datajunction 0.0.199__tar.gz → 0.0.201__tar.gz

{datajunction-0.0.199 → datajunction-0.0.201}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datajunction
-Version: 0.0.199
+Version: 0.0.201
 Summary: DataJunction client library for connecting to a DataJunction server
 Project-URL: repository, https://github.com/DataJunction/dj
 Author-email: DataJunction Authors <yian.shang@gmail.com>

{datajunction-0.0.199 → datajunction-0.0.201}/datajunction/__about__.py

@@ -2,4 +2,4 @@
 Version for Hatch
 """
 
-__version__ = "0.0.199"
+__version__ = "0.0.201"

{datajunction-0.0.199 → datajunction-0.0.201}/datajunction/deployment.py

@@ -46,6 +46,24 @@ class ResultStatus(str, Enum):
     NOOP = "noop"
 
 
+class SpecKind(str, Enum):
+    """
+    Top-level ``kind`` discriminator on a project YAML file.
+
+    A file with no ``kind`` defaults to ``NODE`` (backwards compatible: existing
+    files carry no discriminator). Pre-aggregations live in the same tree but
+    are not nodes, so they declare ``kind: preagg`` and route to the
+    deployment's ``preaggregations`` list.
+    """
+
+    NODE = "node"
+    PREAGG = "preagg"
+
+    @classmethod
+    def default(cls) -> "SpecKind":
+        return cls.NODE
+
+
 class DeploymentService:
     """
     High-level deployment client for exporting and importing DJ namespaces.
@@ -489,7 +507,29 @@ class DeploymentService:
             )
 
         project_metadata = self._read_project_yaml(base_dir)
-        nodes, warnings = self._collect_nodes_from_dir(base_dir)
+        collected, warnings = self._collect_nodes_from_dir(base_dir)
+
+        # Every YAML file declares its ``kind`` (see SpecKind); an absent
+        # discriminator defaults to a node. Pre-aggregation specs live in the
+        # same tree but are not nodes, so they are routed to the deployment's
+        # ``preaggregations`` list. The key is stripped so it never reaches the
+        # node/pre-agg payload.
+        nodes: list[dict[str, Any]] = []
+        preaggregations: list[dict[str, Any]] = []
+        for item in collected:
+            raw_kind = item.pop("kind", SpecKind.default().value)
+            try:
+                kind = SpecKind(raw_kind)
+            except ValueError:
+                raise DJClientException(
+                    f"Unknown kind '{raw_kind}' in "
+                    f"'{item.get('name', '<unnamed>')}'; expected one of "
+                    f"{[k.value for k in SpecKind]}.",
+                )
+            if kind is SpecKind.PREAGG:
+                preaggregations.append(item)
+            else:
+                nodes.append(item)
 
         # Deduplicate nodes by name (keep last occurrence)
         seen_names: dict[str, dict] = {}
@@ -508,6 +548,7 @@ class DeploymentService:
             or project_metadata.get("prefix", ""),
             "nodes": nodes,
             "tags": project_metadata.get("tags", []),
+            "preaggregations": preaggregations,
         }
 
         # Add deployment source if available from env vars

{datajunction-0.0.199 → datajunction-0.0.201}/tests/test_deploy.py

@@ -45,6 +45,46 @@ def test_pull_extracts_zip_from_server(tmp_path):
     assert (tmp_path / "foo" / "bar" / "baz.yaml").exists()
 
 
+def test_reconstruct_deployment_spec_separates_preaggregations(tmp_path):
+    """Files with ``kind: preagg`` route to preaggregations; the rest are nodes."""
+    (tmp_path / "dj.yaml").write_text("namespace: ns\n")
+    (tmp_path / "revenue.yaml").write_text(
+        "name: ns.revenue\nnode_type: metric\nquery: SELECT SUM(amount) FROM ns.fct\n",
+    )
+    (tmp_path / "revenue_by_day.yaml").write_text(
+        "kind: preagg\n"
+        "name: revenue_by_day\n"
+        "metrics:\n  - ns.revenue\n"
+        "dimensions:\n  - ns.date.day\n"
+        "catalog: default\n"
+        "schema: agg\n"
+        "table: revenue_agg\n"
+        "measure_columns:\n  ns.revenue: revenue_sum\n",
+    )
+
+    svc = DeploymentService(MagicMock())
+    spec, _ = svc._reconstruct_deployment_spec(tmp_path)
+
+    # The node file (no kind → node) stays a node.
+    assert [node["name"] for node in spec["nodes"]] == ["ns.revenue"]
+    assert len(spec["preaggregations"]) == 1
+    preagg = spec["preaggregations"][0]
+    assert preagg["name"] == "revenue_by_day"
+    assert preagg["measure_columns"] == {"ns.revenue": "revenue_sum"}
+    # The discriminator is stripped before reaching the payload.
+    assert "kind" not in preagg
+
+
+def test_reconstruct_deployment_spec_rejects_unknown_kind(tmp_path):
+    """An unrecognized ``kind`` fails loudly instead of silently becoming a node."""
+    (tmp_path / "dj.yaml").write_text("namespace: ns\n")
+    (tmp_path / "mystery.yaml").write_text("kind: widget\nname: ns.mystery\n")
+
+    svc = DeploymentService(MagicMock())
+    with pytest.raises(DJClientException, match="Unknown kind 'widget'"):
+        svc._reconstruct_deployment_spec(tmp_path)
+
+
 def test_pull_uploads_existing_yaml_files(tmp_path):
     (tmp_path / "old.yaml").write_text("name: ns.old\n")
     updated_zip = _make_zip({"old.yaml": "name: ns.old\nquery: SELECT 2\n"})