def test_configure():
# Mock info method
magic.info = MagicMock()
# Session not started
conf.override_all({})
magic.configure('{"extra": "yes"}')
assert conf.session_configs() == {"extra": "yes"}
magic.info.assert_called_once_with("")
# Session started - no -f
magic.session_started = True
conf.override_all({})
magic.configure("{\"extra\": \"yes\"}")
assert conf.session_configs() == {}
assert_equals(ipython_display.send_error.call_count, 1)
# Session started - with -f
magic.info.reset_mock()
conf.override_all({})
magic.configure("-f {\"extra\": \"yes\"}")
assert conf.session_configs() == {"extra": "yes"}
spark_controller.delete_session_by_name.assert_called_once_with(
magic.session_name)
spark_controller.add_session.assert_called_once_with(
magic.session_name, magic.connection_string, False, {
"kind": constants.SESSION_KIND_PYSPARK,
"extra": "yes"
})
magic.info.assert_called_once_with("")
def test_configure():
# Mock info method
magic.info = MagicMock()
# Session not started
conf.override_all({})
magic.configure('{"extra": "yes"}')
assert conf.session_configs() == {"extra": "yes"}
magic.info.assert_called_once_with("")
# Session started - no -f
magic.session_started = True
conf.override_all({})
magic.configure("{\"extra\": \"yes\"}")
assert conf.session_configs() == {}
assert_equals(ipython_display.send_error.call_count, 1)
# Session started - with -f
magic.info.reset_mock()
conf.override_all({})
magic.configure("-f {\"extra\": \"yes\"}")
assert conf.session_configs() == {"extra": "yes"}
spark_controller.delete_session_by_name.assert_called_once_with(magic.session_name)
spark_controller.add_session.assert_called_once_with(magic.session_name, magic.endpoint, False,
{"kind": constants.SESSION_KIND_PYSPARK, "extra": "yes"})
magic.info.assert_called_once_with("")
def __init__(self, spark_controller, ipywidget_factory, ipython_display, endpoints_dropdown_widget, refresh_method):
# This is nested
super(CreateSessionWidget, self).__init__(spark_controller, ipywidget_factory, ipython_display, True)
self.refresh_method = refresh_method
self.endpoints_dropdown_widget = endpoints_dropdown_widget
self.session_widget = self.ipywidget_factory.get_text(
description='Name:',
value='session-name'
)
self.lang_widget = self.ipywidget_factory.get_toggle_buttons(
description='Language:',
options=[LANG_SCALA, LANG_PYTHON],
)
self.properties = self.ipywidget_factory.get_text(
description='Properties:',
value=json.dumps(conf.session_configs())
)
self.submit_widget = self.ipywidget_factory.get_submit_button(
description='Create Session'
)
self.children = [self.ipywidget_factory.get_html(value="<br/>", width="600px"), self.endpoints_dropdown_widget,
self.session_widget, self.lang_widget, self.properties,
self.ipywidget_factory.get_html(value="<br/>", width="600px"), self.submit_widget]
for child in self.children:
child.parent_widget = self
def _print_local_info(self):
sessions_info = [" {}".format(i) for i in self.spark_controller.get_manager_sessions_str()]
print("""Info for running Spark:
Sessions:
{}
Session configs:
{}
""".format("\n".join(sessions_info), conf.session_configs()))
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 = "name"
language = "scala"
connection_string = "url=http://location:port;username=name;password=word"
line = " ".join([command, name, language, connection_string])
magic.spark(line)
add_sessions_mock.assert_called_once_with(name, connection_string, False, {"kind": "spark", "extra": "yes"})
conf.load()
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 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 = "name"
language = "scala"
connection_string = "url=http://location:port;username=name;password=word"
line = " ".join([command, name, language, connection_string])
magic.spark(line)
add_sessions_mock.assert_called_once_with(name, connection_string, False, {
"kind": "spark",
"extra": "yes"
})
conf.load()
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. First argument is the name of the session, second argument
is the language, and third argument is the connection string of the Livy endpoint.
A fourth argument specifying if session creation can be skipped if it already exists is optional:
"skip" or empty.
e.g. `%%spark add test python url=https://sparkcluster.net/livy;username=u;password=p skip`
or
e.g. `%%spark add test python url=https://sparkcluster.net/livy;username=u;password=p`
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. Argument is the name of the session to be deleted.
e.g. `%%spark delete 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(self.spark, user_input)
subcommand = args.command[0].lower()
try:
# info
if subcommand == "info":
if len(args.command) == 2:
connection_string = args.command[1]
info_sessions = self.spark_controller.get_all_sessions_endpoint_info(connection_string)
self._print_endpoint_info(info_sessions)
elif len(args.command) == 1:
self._print_local_info()
else:
raise ValueError("Subcommand 'info' requires no value or a connection string to show all sessions.\n"
"{}".format(usage))
# config
elif subcommand == "config":
# Would normally do " ".join(args.command[1:]) but parse_argstring removes quotes...
rest_of_line = user_input[7:]
conf.override(conf.session_configs.__name__, json.loads(rest_of_line))
# add
elif subcommand == "add":
if len(args.command) != 4 and len(args.command) != 5:
raise ValueError("Subcommand 'add' requires three or four arguments.\n{}".format(usage))
name = args.command[1].lower()
language = args.command[2].lower()
connection_string = args.command[3]
if len(args.command) == 5:
skip = args.command[4].lower() == "skip"
else:
skip = False
properties = copy.deepcopy(conf.session_configs())
properties["kind"] = self._get_livy_kind(language)
self.spark_controller.add_session(name, connection_string, skip, properties)
# delete
elif subcommand == "delete":
if len(args.command) == 2:
name = args.command[1].lower()
self.spark_controller.delete_session_by_name(name)
elif len(args.command) == 3:
connection_string = args.command[1]
session_id = args.command[2]
self.spark_controller.delete_session_by_id(connection_string, session_id)
else:
raise ValueError("Subcommand 'delete' requires a session name or a connection string and id.\n{}"
.format(usage))
# cleanup
elif subcommand == "cleanup":
if len(args.command) == 2:
connection_string = args.command[1]
self.spark_controller.cleanup_endpoint(connection_string)
elif len(args.command) == 1:
self.spark_controller.cleanup()
else:
raise ValueError("Subcommand 'cleanup' requires no further values or a connection string to clean up "
"sessions.\n{}".format(usage))
# logs
elif subcommand == "logs":
if len(args.command) == 1:
(success, out) = self.spark_controller.get_logs(args.session)
if success:
self.ipython_display.write(out)
else:
self.ipython_display.send_error(out)
else:
raise ValueError("Subcommand 'logs' requires no further values.\n{}".format(usage))
# run
elif len(subcommand) == 0:
if args.context == Constants.context_name_spark:
(success, out) = self.spark_controller.run_cell(cell, args.session)
if success:
self.ipython_display.write(out)
else:
self.ipython_display.send_error(out)
elif args.context == Constants.context_name_sql:
return self._execute_against_context_that_returns_df(self.spark_controller.run_cell_sql, cell,
args.session, args.output)
elif args.context == Constants.context_name_hive:
return self._execute_against_context_that_returns_df(self.spark_controller.run_cell_hive, cell,
args.session, args.output)
else:
raise ValueError("Context '{}' not found".format(args.context))
# error
else:
raise ValueError("Subcommand '{}' not found. {}".format(subcommand, usage))
except ValueError as err:
self.ipython_display.send_error("{}".format(err))
