feat: support key-only databricks_tags

CHANGELOG.md

@@ -6,6 +6,8 @@
   gated behind the `use_concurrent_microbatch` behavior flag (default: `false`).
   Opt in via `flags: {use_concurrent_microbatch: true}` in `dbt_project.yml`
   ([#914](https://github.com/databricks/dbt-databricks/issues/914))
+- Add support for key-only `databricks_tags` for table and column tagging. This can now be configured by settings
+  tag values as empty strings `""`. ([#1270](https://github.com/databricks/dbt-databricks/issues/1270))
 
 ### Fixes
 

dbt/adapters/databricks/relation_configs/column_tags.py

@@ -55,7 +55,7 @@ def from_relation_results(cls, results: RelationResults) -> ColumnTagsConfig:
                 # row contains [column_name, tag_name, tag_value]
                 column_name = str(row[0])
                 tag_name = str(row[1])
-                tag_value = str(row[2])
+                tag_value = str(row[2] or "")
 
                 if column_name not in set_column_tags:
                     set_column_tags[column_name] = {}
@@ -79,7 +79,7 @@ def from_relation_config(cls, relation_config: RelationConfig) -> ColumnTagsConf
             if databricks_tags:
                 if isinstance(databricks_tags, dict):
                     set_column_tags[col["name"]] = {
-                        str(k): str(v) for k, v in databricks_tags.items()
+                        str(k): str(v or "") for k, v in databricks_tags.items()
                     }
                 else:
                     raise DbtRuntimeError("databricks_tags must be a dictionary")

dbt/adapters/databricks/relation_configs/tags.py

@@ -33,7 +33,7 @@ def from_relation_results(cls, results: RelationResults) -> TagsConfig:
 
         if table:
             for row in table.rows:
-                tags[str(row[0])] = str(row[1])
+                tags[str(row[0])] = str(row[1] or "")
 
         return TagsConfig(set_tags=tags)
 
@@ -43,7 +43,7 @@ def from_relation_config(cls, relation_config: RelationConfig) -> TagsConfig:
         if not tags:
             return TagsConfig(set_tags=dict())
         if isinstance(tags, dict):
-            tags = {str(k): str(v) for k, v in tags.items()}
+            tags = {str(k): str(v or "") for k, v in tags.items()}
             return TagsConfig(set_tags=tags)
         else:
             raise DbtRuntimeError("databricks_tags must be a dictionary")

dbt/include/databricks/macros/relations/tags.sql

@@ -30,8 +30,8 @@
 
 {% macro alter_set_tags(relation, tags) -%}
   ALTER {{ relation.type.render() }} {{ relation.render() }} SET TAGS (
-    {% for tag in tags -%}
-      '{{ tag }}' = '{{ tags[tag] }}' {%- if not loop.last %}, {% endif -%}
+    {% for key, value in tags.items() -%}
+      '{{ key }}' = '{{ value }}' {%- if not loop.last %}, {% endif -%}
     {%- endfor %}
   )
 {%- endmacro -%}

tests/functional/adapter/column_tags/fixtures.py

@@ -13,7 +13,7 @@
       - name: account_number
         databricks_tags:
           pii: "true"
-          sensitive: "true"
+          sensitive: ""
 """
 
 updated_column_tag_model = """
@@ -29,7 +29,7 @@
       - name: account_number
         databricks_tags:
           pii: "true"
-          sensitive: "true"
+          sensitive: ""
 """
 
 column_tags_seed = """

tests/functional/adapter/column_tags/test_column_tags.py

@@ -31,7 +31,7 @@ def test_column_tags(self, project):
         tags = project.run_sql(column_tags_query, fetch="all")
         expected_tags = {
             ("account_number", "pii", "true"),
-            ("account_number", "sensitive", "true"),
+            ("account_number", "sensitive", ""),
         }
         actual_tags = {(row[0], row[1], row[2]) for row in tags}
         assert actual_tags == expected_tags
@@ -51,7 +51,7 @@ def test_column_tags(self, project):
         expected_tags = {
             ("id", "pii", "false"),
             ("account_number", "pii", "true"),
-            ("account_number", "sensitive", "true"),
+            ("account_number", "sensitive", ""),
         }
         actual_tags = {(row[0], row[1], row[2]) for row in tags}
         assert actual_tags == expected_tags

