def handle(self, *args, **kwargs): client = Client("hscic") try: client.create_storage_backed_table( "raw_prescribing_v1", RAW_PRESCRIBING_SCHEMA_V1, "hscic/prescribing_v1/20*Detailed_Prescribing_Information.csv", ) except Conflict: pass try: client.create_storage_backed_table( "raw_prescribing_v2", RAW_PRESCRIBING_SCHEMA_V2, # This pattern may change once the data is published via the # new Open Data Portal. "hscic/prescribing_v2/20*.csv", ) except Conflict: pass for table_name in [ "all_prescribing", "normalised_prescribing", "normalised_prescribing_standard", "raw_prescribing_normalised", ]: self.recreate_table(client, table_name) client = Client("measures") for table_name in [ "dmd_objs_with_form_route", "dmd_objs_hospital_only", "opioid_total_ome", "practice_data_all_low_priority", "pregabalin_total_mg", "vw__median_price_per_unit", "vw__ghost_generic_measure", "vw__herbal_list", # This references pregabalin_total_mg, so must come afterwards "gaba_total_ddd", ]: self.recreate_table(client, table_name) # cmpa_products is a table that has been created and managed by Rich. schema = build_schema( ("bnf_code", "STRING"), ("bnf_name", "STRING"), ("type", "STRING") ) client.get_or_create_table("cmpa_products", schema)
def handle(self, *args, **options): path = options["filename"] head, filename = os.path.split(path) converted_path = "{}_formatted.CSV".format(os.path.splitext(path)[0]) _, year_and_month = os.path.split(head) logger.info("path: %s", path) logger.info("converted_path: %s", converted_path) logger.info("year_and_month: %s", year_and_month) date = year_and_month + "_01" try: datetime.datetime.strptime(date, "%Y_%m_%d") except ValueError: message = ("The file path must have a YYYY_MM " "date component in the containing directory: ") message += path raise CommandError(message) hscic_dataset_client = Client("hscic") tmp_dataset_client = Client("tmp_eu") # Check that we haven't already processed data for this month sql = """SELECT COUNT(*) FROM {dataset}.prescribing WHERE month = TIMESTAMP('{date}')""".format( dataset=hscic_dataset_client.dataset_id, date=date.replace("_", "-")) try: results = hscic_dataset_client.query(sql) assert results.rows[0][0] == 0 except NotFound: pass # Create BQ table backed backed by uploaded source CSV file raw_data_table_name = "raw_prescribing_data_{}".format(year_and_month) gcs_path = "hscic/prescribing/{}/{}".format(year_and_month, filename) logger.info("raw_data_table_name: %s", raw_data_table_name) logger.info("gcs_path: %s", gcs_path) raw_data_table = tmp_dataset_client.create_storage_backed_table( raw_data_table_name, RAW_PRESCRIBING_SCHEMA, gcs_path) # Append aggregated data to prescribing table sql = """ SELECT Area_Team_Code AS sha, LEFT(PCO_Code, 3) AS pct, Practice_Code AS practice, BNF_Code AS bnf_code, BNF_Description AS bnf_name, SUM(Items) AS items, SUM(NIC) AS net_cost, SUM(Actual_Cost) AS actual_cost, SUM(Quantity * Items) AS quantity, TIMESTAMP('%s') AS month, FROM %s WHERE Practice_Code NOT LIKE '%%998' -- see issue #349 GROUP BY bnf_code, bnf_name, pct, practice, sha """ % ( date.replace("_", "-"), raw_data_table.qualified_name, ) logger.info("sql: %s", sql) prescribing_table = hscic_dataset_client.get_table("prescribing") prescribing_table.insert_rows_from_query( sql, legacy=True, write_disposition="WRITE_APPEND") # Write aggregated data to new table, for download sql = """ SELECT LEFT(PCO_Code, 3) AS pct_id, Practice_Code AS practice_code, BNF_Code AS