def get_table_schema(self, table_name, database_name=None):
if database_name is None:
database_name = '.*'
def op():
self._last_operation_string = "RPC_DESCRIBE_TABLE"
self._last_operation_handle = impala.rpc.get_table_schema(
self.service, self.session_handle, table_name, database_name)
self._execute_sync(op)
results = self.fetchall()
if len(results) == 0:
# TODO: the error raised here should be different
raise RPCError("no schema results for table %s.%s" %
(database_name, table_name))
# check that results are derived from a unique table
tables = set()
for col in results:
tables.add((col[1], col[2]))
if len(tables) > 1:
# TODO: the error raised here should be different
raise RPCError("db: %s, table: %s is not unique" %
(database_name, table_name))
return [(r[3], impala.rpc._PrimitiveType_to_TTypeId[r[5]])
for r in results]
def __do_rpc(rpc):
"""Executes the provided callable."""
# if not self.connected:
# raise DisconnectedError(
# "Not connected (use CONNECT to establish a connection)")
# return None, RpcStatus.ERROR
try:
ret = rpc()
status = RpcStatus.OK
# TODO: In the future more advanced error detection/handling can be
# done based on the TStatus return value. For now, just print any
# error(s) that were encountered and validate the result of the
# operation was a success.
if ret is not None and isinstance(ret, TStatus):
if ret.status_code != TStatusCode.OK:
print((ret.error_msgs))
if ret.error_msgs:
raise RPCError('RPC Error: %s' % '\n'.join(ret.error_msgs))
status = RpcStatus.ERROR
return ret, status
except BeeswaxService.QueryNotFoundException:
raise QueryStateError('Error: Stale query handle')
# beeswaxException prints out the entire object, printing
# just the message is far more readable/helpful.
except BeeswaxService.BeeswaxException as b:
raise RPCError("ERROR: %s" % (b.message))
except TTransportException as e:
# issue with the connection with the impalad
raise DisconnectedError("Error communicating with impalad: %s" % e)
except TApplicationException as t:
raise RPCError("Application Exception : %s" % (t))
return None, RpcStatus.ERROR
def close_insert(service, last_query_handle):
"""Fetches the results of an INSERT query"""
rpc_result = __do_rpc(lambda: service.CloseInsert(last_query_handle))
insert_result, status = rpc_result
if status != RpcStatus.OK:
raise RPCError()
num_rows = sum([int(k) for k in insert_result.rows_appended.values()])
return num_rows
def build_default_query_options_dict(service):
# The default query options are retrieved from a hs2_client call, and are dependent
# on the impalad to which a connection has been established. They need to be
# refreshed each time a connection is made. This is particularly helpful when
# there is a version mismatch between the shell and the impalad.
try:
get_default_query_options = service.get_default_configuration(False)
except:
return {}
rpc_result = __do_rpc(lambda: get_default_query_options)
options, status = rpc_result
if status != RpcStatus.OK:
raise RPCError("Unable to retrieve default query options")
return options
def fetch_internal(service, last_query_handle, buffer_size):
"""Fetch all the results.
This function serves a generator to create an iterable of the results.
Result rows are passed to the shell."""
result_rows = []
while True:
rpc_result = __do_rpc(
lambda: service.fetch(last_query_handle, False, buffer_size))
result, status = rpc_result
if status != RpcStatus.OK:
raise RPCError()
result_rows.extend(result.data)
if len(result_rows) >= buffer_size or not result.has_more:
rows = [row.split('\t') for row in result_rows]
return rows
def wrapper(*args, **kwargs):
# get the thrift transport
if 'service' in kwargs:
transport = kwargs['service']._iprot.trans
elif len(args) > 0 and isinstance(args[0], TCLIService.Client):
transport = args[0]._iprot.trans
else:
raise RPCError("RPC function does not have expected 'service' arg")
tries_left = 3
while tries_left > 0:
try:
if not transport.isOpen():
transport.open()
return func(*args, **kwargs)
except socket.error as e:
pass
except TTransportException as e:
pass
except Exception as e:
raise
transport.close()
tries_left -= 1
raise
def execute_statement(service, query):
rpc_result = __do_rpc(lambda: service.query(query))
last_query_handle, status = rpc_result
if status != RpcStatus.OK:
raise RPCError("Error executing the query")
return last_query_handle
print(ret.error_msgs)
if ret.error_msgs:
raise RPCError('RPC Error: %s' % '\n'.join(ret.error_msgs))
status = RpcStatus.ERROR
return ret, status
except BeeswaxService.QueryNotFoundException:
raise QueryStateError('Error: Stale query handle')
# beeswaxException prints out the entire object, printing
# just the message is far more readable/helpful.
except BeeswaxService.BeeswaxException, b:
raise RPCError("ERROR: %s" % (b.message))
except TTransportException, e:
# issue with the connection with the impalad
raise DisconnectedError("Error communicating with impalad: %s" % e)
except TApplicationException, t:
raise RPCError("Application Exception : %s" % (t))
return None, RpcStatus.ERROR
def get_column_names(service, last_query_handle):
rpc_result = __do_rpc(
lambda: service.get_results_metadata(last_query_handle))
metadata, _ = rpc_result
if not metadata is None:
return [fs.name for fs in metadata.schema.fieldSchemas]
def get_results_metadata(service, last_query_handle):
rpc_result = __do_rpc(
lambda: service.get_results_metadata(last_query_handle))
metadata, _ = rpc_result
