feat(components): Support dynamic machine type paramters in CustomTrainingJobOp

sdk/RELEASE.md

@@ -1,6 +1,7 @@
 # Current Version (in development)
 
 ## Features
+* Support dynamic machine type parameters in CustomTrainingJobOp. [\#10883](https://github.com/kubeflow/pipelines/pull/10883)
 
 ## Breaking changes
 * Drop support for Python 3.7 since it has reached end-of-life. [\#10750](https://github.com/kubeflow/pipelines/pull/10750)

sdk/python/kfp/compiler/compiler_utils.py

@@ -772,3 +772,34 @@ def get_dependencies(
             dependencies[downstream_names[0]].add(upstream_names[0])
 
     return dependencies
+
+
+def recursive_replace_placeholders(data: Union[Dict, List], old_value: str,
+                                   new_value: str) -> Union[Dict, List]:
+    """Recursively replaces values in a nested dict/list object.
+
+    This method is used to replace PipelineChannel objects with input parameter
+    placeholders in a nested object like worker_pool_specs for custom jobs.
+
+    Args:
+        data: A nested object that can contain dictionaries and/or lists.
+        old_value: The value that will be replaced.
+        new_value: The value to replace the old value with.
+
+    Returns:
+        A copy of data with all occurences of old_value replaced by new_value.
+    """
+    if isinstance(data, dict):
+        return {
+            k: recursive_replace_placeholders(v, old_value, new_value)
+            for k, v in data.items()
+        }
+    elif isinstance(data, list):
+        return [
+            recursive_replace_placeholders(i, old_value, new_value)
+            for i in data
+        ]
+    else:
+        if isinstance(data, pipeline_channel.PipelineChannel):
+            data = str(data)
+        return new_value if data == old_value else data

sdk/python/kfp/compiler/compiler_utils_test.py

@@ -53,6 +53,67 @@ def test_additional_input_name_for_pipeline_channel(self, channel,
             expected,
             compiler_utils.additional_input_name_for_pipeline_channel(channel))
 
+    @parameterized.parameters(
+        {
+            'data': [{
+                'container_spec': {
+                    'image_uri':
+                        'gcr.io/ml-pipeline/google-cloud-pipeline-components:2.5.0',
+                    'command': ['echo'],
+                    'args': ['foo']
+                },
+                'machine_spec': {
+                    'machine_type':
+                        pipeline_channel.PipelineParameterChannel(
+                            name='Output',
+                            channel_type='String',
+                            task_name='machine-type'),
+                    'accelerator_type':
+                        pipeline_channel.PipelineParameterChannel(
+                            name='Output',
+                            channel_type='String',
+                            task_name='accelerator-type'),
+                    'accelerator_count':
+                        1
+                },
+                'replica_count': 1
+            }],
+            'old_value':
+                '{{channel:task=machine-type;name=Output;type=String;}}',
+            'new_value':
+                '{{$.inputs.parameters['
+                'pipelinechannel--machine-type-Output'
+                ']}}',
+            'expected': [{
+                'container_spec': {
+                    'image_uri':
+                        'gcr.io/ml-pipeline/google-cloud-pipeline-components:2.5.0',
+                    'command': ['echo'],
+                    'args': ['foo']
+                },
+                'machine_spec': {
+                    'machine_type':
+                        '{{$.inputs.parameters['
+                        'pipelinechannel--machine-type-Output'
+                        ']}}',
+                    'accelerator_type':
+                        pipeline_channel.PipelineParameterChannel(
+                            name='Output',
+                            channel_type='String',
+                            task_name='accelerator-type'),
+                    'accelerator_count':
+                        1
+                },
+                'replica_count': 1
+            }],
+        },)
+    def test_recursive_replace_placeholders(self, data, old_value, new_value,
+                                            expected):
+        self.assertEqual(
+            expected,
+            compiler_utils.recursive_replace_placeholders(
+                data, old_value, new_value))
+
 
 if __name__ == '__main__':
     unittest.main()

sdk/python/kfp/compiler/pipeline_spec_builder.py

@@ -238,12 +238,14 @@ def build_task_spec_for_task(
                     input_name].component_input_parameter = (
                         component_input_parameter)
 
-        elif isinstance(input_value, str):
-            # Handle extra input due to string concat
+        elif isinstance(input_value, (str, int, float, bool, dict, list)):
             pipeline_channels = (
                 pipeline_channel.extract_pipeline_channels_from_any(input_value)
             )
             for channel in pipeline_channels:
+                # NOTE: case like this   p3 = print_and_return_str(s='Project = {}'.format(project))
+                # triggers this code
+
                 # value contains PipelineChannel placeholders which needs to be
                 # replaced. And the input needs to be added to the task spec.
 
