def test_add_sessions_command_parses():
# Do not skip and python
add_sessions_mock = MagicMock()
spark_controller.add_session = add_sessions_mock
command = "add"
name = "-s name"
language = "-l python"
connection_string = "-u http://url.com -a sdf -p w"
line = " ".join([command, name, language, connection_string])
magic.spark(line)
add_sessions_mock.assert_called_once_with(
"name", Endpoint("http://url.com", "sdf", "w"), False,
{"kind": "pyspark"})
# Skip and scala - upper case
add_sessions_mock = MagicMock()
spark_controller.add_session = add_sessions_mock
command = "add"
name = "-s name"
language = "-l scala"
connection_string = "--url http://location:port"
line = " ".join([command, name, language, connection_string, "-k"])
magic.spark(line)
add_sessions_mock.assert_called_once_with("name",
Endpoint("http://location:port"),
True, {"kind": "spark"})
def test_cleanup_endpoint_command_parses():
mock_method = MagicMock()
spark_controller.cleanup_endpoint = mock_method
line = "cleanup -u endp"
magic.spark(line)
mock_method.assert_called_once_with(Endpoint("endp"))
line = "cleanup -u endp -a user -p passw"
magic.spark(line)
mock_method.assert_called_with(Endpoint("endp", "user", "passw"))
def refresh_configuration(self):
credentials = getattr(conf,
'kernel_' + self.language + '_credentials')()
(username, password,
url) = (credentials['username'], credentials['password'],
credentials['url'])
self.endpoint = Endpoint(url, username, password)
def test_change_language():
language = constants.LANG_SCALA.upper()
line = "-l {}".format(language)
magic._do_not_call_change_language(line)
assert_equals(constants.LANG_SCALA, magic.language)
assert_equals(Endpoint("new_url"), magic.endpoint)
def test_change_language_not_valid():
language = "not_valid"
line = "-l {}".format(language)
magic._do_not_call_change_language(line)
assert_equals(ipython_display.send_error.call_count, 1)
assert_equals(constants.LANG_PYTHON, magic.language)
assert_equals(Endpoint("url"), magic.endpoint)
def run(self):
endpoint = Endpoint(self.address_widget.value, self.user_widget.value,
self.password_widget.value)
self.endpoints[self.address_widget.value] = endpoint
self.ipython_display.writeln("Added endpoint {}".format(
self.address_widget.value))
# We need to call the refresh method because drop down in Tab 2 for endpoints wouldn't refresh with the new
# value otherwise.
self.refresh_method()
def test_delete_sessions_command_parses():
mock_method = MagicMock()
spark_controller.delete_session_by_name = mock_method
command = "delete -s name"
magic.spark(command)
mock_method.assert_called_once_with("name")
command = "delete -u URL -a username -p password -i 4"
mock_method = MagicMock()
spark_controller.delete_session_by_id = mock_method
magic.spark(command)
mock_method.assert_called_once_with(
Endpoint("URL", "username", "password"), 4)
def test_add_session():
name = "name"
properties = {"kind": "spark"}
endpoint = Endpoint("http://location:port", "name", "word")
session = MagicMock()
controller._livy_session = MagicMock(return_value=session)
controller._http_client = MagicMock(return_value=MagicMock())
controller.add_session(name, endpoint, False, properties)
controller._livy_session.assert_called_once_with(
controller._http_client.return_value, properties, ipython_display)
controller.session_manager.add_session.assert_called_once_with(
name, session)
session.start.assert_called_once_with()
def test_add_sessions_command_exception():
# Do not skip and python
add_sessions_mock = MagicMock(side_effect=BadUserDataException('hehe'))
spark_controller.add_session = add_sessions_mock
command = "add"
name = "-s name"
language = "-l python"
connection_string = "-u http://url.com -a sdf -p w"
line = " ".join([command, name, language, connection_string])
magic.spark(line)
add_sessions_mock.assert_called_once_with(
"name", Endpoint("http://url.com", "sdf", "w"), False,
{"kind": "pyspark"})
ipython_display.send_error.assert_called_once_with(
EXPECTED_ERROR_MSG.format(add_sessions_mock.side_effect))
def test_add_session_throws_when_session_start_fails():
name = "name"
properties = {"kind": "spark"}
endpoint = Endpoint("http://location:port", "name", "word")
session = MagicMock()
controller._livy_session = MagicMock(return_value=session)
controller._http_client = MagicMock(return_value=MagicMock())
e = ValueError(
"Failed to create the SqlContext.\nError, '{}'".format("Exception"))
session.start = MagicMock(side_effect=e)
try:
controller.add_session(name, endpoint, False, properties)
assert False
except ValueError as ex:
assert str(ex) == str(e)
session.start.assert_called_once_with()
def test_add_sessions_command_extra_properties():
conf.override_all({})
magic.spark("config", "{\"extra\": \"yes\"}")
assert conf.session_configs() == {"extra": "yes"}
add_sessions_mock = MagicMock()
spark_controller.add_session = add_sessions_mock
command = "add"
name = "-s name"
language = "-l scala"
connection_string = "-u http://livyendpoint.com"
line = " ".join([command, name, language, connection_string])
magic.spark(line)
add_sessions_mock.assert_called_once_with(
"name", Endpoint("http://livyendpoint.com"), False, {
"kind": "spark",
"extra": "yes"
})
conf.load()
def spark(self, line, cell="", local_ns=None):
"""Magic to execute spark remotely.
