Custom SQL for get source maxLoadedAt

.changes/unreleased/Features-20241217-171631.yaml

@@ -0,0 +1,6 @@
+kind: Features
+body: Calculate source freshness via a SQL query
+time: 2024-12-17T17:16:31.841076-08:00
+custom:
+  Author: ChenyuLInx
+  Issue: "8797"

core/dbt/artifacts/resources/v1/source_definition.py

@@ -59,6 +59,7 @@ class ParsedSourceMandatory(GraphResource, HasRelationMetadata):
 class SourceDefinition(ParsedSourceMandatory):
     quoting: Quoting = field(default_factory=Quoting)
     loaded_at_field: Optional[str] = None
+    loaded_at_query: Optional[str] = None
     freshness: Optional[FreshnessThreshold] = None
     external: Optional[ExternalTable] = None
     description: str = ""

core/dbt/context/providers.py

@@ -880,7 +880,7 @@
 
 # Base context collection, used for parsing configs.
 class ProviderContext(ManifestContext):
-    # subclasses are MacroContext, ModelContext, TestContext
+    # subclasses are MacroContext, ModelContext, TestContext, SourceContext
     def __init__(
         self,
         model,
@@ -893,7 +893,7 @@
             raise DbtInternalError(f"Invalid provider given to context: {provider}")
         # mypy appeasement - we know it'll be a RuntimeConfig
         self.config: RuntimeConfig
-        self.model: Union[Macro, ManifestNode] = model
+        self.model: Union[Macro, ManifestNode, SourceDefinition] = model
         super().__init__(config, manifest, model.package_name)
         self.sql_results: Dict[str, Optional[AttrDict]] = {}
         self.context_config: Optional[ContextConfig] = context_config
@@ -1558,6 +1558,20 @@
             self._search_package = search_package
 
 
+class SourceContext(ProviderContext):
+    # SourceContext is being used to render jinja SQL during execution of
+    # custom SQL in source freshness. It is not used for parsing.
+    model: SourceDefinition
+
+    @contextproperty()
+    def this(self) -> Optional[RelationProxy]:
+        return self.db_wrapper.Relation.create_from(self.config, self.model)
+
+    @contextproperty()
+    def source_node(self) -> SourceDefinition:
+        return self.model
+
+
 class ModelContext(ProviderContext):
     model: ManifestNode
 

core/dbt/contracts/graph/unparsed.py

@@ -320,6 +320,7 @@ class UnparsedSourceTableDefinition(HasColumnTests, HasColumnAndTestProps):
     config: Dict[str, Any] = field(default_factory=dict)
     loaded_at_field: Optional[str] = None
     loaded_at_field_present: Optional[bool] = None
+    loaded_at_query: Optional[str] = None
     identifier: Optional[str] = None
     quoting: Quoting = field(default_factory=Quoting)
     freshness: Optional[FreshnessThreshold] = field(default_factory=FreshnessThreshold)
@@ -345,6 +346,7 @@ class UnparsedSourceDefinition(dbtClassMixin):
     freshness: Optional[FreshnessThreshold] = field(default_factory=FreshnessThreshold)
     loaded_at_field: Optional[str] = None
     loaded_at_field_present: Optional[bool] = None
+    loaded_at_query: Optional[str] = None
     tables: List[UnparsedSourceTableDefinition] = field(default_factory=list)
     tags: List[str] = field(default_factory=list)
     config: Dict[str, Any] = field(default_factory=dict)
@@ -382,6 +384,7 @@ class SourceTablePatch(dbtClassMixin):
     docs: Optional[Docs] = None
     loaded_at_field: Optional[str] = None
     loaded_at_field_present: Optional[bool] = None
+    loaded_at_query: Optional[str] = None
     identifier: Optional[str] = None
     quoting: Quoting = field(default_factory=Quoting)
     freshness: Optional[FreshnessThreshold] = field(default_factory=FreshnessThreshold)
@@ -425,6 +428,7 @@ class SourcePatch(dbtClassMixin):
     freshness: Optional[Optional[FreshnessThreshold]] = field(default_factory=FreshnessThreshold)
     loaded_at_field: Optional[str] = None
     loaded_at_field_present: Optional[bool] = None
+    loaded_at_query: Optional[str] = None
     tables: Optional[List[SourceTablePatch]] = None
     tags: Optional[List[str]] = None
 

core/dbt/parser/schema_renderer.py

@@ -37,11 +37,21 @@ def _is_norender_key(self, keypath: Keypath) -> bool:
         "tests" and "data_tests" are both currently supported but "tests" has been deprecated
         """
         # top level descriptions and data_tests
-        if len(keypath) >= 1 and keypath[0] in ("tests", "data_tests", "description"):
+        if len(keypath) >= 1 and keypath[0] in (
+            "tests",
+            "data_tests",
+            "description",
+            "loaded_at_query",
+        ):
             return True
 
