def get_partitions(self, db_name, table, partition_spec=None, max_parts=None, reverse_sort=True):
if max_parts is None or max_parts > LIST_PARTITIONS_LIMIT.get():
max_parts = LIST_PARTITIONS_LIMIT.get()
return self.client.get_partitions(
db_name, table.name, partition_spec, max_parts=max_parts, reverse_sort=reverse_sort
)
def get_partitions(self,
db_name,
table,
partition_spec=None,
max_parts=None,
reverse_sort=True):
if max_parts is None or max_parts > LIST_PARTITIONS_LIMIT.get():
max_parts = LIST_PARTITIONS_LIMIT.get()
return self.client.get_partitions(db_name,
table.name,
partition_spec,
max_parts=max_parts,
reverse_sort=reverse_sort)
def get_partitions(self,
database,
table_name,
partition_spec=None,
max_parts=None,
reverse_sort=True):
table = self.get_table(database, table_name)
query = 'SHOW PARTITIONS `%s`.`%s`' % (database, table_name)
if partition_spec:
query += ' PARTITION(%s)' % partition_spec
# We fetch N partitions then reverse the order later and get the max_parts. Use partition_spec to refine more the initial list.
# Need to fetch more like this until SHOW PARTITIONS offers a LIMIT and ORDER BY
partition_table = self.execute_query_statement(query, max_rows=10000)
partitions = [
PartitionValueCompatible(partition, table)
for partition in partition_table.rows()
]
if reverse_sort:
partitions.reverse()
if max_parts is None or max_parts <= 0:
max_parts = LIST_PARTITIONS_LIMIT.get()
return partitions[:max_parts]
def get_partitions(self, database, table_name, partition_spec=None, max_parts=None, reverse_sort=True):
table = self.get_table(database, table_name)
query = 'SHOW PARTITIONS `%s`.`%s`' % (database, table_name)
if self.query_server['server_name'] == 'beeswax' and partition_spec:
query += ' PARTITION(%s)' % partition_spec
# We fetch N partitions then reverse the order later and get the max_parts. Use partition_spec to refine more the initial list.
# Need to fetch more like this until SHOW PARTITIONS offers a LIMIT and ORDER BY
partition_table = self.execute_query_statement(query, max_rows=10000, orientation=TFetchOrientation.FETCH_NEXT, close_operation=True)
if self.query_server['server_name'].startswith('impala'):
try:
# Fetch all partition key names, which are listed before the #Rows column
cols = [col.name for col in partition_table.cols()]
stop = cols.index('#Rows')
partition_keys = cols[:stop]
num_parts = len(partition_keys)
# Get all partition values
rows = partition_table.rows()
partition_values = [partition[:num_parts] for partition in rows]
# Truncate last row which is the Total
partition_values = partition_values[:-1]
partitions_formatted = []
# Format partition key and values into Hive format: [key1=val1/key2=value2]
for values in partition_values:
zipped_parts = izip(partition_keys, values)
partitions_formatted.append(['/'.join(['%s=%s' % (str(part[0]), str(part[1])) for part in zipped_parts if all(part)])])
partitions = [PartitionValueCompatible(partition, table) for partition in partitions_formatted]
except Exception:
raise ValueError(_('Failed to determine partition keys for Impala table: `%s`.`%s`') % (database, table_name))
else:
partitions = [PartitionValueCompatible(partition, table) for partition in partition_table.rows()]
if reverse_sort:
partitions.reverse()
if max_parts is None or max_parts <= 0:
max_parts = LIST_PARTITIONS_LIMIT.get()
return partitions[:max_parts]
def get_partitions(self, database, table_name, partition_spec=None, max_parts=None, reverse_sort=True):
table = self.get_table(database, table_name)
query = 'SHOW PARTITIONS `%s`.`%s`' % (database, table_name)
if partition_spec:
query += ' PARTITION(%s)' % partition_spec
# We fetch N partitions then reverse the order later and get the max_parts. Use partition_spec to refine more the initial list.
# Need to fetch more like this until SHOW PARTITIONS offers a LIMIT and ORDER BY
partition_table = self.execute_query_statement(query, max_rows=10000)
partitions = [PartitionValueCompatible(partition, table) for partition in partition_table.rows()]
if reverse_sort:
partitions.reverse()
if max_parts is None or max_parts <= 0:
max_parts = LIST_PARTITIONS_LIMIT.get()
return partitions[:max_parts]
# Format partition key and values into Hive format: [key1=val1/key2=value2]
for values in partition_values:
zipped_parts = izip(partition_keys, values)
partitions_formatted.append(['/'.join(['%s=%s' % (str(part[0]), str(part[1])) for part in zipped_parts if all(part)])])
partitions = [PartitionValueCompatible(partition, table) for partition in partitions_formatted]
except Exception, e:
raise ValueError(_('Failed to determine partition keys for Impala table: `%s`.`%s`') % (database, table_name))
else:
partitions = [PartitionValueCompatible(partition, table) for partition in partition_table.rows()]
if reverse_sort:
partitions.reverse()
if max_parts is None or max_parts <= 0:
max_parts = LIST_PARTITIONS_LIMIT.get()
return partitions[:max_parts]
def get_configuration(self):
configuration = {}
if self.query_server['server_name'] == 'impala': # Return all configuration settings
query = 'SET'
results = self.execute_query_statement(query, orientation=TFetchOrientation.FETCH_NEXT, close_operation=True)
configuration = dict((row[0], row[1]) for row in results.rows())
else: # For Hive, only return white-listed configurations
query = 'SET -v'
results = self.execute_query_statement(query, orientation=TFetchOrientation.FETCH_FIRST, max_rows=-1, close_operation=True)
config_whitelist = [config.lower() for config in CONFIG_WHITELIST.get()]
# Format partition key and values into Hive format: [key1=val1/key2=value2]
for values in partition_values:
zipped_parts = izip(partition_keys, values)
partitions_formatted.append(['/'.join(['%s=%s' % (part[0], part[1]) for part in zipped_parts])])
partitions = [PartitionValueCompatible(partition, table) for partition in partitions_formatted]
except Exception, e:
raise ValueError(_('Failed to determine partition keys for Impala table: `%s`.`%s`') % (database, table_name))
else:
partitions = [PartitionValueCompatible(partition, table) for partition in partition_table.rows()]
if reverse_sort:
partitions.reverse()
if max_parts is None or max_parts <= 0:
max_parts = LIST_PARTITIONS_LIMIT.get()
return partitions[:max_parts]
def get_configuration(self):
configuration = {}
if self.query_server['server_name'] == 'impala': # Return all configuration settings
query = 'SET'
results = self.execute_query_statement(query, orientation=TFetchOrientation.FETCH_NEXT)
configuration = dict((row[0], row[1]) for row in results.rows())
else: # For Hive, only return white-listed configurations
query = 'SET -v'
results = self.execute_query_statement(query, orientation=TFetchOrientation.FETCH_FIRST)
config_whitelist = [config.lower() for config in CONFIG_WHITELIST.get()]