robertwb commented on code in PR #28754: URL: https://github.com/apache/beam/pull/28754#discussion_r1349171107
########## sdks/python/apache_beam/yaml/json_utils.py: ########## @@ -0,0 +1,167 @@ +# +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# + +"""Utilities for converting between JSON and Beam Schema'd data. Review Comment: These may be fine to expose, but added a disclaimer just in case. ########## sdks/python/apache_beam/yaml/json_utils.py: ########## @@ -0,0 +1,167 @@ +# +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# + +"""Utilities for converting between JSON and Beam Schema'd data. +""" + +import json +from typing import Any +from typing import Callable +from typing import Dict + +import apache_beam as beam +from apache_beam.portability.api import schema_pb2 +from apache_beam.typehints import schemas + +JSON_ATOMIC_TYPES_TO_BEAM = { + 'boolean': schema_pb2.BOOLEAN, + 'integer': schema_pb2.INT64, + 'number': schema_pb2.DOUBLE, + 'string': schema_pb2.STRING, +} + + +def json_schema_to_beam_schema( + json_schema: Dict[str, Any]) -> schema_pb2.Schema: + def maybe_nullable(beam_type, nullable): + if nullable: + beam_type.nullable = True + return beam_type + + json_type = json_schema.get('type', None) + if json_type != 'object': + raise ValueError('Expected object type, got {json_type}.') + if 'properties' not in json_schema: + # Technically this is a valid (vacuous) schema, but as it's not generally + # meaningful, throw an informative error instead. + # (We could add a flag to allow this degenerate case.) + raise ValueError('Missing properties for {json_schema}.') + required = set(json_schema.get('required', [])) + return schema_pb2.Schema( + fields=[ + schemas.schema_field( + name, + maybe_nullable(json_type_to_beam_type(t), name not in required)) + for (name, t) in json_schema['properties'].items() + ]) + + +def json_type_to_beam_type(json_type: Dict[str, Any]) -> schema_pb2.FieldType: + if not isinstance(json_type, dict) or 'type' not in json_type: + raise ValueError(f'Malformed type {json_type}.') + type_name = json_type['type'] + if type_name in JSON_ATOMIC_TYPES_TO_BEAM: + return schema_pb2.FieldType( + atomic_type=JSON_ATOMIC_TYPES_TO_BEAM[type_name]) + elif type_name == 'array': + return schema_pb2.FieldType( + array_type=schema_pb2.ArrayType( + element_type=json_type_to_beam_type(json_type['items']))) + elif type_name == 'object': + if 'properties' in json_type: + return schema_pb2.FieldType( + row_type=schema_pb2.RowType( + schema=json_schema_to_beam_schema(json_type))) + elif 'additionalProperties' in json_type: + return schema_pb2.FieldType( + map_type=schema_pb2.MapType( + key_type=schema_pb2.FieldType(atomic_type=schema_pb2.STRING), + value_type=json_type_to_beam_type( + json_type['additionalProperties']))) + else: + raise ValueError( + f'Object type must have either properties or additionalProperties, ' + f'got {json_type}.') + else: + raise ValueError(f'Unable to convert {json_type} to a Beam schema.') + + +def json_to_row(beam_type: schema_pb2.FieldType) -> Callable[[Any], Any]: + type_info = beam_type.WhichOneof("type_info") + if type_info == "atomic_type": + return lambda value: value + elif type_info == "array_type": + element_converter = json_to_row(beam_type.array_type.element_type) + return lambda value: [element_converter(e) for e in value] + elif type_info == "iterable_type": + element_converter = json_to_row(beam_type.iterable_type.element_type) + return lambda value: [element_converter(e) for e in value] + elif type_info == "map_type": + if beam_type.map_type.key_type.atomic_type != schema_pb2.STRING: + raise TypeError( + f'Only strings allowd as map keys when converting from JSON, ' + f'found {beam_type}') + value_converter = json_to_row(beam_type.map_type.value_type) + return lambda value: {k: value_converter(v) for (k, v) in value.items()} + elif type_info == "row_type": + converters = { + field.name: json_to_row(field.type) + for field in beam_type.row_type.schema.fields + } + return lambda value: beam.Row( + ** + {name: convert(value[name]) + for (name, convert) in converters.items()}) + elif type_info == "logical_type": + return lambda value: value + else: + raise ValueError(f"Unrecognized type_info: {type_info!r}") + + +def json_parser(beam_schema: schema_pb2.Schema) -> Callable[[bytes], beam.Row]: Review Comment: Added basic docstrings to all of these methods. Hopefully the name and signatures should make them clear as well. ########## sdks/python/apache_beam/yaml/yaml_io_test.py: ########## @@ -167,6 +167,46 @@ def test_read_with_id_attribute(self): result, equal_to([beam.Row(payload=b'msg1'), beam.Row(payload=b'msg2')])) + def test_read_json(self): + with beam.Pipeline(options=beam.options.pipeline_options.PipelineOptions( + pickle_library='cloudpickle')) as p: + with mock.patch('apache_beam.io.ReadFromPubSub', + FakeReadFromPubSub( + topic='my_topic', + messages=[PubsubMessage( + b'{"generator": {"x": 0, "y": 0}, "rank": 1}', + {'weierstrass': 'y^2+y=x^3-x', 'label': '37a'}) + ])): + result = p | YamlTransform( Review Comment: Done. This actually uncovered a bug when transforms are explicitly named in some cases. Thanks. -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
