+    assert _convert_jinja_test_to_macro(macro_input) == macro_input