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_r395944333
##########
File path: sdks/python/apache_beam/pipeline.py
##########
@@ -1127,30 +1133,79 @@ def transform_to_runner_api(transform, # type:
Optional[ptransform.PTransform]
def from_runner_api(proto, # type: beam_runner_api_pb2.PTransform
context # type: PipelineContext
):
+ side_input_tags = []
+ if common_urns.primitives.PAR_DO.urn == proto.spec.urn:
+ # Preserving side input tags.
+ from apache_beam.utils import proto_utils
+ from apache_beam.portability.api import beam_runner_api_pb2
+ payload = (
+ proto_utils.parse_Bytes(
+ proto.spec.payload, beam_runner_api_pb2.ParDoPayload))
+ for tag, si in payload.side_inputs.items():
+ side_input_tags.append(tag)
+
# type: (...) -> AppliedPTransform
- def is_side_input(tag):
- # type: (str) -> bool
+ def is_python_side_input(tag):
# As per named_inputs() above.
- return tag.startswith('side')
+ return re.match(SIDE_INPUT_REGEX, tag)
+
+ all_input_tags = [tag for tag, id in proto.inputs.items()]
+
+ # All side inputs have to be available in input tags
+ python_indexed_side_inputs = False
+ for side_tag in side_input_tags:
+ if side_tag not in all_input_tags:
+ raise Exception(
+ 'Side input tag %s is not available in list of input tags %r' %
+ (side_tag, all_input_tags))
+
+ # We process Python and external side inputs differently. We fail early
+ # here if we cannot decide which way to go.
+ if is_python_side_input(side_tag):
+ python_indexed_side_inputs = True
+ else:
+ if python_indexed_side_inputs:
+ raise Exception(
+ 'Cannot process side inputs due to inconsistent sideinput '
+ 'naming. If using an external transform consider re-naming side '
+ 'inputs to not match Python indexed format %s' %
+ SIDE_INPUT_REGEX)
main_inputs = [
context.pcollections.get_by_id(id) for tag,
- id in proto.inputs.items() if not is_side_input(tag)
+ id in proto.inputs.items() if tag not in side_input_tags
]
- # Ordering is important here.
- indexed_side_inputs = [
- (get_sideinput_index(tag), context.pcollections.get_by_id(id)) for tag,
- id in proto.inputs.items() if is_side_input(tag)
- ]
- side_inputs = [si for _, si in sorted(indexed_side_inputs)]
+ if python_indexed_side_inputs:
+ # Ordering is important here.
Review comment:
This all seems rather fragile. Would it be possible to just make side_inputs
a dict everywhere in the internal SDK representation? (Or is there
introspection with the legacy worker code that would make this hard?)
----------------------------------------------------------------
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:
[email protected]
With regards,
Apache Git Services