         # columns descriptions and data_tests
-        if len(keypath) == 2 and keypath[1] in ("tests", "data_tests", "description"):
+        if len(keypath) == 2 and keypath[1] in (
+            "tests",
+            "data_tests",
+            "description",
+            "loaded_at_query",
+        ):
             return True
 
         # pre- and post-hooks
@@ -69,9 +79,8 @@ def _is_norender_key(self, keypath: Keypath) -> bool:
     def should_render_keypath(self, keypath: Keypath) -> bool:
         if len(keypath) < 1:
             return True
-
         if self.key == "sources":
-            if keypath[0] == "description":
+            if keypath[0] in ("description", "loaded_at_query"):
                 return False
             if keypath[0] == "tables":
                 if self._is_norender_key(keypath[2:]):

core/dbt/parser/sources.py

@@ -26,6 +26,7 @@
     UnparsedSourceTableDefinition,
 )
 from dbt.events.types import FreshnessConfigProblem, UnusedTables
+from dbt.exceptions import ParsingError
 from dbt.node_types import NodeType
 from dbt.parser.common import ParserRef
 from dbt.parser.schema_generic_tests import SchemaGenericTestParser
@@ -131,11 +132,28 @@ def parse_source(self, target: UnpatchedSourceDefinition) -> SourceDefinition:
         # We need to be able to tell the difference between explicitly setting the loaded_at_field to None/null
         # and when it's simply not set.  This allows a user to override the source level loaded_at_field so that
         # specific table can default to metadata-based freshness.
+        if table.loaded_at_field_present and table.loaded_at_query:
+            raise ParsingError(
+                "Cannot specify both loaded_at_field and loaded_at_query at table level."
+            )
+        if source.loaded_at_field and source.loaded_at_query:
+            raise ParsingError(
+                "Cannot specify both loaded_at_field and loaded_at_query at source level."
+            )
+
         if table.loaded_at_field_present or table.loaded_at_field is not None:
             loaded_at_field = table.loaded_at_field
         else:
             loaded_at_field = source.loaded_at_field  # may be None, that's okay
 
+        loaded_at_query: Optional[str]
+        if table.loaded_at_query is not None:
+            loaded_at_query = table.loaded_at_query
+        else:
+            if table.loaded_at_field_present:
+                loaded_at_query = None
+            else:
+                loaded_at_query = source.loaded_at_query
         freshness = merge_freshness(source.freshness, table.freshness)
         quoting = source.quoting.merged(table.quoting)
         # path = block.path.original_file_path
@@ -185,6 +203,7 @@ def parse_source(self, target: UnpatchedSourceDefinition) -> SourceDefinition:
             meta=meta,
             loader=source.loader,
             loaded_at_field=loaded_at_field,
+            loaded_at_query=loaded_at_query,
             freshness=freshness,
             quoting=quoting,
             resource_type=NodeType.Source,

core/dbt/task/freshness.py

@@ -15,6 +15,8 @@
     PartialSourceFreshnessResult,
     SourceFreshnessResult,
 )
+from dbt.clients import jinja
+from dbt.context.providers import RuntimeProvider, SourceContext
 from dbt.contracts.graph.manifest import Manifest
 from dbt.contracts.graph.nodes import HookNode, SourceDefinition
 from dbt.contracts.results import RunStatus
@@ -114,7 +116,22 @@
             adapter_response: Optional[AdapterResponse] = None
             freshness: Optional[FreshnessResponse] = None
 
