def check_surrounding_events(self, threshold_ids):
"""Checks that the events surrounding the first event in a new session
are allocated correctly.
Args:
threshold_ids: a list of IDs of the first events in the sessions.
"""
ds = MockDataStore('test', 0)
session_no = 1
last_id = threshold_ids[-1]
for threshold_id in threshold_ids:
if threshold_id != 0:
# check previous event is in the previous session
event = (ds.get_event('test_index', str(threshold_id - 1),
stored_events=True))
self.assertEqual(event['_source']['session_number'],
session_no)
if threshold_id != last_id:
# check next event is in the same session (as the event with
# threshold id)
session_no += 1
event = (ds.get_event('test_index', str(threshold_id + 1),
stored_events=True))
self.assertEqual(event['_source']['session_number'],
session_no)
def test_edge_time_diff(self):
"""Test events with the edge time difference between them are
allocated correctly."""
with mock.patch.object(
self.analyzer_class,
'event_stream',
return_value=_create_mock_event(
0,
2,
time_diffs=[self.analyzer_class.max_time_diff_micros])):
index = 'test_index'
sketch_id = 1
analyzer = self.analyzer_class(index, sketch_id)
message = analyzer.run()
self.assertEqual(
message,
'Sessionizing completed, number of session created: 1')
ds = MockDataStore('test', 0)
event1 = (ds.get_event('test_index', '0', stored_events=True))
self.assertEqual(event1['_source']['session_id'],
{analyzer.session_type: 1})
event2 = (ds.get_event('test_index', '101', stored_events=True))
self.assertEqual(event2['_source']['session_id'],
{analyzer.session_type: 1})
def test_diff_session(self):
"""Test multiple events in different sessions are allocated
correctly."""
with mock.patch.object(
self.analyzer_class,
'event_stream',
return_value=_create_mock_event(
0,
2,
time_diffs=[
self.analyzer_class.max_time_diff_micros +
(self.analyzer_class.max_time_diff_micros / 2)
])):
index = 'test_index'
sketch_id = 1
analyzer = self.analyzer_class(index, sketch_id)
message = analyzer.run()
self.assertEqual(
message,
'Sessionizing completed, number of session created: 2')
ds = MockDataStore('test', 0)
event1 = (ds.get_event('test_index', '0', stored_events=True))
self.assertEqual(event1['_source']['session_id'],
{analyzer.session_type: 1})
event2 = (ds.get_event('test_index', '101', stored_events=True))
self.assertEqual(event2['_source']['session_id'],
{analyzer.session_type: 2})
self._check_surrounding_events([101], analyzer.session_type)
def test_same_session(self):
"""Test multiple events in the same session are allocated correctly."""
with mock.patch.object(
self.analyzer_class,
'event_stream',
return_value=_create_mock_event(
0,
2,
time_diffs=[self.analyzer_class.max_time_diff_micros / 2
])):
index = 'test_index'
sketch_id = 1
analyzer = self.analyzer_class(index, sketch_id)
message = analyzer.run()
self.assertEqual(
message,
'Sessionizing completed, number of session created: 1')
ds = MockDataStore('test', 0)
event1 = (ds.get_event('test_index', '0', stored_events=True))
self.assertEqual(event1['_source']['session_id'],
{analyzer.session_type: 1})
# checking event with id '101' as 100 events have been inserted
# as 'padding' (see _create_mock_event())
event2 = (ds.get_event('test_index', '101', stored_events=True))
self.assertEqual(event2['_source']['session_id'],
{analyzer.session_type: 1})
def test_zero_time_diff(self):
"""Test events with no time difference between them are allocated
correctly."""
with mock.patch.object(SessionizerSketchPlugin,
'event_stream',
return_value=_create_mock_event(0,
2,
time_diffs=[0
])):
index = 'test_index'
sketch_id = 1
analyser = SessionizerSketchPlugin(index, sketch_id)
message = analyser.run()
self.assertEqual(
message,
'Sessionizing completed, number of session created: 1')
ds = MockDataStore('test', 0)
event1 = (ds.get_event('test_index', '0', stored_events=True))
self.assertEqual(event1['_source']['session_id'],
{'all_events': 1})
event1 = (ds.get_event('test_index', '100', stored_events=True))
self.assertEqual(event1['_source']['session_id'],
{'all_events': 1})
event2 = (ds.get_event('test_index', '101', stored_events=True))
self.assertEqual(event2['_source']['session_id'],
{'all_events': 1})
def test_multiple_sessions(self):
"""Test multiple events, two of which are in the same session and
one in a different session."""
