robertwb commented on a change in pull request #11185: [BEAM-8019] Some
generalizations to support cross-language transforms.
URL: https://github.com/apache/beam/pull/11185#discussion_r397540960
##########
File path: sdks/python/apache_beam/coders/coders.py
##########
@@ -1383,22 +1375,74 @@ def from_runner_api_parameter(payload, components,
context):
write_state_threshold=int(payload))
-class RunnerAPICoderHolder(Coder):
+class ElementTypeHolder(typehints.TypeConstraint):
+ """A dummy element type for external coders that cannot be parsed in
Python"""
+ def __init__(self, coder, context):
+ self.coder = coder
+ self.context = context
+
+
+class ExternalCoder(Coder):
"""A `Coder` that holds a runner API `Coder` proto.
This is used for coders for which corresponding objects cannot be
initialized in Python SDK. For example, coders for remote SDKs that may
be available in Python SDK transform graph when expanding a cross-language
transform.
"""
- def __init__(self, proto):
- self._proto = proto
- def proto(self):
- return self._proto
+ coder_count = 0
- def to_runner_api(self, context):
- return self._proto
+ def __init__(self, element_type_holder):
+ self.element_type_holder = element_type_holder
- def to_type_hint(self):
- return Any
+ def as_cloud_object(self, coders_context=None):
+ if not coders_context:
+ raise Exception(
+ 'coders_context must be specified to correctly encode external
coders'
+ )
+ coder_id = coders_context.get_by_proto(
+ self.element_type_holder.coder, deduplicate=True)
+
+ coder_proto = self.element_type_holder.coder
+
+ kind_str = 'kind:external' + str(ExternalCoder.coder_count)
+ ExternalCoder.coder_count = ExternalCoder.coder_count + 1
+ component_encodings = []
+ if coder_proto.spec.urn == 'beam:coder:kv:v1':
+ kind_str = 'kind:pair'
+ for component_coder_id in coder_proto.component_coder_ids:
+ component_encodings.append({
+ '@type': 'kind:external' + str(ExternalCoder.coder_count),
+ 'pipeline_proto_coder_id': component_coder_id
+ })
+ ExternalCoder.coder_count = ExternalCoder.coder_count + 1
+
+ value = {
+ # This is a placeholder type. Dataflow will get the actual coder from
+ # pipeline proto using the pipeline_proto_coder_id property.
+ '@type': kind_str,
+ 'pipeline_proto_coder_id': coder_id
+ }
+ if component_encodings:
+ value['is_pair_like'] = True
+ value['component_encodings'] = component_encodings
+
+ return value
+
+ @staticmethod
+ def from_type_hint(typehint, unused_registry):
+ if isinstance(typehint, ElementTypeHolder):
+ return ExternalCoder(typehint)
+ else:
+ raise ValueError((
+ 'Expected an instance of ElementTypeHolder'
+ ', but got a %s' % typehint))
+
+ def to_runner_api_parameter(self, context):
+ if self.element_type_holder.coder.component_coder_ids:
+ raise NotImplementedError
Review comment:
Why raise NotImplementedError here instead of returning this in place of the
empty tuple below?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