-            if compiled_node.loaded_at_field is not None:
+            if compiled_node.loaded_at_query is not None:
+                # within the context user can have access to `this`, `source_node`(`model` will point to the same thing),  etc
+                compiled_code = jinja.get_rendered(
+                    compiled_node.loaded_at_query,
+                    SourceContext(
+                        compiled_node, self.config, manifest, RuntimeProvider(), None
+                    ).to_dict(),
+                    compiled_node,
+                )
+                adapter_response, freshness = self.adapter.calculate_freshness_from_custom_sql(
+                    relation,
+                    compiled_code,
+                    macro_resolver=manifest,
+                )
+                status = compiled_node.freshness.status(freshness["age"])
+            elif compiled_node.loaded_at_field is not None:
                 adapter_response, freshness = self.adapter.calculate_freshness(
                     relation,
                     compiled_node.loaded_at_field,
@@ -146,7 +163,6 @@
                 raise DbtRuntimeError(
                     f"Could not compute freshness for source {compiled_node.name}: no 'loaded_at_field' provided and {self.adapter.type()} adapter does not support metadata-based freshness checks."
                 )
-
         # adapter_response was not returned in previous versions, so this will be None
         # we cannot call to_dict() on NoneType
         if adapter_response:

core/setup.py

@@ -72,7 +72,7 @@
         "dbt-semantic-interfaces>=0.8.3,<0.9",
         # Minor versions for these are expected to be backwards-compatible
         "dbt-common>=1.13.0,<2.0",
-        "dbt-adapters>=1.10.1,<2.0",
+        "dbt-adapters>=1.13.0,<2.0",
         # ----
         # Expect compatibility with all new versions of these packages, so lower bounds only.
         "packaging>20.9",

schemas/dbt/manifest/v12.json

@@ -7752,6 +7752,17 @@
             ],
             "default": null
           },
+          "loaded_at_query": {
+            "anyOf": [
+              {
+                "type": "string"
+              },
+              {
+                "type": "null"
+              }
+            ],
+            "default": null
+          },
           "freshness": {
             "anyOf": [
               {
@@ -17540,6 +17551,17 @@
                       ],
                       "default": null
                     },
+                    "loaded_at_query": {
+                      "anyOf": [
+                        {
+                          "type": "string"
+                        },
+                        {
+                          "type": "null"
+                        }
+                      ],
+                      "default": null
+                    },
                     "freshness": {
                       "anyOf": [
                         {

tests/functional/artifacts/expected_manifest.py

@@ -793,6 +793,7 @@ def expected_seeded_manifest(project, model_database=None, quote_model=False):
                 },
                 "identifier": "seed",
                 "loaded_at_field": None,
+                "loaded_at_query": None,
                 "loader": "a_loader",
                 "meta": {},
                 "name": "my_table",
@@ -1299,6 +1300,7 @@ def expected_references_manifest(project):
                 },
                 "identifier": "seed",
                 "loaded_at_field": None,
+                "loaded_at_query": None,
                 "loader": "a_loader",
                 "meta": {},
                 "name": "my_table",

tests/functional/sources/fixtures.py

@@ -472,3 +472,26 @@
       - name: test_table
         identifier: source
 """
+
+freshness_via_custom_sql_schema_yml = """version: 2
+sources:
+  - name: test_source
+    freshness:
+      warn_after: {count: 10, period: hour}
+    schema: "{{ var(env_var('DBT_TEST_SCHEMA_NAME_VARIABLE')) }}"
+    quoting:
+      identifier: True
+    tags:
+      - my_test_source_tag
+    tables:
+      - name: source_a
+        identifier: source
+        loaded_at_field: "{{ var('test_loaded_at') | as_text }}"
+      - name: source_b
+        identifier: source
+        loaded_at_query: "select max({{ var('test_loaded_at') | as_text }}) from {{this}}"
+      - name: source_c
+        identifier: source
+        loaded_at_query: "select {{current_timestamp()}}"
+
+"""

tests/functional/sources/test_source_freshness.py

@@ -17,6 +17,7 @@
     error_models_model_sql,
     error_models_schema_yml,
     filtered_models_schema_yml,
+    freshness_via_custom_sql_schema_yml,
     freshness_via_metadata_schema_yml,
     override_freshness_models_schema_yml,
 )
@@ -578,3 +579,18 @@ def test_hooks_do_not_run_for_source_freshness(
         )
         # default behaviour - no hooks run in source freshness
         self._assert_project_hooks_not_called(log_output)
+
+
+class TestSourceFreshnessCustomSQL(SuccessfulSourceFreshnessTest):
+    @pytest.fixture(scope="class")
+    def models(self):
+        return {"schema.yml": freshness_via_custom_sql_schema_yml}
+
+    def test_source_freshness_custom_sql(self, project):
+        result = self.run_dbt_with_vars(project, ["source", "freshness"], expect_pass=True)
+        # They are the same source but different queries were executed for each
+        assert {r.node.name: r.status for r in result} == {
+            "source_a": "warn",
+            "source_b": "warn",
+            "source_c": "pass",
+        }