Use pipeline default tasks for task include/exclude check

hrqb/base/task.py

@@ -334,8 +334,12 @@ def requires(self) -> Iterator[luigi.Task]:
     def verify_include_and_exclude_tasks_exist(self) -> None:
         """Verify that included and excluded tasks exist in the pipeline."""
         filter_tasks = (self.include_tasks or ()) + (self.exclude_tasks or ())
+        all_pipeline_tasks = {
+            task.name: task
+            for _, task in self.pipeline_tasks_iter(use_default_requires=True)
+        }
         for task_name in filter_tasks:
-            if self.get_task(task_name) is None:
+            if task_name not in all_pipeline_tasks:
                 raise TaskNotInPipelineScopeError(task_name)
 
     @staticmethod

hrqb/cli.py

@@ -138,7 +138,7 @@ def pipeline(
             pipeline_parameters=pipeline_parameters,
         )
     except TaskNotInPipelineScopeError as exc:
-        message = f"--include-tasks or --exclude-task are invalid: {exc}, exiting"
+        message = f"CLI options --include or --exclude are invalid: {exc} Exiting."
         raise click.ClickException(message) from exc
 
     ctx.obj["PIPELINE_TASK"] = pipeline_task
@@ -161,6 +161,8 @@ def remove_data(ctx: click.Context) -> None:
     pipeline_task = ctx.obj["PIPELINE_TASK"]
     pipeline_task.remove_pipeline_targets()
     logger.info("Successfully removed target data(s).")
+    logger.info("Updated status after data cleanup:")
+    logger.info(pipeline_task.pipeline_as_ascii())
 
 
 @pipeline.command()
@@ -183,5 +185,3 @@ def run(
 
     if cleanup:
         ctx.invoke(remove_data)
-        logger.info("Updated status after data cleanup:")
-        logger.info(pipeline_task.pipeline_as_ascii())

tests/test_cli.py

@@ -105,8 +105,8 @@ def test_cli_pipeline_remove_data_task_not_found(
     result = runner.invoke(cli.main, args)
     assert result.exit_code == ERROR_RESULT_CODE
     assert (
-        "--include-tasks or --exclude-task are invalid: Task 'BadTask' not found in "
-        "pipeline" in result.output
+        "CLI options --include or --exclude are invalid: Task 'BadTask' not found in "
+        "pipeline. Exiting." in result.output
     )
 
 
@@ -211,8 +211,8 @@ def test_cli_pipeline_run_start_task_not_found_error(caplog, runner):
     result = runner.invoke(cli.main, args)
     assert result.exit_code == ERROR_RESULT_CODE
     assert (
-        "--include-tasks or --exclude-task are invalid: Task 'BadTask' not found in "
-        "pipeline" in result.output
+        "CLI options --include or --exclude are invalid: Task 'BadTask' not found in "
+        "pipeline. Exiting." in result.output
     )
 
 

tests/test_exceptions.py

@@ -0,0 +1,16 @@
+from hrqb.exceptions import ExcludedTaskRequiredError, TaskNotInPipelineScopeError
+
+TASK_NAME = "TaskFoo"
+
+
+def test_excluded_task_required_exception_message():
+    exc = ExcludedTaskRequiredError("TaskFoo")
+    assert str(exc) == (
+        f"Task '{TASK_NAME}' was required by pipeline but is explicitly "
+        "excluded and does not have pre-existing target data."
+    )
+
+
+def test_task_not_in_pipeline_scope_exception_message():
+    exc = TaskNotInPipelineScopeError("TaskFoo")
+    assert str(exc) == f"Task '{TASK_NAME}' not found in pipeline."