Пример #1
0
 def test_mark_object_as_blank(self, test_id, dynamo_test_environment):
     function_name = 'mark_object_as_blank'
     dynamo_driver = LeechDriver(table_name=blank_table_name)
     test_identifier_stem = test_id[0]
     test_id_value = test_id[1]
     dynamo_driver.mark_object_as_blank(identifier_stem=test_identifier_stem, id_value=test_id_value)
     self._assert_dynamo_call(function_name, test_id_value, test_identifier_stem, dynamo_test_environment,
                              stage_name='extraction')
Пример #2
0
class Dentist:
    def __init__(self, metal_order, **kwargs):
        self._extraction_order = metal_order
        self._extraction_function_name = metal_order.extraction_function_name
        self._extraction_properties = metal_order.extraction_properties
        self._schema_entry = metal_order.schema_entry
        self._dynamo_driver = LeechDriver()
        self._transform_queue = kwargs.get(
            'transform_queue', ForgeQueue.get_for_transform_queue(**kwargs))

    @classmethod
    def extract_bulk(cls, metal_orders):
        results = []
        for metal_order in metal_orders:
            dentist = cls(metal_order)
            result = dentist.extract()
            results.append(result)
        return results

    def extract(self):
        extracted_data = StageManager.run_extraction(
            self._extraction_function_name, self._extraction_properties)
        source_data = extracted_data['source']
        if len(source_data) > 1:
            raise InvalidExtractionMultipleSourceException(
                self._extraction_function_name, self._extraction_order)
        if not source_data:
            return self._mark_object_blank()
        for entry in source_data:
            if not entry:
                return self._mark_object_blank()
            extracted_data['source'] = entry
            break
        transform_order = TransformObjectOrder(
            self._extraction_order.identifier_stem,
            self._extraction_order.id_value, extracted_data,
            self._schema_entry)
        self._transform_queue.add_order(transform_order)
        self._dynamo_driver.set_extraction_results(
            extracted_data,
            identifier_stem=self._extraction_order.identifier_stem,
            id_value=self._extraction_order.id_value)
        self._transform_queue.push_orders()

    def _mark_object_blank(self):
        return self._dynamo_driver.mark_object_as_blank(
            identifier_stem=self._extraction_order.identifier_stem,
            id_value=self._extraction_order.id_value)