tests/functional/adapter/tags/fixtures.py

@@ -1,7 +1,7 @@
 tags_sql = """
 {{ config(
     materialized = 'table',
-    databricks_tags = {'a': 'b', 'c': 'd'},
+    databricks_tags = {'a': 'b', 'c': 'd', 'k': ''},
 ) }}
 
 select cast(1 as bigint) as id, 'hello' as msg, 'blue' as color
@@ -19,7 +19,7 @@
 streaming_table_tags_sql = """
 {{ config(
     materialized='streaming_table',
-    databricks_tags = {'a': 'b', 'c': 'd'},
+    databricks_tags = {'a': 'b', 'c': 'd', 'k': ''},
 ) }}
 
 select * from stream {{ ref('my_seed') }}
@@ -54,4 +54,5 @@ def model(dbt, spark):
       databricks_tags:
         a: b
         c: d
+        k: ""
 """

tests/functional/adapter/tags/test_databricks_tags.py

@@ -23,8 +23,8 @@ def test_tags(self, project):
             " where schema_name = '{schema}' and table_name='tags'",
             fetch="all",
         )
-        assert len(results) == 2
-        expected_tags = {("a", "b"), ("c", "d")}
+        assert len(results) == 3
+        expected_tags = {("a", "b"), ("c", "d"), ("k", "")}
         actual_tags = set((row[0], row[1]) for row in results)
         assert actual_tags == expected_tags
 
@@ -56,8 +56,8 @@ def test_updated_tags(self, project):
             " where schema_name = '{schema}' and table_name='tags'",
             fetch="all",
         )
-        assert len(results) == 3
-        expected_tags = {("a", "b"), ("c", "d"), ("e", "f")}
+        assert len(results) == 4
+        expected_tags = {("a", "b"), ("c", "d"), ("k", ""), ("e", "f")}
         actual_tags = set((row[0], row[1]) for row in results)
         assert actual_tags == expected_tags
 
@@ -151,7 +151,7 @@ def test_updated_tags(self, project):
             " where schema_name = '{schema}' and table_name='tags'",
             fetch="all",
         )
-        assert len(results) == 3
+        assert len(results) == 4
 
 
 @pytest.mark.python

tests/unit/relation_configs/test_column_tags_config.py

@@ -26,7 +26,7 @@ def test_from_relation_results__some(self):
             "information_schema.column_tags": Table(
                 rows=[
                     ["col1", "tag_a", "value_a"],
-                    ["col1", "tag_b", "value_b"],
+                    ["col1", "tag_b", ""],  # key-only tag
                     ["col2", "tag_c", "value_c"],
                 ],
                 column_names=["column_name", "tag_name", "tag_value"],
@@ -35,7 +35,7 @@ def test_from_relation_results__some(self):
         spec = ColumnTagsProcessor.from_relation_results(results)
         assert spec == ColumnTagsConfig(
             set_column_tags={
-                "col1": {"tag_a": "value_a", "tag_b": "value_b"},
+                "col1": {"tag_a": "value_a", "tag_b": ""},
                 "col2": {"tag_c": "value_c"},
             }
         )
@@ -54,14 +54,14 @@ def test_from_relation_config__without_column_tags(self):
     def test_from_relation_config__with_dict(self):
         model = Mock()
         model.columns = {
-            "email": {"_extra": {"databricks_tags": {"pii": "true", "env": "prod"}}},
+            "email": {"_extra": {"databricks_tags": {"pii": "", "env": "prod"}}},
             "id": {"_extra": {}},
             "created_at": {},
         }
         spec = ColumnTagsProcessor.from_relation_config(model)
         assert spec == ColumnTagsConfig(
             set_column_tags={
-                "email": {"pii": "true", "env": "prod"},
+                "email": {"pii": "", "env": "prod"},
             }
         )
 
