def validate(cls, sql: str, schema: str,
database: Any) -> List[SQLValidationAnnotation]:
"""
Presto supports query-validation queries by running them with a
prepended explain.
For example, "SELECT 1 FROM default.mytable" becomes "EXPLAIN (TYPE
VALIDATE) SELECT 1 FROM default.mytable.
"""
user_name = g.user.username if g.user else None
parsed_query = ParsedQuery(sql)
statements = parsed_query.get_statements()
logger.info(f"Validating {len(statements)} statement(s)")
engine = database.get_sqla_engine(
schema=schema,
nullpool=True,
user_name=user_name,
source=sources.get("sql_lab", None),
)
# Sharing a single connection and cursor across the
# execution of all statements (if many)
annotations: List[SQLValidationAnnotation] = []
with closing(engine.raw_connection()) as conn:
with closing(conn.cursor()) as cursor:
for statement in parsed_query.get_statements():
annotation = cls.validate_statement(
statement, database, cursor, user_name)
if annotation:
annotations.append(annotation)
logger.debug(f"Validation found {len(annotations)} error(s)")
return annotations
def cancel_query(query, user_name):
if not query.connection_id:
return
database = query.database
engine = database.get_sqla_engine(
schema=query.schema,
nullpool=True,
user_name=user_name,
source=sources.get('sql_lab', None),
)
db_engine_spec = database.db_engine_spec
with closing(engine.raw_connection()) as conn:
with closing(conn.cursor()) as cursor:
db_engine_spec.cancel_query(cursor, query)
def cancel_query(query, user_name):
logging.info(
"Query with id `%s` has connection id `%s`", query.id, query.connection_id
)
if not query.connection_id:
logging.info("No connection id found, query cancellation skipped")
return
database = query.database
engine = database.get_sqla_engine(
schema=query.schema,
nullpool=True,
user_name=user_name,
source=sources.get("sql_lab", None),
)
db_engine_spec = database.db_engine_spec
with closing(engine.raw_connection()) as conn:
with closing(conn.cursor()) as cursor:
logging.info("Calling `cancel_query` on db engine")
db_engine_spec.cancel_query(cursor, query)
def execute_sql_statements(
query_id,
rendered_query,
return_results=True,
store_results=False,
user_name=None,
session=None,
start_time=None,
expand_data=False,
): # pylint: disable=too-many-arguments, too-many-locals, too-many-statements
"""Executes the sql query returns the results."""
if store_results and start_time:
# only asynchronous queries
stats_logger.timing("sqllab.query.time_pending",
now_as_float() - start_time)
query = get_query(query_id, session)
payload = dict(query_id=query_id)
database = query.database
db_engine_spec = database.db_engine_spec
db_engine_spec.patch()
if database.allow_run_async and not results_backend:
raise SqlLabException("Results backend isn't configured.")
# Breaking down into multiple statements
parsed_query = ParsedQuery(rendered_query)
statements = parsed_query.get_statements()
logger.info(f"Query {query_id}: Executing {len(statements)} statement(s)")
logger.info(f"Query {query_id}: Set query to 'running'")
query.status = QueryStatus.RUNNING
query.start_running_time = now_as_float()
session.commit()
engine = database.get_sqla_engine(
schema=query.schema,
nullpool=True,
user_name=user_name,
source=sources.get("sql_lab", None),
)
# Sharing a single connection and cursor across the
# execution of all statements (if many)
with closing(engine.raw_connection()) as conn:
with closing(conn.cursor()) as cursor:
statement_count = len(statements)
for i, statement in enumerate(statements):
# Check if stopped
query = get_query(query_id, session)
if query.status == QueryStatus.STOPPED:
return None
# Run statement
msg = f"Running statement {i+1} out of {statement_count}"
logger.info(f"Query {query_id}: {msg}")
query.set_extra_json_key("progress", msg)
session.commit()
try:
cdf = execute_sql_statement(statement, query, user_name,
session, cursor)
except Exception as e: # pylint: disable=broad-except
msg = str(e)
if statement_count > 1:
msg = f"[Statement {i+1} out of {statement_count}] " + msg
payload = handle_query_error(msg, query, session, payload)
return payload
# Success, updating the query entry in database
query.rows = cdf.size
query.progress = 100
query.set_extra_json_key("progress", None)
if query.select_as_cta:
query.select_sql = database.select_star(
query.tmp_table_name,
limit=query.limit,
schema=database.force_ctas_schema,
