def test_get_update_table_unsupported_column_type(
update_table_strategy_unknown):
"""
get_update_table should raise UnsupportedColumnStrategyError if presented with an unsupported column type
"""
with pytest.raises(UnsupportedColumnStrategyError):
query_factory.get_update_table("seed_table",
update_table_strategy_unknown)
def test_get_update_table_literal(literal_strategy):
result_queries = query_factory.get_update_table("seed_table", UpdateColumnsTableStrategy("anon_table", [
LiteralUpdateColumnStrategy("literal_column", "RANDOM()")
]))
assert result_queries == [
"UPDATE anon_table AS \"updatetarget\" SET \"literal_column\" = RANDOM();"
]
def anonymize_database(self, database_strategy):
"""
Anonymize a restored database using the passed database strategy
:param database_strategy: a strategy.DatabaseStrategy configuration
:return:
"""
qualifier_map = database_strategy.fake_update_qualifier_map
if len(qualifier_map) > 0:
self.logger.info("creating seed table with %d columns",
len(qualifier_map))
create_seed_table_sql = query_factory.get_create_seed_table(
SEED_TABLE_NAME, qualifier_map)
self.__runner.db_execute(create_seed_table_sql)
self.logger.info("Inserting seed data")
self.__seed(qualifier_map)
self.__run_scripts(database_strategy.before_scripts, "before")
table_strategies = database_strategy.table_strategies
self.logger.info("Anonymizing %d tables", len(table_strategies))
with tqdm(desc="Anonymizing database",
total=len(table_strategies)) as progressbar:
for table_strategy in table_strategies:
if table_strategy.strategy_type == TableStrategyTypes.TRUNCATE:
progressbar.set_description("Truncating {}".format(
table_strategy.qualified_name))
self.__runner.db_execute(
query_factory.get_truncate_table(table_strategy))
elif table_strategy.strategy_type == TableStrategyTypes.DELETE:
progressbar.set_description("Deleting {}".format(
table_strategy.qualified_name))
self.__runner.db_execute(
query_factory.get_delete_table(table_strategy))
elif table_strategy.strategy_type == TableStrategyTypes.UPDATE_COLUMNS:
progressbar.set_description("Anonymizing {}".format(
table_strategy.qualified_name))
statements = query_factory.get_update_table(
SEED_TABLE_NAME, table_strategy)
self.__runner.db_execute(statements)
else:
raise UnsupportedTableStrategyError(table_strategy)
progressbar.update()
self.__run_scripts(database_strategy.after_scripts, "after")
self.logger.info("dropping seed table")
self.__runner.db_execute(
query_factory.get_drop_seed_table(SEED_TABLE_NAME))
def test_get_update_table_literal(literal_strategy):
result_queries = query_factory.get_update_table(
"seed_table",
UpdateColumnsTableStrategy(
"anon_table", [LiteralUpdateColumnStrategy("literal_column", "RANDOM()")]
),
)
assert result_queries == [
'UPDATE "anon_table" AS "updatetarget" SET "literal_column" = RANDOM();'
]
def test_get_update_table_fake_column(column_strategy_list):
update_table_all = query_factory.get_update_table("seed_table", UpdateColumnsTableStrategy("anon_table", column_strategy_list))
assert update_table_all == [
"UPDATE anon_table AS \"updatetarget\" SET "
"\"test_column1\" = ( SELECT first_name FROM seed_table ORDER BY RANDOM(), MD5(\"updatetarget\"::text) LIMIT 1),"
"\"test_column2\" = ( SELECT last_name FROM seed_table ORDER BY RANDOM(), MD5(\"updatetarget\"::text) LIMIT 1),"
"\"test_column7\" = ( SELECT user_id::UUID FROM seed_table ORDER BY RANDOM(), MD5(\"updatetarget\"::text) LIMIT 1),"
"\"test_column3\" = (''),"
"\"test_column4\" = ( SELECT md5(random()::text) ORDER BY MD5(\"updatetarget\"::text) LIMIT 1),"
"\"test_column5\" = ( SELECT CONCAT(md5(random()::text), '@', md5(random()::text), '.com') ORDER BY MD5(\"updatetarget\"::text) LIMIT 1),"
"\"test_column6\" = RANDOM();"
]
def test_get_update_table_fake_column(column_strategy_list):
update_table_all = query_factory.get_update_table(
"seed_table",
UpdateColumnsTableStrategy("anon_table", column_strategy_list))
assert update_table_all == [
"UPDATE anon_table AS \"updatetarget\" SET "
"test_column1 = ( SELECT first_name FROM seed_table WHERE \"updatetarget\"=\"updatetarget\" ORDER BY RANDOM() LIMIT 1),"
"test_column2 = ( SELECT last_name FROM seed_table WHERE \"updatetarget\"=\"updatetarget\" ORDER BY RANDOM() LIMIT 1),"
"test_column3 = (''),"
"test_column4 = ( SELECT md5(random()::text) WHERE \"updatetarget\"=\"updatetarget\" ),"
"test_column5 = ( SELECT CONCAT(md5(random()::text), '@', md5(random()::text), '.com') WHERE \"updatetarget\"=\"updatetarget\" ),"
"test_column6 = RANDOM();"
]
def test_get_update_table_fake_column(column_strategy_list):
update_table_all = query_factory.get_update_table(
"seed_table", UpdateColumnsTableStrategy("anon_table", column_strategy_list)
)
assert update_table_all == [
'UPDATE "anon_table" AS "updatetarget" SET '
'"test_column1" = ( SELECT "first_name" FROM "seed_table" ORDER BY RANDOM(), MD5("updatetarget"::text) LIMIT 1),'
'"test_column2" = ( SELECT "last_name" FROM "seed_table" ORDER BY RANDOM(), MD5("updatetarget"::text) LIMIT 1),'
'"test_column7" = ( SELECT "user_id"::UUID FROM "seed_table" ORDER BY RANDOM(), MD5("updatetarget"::text) LIMIT 1),'
"\"test_column3\" = (''),"
'"test_column4" = ( SELECT md5(random()::text) ORDER BY MD5("updatetarget"::text) LIMIT 1),'
"\"test_column5\" = ( SELECT CONCAT(md5(random()::text), '@', md5(random()::text), '.com') ORDER BY MD5(\"updatetarget\"::text) LIMIT 1),"
'"test_column6" = RANDOM();'
]