@@ -265,8 +267,14 @@ def build_task_spec_for_task(
 
                 additional_input_placeholder = placeholders.InputValuePlaceholder(
                     additional_input_name)._to_string()
-                input_value = input_value.replace(channel.pattern,
-                                                  additional_input_placeholder)
+
+                if isinstance(input_value, str):
+                    input_value = input_value.replace(
+                        channel.pattern, additional_input_placeholder)
+                else:
+                    input_value = compiler_utils.recursive_replace_placeholders(
+                        input_value, channel.pattern,
+                        additional_input_placeholder)
 
                 if channel.task_name:
                     # Value is produced by an upstream task.
@@ -299,11 +307,6 @@ def build_task_spec_for_task(
                         additional_input_name].component_input_parameter = (
                             component_input_parameter)
 
-            pipeline_task_spec.inputs.parameters[
-                input_name].runtime_value.constant.string_value = input_value
-
-        elif isinstance(input_value, (str, int, float, bool, dict, list)):
-
             pipeline_task_spec.inputs.parameters[
                 input_name].runtime_value.constant.CopyFrom(
                     to_protobuf_value(input_value))

sdk/python/kfp/dsl/pipeline_channel.py

@@ -586,7 +586,7 @@ def extract_pipeline_channels_from_string(
 
 
 def extract_pipeline_channels_from_any(
-    payload: Union[PipelineChannel, str, list, tuple, dict]
+    payload: Union[PipelineChannel, str, int, float, bool, list, tuple, dict]
 ) -> List[PipelineChannel]:
     """Recursively extract PipelineChannels from any object or list of objects.
 

sdk/python/test_data/pipelines/pipeline_with_condition_dynamic_task_output_custom_training_job.py

@@ -0,0 +1,64 @@
+import google_cloud_pipeline_components.v1.custom_job as custom_job
+from kfp import dsl
+
+
+@dsl.component
+def flip_biased_coin_op() -> str:
+    """Flip a coin and output heads."""
+    return 'heads'
+
+
+@dsl.component
+def machine_type() -> str:
+    return 'n1-standard-4'
+
+
+@dsl.component
+def accelerator_type() -> str:
+    return 'NVIDIA_TESLA_P4'
+
+
+@dsl.component
+def accelerator_count() -> int:
+    return 1
+
+
+@dsl.pipeline
+def pipeline(
+    project: str,
+    location: str,
+    encryption_spec_key_name: str = '',
+):
+    flip1 = flip_biased_coin_op().set_caching_options(False)
+    machine_type_task = machine_type()
+    accelerator_type_task = accelerator_type()
+    accelerator_count_task = accelerator_count()
+
+    with dsl.Condition(flip1.output == 'heads'):
+        custom_job.CustomTrainingJobOp(
+            display_name='add-numbers',
+            worker_pool_specs=[{
+                'container_spec': {
+                    'image_uri': (
+                        'gcr.io/ml-pipeline/google-cloud-pipeline-components:2.5.0'
+                    ),
+                    'command': ['echo'],
+                    'args': ['foo'],
+                },
+                'machine_spec': {
+                    'machine_type': machine_type_task.output,
+                    'accelerator_type': accelerator_type_task.output,
+                    'accelerator_count': accelerator_count_task.output,
+                },
+                'replica_count': 1,
+            }],
+            project=project,
+            location=location,
+            encryption_spec_key_name=encryption_spec_key_name,
+        )
+
+
+if __name__ == '__main__':
+    from kfp import compiler
+    compiler.Compiler().compile(
+        pipeline_func=pipeline, package_path=__file__.replace('.py', '.yaml'))