show_cols=False,
latest_partition=False,
)
query.end_time = now_as_float()
data, selected_columns, all_columns, expanded_columns = _serialize_and_expand_data(
cdf, db_engine_spec, store_results and results_backend_use_msgpack,
expand_data)
payload.update({
"status": QueryStatus.SUCCESS,
"data": data,
"columns": all_columns,
"selected_columns": selected_columns,
"expanded_columns": expanded_columns,
"query": query.to_dict(),
})
payload["query"]["state"] = QueryStatus.SUCCESS
if store_results and results_backend:
key = str(uuid.uuid4())
logger.info(
f"Query {query_id}: Storing results in results backend, key: {key}"
)
with stats_timing("sqllab.query.results_backend_write", stats_logger):
with stats_timing(
"sqllab.query.results_backend_write_serialization",
stats_logger):
serialized_payload = _serialize_payload(
payload, results_backend_use_msgpack)
cache_timeout = database.cache_timeout
if cache_timeout is None:
cache_timeout = config["CACHE_DEFAULT_TIMEOUT"]
compressed = zlib_compress(serialized_payload)
logger.debug(
f"*** serialized payload size: {getsizeof(serialized_payload)}"
)
logger.debug(
f"*** compressed payload size: {getsizeof(compressed)}")
results_backend.set(key, compressed, cache_timeout)
query.results_key = key
query.status = QueryStatus.SUCCESS
session.commit()
if return_results:
return payload
return None
def execute_sql_statements(
ctask,
query_id,
rendered_query,
return_results=True,
store_results=False,
user_name=None,
session=None,
start_time=None,
):
"""Executes the sql query returns the results."""
if store_results and start_time:
# only asynchronous queries
stats_logger.timing("sqllab.query.time_pending", now_as_float() - start_time)
query = get_query(query_id, session)
payload = dict(query_id=query_id)
database = query.database
db_engine_spec = database.db_engine_spec
db_engine_spec.patch()
if store_results and not results_backend:
raise SqlLabException("Results backend isn't configured.")
# Breaking down into multiple statements
parsed_query = ParsedQuery(rendered_query)
statements = parsed_query.get_statements()
logging.info(f"Executing {len(statements)} statement(s)")
logging.info("Set query to 'running'")
query.status = QueryStatus.RUNNING
query.start_running_time = now_as_float()
engine = database.get_sqla_engine(
schema=query.schema,
nullpool=True,
user_name=user_name,
source=sources.get("sql_lab", None),
)
# Sharing a single connection and cursor across the
# execution of all statements (if many)
with closing(engine.raw_connection()) as conn:
with closing(conn.cursor()) as cursor:
query.connection_id = db_engine_spec.get_connection_id(cursor)
session.commit()
statement_count = len(statements)
for i, statement in enumerate(statements):
# check if the query was stopped
session.refresh(query)
if query.status == QueryStatus.STOPPED:
payload.update({"status": query.status})
return payload
msg = f"Running statement {i+1} out of {statement_count}"
logging.info(msg)
query.set_extra_json_key("progress", msg)
session.commit()
try:
cdf = execute_sql_statement(
statement, query, user_name, session, cursor
)
msg = f"Running statement {i+1} out of {statement_count}"
except Exception as e:
# query can be stopped in another thread/worker
# but in synchronized mode it may lead to an error
# skip error the error in such case
session.refresh(query)
if query.status == QueryStatus.STOPPED:
payload.update({"status": query.status})
return payload
msg = str(e)
if statement_count > 1:
msg = f"[Statement {i+1} out of {statement_count}] " + msg
payload = handle_query_error(msg, query, session, payload)
return payload
# Success, updating the query entry in database
query.rows = cdf.size
query.progress = 100
query.set_extra_json_key("progress", None)
query.connection_id = None
if query.select_as_cta:
query.select_sql = database.select_star(
query.tmp_table_name,
limit=query.limit,
schema=database.force_ctas_schema,
show_cols=False,
latest_partition=False,
)
query.end_time = now_as_float()
selected_columns = cdf.columns or []
data = cdf.data or []
all_columns, data, expanded_columns = db_engine_spec.expand_data(
selected_columns, data
)
payload.update(
{
"status": QueryStatus.SUCCESS,
"data": data,
"columns": all_columns,
"selected_columns": selected_columns,
"expanded_columns": expanded_columns,
"query": query.to_dict(),
}
)
payload["query"]["state"] = QueryStatus.SUCCESS
