def run(self, events_table):
if not events_table:
raise ValueError(
dedent('''Section: events_table -
Section not found. You must define an events table.'''))
table_should_have_data(events_table, self.db_engine)
column_should_be_intlike(events_table, 'entity_id', self.db_engine)
column_should_be_timelike(events_table, 'outcome_date', self.db_engine)
column_should_be_booleanlike(events_table, 'outcome', self.db_engine)
def validate(self, db_engine):
logging.info('Validating from_obj %s', self.materialized_table)
table_should_exist(self.materialized_table, db_engine)
logging.info('Table %s successfully found', self.materialized_table)
table_should_have_column(self.materialized_table, 'entity_id',
db_engine)
logging.info('Successfully found entity_id column in %s',
self.materialized_table)
table_should_have_column(self.materialized_table,
self.knowledge_date_column, db_engine)
column_should_be_timelike(self.materialized_table,
self.knowledge_date_column, db_engine)
logging.info(
'Successfully found configured knowledge date column in %s',
self.materialized_table)
def validate(self, db_engine):
logger.spam(f"Validating from_obj {self.materialized_table}")
table_should_exist(self.materialized_table, db_engine)
logger.spam(f"Table {self.materialized_table} successfully found")
table_should_have_column(self.materialized_table, 'entity_id',
db_engine)
logger.spam(
f"Successfully found entity_id column in {self.materialized_table}"
)
table_should_have_column(self.materialized_table,
self.knowledge_date_column, db_engine)
column_should_be_timelike(self.materialized_table,
self.knowledge_date_column, db_engine)
logger.spam(
f"Successfully found configured knowledge date column in {self.materialized_table}"
)
def run(self, state_config):
if 'table_name' in state_config:
dense_state_table = state_config['table_name']
table_should_have_data(dense_state_table, self.db_engine)
column_should_be_intlike(dense_state_table, 'entity_id',
self.db_engine)
column_should_be_stringlike(dense_state_table, 'state',
self.db_engine)
column_should_be_timelike(dense_state_table, 'start_time',
self.db_engine)
column_should_be_timelike(dense_state_table, 'end_time',
self.db_engine)
if 'state_filters' not in state_config or len(
state_config['state_filters']) < 1:
raise ValueError(
dedent('''Section: state_config -
If a table_name is given in state_config,
at least one state filter must be present'''))
else:
logging.warning('No table_name found in state_config.' +
'The provided events table will be used, which ' +
'may result in unnecessarily large matrices')
def _run(self, cohort_config):
mutex_keys = set(['dense_states', 'entities_table', 'query'])
available_keys = mutex_keys | set(['name'])
used_keys = set(cohort_config.keys())
bad_keys = used_keys - available_keys
if bad_keys:
raise ValueError(
dedent('''
Section: cohort_config -
The following given keys: '{}'
are invalid. Available keys are: '{}'
'''.format(bad_keys, available_keys)))
used_mutex_keys = mutex_keys & used_keys
if len(used_mutex_keys) > 1:
raise ValueError(
dedent('''
Section: cohort_config -
Only one of the following keys can be sent: '{}'
Found '{}'
'''.format(mutex_keys, used_mutex_keys)))
if 'dense_states' in cohort_config:
state_config = cohort_config['dense_states']
if 'table_name' not in state_config:
raise ValueError(
dedent('''
Section: cohort_config -
If 'dense_states' is used as cohort config,
a table name must be present'''))
dense_state_table = state_config['table_name']
table_should_have_data(dense_state_table, self.db_engine)
column_should_be_intlike(dense_state_table, 'entity_id',
self.db_engine)
column_should_be_stringlike(dense_state_table, 'state',
self.db_engine)
column_should_be_timelike(dense_state_table, 'start_time',
self.db_engine)
column_should_be_timelike(dense_state_table, 'end_time',
self.db_engine)
if 'state_filters' not in state_config or len(
state_config['state_filters']) < 1:
raise ValueError(
dedent('''
Section: cohort_config -
If 'dense_states' is used as cohort config,
at least one state filter must be present'''))
elif 'entities_table' in cohort_config:
entities_table = cohort_config['entities_table']
table_should_have_data(entities_table, self.db_engine)
column_should_be_intlike(entities_table, 'entity_id',
self.db_engine)
elif 'query' in cohort_config:
query = cohort_config['query']
if '{as_of_date}' not in query:
raise ValueError(
dedent('''
Section: cohort_config -
If 'query' is used as cohort_config,
{as_of_date} must be present'''))
dated_query = query.replace('{as_of_date}',
'2016-01-01::timestamp')
conn = self.db_engine.connect()
logging.info('Validating cohort query')
try:
conn.execute('explain {}'.format(dated_query))
except Exception as e:
raise ValueError(
dedent('''
Section: cohort_config -
given query can not be run with a sample as_of_date .
query: "{}"
Full error: {}'''.format(query, e)))
def _run(self, cohort_config):
mutex_keys = set(["dense_states", "entities_table", "query"])
available_keys = mutex_keys | set(["name"])
used_keys = set(cohort_config.keys())
bad_keys = used_keys - available_keys
if bad_keys:
raise ValueError(
dedent("""
Section: cohort_config -
The following given keys: '{}'
are invalid. Available keys are: '{}'
""".format(bad_keys, available_keys)))
used_mutex_keys = mutex_keys & used_keys
if len(used_mutex_keys) > 1:
raise ValueError(
dedent("""
Section: cohort_config -
Only one of the following keys can be sent: '{}'
Found '{}'
""".format(mutex_keys, used_mutex_keys)))
if "dense_states" in cohort_config:
state_config = cohort_config["dense_states"]
if "table_name" not in state_config:
raise ValueError(
dedent("""
Section: cohort_config -
If 'dense_states' is used as cohort config,
a table name must be present"""))
dense_state_table = state_config["table_name"]
table_should_have_data(dense_state_table, self.db_engine)
column_should_be_intlike(dense_state_table, "entity_id",
self.db_engine)
column_should_be_stringlike(dense_state_table, "state",
self.db_engine)
column_should_be_timelike(dense_state_table, "start_time",
self.db_engine)
column_should_be_timelike(dense_state_table, "end_time",
self.db_engine)
if ("state_filters" not in state_config
or len(state_config["state_filters"]) < 1):
raise ValueError(
dedent("""
Section: cohort_config -
If 'dense_states' is used as cohort config,
at least one state filter must be present"""))
elif "entities_table" in cohort_config:
entities_table = cohort_config["entities_table"]
table_should_have_data(entities_table, self.db_engine)
column_should_be_intlike(entities_table, "entity_id",
self.db_engine)
elif "query" in cohort_config:
query = cohort_config["query"]
if "{as_of_date}" not in query:
raise ValueError(
dedent("""
Section: cohort_config -
If 'query' is used as cohort_config,
{as_of_date} must be present"""))
dated_query = query.replace("{as_of_date}", "2016-01-01")
conn = self.db_engine.connect()
logging.info("Validating cohort query")
try:
conn.execute(f"explain {dated_query}")
except Exception as e:
raise ValueError(
dedent(f"""
Section: cohort_config -
given query can not be run with a sample as_of_date .
query: "{query}"
Full error: {e}"""))