with mock.patch.object(SessionizerSketchPlugin,
'event_stream',
return_value=_create_mock_event(
0, 3, time_diffs=[3000, 400000000])):
index = 'test_index'
sketch_id = 1
analyser = SessionizerSketchPlugin(index, sketch_id)
message = analyser.run()
self.assertEqual(
message,
'Sessionizing completed, number of session created: 2')
ds = MockDataStore('test', 0)
event1 = (ds.get_event('test_index', '0', stored_events=True))
self.assertEqual(event1['_source']['session_id'],
{'all_events': 1})
event2 = (ds.get_event('test_index', '101', stored_events=True))
self.assertEqual(event2['_source']['session_id'],
{'all_events': 1})
event3 = (ds.get_event('test_index', '202', stored_events=True))
self.assertEqual(event3['_source']['session_id'],
{'all_events': 2})
self._check_surrounding_events([202], 'all_events')
def _create_mock_event(event_id, quantity, time_diffs=None, source_attrs=None):
"""
Returns an instance of Event, based on the MockDataStore event_dict
example.
Args:
event_id: Desired ID for the Event.
quantity: The number of Events to be generated.
time_diffs: A list of time differences between the generated
Events.
source_attrs: Dictionary of attributes to add to the source of the
generated events.
Returns:
A generator of Event objects.
"""
if not time_diffs:
time_diffs = [0]
if quantity < 0:
quantity = abs(quantity)
# If the list of time differences is too small to be compatible
# with the quantity of events, then extend the list with the last
# value for as many items as necessary.
if quantity - len(time_diffs) > 0:
time_diffs.extend([time_diffs[len(time_diffs) - 1]] *
(quantity - len(time_diffs)))
# Setup for Event object initialisation
ds = MockDataStore('test', 0)
user = User('test_user')
sketch = Sketch('test_sketch', 'description', user)
label = sketch.Label(label='Test label', user=user)
sketch.labels.append(label)
event_timestamp = 1410895419859714
event_template = ds.get_event('test', 'test')
for i in range(quantity):
eventObj = _create_eventObj(ds, sketch, event_template, event_id,
event_timestamp, source_attrs)
yield eventObj
# adding extra events after every requested event for better
# simulation of real timeline data i.e. working with a larger
# dataset
for _ in range(100):
event_timestamp += 1
event_id += 1
eventObj = _create_eventObj(ds, sketch, event_template, event_id,
event_timestamp, source_attrs)
yield eventObj
event_timestamp += abs(time_diffs[i])
event_id += 1
def test_diff_session(self):
"""Test multiple events in different sessions are allocated
correctly."""
with mock.patch.object(SessionizerSketchPlugin,
'event_stream',
return_value=_create_mock_event(
0, 2, time_diffs=[400000000])):
index = 'test_index'
sketch_id = 1
analyser = sessionizer.SessionizerSketchPlugin(index, sketch_id)
message = analyser.run()
self.assertEqual(
message,
'Sessionizing completed, number of session created: 2')
ds = MockDataStore('test', 0)
event1 = (ds.get_event('test_index', '0', stored_events=True))
self.assertEqual(event1['_source']['session_number'], 1)
event2 = (ds.get_event('test_index', '101', stored_events=True))
self.assertEqual(event2['_source']['session_number'], 2)
self.check_surrounding_events([101])
def test_one_event(self):
"""Test the behaviour of the sessionizer when given one event."""
with mock.patch.object(SessionizerSketchPlugin,
'event_stream',
return_value=_create_mock_event(0, 1)):
index = 'test_index'
sketch_id = 1
analyser = sessionizer.SessionizerSketchPlugin(index, sketch_id)
message = analyser.run()
self.assertEqual(
message,
'Sessionizing completed, number of session created: 1')
ds = MockDataStore('test', 0)
event1 = (ds.get_event('test_index', '0', stored_events=True))
self.assertEqual(event1['_source']['session_number'], 1)