# go over each row, find bytes columns that start with the magic UAST
# sequence b'\x00bgr', and replace it with a string containing the
# UAST in JSON
for row in payload["data"]:
for k, v in row.items():
if isinstance(v, bytes) and len(v) > 4 and v[0:4] == b"\x00bgr":
try:
ctx = decode(v, format=0)
row[k] = json.dumps(ctx.load())
except Exception:
pass
if store_results:
key = str(uuid.uuid4())
logging.info(f"Storing results in results backend, key: {key}")
with stats_timing("sqllab.query.results_backend_write", stats_logger):
json_payload = json.dumps(
payload, default=json_iso_dttm_ser, ignore_nan=True
)
cache_timeout = database.cache_timeout
if cache_timeout is None:
cache_timeout = config.get("CACHE_DEFAULT_TIMEOUT", 0)
results_backend.set(key, zlib_compress(json_payload), cache_timeout)
query.results_key = key
query.status = QueryStatus.SUCCESS
session.commit()
if return_results:
return payload
def execute_sql_statements(
ctask,
query_id,
rendered_query,
return_results=True,
store_results=False,
user_name=None,
session=None,
start_time=None,
):
"""Executes the sql query returns the results."""
if store_results and start_time:
# only asynchronous queries
stats_logger.timing('sqllab.query.time_pending',
now_as_float() - start_time)
query = get_query(query_id, session)
payload = dict(query_id=query_id)
database = query.database
db_engine_spec = database.db_engine_spec
db_engine_spec.patch()
if store_results and not results_backend:
raise SqlLabException("Results backend isn't configured.")
# Breaking down into multiple statements
parsed_query = ParsedQuery(rendered_query)
statements = parsed_query.get_statements()
logging.info(f'Executing {len(statements)} statement(s)')
logging.info("Set query to 'running'")
query.status = QueryStatus.RUNNING
query.start_running_time = now_as_float()
engine = database.get_sqla_engine(
schema=query.schema,
nullpool=True,
user_name=user_name,
source=sources.get('sql_lab', None),
)
# Sharing a single connection and cursor across the
# execution of all statements (if many)
with closing(engine.raw_connection()) as conn:
with closing(conn.cursor()) as cursor:
statement_count = len(statements)
for i, statement in enumerate(statements):
# TODO CHECK IF STOPPED
msg = f'Running statement {i+1} out of {statement_count}'
logging.info(msg)
query.set_extra_json_key('progress', msg)
session.commit()
try:
cdf = execute_sql_statement(statement, query, user_name,
session, cursor)
msg = f'Running statement {i+1} out of {statement_count}'
except Exception as e:
msg = str(e)
if statement_count > 1:
msg = f'[Statement {i+1} out of {statement_count}] ' + msg
payload = handle_query_error(msg, query, session, payload)
return payload
# Success, updating the query entry in database
query.rows = cdf.size
query.progress = 100
query.set_extra_json_key('progress', None)
query.status = QueryStatus.SUCCESS
if query.select_as_cta:
query.select_sql = database.select_star(
query.tmp_table_name,
limit=query.limit,
schema=database.force_ctas_schema,
show_cols=False,
latest_partition=False)
query.end_time = now_as_float()
payload.update({
'status': query.status,
'data': cdf.data if cdf.data else [],
'columns': cdf.columns if cdf.columns else [],
'query': query.to_dict(),
})
if store_results:
key = str(uuid.uuid4())
logging.info(f'Storing results in results backend, key: {key}')
with stats_timing('sqllab.query.results_backend_write', stats_logger):
json_payload = json.dumps(payload,
default=json_iso_dttm_ser,
ignore_nan=True)
cache_timeout = database.cache_timeout
if cache_timeout is None:
cache_timeout = config.get('CACHE_DEFAULT_TIMEOUT', 0)
results_backend.set(key, zlib_compress(json_payload),
cache_timeout)
query.results_key = key
session.commit()
if return_results:
return payload
def execute_sql_statements(
ctask, query_id, rendered_query, return_results=True, store_results=False,
user_name=None, session=None, start_time=None,
):
"""Executes the sql query returns the results."""
if store_results and start_time:
# only asynchronous queries
stats_logger.timing(
'sqllab.query.time_pending', now_as_float() - start_time)
query = get_query(query_id, session)
payload = dict(query_id=query_id)
database = query.database
db_engine_spec = database.db_engine_spec
db_engine_spec.patch()
if store_results and not results_backend:
raise SqlLabException("Results backend isn't configured.")