presentation_code, SUM(Items) AS total_items, SUM(NIC) AS net_cost, SUM(Actual_Cost) AS actual_cost, SUM(Quantity * Items) AS quantity, '%s' AS processing_date, FROM %s WHERE Practice_Code NOT LIKE '%%998' -- see issue #349 GROUP BY presentation_code, pct_id, practice_code """ % ( date, raw_data_table.qualified_name, ) fmtd_data_table_name = "formatted_prescribing_%s" % year_and_month logger.info("sql: %s", sql) logger.info("fmtd_data_table_name: %s", fmtd_data_table_name) fmtd_data_table = tmp_dataset_client.get_table(fmtd_data_table_name) fmtd_data_table.insert_rows_from_query(sql, legacy=True) # Export new table to storage, and download exporter = TableExporter(fmtd_data_table, gcs_path + "_formatted-") exporter.export_to_storage(print_header=False) with tempfile.NamedTemporaryFile(dir=head) as f: exporter.download_from_storage_and_unzip(f) # Sort the output. # # Why? Because this is equivalent to CLUSTERing the table on # loading, but less resource-intensive than doing it in # Postgres. And the table is too big to sort within BigQuery. subprocess.call( "ionice -c 2 nice -n 10 sort -k3,3 -k1,1 -k2,2 -t, %s > %s" % (f.name, converted_path), shell=True, )
def handle(self, *args, **options): path = options["filename"] head, filename = os.path.split(path) _, year_and_month = os.path.split(head) logger.info("path: %s", path) logger.info("year_and_month: %s", year_and_month) date = year_and_month + "_01" try: datetime.datetime.strptime(date, "%Y_%m_%d") except ValueError: message = ("The file path must have a YYYY_MM " "date component in the containing directory: ") message += path raise CommandError(message) hscic_dataset_client = Client("hscic") tmp_dataset_client = Client("tmp_eu") # Check that we haven't already processed data for this month sql = """SELECT COUNT(*) FROM {hscic}.prescribing_v2 WHERE month = TIMESTAMP('{date}')""" try: results = hscic_dataset_client.query( sql, substitutions={"date": date.replace("_", "-")}) assert results.rows[0][0] == 0 except NotFound: pass # Create BQ table backed backed by uploaded source CSV file raw_data_table_name = "raw_prescribing_data_{}".format(year_and_month) gcs_path = "hscic/prescribing_v2/{}/{}".format(year_and_month, filename) logger.info("raw_data_table_name: %s", raw_data_table_name) logger.info("gcs_path: %s", gcs_path) raw_data_table = tmp_dataset_client.create_storage_backed_table( raw_data_table_name, RAW_PRESCRIBING_SCHEMA_V2, gcs_path) # Append aggregated data to prescribing table sql = """ SELECT AREA_TEAM_CODE AS sha, LEFT(PCO_CODE, 3) AS pct, PRACTICE_CODE AS practice, BNF_CODE AS bnf_code, BNF_DESCRIPTION AS bnf_name, SUM(ITEMS) AS items, SUM(NIC) AS net_cost, SUM(ACTUAL_COST) AS actual_cost, SUM(TOTAL_QUANTITY) AS quantity, TIMESTAMP('%s') AS month, FROM %s WHERE PRACTICE_CODE NOT LIKE '%%998' -- see issue #349 GROUP BY bnf_code, bnf_name, pct, practice, sha """ % ( date.replace("_", "-"), raw_data_table.qualified_name, ) logger.info("sql: %s", sql) prescribing_table = hscic_dataset_client.get_table("prescribing_v2") prescribing_table.insert_rows_from_query( sql, legacy=True, write_disposition="WRITE_APPEND") ImportLog.objects.create(current_at=date.replace("_", "-"), filename=filename, category="prescribing")
