[Bigquery] unit test tests + support for complex types (#1031)

* first pass: unit test typing

* expand test coverage, update safe_cast

* TestBigQueryUnitTestCaseInsensitivity

* changelog entry

* structs

* structs of structs, arrays of structs, clean up safe_cast

* typos

* restore dev-requirements

* changelog entry

.changes/unreleased/Features-20240205-174614.yaml

@@ -0,0 +1,7 @@
+kind: Features
+body: Support all types for unit testing in dbt-bigquery, expand coverage of
+  safe_cast macro
+time: 2024-02-05T17:46:14.505597-05:00
+custom:
+  Author: michelleark
+  Issue: "1090"

dbt/include/bigquery/macros/utils/safe_cast.sql

@@ -1,3 +1,27 @@
 {% macro bigquery__safe_cast(field, type) %}
+{%- if type.lower().startswith('array') and field is iterable and (field is not string and field is not mapping) and field | length > 0 -%}
+    {#-- Extract nested type from 'array<nested_type>' --#}
+    {% set nested_type = type.lower()[6:-1] %}
+    {#-- BigQuery does not support direct casts to arrays. instead, each element must be cast individually + reaggregated into an array --#}
+    {%- if cast_from_string_unsupported_for(nested_type)  %}
+        (select array_agg(safe_cast(i as {{ nested_type }})) from unnest([
+            {%- for nested_field in field %}
+                {{ nested_field.strip('"').strip("'") }}{{ ',' if not loop.last }}
+            {%- endfor %}
+        ]) i)
+    {%- else -%}
+        (select array_agg(safe_cast(i as {{nested_type}})) from unnest({{field}}) i)
+    {%- endif -%}
+
+{%- elif type.lower() == 'json' and field is mapping -%}
+    safe_cast(json {{ dbt.string_literal(tojson(field)) }} as json)
+{%- elif cast_from_string_unsupported_for(type) and field is string -%}
+    safe_cast({{field.strip('"').strip("'")}} as {{type}})
+{%- else -%}
     safe_cast({{field}} as {{type}})
+{%- endif -%}
+{% endmacro %}
+
+{% macro cast_from_string_unsupported_for(type) %}
+    {{ return(type.lower().startswith('struct') or type.lower() == 'geography') }}
 {% endmacro %}

tests/functional/adapter/unit_testing/test_unit_testing.py

@@ -0,0 +1,64 @@
+import pytest
+from dbt.tests.adapter.unit_testing.test_types import BaseUnitTestingTypes
+from dbt.tests.adapter.unit_testing.test_case_insensitivity import BaseUnitTestCaseInsensivity
+from dbt.tests.adapter.unit_testing.test_invalid_input import BaseUnitTestInvalidInput
+
+
+class TestBigQueryUnitTestingTypes(BaseUnitTestingTypes):
+    @pytest.fixture
+    def data_types(self):
+        # sql_value, yaml_value
+        return [
+            ["1", "1"],
+            ["'1'", "1"],
+            ["cast('true' as boolean)", "true"],
+            ["1.0", "1.0"],
+            ["'string value'", "string value"],
+            ["cast(1.0 as numeric)", "1.0"],
+            ["cast(1 as bigint)", 1],
+            ["cast('2019-01-01' as date)", "2019-01-01"],
+            ["cast('2013-11-03 00:00:00-07' as timestamp)", "2013-11-03 00:00:00-07"],
+            ["st_geogpoint(75, 45)", "'st_geogpoint(75, 45)'"],
+            # arrays
+            ["cast(['a','b','c'] as array<string>)", "['a','b','c']"],
+            ["cast([1,2,3] as array<int>)", "[1,2,3]"],
+            ["cast([true,true,false] as array<bool>)", "[true,true,false]"],
+            # array of date
+            ["[date '2019-01-01']", "['2020-01-01']"],
+            ["[date '2019-01-01']", "[]"],
+            ["[date '2019-01-01']", "null"],
+            # array of timestamp
+            ["[timestamp '2019-01-01']", "['2020-01-01']"],
+            ["[timestamp '2019-01-01']", "[]"],
+            ["[timestamp '2019-01-01']", "null"],
+            # json
+            [
+                """json '{"name": "Cooper", "forname": "Alice"}'""",
+                """{"name": "Cooper", "forname": "Alice"}""",
+            ],
+            ["""json '{"name": "Cooper", "forname": "Alice"}'""", "{}"],
+            # structs
+            ["struct('Isha' as name, 22 as age)", """'struct("Isha" as name, 22 as age)'"""],
+            [
+                "struct('Kipketer' AS name, [23.2, 26.1, 27.3, 29.4] AS laps)",
+                """'struct("Kipketer" AS name, [23.2, 26.1, 27.3, 29.4] AS laps)'""",
+            ],
+            # struct of struct
+            [
+                "struct(struct(1 as id, 'blue' as color) as my_struct)",
+                """'struct(struct(1 as id, "blue" as color) as my_struct)'""",
+            ],
+            # array of struct
+            [
+                "[struct(st_geogpoint(75, 45) as my_point), struct(st_geogpoint(75, 35) as my_point)]",
+                "['struct(st_geogpoint(75, 45) as my_point)', 'struct(st_geogpoint(75, 35) as my_point)']",
+            ],
+        ]
+
+
+class TestBigQueryUnitTestCaseInsensitivity(BaseUnitTestCaseInsensivity):
+    pass
+
+
+class TestBigQueryUnitTestInvalidInput(BaseUnitTestInvalidInput):
+    pass