# Breaking down into multiple statements
parsed_query = ParsedQuery(rendered_query)
statements = parsed_query.get_statements()
logging.info(f'Executing {len(statements)} statement(s)')
logging.info("Set query to 'running'")
query.status = QueryStatus.RUNNING
query.start_running_time = now_as_float()
engine = database.get_sqla_engine(
schema=query.schema,
nullpool=True,
user_name=user_name,
source=sources.get('sql_lab', None),
)
# Sharing a single connection and cursor across the
# execution of all statements (if many)
with closing(engine.raw_connection()) as conn:
with closing(conn.cursor()) as cursor:
statement_count = len(statements)
for i, statement in enumerate(statements):
# TODO CHECK IF STOPPED
msg = f'Running statement {i+1} out of {statement_count}'
logging.info(msg)
query.set_extra_json_key('progress', msg)
session.commit()
is_last_statement = i == len(statements) - 1
try:
cdf = execute_sql_statement(
statement, query, user_name, session, cursor,
return_results=is_last_statement and return_results)
msg = f'Running statement {i+1} out of {statement_count}'
except Exception as e:
msg = str(e)
if statement_count > 1:
msg = f'[Statement {i+1} out of {statement_count}] ' + msg
payload = handle_query_error(msg, query, session, payload)
return payload
# Success, updating the query entry in database
query.rows = cdf.size
query.progress = 100
query.set_extra_json_key('progress', None)
query.status = QueryStatus.SUCCESS
if query.select_as_cta:
query.select_sql = database.select_star(
query.tmp_table_name,
limit=query.limit,
schema=database.force_ctas_schema,
show_cols=False,
latest_partition=False)
query.end_time = now_as_float()
session.commit()
payload.update({
'status': query.status,
'data': cdf.data if cdf.data else [],
'columns': cdf.columns if cdf.columns else [],
'query': query.to_dict(),
})
if store_results:
key = str(uuid.uuid4())
logging.info(f'Storing results in results backend, key: {key}')
with stats_timing('sqllab.query.results_backend_write', stats_logger):
json_payload = json.dumps(
payload, default=json_iso_dttm_ser, ignore_nan=True)
cache_timeout = database.cache_timeout
if cache_timeout is None:
cache_timeout = config.get('CACHE_DEFAULT_TIMEOUT', 0)
results_backend.set(key, zlib_compress(json_payload), cache_timeout)
query.results_key = key
session.commit()
if return_results:
return payload
def execute_sql(
ctask, query_id, rendered_query, return_results=True, store_results=False,
user_name=None, session=None, start_time=None,
):
"""Executes the sql query returns the results."""
if store_results and start_time:
# only asynchronous queries
stats_logger.timing(
'sqllab.query.time_pending', now_as_float() - start_time)
query = get_query(query_id, session)
payload = dict(query_id=query_id)
database = query.database
db_engine_spec = database.db_engine_spec
db_engine_spec.patch()
def handle_error(msg):
"""Local method handling error while processing the SQL"""
troubleshooting_link = config['TROUBLESHOOTING_LINK']
query.error_message = msg
query.status = QueryStatus.FAILED
query.tmp_table_name = None
session.commit()
payload.update({
'status': query.status,
'error': msg,
})
if troubleshooting_link:
payload['link'] = troubleshooting_link
return payload
if store_results and not results_backend:
return handle_error("Results backend isn't configured.")
