コード例 #1
0
ファイル: ehr_union_test.py プロジェクト: ksdkalluri/curation
 def _test_mapping_query(self):
     table = 'measurement'
     hpo_ids = ['chs', 'pitt']
     project_id = bq_utils.app_identity.get_application_id()
     dataset_id = bq_utils.get_dataset_id()
     created_tables = []
     for hpo_id in hpo_ids:
         hpo_table = self._create_hpo_table(hpo_id, table, dataset_id)
         created_tables.append(hpo_table)
     q = ehr_union.mapping_query(table, hpo_ids, dataset_id, project_id)
コード例 #2
0
    def test_mapping_query(self):
        table = 'measurement'
        hpo_ids = ['chs', 'pitt']
        mapping_msg = 'Expected mapping subquery count %s but got %s for hpo_id %s'
        project_id = bq_utils.app_identity.get_application_id()
        dataset_id = bq_utils.get_dataset_id()
        created_tables = []
        for hpo_id in hpo_ids:
            hpo_table = self._create_hpo_table(hpo_id, table, dataset_id)
            created_tables.append(hpo_table)
        query = ehr_union.mapping_query(table, hpo_ids, dataset_id, project_id)
        dataset_id = os.environ.get('BIGQUERY_DATASET_ID')
        app_id = os.getenv('APPLICATION_ID')
        #testing the query string
        expected_query = '''
            WITH all_measurement AS (
      
                (SELECT 'chs_measurement' AS src_table_id,
                  measurement_id AS src_measurement_id,
                  ROW_NUMBER() over() + 100000000 as measurement_id
                  FROM `{app_id}.{dataset_id}.chs_measurement`)
                

        UNION ALL
        

                (SELECT 'pitt_measurement' AS src_table_id,
                  measurement_id AS src_measurement_id,
                  ROW_NUMBER() over() + 200000000 as measurement_id
                  FROM `{app_id}.{dataset_id}.pitt_measurement`)
                
    )
    SELECT 
        src_table_id,
        src_measurement_id,
        measurement_id,
        SUBSTR(src_table_id, 1, STRPOS(src_table_id, "_measurement")-1) AS src_hpo_id
    FROM all_measurement
    '''.format(dataset_id=dataset_id, app_id=app_id)
        self.assertEqual(
            expected_query.strip(), query.strip(),
            "Mapping query for \n {q} \n to is not as expected".format(
                q=query))
コード例 #3
0
    def test_mapping_query(self):
        table = 'measurement'
        hpo_ids = ['nyc', 'pitt']
        mapping_msg = 'Expected mapping subquery count %s but got %s for hpo_id %s'
        project_id = bq_utils.app_identity.get_application_id()
        dataset_id = bq_utils.get_dataset_id()
        created_tables = []
        for hpo_id in hpo_ids:
            hpo_table = self._create_hpo_table(hpo_id, table, dataset_id)
            created_tables.append(hpo_table)
        query = ehr_union.mapping_query(table, hpo_ids, dataset_id, project_id)
        # testing the query string
        expected_query = '''
            WITH all_measurement AS (
      
                (SELECT 'nyc_measurement' AS src_table_id,
                  measurement_id AS src_measurement_id,
                  measurement_id + 3000000000000000 as measurement_id
                  FROM `{project_id}.{dataset_id}.nyc_measurement`)
                

        UNION ALL
        

                (SELECT 'pitt_measurement' AS src_table_id,
                  measurement_id AS src_measurement_id,
                  measurement_id + 4000000000000000 as measurement_id
                  FROM `{project_id}.{dataset_id}.pitt_measurement`)
                
    )
    SELECT DISTINCT
        src_table_id,
        src_measurement_id,
        measurement_id,
        SUBSTR(src_table_id, 1, STRPOS(src_table_id, "_measurement")-1) AS src_hpo_id
    FROM all_measurement
    '''.format(dataset_id=dataset_id, project_id=project_id)
        self.assertEqual(
            expected_query.strip(), query.strip(),
            "Mapping query for \n {q} \n to is not as expected".format(
                q=query))
コード例 #4
0
    def test_mapping_query(self, mock_list_all_table_ids, mock_hpo_info):
        mock_hpo_info.return_value = [{
            'hpo_id': hpo_id
        } for hpo_id in self.hpo_ids]
        mock_list_all_table_ids.return_value = [
            f'{self.FAKE_SITE_1}_measurement', f'{self.FAKE_SITE_2}_measurement'
        ]
        dataset_id = 'fake_dataset'
        project_id = 'fake_project'
        table = 'measurement'
        query = eu.mapping_query(table, self.hpo_ids, dataset_id, project_id)
        # testing the query string
        expected_query = f'''
            WITH all_measurement AS (
    
    (SELECT '{self.FAKE_SITE_1}_measurement' AS src_table_id,
      measurement_id AS src_measurement_id,
      measurement_id + 3000000000000000 AS measurement_id
      FROM `{project_id}.{dataset_id}.{self.FAKE_SITE_1}_measurement`)
    

        UNION ALL


    (SELECT '{self.FAKE_SITE_2}_measurement' AS src_table_id,
      measurement_id AS src_measurement_id,
      measurement_id + 4000000000000000 AS measurement_id
      FROM `{project_id}.{dataset_id}.{self.FAKE_SITE_2}_measurement`)
    
    )
    SELECT DISTINCT
        src_table_id,
        src_measurement_id,
        measurement_id,
        SUBSTR(src_table_id, 1, STRPOS(src_table_id, "_measurement")-1) AS src_hpo_id,
        '{dataset_id}' as src_dataset_id
    FROM all_measurement
    '''.format(dataset_id=dataset_id, project_id=project_id)
        self.assertEqual(
            expected_query.strip(), query.strip(),
            "Mapping query for \n {q} \n to is not as expected".format(q=query))