Improve schedule YAML missing-file errors by surfacing a direct validation failure

sdk/ml/azure-ai-ml/azure/ai/ml/_schema/core/fields.py

@@ -45,6 +45,7 @@
 
 module_logger = logging.getLogger(__name__)
 T = typing.TypeVar("T")
+NO_SUCH_FILE_OR_DIRECTORY_ERROR = "No such file or directory"
 
 
 class StringTransformedEnum(Field):
@@ -369,7 +370,10 @@ def _deserialize(self, value, attr, data, **kwargs):
             if not path.is_absolute():
                 path = base_path / path
                 path.resolve()
-            data = load_file(path)
+            try:
+                data = load_file(path)
+            except (FileNotFoundError, OSError) as e:
+                raise ValidationError(f"{NO_SUCH_FILE_OR_DIRECTORY_ERROR}: {path}") from e
             return data
         raise ValidationError(f"Not supporting non file for {attr}")
 
@@ -451,6 +455,16 @@ def union_fields(self):
     def insert_union_field(self, field):
         self._union_fields.insert(0, field)
 
+    @staticmethod
+    def _contains_file_not_found_error(error_message):
+        if isinstance(error_message, str):
+            return NO_SUCH_FILE_OR_DIRECTORY_ERROR in error_message
+        if isinstance(error_message, dict):
+            return any(UnionField._contains_file_not_found_error(value) for value in error_message.values())
+        if isinstance(error_message, list):
+            return any(UnionField._contains_file_not_found_error(value) for value in error_message)
+        return False
+
     # This sets the parent for the schema and also handles nesting.
     def _bind_to_schema(self, field_name, schema):
         super()._bind_to_schema(field_name, schema)
@@ -484,7 +498,10 @@ def _deserialize(self, value, attr, data, **kwargs):
             try:
                 return schema.deserialize(value, attr, data, **kwargs)
             except ValidationError as e:
-                errors.append(e.normalized_messages())
+                normalized_messages = e.normalized_messages()
+                if self._contains_file_not_found_error(normalized_messages):
+                    raise ValidationError(normalized_messages, field_name=attr) from e
+                errors.append(normalized_messages)
             except ValidationException as e:
                 # ValidationException is explicitly raised in project code so usually easy to locate with error message
                 errors.append([str(e)])

sdk/ml/azure-ai-ml/tests/schedule/unittests/test_schedule_entity.py

@@ -1,8 +1,10 @@
 from datetime import datetime
 
 import pytest
+from marshmallow import ValidationError
 from test_utilities.utils import verify_entity_load_and_dump
 
+from azure.ai.ml._schema.core.fields import NO_SUCH_FILE_OR_DIRECTORY_ERROR
 from azure.ai.ml.constants import TimeZone
 from azure.ai.ml.entities import CronTrigger, JobSchedule, PipelineJob, RecurrencePattern, RecurrenceTrigger
 from azure.ai.ml.entities._load_functions import load_job, load_schedule
@@ -189,3 +191,15 @@ def test_schedule_create_out_of_box_monitoring_job(self):
         test_path = "./tests/test_configs/schedule/out_of_box_monitoring.yaml"
         schedule = load_schedule(test_path)
         assert "genai_app_monitoring", schedule.name
+
+    def test_load_schedule_with_nonexistent_pipeline_file_reference(self):
+        test_path = "./tests/test_configs/schedule/hello_cron_schedule_with_file_reference.yml"
+        params_override = [{"create_job.job": "../pipeline_jobs/not_exists_pipeline.yml"}]
+
+        with pytest.raises(ValidationError) as e:
+            load_schedule(test_path, params_override=params_override)
+
+        message = str(e.value)
+        assert NO_SUCH_FILE_OR_DIRECTORY_ERROR in message
+        assert message.count(NO_SUCH_FILE_OR_DIRECTORY_ERROR) == 1
+        assert "Not supporting non file for create_job" not in message