Address GitHub feedback: rename global_tags to column_tags and add test for disabled flag

- Rename global_tags variable to column_tags in is_pii_column.sql for clarity
- Add test_column_pii_sampling_tags_exist_but_flag_disabled test case
- Verify that when PII tags exist but disable_samples_on_pii_columns=false,
  samples are collected normally including PII columns

Addresses GitHub comments from arbiv on PR #833

Co-Authored-By: Yosef Arbiv <[email protected]>

Author: devin-ai-integration[bot]

integration_tests/tests/test_column_pii_sampling.py

@@ -94,6 +94,46 @@ def test_column_pii_sampling_disabled(test_id: str, dbt_project: DbtProject):
         assert SAFE_COLUMN in sample
 
 
+@pytest.mark.skip_targets(["clickhouse"])
+def test_column_pii_sampling_tags_exist_but_flag_disabled(
+    test_id: str, dbt_project: DbtProject
+):
+    """Test that when PII tags exist but disable_samples_on_pii_columns is false, samples are collected normally"""
+    data = [
+        {SENSITIVE_COLUMN: f"user{i}@example.com", SAFE_COLUMN: None} for i in range(10)
+    ]
+
+    test_result = dbt_project.test(
+        test_id,
+        "not_null",
+        test_args=dict(column_name=SAFE_COLUMN),
+        data=data,
+        columns=[
+            {"name": SENSITIVE_COLUMN, "config": {"tags": ["pii"]}},
+            {"name": SAFE_COLUMN},
+        ],
+        test_vars={
+            "enable_elementary_test_materialization": True,
+            "test_sample_row_count": TEST_SAMPLE_ROW_COUNT,
+            "disable_samples_on_pii_columns": False,  # Flag is disabled
+            "pii_column_tags": ["pii"],
+        },
+    )
+    assert test_result["status"] == "fail"
+
+    samples = [
+        json.loads(row["result_row"])
+        for row in dbt_project.run_query(SAMPLES_QUERY.format(test_id=test_id))
+    ]
+
+    assert len(samples) == TEST_SAMPLE_ROW_COUNT
+    for sample in samples:
+        assert (
+            SENSITIVE_COLUMN in sample
+        )  # PII column should be included when flag is disabled
+        assert SAFE_COLUMN in sample
+
+
 @pytest.mark.skip_targets(["clickhouse"])
 def test_column_pii_sampling_all_columns_pii(test_id: str, dbt_project: DbtProject):
     """Test behavior when all columns are tagged as PII"""

macros/edr/system/system_utils/is_pii_column.sql

@@ -25,10 +25,10 @@
   {% for column_node in column_nodes.values() %}
     {% set config_dict = column_node.get('config', {}) %}
     {% set config_tags = config_dict.get('tags', []) %}
-    {% set global_tags = column_node.get('tags', []) %}
+    {% set column_tags = column_node.get('tags', []) %}
     {% set meta_dict = column_node.get('meta', {}) %}
     {% set meta_tags = meta_dict.get('tags', []) %}
-    {% set all_column_tags = config_tags + global_tags + meta_tags %}
+    {% set all_column_tags = config_tags + column_tags + meta_tags %}
 
     {% for pii_tag in pii_column_tags %}
       {% if pii_tag in all_column_tags %}