def _create_events_table()

in Gems/AWSMetrics/cdk/aws_metrics/data_lake_integration.py [0:0]

• 78 lines of code
• 1 McCabe index (conditional complexity)

    def _create_events_table(self) -> None:
        """
        Create the Glue table for metrics events. This table is used by the Kinesis Data Firehose
        to convert data from the JSON format to the Parquet format before writing it to Amazon S3.
        """
        self._events_table = glue.CfnTable(
            self._stack,
            id=f'EventsTable',
            catalog_id=self._stack.account,
            database_name=self._events_database.ref,
            table_input=glue.CfnTable.TableInputProperty(
                description=f'Stores metrics event data from the analytics pipeline for stack {self._stack.stack_name}',
                name=aws_metrics_constants.GLUE_TABLE_NAME,
                table_type='EXTERNAL_TABLE',
                partition_keys=[
                    glue.CfnTable.ColumnProperty(
                        name='year',
                        type='string'
                    ),
                    glue.CfnTable.ColumnProperty(
                        name='month',
                        type='string'
                    ),
                    glue.CfnTable.ColumnProperty(
                        name='day',
                        type='string'
                    ),
                ],
                parameters={
                    'classification': 'parquet',
                    'compressionType': 'none',
                    'typeOfData': 'file'
                },
                storage_descriptor=glue.CfnTable.StorageDescriptorProperty(
                    input_format=aws_metrics_constants.GLUE_TABLE_INPUT_FORMAT,
                    output_format=aws_metrics_constants.GLUE_TABLE_OUTPUT_FORMAT,
                    serde_info=glue.CfnTable.SerdeInfoProperty(
                        serialization_library=aws_metrics_constants.GLUE_TABLE_SERIALIZATION_LIBRARY,
                        parameters={
                            'serialization.format':
                                aws_metrics_constants.GLUE_TABLE_SERIALIZATION_LIBRARY_SERIALIZATION_FORMAT
                        }
                    ),
                    stored_as_sub_directories=False,
                    location=f's3://{self._analytics_bucket.bucket_name}/{aws_metrics_constants.GLUE_TABLE_NAME}/',
                    columns=[
                        glue.CfnTable.ColumnProperty(
                            name='event_id',
                            type='string'
                        ),
                        glue.CfnTable.ColumnProperty(
                            name='event_type',
                            type='string'
                        ),
                        glue.CfnTable.ColumnProperty(
                            name='event_name',
                            type='string'
                        ),
                        glue.CfnTable.ColumnProperty(
                            name='event_timestamp',
                            type='string'
                        ),
                        glue.CfnTable.ColumnProperty(
                            name='event_version',
                            type='string'
                        ),
                        glue.CfnTable.ColumnProperty(
                            name='event_source',
                            type='string'
                        ),
                        glue.CfnTable.ColumnProperty(
                            name='application_id',
                            type='string'
                        ),
                        glue.CfnTable.ColumnProperty(
                            name='event_data',
                            type='string'
                        )
                    ]
                )
            )
        )