Sequence tasks with dependencies

Author: JCZuurmond

src/databricks/labs/ucx/assessment/sequencing.py

@@ -193,6 +193,17 @@ def register_job(self, job: jobs.Job) -> MaybeMigrationNode:
                 problems.extend(maybe_task_node.problems)
                 if maybe_task_node.node:
                     self._outgoing_references[job_node.key].add(maybe_task_node.node)
+            # Only after registering all tasks, we can resolve the task dependencies
+            for task in job.settings.tasks or []:
+                task_key = ("TASK", f"{job.job_id}/{task.task_key}")
+                for task_dependency in task.depends_on or []:
+                    task_dependency_key = ("TASK", f"{job.job_id}/{task_dependency.task_key}")
+                    maybe_task_dependency = self._nodes.get(task_dependency_key)
+                    if maybe_task_dependency:
+                        self._outgoing_references[task_key].add(maybe_task_dependency)
+                    else:
+                        problem = DependencyProblem('task-dependency-not-found', f"Could not find task: {task_dependency}")
+                        problems.append(problem)
         return MaybeMigrationNode(job_node, problems)
 
     def _register_workflow_task(self, task: jobs.Task, parent: MigrationNode) -> MaybeMigrationNode:

tests/unit/assessment/test_sequencing.py

@@ -316,7 +316,7 @@ def test_sequence_steps_from_job_task_referencing_other_task(ws, admin_locator)
     steps = list(sequencer.generate_steps())
     assert steps == [
         MigrationStep(
-            step_id=1,
+            step_id=2,
             step_number=0,
             object_type="TASK",
             object_id="1234/task1",
@@ -325,17 +325,17 @@ def test_sequence_steps_from_job_task_referencing_other_task(ws, admin_locator)
             required_step_ids=[],
         ),
         MigrationStep(
-            step_id=2,
-            step_number=0,
+            step_id=1,
+            step_number=1,
             object_type="TASK",
             object_id="1234/task2",
             object_name="task2",
             object_owner="John Doe",
-            required_step_ids=[1],
+            required_step_ids=[2],
         ),
         MigrationStep(
             step_id=0,
-            step_number=1,
+            step_number=2,
             object_type="JOB",
             object_id="1234",
             object_name="job",