def handle(self, *args, **options): path = options['filename'] head, filename = os.path.split(path) converted_path = '{}_formatted.CSV'.format(os.path.splitext(path)[0]) _, year_and_month = os.path.split(head) logger.info('path: %s', path) logger.info('converted_path: %s', converted_path) logger.info('year_and_month: %s', year_and_month) date = year_and_month + '_01' try: datetime.datetime.strptime(date, '%Y_%m_%d') except ValueError: message = ('The file path must have a YYYY_MM ' 'date component in the containing directory: ') message += path raise CommandError(message) hscic_dataset_client = Client('hscic') tmp_dataset_client = Client('tmp_eu') # Check that we haven't already processed data for this month sql = '''SELECT COUNT(*) FROM {dataset}.prescribing WHERE month = TIMESTAMP('{date}')'''.format( dataset=hscic_dataset_client.dataset_id, date=date.replace('_', '-'), ) try: results = hscic_dataset_client.query(sql) assert results.rows[0][0] == 0 except NotFound: pass # Create BQ table backed backed by uploaded source CSV file raw_data_table_name = 'raw_prescribing_data_{}'.format(year_and_month) gcs_path = 'hscic/prescribing/{}/{}'.format(year_and_month, filename) logger.info('raw_data_table_name: %s', raw_data_table_name) logger.info('gcs_path: %s', gcs_path) schema = [ {'name': 'Regional_Office_Name', 'type': 'string'}, {'name': 'Regional_Office_Code', 'type': 'string'}, {'name': 'Area_Team_Name', 'type': 'string'}, {'name': 'Area_Team_Code', 'type': 'string', 'mode': 'required'}, {'name': 'PCO_Name', 'type': 'string'}, {'name': 'PCO_Code', 'type': 'string'}, {'name': 'Practice_Name', 'type': 'string'}, {'name': 'Practice_Code', 'type': 'string', 'mode': 'required'}, {'name': 'BNF_Code', 'type': 'string', 'mode': 'required'}, {'name': 'BNF_Description', 'type': 'string', 'mode': 'required'}, {'name': 'Items', 'type': 'integer', 'mode': 'required'}, {'name': 'Quantity', 'type': 'integer', 'mode': 'required'}, {'name': 'ADQ_Usage', 'type': 'float'}, {'name': 'NIC', 'type': 'float', 'mode': 'required'}, {'name': 'Actual_Cost', 'type': 'float', 'mode': 'required'}, ] raw_data_table = tmp_dataset_client.create_storage_backed_table( raw_data_table_name, schema, gcs_path ) # Append aggregated data to prescribing table sql = ''' SELECT Area_Team_Code AS sha, LEFT(PCO_Code, 3) AS pct, Practice_Code AS practice, BNF_Code AS bnf_code, BNF_Description AS bnf_name, SUM(Items) AS items, SUM(NIC) AS net_cost, SUM(Actual_Cost) AS actual_cost, SUM(Quantity * Items) AS quantity, TIMESTAMP('%s') AS month, FROM %s WHERE Practice_Code NOT LIKE '%%998' -- see issue #349 GROUP BY bnf_code, bnf_name, pct, practice, sha ''' % (date.replace('_', '-'), raw_data_table.qualified_name) logger.info('sql: %s', sql) prescribing_table = hscic_dataset_client.get_table('prescribing') prescribing_table.insert_rows_from_query( sql, legacy=True, write_disposition='WRITE_APPEND' ) # Write aggregated data to new table, for download sql = ''' SELECT LEFT(PCO_Code, 3) AS pct_id, Practice_Code AS practice_code, BNF_Code AS presentation_code, SUM(Items) AS total_items, SUM(NIC) AS net_cost, SUM(Actual_Cost) AS actual_cost, SUM(Quantity * Items) AS quantity, '%s' AS processing_date, FROM %s WHERE Practice_Code NOT LIKE '%%998' -- see issue #349 GROUP BY presentation_code, pct_id, practice_code ''' % (date, raw_data_table.qualified_name) fmtd_data_table_name = 'formatted_prescribing_%s' % year_and_month logger.info('sql: %s', sql) logger.info('fmtd_data_table_name: %s', fmtd_data_table_name) fmtd_data_table = tmp_dataset_client.get_table(fmtd_data_table_name) fmtd_data_table.insert_rows_from_query(sql, legacy=True) # Export new table to storage, and download exporter = TableExporter(fmtd_data_table, gcs_path + '_formatted-') exporter.export_to_storage(print_header=False) with tempfile.NamedTemporaryFile(dir=head) as f: exporter.download_from_storage_and_unzip(f) # Sort the output. # # Why? Because this is equivalent to CLUSTERing the table on # loading, but less resource-intensive than doing it in # Postgres. And the table is too big to sort within BigQuery. subprocess.call( "ionice -c 2 nice -n 10 sort -k3,3 -k1,1 -k2,2 -t, %s > %s" % ( f.name, converted_path), shell=True)