# Limit enforced only for retrieving the data, not for the CTA queries.
superset_query = SupersetQuery(rendered_query)
executed_sql = superset_query.stripped()
SQL_MAX_ROWS = app.config.get('SQL_MAX_ROW')
if not superset_query.is_readonly() and not database.allow_dml:
return handle_error(
'Only `SELECT` statements are allowed against this database')
if query.select_as_cta:
if not superset_query.is_select():
return handle_error(
'Only `SELECT` statements can be used with the CREATE TABLE '
'feature.')
if not query.tmp_table_name:
start_dttm = datetime.fromtimestamp(query.start_time)
query.tmp_table_name = 'tmp_{}_table_{}'.format(
query.user_id, start_dttm.strftime('%Y_%m_%d_%H_%M_%S'))
executed_sql = superset_query.as_create_table(query.tmp_table_name)
query.select_as_cta_used = True
if superset_query.is_select():
if SQL_MAX_ROWS and (not query.limit or query.limit > SQL_MAX_ROWS):
query.limit = SQL_MAX_ROWS
if query.limit:
executed_sql = database.apply_limit_to_sql(executed_sql, query.limit)
# Hook to allow environment-specific mutation (usually comments) to the SQL
SQL_QUERY_MUTATOR = config.get('SQL_QUERY_MUTATOR')
if SQL_QUERY_MUTATOR:
executed_sql = SQL_QUERY_MUTATOR(
executed_sql, user_name, security_manager, database)
query.executed_sql = executed_sql
query.status = QueryStatus.RUNNING
query.start_running_time = now_as_float()
session.merge(query)
session.commit()
logging.info("Set query to 'running'")
conn = None
try:
engine = database.get_sqla_engine(
schema=query.schema,
nullpool=True,
user_name=user_name,
source=sources.get('sql_lab', None),
)
conn = engine.raw_connection()
cursor = conn.cursor()
logging.info('Running query: \n{}'.format(executed_sql))
logging.info(query.executed_sql)
query_start_time = now_as_float()
db_engine_spec.execute(cursor, query.executed_sql, async_=True)
logging.info('Handling cursor')
db_engine_spec.handle_cursor(cursor, query, session)
logging.info('Fetching data: {}'.format(query.to_dict()))
stats_logger.timing(
'sqllab.query.time_executing_query',
now_as_float() - query_start_time)
fetching_start_time = now_as_float()
data = db_engine_spec.fetch_data(cursor, query.limit)
stats_logger.timing(
'sqllab.query.time_fetching_results',
now_as_float() - fetching_start_time)
except SoftTimeLimitExceeded as e:
logging.exception(e)
if conn is not None:
conn.close()
return handle_error(
"SQL Lab timeout. This environment's policy is to kill queries "
'after {} seconds.'.format(SQLLAB_TIMEOUT))
except Exception as e:
logging.exception(e)
if conn is not None:
conn.close()
return handle_error(db_engine_spec.extract_error_message(e))
logging.info('Fetching cursor description')
cursor_description = cursor.description
if conn is not None:
conn.commit()
conn.close()
if query.status == QueryStatus.STOPPED:
return handle_error('The query has been stopped')
logging.debug('Cursor description: %s', cursor_description)
cdf = dataframe.SupersetDataFrame(data, cursor_description, db_engine_spec)
query.rows = cdf.size
query.progress = 100
query.status = QueryStatus.SUCCESS
if query.select_as_cta:
query.select_sql = '{}'.format(
database.select_star(
query.tmp_table_name,
limit=query.limit,
schema=database.force_ctas_schema,
show_cols=False,
latest_partition=False))
query.end_time = now_as_float()
session.merge(query)
session.flush()
payload.update({
'status': query.status,
'data': cdf.data if cdf.data else [],
'columns': cdf.columns if cdf.columns else [],
'query': query.to_dict(),
})
if store_results:
key = '{}'.format(uuid.uuid4())
logging.info('Storing results in results backend, key: {}'.format(key))
write_to_results_backend_start = now_as_float()
json_payload = json.dumps(
payload, default=json_iso_dttm_ser, ignore_nan=True)
cache_timeout = database.cache_timeout
if cache_timeout is None:
cache_timeout = config.get('CACHE_DEFAULT_TIMEOUT', 0)
results_backend.set(key, zlib_compress(json_payload), cache_timeout)
query.results_key = key
stats_logger.timing(
'sqllab.query.results_backend_write',
now_as_float() - write_to_results_backend_start)
session.merge(query)
session.commit()
if return_results:
return payload
