[azure-ai-ml] Fix deployment_templates.get(name) resolving latest version (Bug 5402671)

sdk/ml/azure-ai-ml/CHANGELOG.md

@@ -8,6 +8,7 @@
 - Fixed `MLClient.jobs.create_or_update`, `archive`, and `restore` failing for previously-fetched jobs across all job types by routing metadata-only edits through the RunHistory PATCH endpoint.
 - Fixed `DeploymentTemplate.creation_context` always being `None` when retrieved via `get()` or `list()`. The created/modified timestamps and identity returned by the service (as `createdTime` / `modifiedTime` / `createdBy`) are now populated on `creation_context`, making `DeploymentTemplate` consistent with `Model` and `Environment`.
 - Fixed `deployment_templates.list(name=...)` raising `AttributeError: 'str' object has no attribute 'request_timeout'`. In the list response, `requestSettings` / `livenessProbe` / `readinessProbe` arrive as stringified dicts nested under `properties`; these are now parsed before conversion, giving `list()` parity with `models.list()` / `environments.list()`.
+- Fixed `deployment_templates.get(name)` failing with a 404 (`DeploymentTemplate {name}:latest not found`) when no version was supplied, because the literal string `"latest"` was sent as the version. The latest version is now resolved client-side (the service exposes no `latest` label and no server-side ordering), and `get()` accepts a `label` keyword (`label="latest"` resolves to the latest version) mirroring `models.get()`. `delete(name)` resolves the latest version the same way.
 
 ### Other Changes
 

sdk/ml/azure-ai-ml/api.md

@@ -10419,6 +10419,7 @@ namespace azure.ai.ml.operations
                 self, 
                 name: str, 
                 version: Optional[str] = None, 
+                label: Optional[str] = None, 
                 **kwargs: Any
             ) -> DeploymentTemplate: ...
 

sdk/ml/azure-ai-ml/api.metadata.yml

@@ -1,3 +1,3 @@
-apiMdSha256: 7262033cadec8b494dfa9cca72d97a9dad84e1e465ce8c455782d9fab275a81f
+apiMdSha256: 81371e37ca26ea3c761bdb2104b7a0b89a0026216a97510bc16f533cf936e2d9
 parserVersion: 0.3.28
 pythonVersion: 3.12.10

sdk/ml/azure-ai-ml/azure/ai/ml/operations/_deployment_template_operations.py

@@ -318,33 +318,81 @@ def list(
             ),
         )
 
+    def _resolve_latest_version(self, name: str) -> str:
+        """Resolve the latest version of a deployment template by name.
+
+        The deployment-template service does not expose a ``latest`` label and its list endpoint
+        has no server-side ordering, so the latest version is resolved client-side by enumerating
+        the available (active) versions and returning the highest one.
+
+        :param name: Name of the deployment template.
+        :type name: str
+        :return: The highest available version.
+        :rtype: str
+        :raises: ~azure.core.exceptions.ResourceNotFoundError if no versions exist for the name.
+        """
+        versions = [template.version for template in self.list(name=name) if template.version is not None]
+        if not versions:
+            raise ResourceNotFoundError(f"DeploymentTemplate {name} not found: no versions available.")
+
+        def _version_sort_key(version: str) -> tuple:
+            # Numeric versions sort above (and among) non-numeric ones so the highest number wins,
+            # while non-numeric versions still resolve deterministically via string comparison.
+            try:
+                return (1, int(version))
+            except (TypeError, ValueError):
+                return (0, str(version))
+
+        return max(versions, key=_version_sort_key)
+
     @distributed_trace
     @monitor_with_telemetry_mixin(ops_logger, "DeploymentTemplate.Get", ActivityType.PUBLICAPI)
     @experimental
-    def get(self, name: str, version: Optional[str] = None, **kwargs: Any) -> DeploymentTemplate:
-        """Get a deployment template by name and version.
+    def get(
+        self,
+        name: str,
+        version: Optional[str] = None,
+        label: Optional[str] = None,
+        **kwargs: Any,
+    ) -> DeploymentTemplate:
+        """Get a deployment template by name and version (or label).
 