This magic allows you to create a Livy Scala or Python session against a Livy endpoint. Every session can
be used to execute either Spark code or SparkSQL code by executing against the SQL context in the session.
When the SQL context is used, the result will be a Pandas dataframe of a sample of the results.
If invoked with no subcommand, the cell will be executed against the specified session.
Subcommands
-----------
info
Display the available Livy sessions and other configurations for sessions.
add
Add a Livy session given a session name (-s), language (-l), and endpoint credentials.
The -k argument, if present, will skip adding this session if it already exists.
e.g. `%spark add -s test -l python -u https://sparkcluster.net/livy -a u -p -k`
config
Override the livy session properties sent to Livy on session creation. All session creations will
contain these config settings from then on.
Expected value is a JSON key-value string to be sent as part of the Request Body for the POST /sessions
endpoint in Livy.
e.g. `%%spark config`
`{"driverMemory":"1000M", "executorCores":4}`
run
Run Spark code against a session.
e.g. `%%spark -s testsession` will execute the cell code against the testsession previously created
e.g. `%%spark -s testsession -c sql` will execute the SQL code against the testsession previously created
e.g. `%%spark -s testsession -c sql -o my_var` will execute the SQL code against the testsession
previously created and store the pandas dataframe created in the my_var variable in the
Python environment.
logs
Returns the logs for a given session.
e.g. `%spark logs -s testsession` will return the logs for the testsession previously created
delete
Delete a Livy session.
e.g. `%spark delete -s defaultlivy`
cleanup
Delete all Livy sessions created by the notebook. No arguments required.
e.g. `%spark cleanup`
"""
usage = "Please look at usage of %spark by executing `%spark?`."
user_input = line
args = parse_argstring_or_throw(self.spark, user_input)
subcommand = args.command[0].lower()
# info
if subcommand == "info":
if args.url is not None:
endpoint = Endpoint(args.url, args.user, args.password)
info_sessions = self.spark_controller.get_all_sessions_endpoint_info(endpoint)
self.print_endpoint_info(info_sessions)
else:
self._print_local_info()
# config
elif subcommand == "config":
conf.override(conf.session_configs.__name__, json.loads(cell))
# add
elif subcommand == "add":
if args.url is None:
self.ipython_display.send_error("Need to supply URL argument (e.g. -u https://example.com/livyendpoint)")
return
name = args.session
language = args.language
endpoint = Endpoint(args.url, args.user, args.password)
skip = args.skip
properties = conf.get_session_properties(language)
self.spark_controller.add_session(name, endpoint, skip, properties)
# delete
elif subcommand == "delete":
if args.session is not None:
self.spark_controller.delete_session_by_name(args.session)
elif args.url is not None:
if args.id is None:
self.ipython_display.send_error("Must provide --id or -i option to delete session at endpoint from URL")
return
endpoint = Endpoint(args.url, args.user, args.password)
session_id = args.id
self.spark_controller.delete_session_by_id(endpoint, session_id)
else:
self.ipython_display.send_error("Subcommand 'delete' requires a session name or a URL and session ID")
# cleanup
elif subcommand == "cleanup":
if args.url is not None:
endpoint = Endpoint(args.url, args.user, args.password)
self.spark_controller.cleanup_endpoint(endpoint)
else:
self.spark_controller.cleanup()
# logs
elif subcommand == "logs":
if args.session is None:
self.ipython_display.send_error("Need to provide session argument (-s SESSION_NAME)")
return
self.ipython_display.write(self.spark_controller.get_logs(args.session))
# run
elif len(subcommand) == 0:
if args.session is None:
self.ipython_display.send_error("Need to provide session argument (-s SESSION_NAME)")
return
if args.context == CONTEXT_NAME_SPARK:
(success, out) = self.spark_controller.run_command(Command(cell), args.session)
if success:
self.ipython_display.write(out)
else:
self.ipython_display.send_error(out)
elif args.context == CONTEXT_NAME_SQL:
return self.execute_sqlquery(cell, args.samplemethod, args.maxrows, args.samplefraction,
args.session, args.output, args.quiet)
else:
self.ipython_display.send_error("Context '{}' not found".format(args.context))
# error
else:
self.ipython_display.send_error("Subcommand '{}' not found. {}".format(subcommand, usage))
# Copyright (c) 2015 [email protected] # Distributed under the terms of the Modified BSD License. from mock import patch, PropertyMock, MagicMock from nose.tools import raises, assert_equals, with_setup import requests from remotespark.livyclientlib.endpoint import Endpoint from remotespark.livyclientlib.exceptions import HttpClientException from remotespark.livyclientlib.linearretrypolicy import LinearRetryPolicy from remotespark.livyclientlib.reliablehttpclient import ReliableHttpClient retry_policy = None sequential_values = [] endpoint = Endpoint("http://url.com", "username", "password") def _setup(): global retry_policy retry_policy = LinearRetryPolicy(0.01, 5) def _teardown(): pass def return_sequential(): global sequential_values val = sequential_values[0] sequential_values = sequential_values[1:]
def refresh_configuration(self):
self.endpoint = Endpoint("new_url")
def __init__(self, shell, data=None, spark_events=None):
super(TestKernelMagics, self).__init__(shell,
spark_events=spark_events)
self.language = constants.LANG_PYTHON
self.endpoint = Endpoint("url")