@@ -71,14 +71,14 @@ def test_from_relation_config__with_column_info(self):
             "id": ColumnInfo(name="id", _extra={}),
             "email": ColumnInfo(
                 name="email",
-                _extra={"databricks_tags": {"pii": "true", "env": "prod"}},
+                _extra={"databricks_tags": {"pii": "", "env": "prod"}},
             ),
             "created_at": ColumnInfo(name="created_at"),
         }
         spec = ColumnTagsProcessor.from_relation_config(model)
         assert spec == ColumnTagsConfig(
             set_column_tags={
-                "email": {"pii": "true", "env": "prod"},
+                "email": {"pii": "", "env": "prod"},
             }
         )
 

tests/unit/relation_configs/test_tags.py

@@ -24,6 +24,15 @@ def test_from_relation_results__some(self):
         spec = TagsProcessor.from_relation_results(results)
         assert spec == TagsConfig(set_tags={"a": "valA", "b": "valB"})
 
+    def test_from_relation_results__key_only(self):
+        results = {
+            "information_schema.tags": Table(
+                rows=[["a", ""]], column_names=["tag_name", "tag_value"]
+            )
+        }
+        spec = TagsProcessor.from_relation_results(results)
+        assert spec == TagsConfig(set_tags={"a": ""})
+
     def test_from_relation_config__without_tags(self):
         model = Mock()
         model.config.extra = {}
@@ -36,6 +45,12 @@ def test_from_relation_config__with_tags(self):
         spec = TagsProcessor.from_relation_config(model)
         assert spec == TagsConfig(set_tags={"a": "valA", "b": "1"})
 
+    def test_from_relation_config__with_key_only_tags(self):
+        model = Mock()
+        model.config.extra = {"databricks_tags": {"a": "", "b": None}}
+        spec = TagsProcessor.from_relation_config(model)
+        assert spec == TagsConfig(set_tags={"a": "", "b": ""})
+
     def test_from_relation_config__with_incorrect_tags(self):
         model = Mock()
         model.config.extra = {"databricks_tags": ["a", "b"]}
@@ -52,25 +67,25 @@ def test_get_diff__empty_and_some_exist(self):
         # Tags are "set only" - when config has no tags and relation has tags,
         # we don't unset the existing tags
         config = TagsConfig(set_tags={})
-        other = TagsConfig(set_tags={"tag": "value"})
-        diff = config.get_diff(other)
+        config_old = TagsConfig(set_tags={"tag": "value"})
+        diff = config.get_diff(config_old)
         assert diff is None  # No changes needed since we don't unset tags
 
     def test_get_diff__some_new_and_empty_existing(self):
         config = TagsConfig(set_tags={"tag": "value"})
-        other = TagsConfig(set_tags={})
-        diff = config.get_diff(other)
+        config_old = TagsConfig(set_tags={})
+        diff = config.get_diff(config_old)
         assert diff == TagsConfig(set_tags={"tag": "value"})
 
     def test_get_diff__mixed_case(self):
         # Tags are "set only" - only the new/updated tags are included
         config = TagsConfig(set_tags={"a": "value", "b": "value"})
-        other = TagsConfig(set_tags={"b": "other_value", "c": "value"})
-        diff = config.get_diff(other)
+        config_old = TagsConfig(set_tags={"b": "other_value", "c": "value"})
+        diff = config.get_diff(config_old)
         assert diff == TagsConfig(set_tags={"a": "value", "b": "value"})
 
     def test_get_diff__no_changes(self):
         config = TagsConfig(set_tags={"tag": "value"})
-        other = TagsConfig(set_tags={"tag": "value"})
-        diff = config.get_diff(other)
+        config_old = TagsConfig(set_tags={"tag": "value"})
+        diff = config.get_diff(config_old)
         assert diff is None