         :param name: Name of the deployment template.
         :type name: str
-        :param version: Version of the deployment template. If not provided, gets the latest version.
+        :param version: Version of the deployment template. If neither ``version`` nor ``label`` is
+            provided, the latest version is returned.
         :type version: Optional[str]
+        :param label: Label of the deployment template. Only the ``latest`` label is supported, which
+            resolves to the latest version. Cannot be used together with ``version``.
+        :type label: Optional[str]
         :return: DeploymentTemplate object.
         :rtype: ~azure.ai.ml.entities.DeploymentTemplate
         :raises: ~azure.core.exceptions.ResourceNotFoundError if deployment template not found.
         """
-        version = version or "latest"
+        if version and label:
+            raise ValueError("Cannot specify both version and label.")
+
+        if label is not None and label != "latest":
+            raise ResourceNotFoundError(
+                f"DeploymentTemplate {name} with label '{label}' not found. Only the 'latest' label is supported."
+            )
+
+        # No explicit version (or label='latest') -> resolve the latest version client-side.
+        resolved_version = version or self._resolve_latest_version(name)
 
         try:
             result = self._service_client.deployment_templates.get(
                 registry_name=self._operation_scope.registry_name,
                 name=name,
-                version=version,
+                version=resolved_version,
                 **kwargs,
             )
             return DeploymentTemplate._from_rest_object(result)
+        except ResourceNotFoundError:
+            raise
         except Exception as e:
             module_logger.debug("DeploymentTemplate get operation failed: %s", e)
-            raise ResourceNotFoundError(f"DeploymentTemplate {name}:{version} not found") from e
+            raise ResourceNotFoundError(f"DeploymentTemplate {name}:{resolved_version} not found") from e
 
     @distributed_trace
     @monitor_with_telemetry_mixin(ops_logger, "DeploymentTemplate.CreateOrUpdate", ActivityType.PUBLICAPI)
@@ -389,7 +437,7 @@ def delete(self, name: str, version: Optional[str] = None, **kwargs: Any) -> Non
         :param version: Version of the deployment template to delete. If not provided, deletes the latest version.
         :type version: Optional[str]
         """
-        version = version or "latest"
+        version = version or self._resolve_latest_version(name)
 
         try:
             self._service_client.deployment_templates.delete(

sdk/ml/azure-ai-ml/tests/deployment_template/unittests/test_deployment_template_operations.py

@@ -243,7 +243,7 @@ def test_get_deployment_template_not_found(self, deployment_template_ops):
         )
 
         with pytest.raises(HttpResponseError):
-            deployment_template_ops.get("nonexistent-template")
+            deployment_template_ops.get("nonexistent-template", "1.0")
 
     def test_list_deployment_templates(self, deployment_template_ops, sample_rest_template):
         """Test list operation for deployment templates."""
@@ -331,7 +331,74 @@ def test_delete_deployment_template_not_found(self, deployment_template_ops):
         deployment_template_ops._operation_scope.workspace_name = "test-workspace"
 
         with pytest.raises(HttpResponseError):
-            deployment_template_ops.delete("nonexistent-template")
+            deployment_template_ops.delete("nonexistent-template", "1.0")
+
+    def test_get_no_version_resolves_latest(self, deployment_template_ops, sample_rest_template):
+        """get() without a version resolves the highest available version instead of sending 'latest'."""
+        deployment_template_ops._service_client.deployment_templates.get = Mock(return_value=sample_rest_template)
+        # list() returns versions out of order; the highest (5) must be selected.
+        deployment_template_ops.list = Mock(
+            return_value=[
+                DeploymentTemplate(name="test-template", version="2"),
+                DeploymentTemplate(name="test-template", version="5"),
+                DeploymentTemplate(name="test-template", version="3"),
+            ]
+        )
+
+        result = deployment_template_ops.get("test-template")
+
+        deployment_template_ops.list.assert_called_once_with(name="test-template")
+        call_args = deployment_template_ops._service_client.deployment_templates.get.call_args
+        assert call_args[1]["version"] == "5"
+        assert isinstance(result, DeploymentTemplate)
+
+    def test_get_label_latest_resolves_latest(self, deployment_template_ops, sample_rest_template):
+        """get(label='latest') resolves to the highest available version."""
+        deployment_template_ops._service_client.deployment_templates.get = Mock(return_value=sample_rest_template)
+        deployment_template_ops.list = Mock(
+            return_value=[
+                DeploymentTemplate(name="test-template", version="1"),
+                DeploymentTemplate(name="test-template", version="4"),
+            ]
+        )
+
+        result = deployment_template_ops.get("test-template", label="latest")
+
+        call_args = deployment_template_ops._service_client.deployment_templates.get.call_args
+        assert call_args[1]["version"] == "4"
+        assert isinstance(result, DeploymentTemplate)
+
+    def test_get_version_and_label_raises(self, deployment_template_ops):
+        """get() with both version and label is rejected."""
+        with pytest.raises(ValueError):
+            deployment_template_ops.get("test-template", version="1", label="latest")
+
+    def test_get_unsupported_label_raises(self, deployment_template_ops):
+        """Only the 'latest' label is supported; other labels raise ResourceNotFoundError."""
+        with pytest.raises(ResourceNotFoundError):
+            deployment_template_ops.get("test-template", label="production")
+
+    def test_get_no_versions_raises_not_found(self, deployment_template_ops):
+        """get() without a version raises when no versions exist to resolve."""
+        deployment_template_ops.list = Mock(return_value=[])
+
+        with pytest.raises(ResourceNotFoundError):
+            deployment_template_ops.get("test-template")
+
+    def test_delete_no_version_resolves_latest(self, deployment_template_ops):
+        """delete() without a version resolves the highest available version instead of sending 'latest'."""
+        deployment_template_ops._service_client.deployment_templates.delete = Mock(return_value=None)
+        deployment_template_ops.list = Mock(
+            return_value=[
+                DeploymentTemplate(name="test-template", version="2"),
+                DeploymentTemplate(name="test-template", version="5"),
+            ]
+        )
+
+        deployment_template_ops.delete("test-template")
+
+        call_args = deployment_template_ops._service_client.deployment_templates.delete.call_args
+        assert call_args[1]["version"] == "5"
 