def test_the_lot(self): client = Client('test') schema = build_schema( ('a', 'INTEGER'), ('b', 'STRING'), ) headers = ['a', 'b'] rows = [ (1, 'apple'), (2, 'banana'), (3, 'coconut'), ] t1 = client.get_or_create_table('t1', schema) t1_qname = t1.qualified_name # Test Table.insert_rows_from_csv t1.insert_rows_from_csv('gcutils/tests/test_table.csv') self.assertEqual(sorted(t1.get_rows()), rows) # Test Table.insert_rows_from_query t2 = client.get_table('t2') sql = 'SELECT * FROM {} WHERE a > 1'.format(t1_qname) t2.insert_rows_from_query(sql) self.assertEqual(sorted(t2.get_rows()), rows[1:]) # Test Client.query sql = 'SELECT * FROM {} WHERE a > 2'.format(t1_qname) results = client.query(sql) self.assertEqual(sorted(results.rows), rows[2:]) # Test Client.query_into_dataframe sql = 'SELECT * FROM {} WHERE a > 2'.format(t1_qname) df = client.query_into_dataframe(sql) self.assertEqual(df.values.tolist(), [list(rows[2])]) # Test TableExporter.export_to_storage and # TableExporter.download_from_storage_and_unzip t1_exporter = TableExporter(t1, self.storage_prefix + 'test_table-') t1_exporter.export_to_storage() with tempfile.NamedTemporaryFile(mode='r+') as f: t1_exporter.download_from_storage_and_unzip(f) f.seek(0) reader = csv.reader(f) data = [reader.next()] + sorted(reader) self.assertEqual(data, [map(str, row) for row in [headers] + rows]) # Test Table.insert_rows_from_storage storage_path = self.storage_prefix + 'test_table.csv' self.upload_to_storage('gcutils/tests/test_table.csv', storage_path) t2.insert_rows_from_storage(storage_path) self.assertEqual(sorted(t2.get_rows()), rows) # Test Client.create_storage_backed_table storage_path = self.storage_prefix + 'test_table_headers.csv' self.upload_to_storage( 'gcutils/tests/test_table_headers.csv', storage_path ) schema = [ {'name': 'a', 'type': 'integer'}, {'name': 'b', 'type': 'string'}, ] t3 = client.create_storage_backed_table( 't3', schema, storage_path ) results = client.query('SELECT * FROM {}'.format(t3.qualified_name)) self.assertEqual(sorted(results.rows), rows) self.upload_to_storage( 'gcutils/tests/test_table_headers_2.csv', storage_path ) results = client.query('SELECT * FROM {}'.format(t3.qualified_name)) self.assertEqual(sorted(results.rows), rows + [(4, u'damson')]) # Test Client.create_table_with_view sql = 'SELECT * FROM {{project}}.{} WHERE a > 1'.format(t1_qname) t4 = client.create_table_with_view('t4', sql, False) results = client.query('SELECT * FROM {}'.format(t4.qualified_name)) self.assertEqual(sorted(results.rows), rows[1:]) # Test Client.insert_rows_from_pg PCT.objects.create(code='ABC', name='CCG 1') PCT.objects.create(code='XYZ', name='CCG 2') def transformer(row): return [ord(row[0][0]), row[1]] t1.insert_rows_from_pg(PCT, ['code', 'name'], transformer) self.assertEqual(sorted(t1.get_rows()), [(65, 'CCG 1'), (88, 'CCG 2')]) # Test Table.delete_all_rows t1.delete_all_rows() self.assertEqual(list(t1.get_rows()), [])
