def _get_sample_data(db,
database,
table,
column,
is_async=False,
cluster=None,
operation=None):
if operation == 'hello':
table_obj = None
else:
table_obj = db.get_table(database, table)
if table_obj.is_impala_only and db.client.query_server[
'server_name'] != 'impala': # Kudu table, now Hive should support it though
query_server = get_query_server_config('impala', connector=cluster)
db = dbms.get(db.client.user, query_server, cluster=cluster)
sample_data = db.get_sample(database,
table_obj,
column,
generate_sql_only=is_async,
operation=operation)
response = {'status': -1}
if sample_data:
response['status'] = 0
if is_async:
notebook = make_notebook(name=_(
'Table sample for `%(database)s`.`%(table)s`.`%(column)s`') % {
'database': database,
'table': table,
'column': column
},
editor_type=_get_servername(db),
statement=sample_data,
status='ready-execute',
skip_historify=True,
is_task=False,
compute=cluster if cluster else None)
response['result'] = notebook.execute(
request=MockedDjangoRequest(user=db.client.user), batch=False)
if table_obj.is_impala_only:
response['result']['type'] = 'impala'
else:
sample = escape_rows(sample_data.rows(), nulls_only=True)
if column:
sample = set([row[0] for row in sample])
sample = [[item] for item in sorted(list(sample))]
response['headers'] = sample_data.cols()
response['full_headers'] = sample_data.full_cols()
response['rows'] = sample
else:
response['message'] = _('Failed to get sample data.')
return response
def _get_sample_data(db, database, table, column, async=False, cluster=None, operation=None):
table_obj = db.get_table(database, table)
if table_obj.is_impala_only and db.client.query_server['server_name'] != 'impala':
query_server = get_query_server_config('impala', cluster=cluster)
db = dbms.get(db.client.user, query_server, cluster=cluster)
sample_data = db.get_sample(database, table_obj, column, generate_sql_only=async, operation=operation)
response = {'status': -1}
if sample_data:
response['status'] = 0
if async:
notebook = make_notebook(
name=_('Table sample for `%(database)s`.`%(table)s`.`%(column)s`') % {'database': database, 'table': table, 'column': column},
editor_type=_get_servername(db),
statement=sample_data,
status='ready-execute',
skip_historify=True,
is_task=False,
compute=cluster if cluster else None
)
response['result'] = notebook.execute(request=MockedDjangoRequest(user=db.client.user), batch=False)
if table_obj.is_impala_only:
response['result']['type'] = 'impala'
else:
sample = escape_rows(sample_data.rows(), nulls_only=True)
if column:
sample = set([row[0] for row in sample])
sample = [[item] for item in sorted(list(sample))]
def _get_sample_data(db, database, table, column, async=False, cluster=None, operation=None):
table_obj = db.get_table(database, table)
if table_obj.is_impala_only and db.client.query_server['server_name'] != 'impala':
query_server = get_query_server_config('impala', cluster=cluster)
db = dbms.get(db.client.user, query_server, cluster=cluster)
sample_data = db.get_sample(database, table_obj, column, generate_sql_only=async, operation=operation)
response = {'status': -1}
if sample_data:
response['status'] = 0
if async:
notebook = make_notebook(
name=_('Table sample for `%(database)s`.`%(table)s`.`%(column)s`') % {'database': database, 'table': table, 'column': column},
editor_type=_get_servername(db),
statement=sample_data,
status='ready-execute',
skip_historify=True,
is_task=False,
compute=cluster if cluster else None
)
response['result'] = notebook.execute(request=MockedDjangoRequest(user=db.client.user), batch=False)
if table_obj.is_impala_only:
response['result']['type'] = 'impala'
else:
sample = escape_rows(sample_data.rows(), nulls_only=True)
if column:
sample = set([row[0] for row in sample])
sample = [[item] for item in sorted(list(sample))]