     def test_create_or_update_with_none_input(self, deployment_template_ops):
         """Test create_or_update with None input."""
@@ -416,7 +483,7 @@ def test_get_with_invalid_name(self, deployment_template_ops):
         )
 
         with pytest.raises(HttpResponseError):
-            deployment_template_ops.get("")
+            deployment_template_ops.get("", "1.0")
 
     def test_delete_with_invalid_name(self, deployment_template_ops):
         """Test delete operation with invalid template name."""
@@ -434,7 +501,7 @@ def test_delete_with_invalid_name(self, deployment_template_ops):
         deployment_template_ops._operation_scope.workspace_name = "test-workspace"
 
         with pytest.raises(HttpResponseError):
-            deployment_template_ops.delete("")
+            deployment_template_ops.delete("", "1.0")
 
     @patch("azure.ai.ml.entities._load_functions.load_deployment_template")
     def test_create_or_update_yaml_file_not_found(self, mock_load, deployment_template_ops):
@@ -813,27 +880,39 @@ def test_create_or_update_with_invalid_type(self, deployment_template_ops):
             deployment_template_ops.create_or_update({"name": "test"})
 
     def test_delete_default_version(self, deployment_template_ops):
-        """Test delete operation with default version."""
+        """Test delete resolves the latest version when no version is provided."""
         deployment_template_ops._service_client.deployment_templates.delete = Mock()
+        deployment_template_ops.list = Mock(
+            return_value=[
+                DeploymentTemplate(name="test-template", version="1"),
+                DeploymentTemplate(name="test-template", version="3"),
+            ]
+        )
         deployment_template_ops._operation_scope.subscription_id = "test-sub"
         deployment_template_ops._operation_scope.resource_group_name = "test-rg"
         deployment_template_ops._operation_scope.registry_name = "test-registry"
 
         deployment_template_ops.delete("test-template")
 
-        # Verify version defaults to "latest"
+        # Verify version defaults to the highest available version (not the literal string "latest")
         call_args = deployment_template_ops._service_client.deployment_templates.delete.call_args
-        assert call_args[1]["version"] == "latest"
+        assert call_args[1]["version"] == "3"
 
     def test_get_default_version(self, deployment_template_ops, sample_rest_template):
-        """Test get operation with default version."""
+        """Test get resolves the latest version when no version is provided."""
         deployment_template_ops._service_client.deployment_templates.get = Mock(return_value=sample_rest_template)
+        deployment_template_ops.list = Mock(
+            return_value=[
+                DeploymentTemplate(name="test-template", version="1"),
+                DeploymentTemplate(name="test-template", version="3"),
+            ]
+        )
         deployment_template_ops._operation_scope.subscription_id = "test-sub"
         deployment_template_ops._operation_scope.resource_group_name = "test-rg"
         deployment_template_ops._operation_scope.registry_name = "test-registry"
 
         deployment_template_ops.get("test-template")
 
-        # Verify version defaults to "latest"
+        # Verify version defaults to the highest available version (not the literal string "latest")
         call_args = deployment_template_ops._service_client.deployment_templates.get.call_args
-        assert call_args[1]["version"] == "latest"
+        assert call_args[1]["version"] == "3"