def test_the_lot(self): client = Client("test") archive_client = Client("archive") orig_schema = build_schema(("a", "STRING"), ("b", "INTEGER")) schema = build_schema(("a", "INTEGER"), ("b", "STRING")) headers = ["a", "b"] rows = [(1, "apple"), (2, "banana"), (3, "coconut")] t1 = client.get_or_create_table("t1", orig_schema) t1_qname = t1.qualified_name # Test Table.insert_rows_from_csv t1.insert_rows_from_csv("gcutils/tests/test_table.csv", schema) self.assertEqual(sorted(t1.get_rows()), rows) # Test Table.insert_rows_from_query t2 = client.get_table("t2") sql = "SELECT * FROM {} WHERE a > 1".format(t1_qname) t2.insert_rows_from_query(sql) self.assertEqual(sorted(t2.get_rows()), rows[1:]) # Test Client.query sql = "SELECT * FROM {} WHERE a > 2".format(t1_qname) results = client.query(sql) self.assertEqual(sorted(results.rows), rows[2:]) # Test Client.query_into_dataframe sql = "SELECT * FROM {} WHERE a > 2".format(t1_qname) df = client.query_into_dataframe(sql) self.assertEqual(df.values.tolist(), [list(rows[2])]) # Test TableExporter.export_to_storage and # TableExporter.download_from_storage_and_unzip t1_exporter = TableExporter(t1, self.storage_prefix + "test_table-") t1_exporter.export_to_storage() with tempfile.NamedTemporaryFile(mode="r+") as f: t1_exporter.download_from_storage_and_unzip(f) f.seek(0) reader = csv.reader(f) data = [next(reader)] + sorted(reader) self.assertEqual(data, [list(map(str, row)) for row in [headers] + rows]) # Test Table.insert_rows_from_storage storage_path = self.storage_prefix + "test_table.csv" self.upload_to_storage("gcutils/tests/test_table.csv", storage_path) t2.insert_rows_from_storage(storage_path) self.assertEqual(sorted(t2.get_rows()), rows) # Test Client.create_storage_backed_table storage_path = self.storage_prefix + "test_table_headers.csv" self.upload_to_storage("gcutils/tests/test_table_headers.csv", storage_path) schema = build_schema(("a", "INTEGER"), ("b", "STRING")) t3 = client.create_storage_backed_table("t3", schema, storage_path) results = client.query("SELECT * FROM {}".format(t3.qualified_name)) self.assertEqual(sorted(results.rows), rows) self.upload_to_storage("gcutils/tests/test_table_headers_2.csv", storage_path) results = client.query("SELECT * FROM {}".format(t3.qualified_name)) self.assertEqual(sorted(results.rows), rows + [(4, "damson")]) # Test Client.create_table_with_view sql = "SELECT * FROM {{project}}.{} WHERE a > 1".format(t1_qname) t4 = client.create_table_with_view("t4", sql, False) results = client.query("SELECT * FROM {}".format(t4.qualified_name)) self.assertEqual(sorted(results.rows), rows[1:]) # Test Table.copy_to_new_dataset t1.copy_to_new_dataset("archive") t1_archived = archive_client.get_table("t1") self.assertEqual(sorted(t1_archived.get_rows()), rows) self.assertEqual(sorted(t1.get_rows()), rows) # Test Table.move_to_new_dataset t2.move_to_new_dataset("archive") t2_archived = archive_client.get_table("t2") self.assertEqual(sorted(t2_archived.get_rows()), rows) with self.assertRaises(NotFound): list(t2.get_rows()) # Test Client.insert_rows_from_pg PCT.objects.create(code="ABC", name="CCG 1") PCT.objects.create(code="XYZ", name="CCG 2") def transformer(row): return [ord(row[0][0]), row[1]] t1.insert_rows_from_pg( PCT, build_schema(("code", "INTEGER"), ("name", "STRING")), transformer=transformer, ) self.assertEqual(sorted(t1.get_rows()), [(65, "CCG 1"), (88, "CCG 2")]) # Test Table.delete_all_rows t1.delete_all_rows() self.assertEqual(list(t1.get_rows()), [])