def test_requires(self, connect_s3_mock):
s3_conn_mock = connect_s3_mock.return_value
bucket_mock = s3_conn_mock.get_bucket.return_value
class FakeKey(object):
"""A test double of the structure returned by boto when listing keys in an S3 bucket."""
def __init__(self, path):
self.key = path
self.size = 10
bucket_mock.list.return_value = [FakeKey(path) for path in self.SAMPLE_KEY_PATHS]
task = PathSelectionByDateIntervalTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[r'.*?FakeServerGroup/tracking.log-(?P<date>\d{8}).*\.gz'],
expand_interval=datetime.timedelta(0),
)
expected_paths = [
'FakeServerGroup/tracking.log-20140318.gz',
'FakeServerGroup/tracking.log-20140319-1395256622.gz',
]
self.assertItemsEqual(
task.requires(),
[UncheckedExternalURL(source + path) for path in expected_paths for source in self.SOURCE]
)
def test_requires(self, connect_s3_mock):
s3_conn_mock = connect_s3_mock
bucket_mock = s3_conn_mock.get_bucket.return_value
class FakeKey(object):
"""A test double of the structure returned by boto when listing keys in an S3 bucket."""
def __init__(self, path):
self.key = path
self.size = 10
bucket_mock.list.return_value = [FakeKey(path) for path in self.SAMPLE_KEY_PATHS]
task = PathSelectionByDateIntervalTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[r'.*?FakeServerGroup/tracking.log-(?P<date>\d{8}).*\.gz'],
expand_interval=datetime.timedelta(0),
)
expected_paths = [
'FakeServerGroup/tracking.log-20140318.gz',
'FakeServerGroup/tracking.log-20140319-1395256622.gz',
]
self.assertItemsEqual(
task.requires(),
[UncheckedExternalURL(source + path) for path in expected_paths for source in self.SOURCE]
)
def test_timestamped_urls(self):
task = PathSelectionByDateIntervalTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[r'.*?FakeServerGroup/tracking.log-.*-(?P<timestamp>\d{10})\.gz'],
expand_interval=datetime.timedelta(0),
)
self.assert_only_matched(task, [
'FakeServerGroup/tracking.log-20140319-1395256622.gz',
])
task = PathSelectionByDateIntervalTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[r'.*?FakeServerGroup/tracking.log-.*-(?P<timestamp>\d{10})\.gz'],
expand_interval=datetime.timedelta(1),
)
self.assert_only_matched(task, [
'FakeServerGroup/tracking.log-20140319-1395256622.gz',
'FakeServerGroup/tracking.log-20140401-1396379384.gz',
])
def test_edge_urls(self):
task = PathSelectionByDateIntervalTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[r'.*?FakeEdgeServerGroup/tracking.log-(?P<date>\d{8}).*\.gz'],
expand_interval=datetime.timedelta(0),
)
self.assert_only_matched(task, [
'FakeEdgeServerGroup/tracking.log-20140324-1395670621.gz',
])
def test_timestamped_urls(self):
task = PathSelectionByDateIntervalTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[r'.*?FakeServerGroup/tracking.log-.*-(?P<timestamp>\d{10})\.gz'],
expand_interval=datetime.timedelta(0),
)
self.assert_only_matched(task, [
'FakeServerGroup/tracking.log-20140319-1395256622.gz',
])
task = PathSelectionByDateIntervalTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[r'.*?FakeServerGroup/tracking.log-.*-(?P<timestamp>\d{10})\.gz'],
expand_interval=datetime.timedelta(1),
)
self.assert_only_matched(task, [
'FakeServerGroup/tracking.log-20140319-1395256622.gz',
'FakeServerGroup/tracking.log-20140401-1396379384.gz',
])
def test_edge_urls(self):
task = PathSelectionByDateIntervalTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[r'.*?FakeEdgeServerGroup/tracking.log-(?P<date>\d{8}).*\.gz'],
expand_interval=datetime.timedelta(0),
)
self.assert_only_matched(task, [
'FakeEdgeServerGroup/tracking.log-20140324-1395670621.gz',
])
def test_filtering_of_urls(self):
task = EventLogSelectionTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=r'.*?FakeServerGroup/tracking.log-(?P<date>\d{8}).*\.gz',
expand_interval=datetime.timedelta(0),
)
self.assert_only_matched(task, [
'FakeServerGroup/tracking.log-20140318.gz',
'FakeServerGroup/tracking.log-20140319-1395256622.gz',
])
def test_filtering_of_urls(self):
task = EventLogSelectionTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[r'.*?FakeServerGroup/tracking.log-(?P<date>\d{8}).*\.gz'],
expand_interval=datetime.timedelta(0),
)
self.assert_only_matched(task, [
'FakeServerGroup/tracking.log-20140318.gz',
'FakeServerGroup/tracking.log-20140319-1395256622.gz',
])
def test_expanded_interval(self):
task = PathSelectionByDateIntervalTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[r'.*?FakeServerGroup/tracking.log-(?P<date>\d{8}).*\.gz'],
expand_interval=datetime.timedelta(1),
)
self.assert_only_matched(task, [
'FakeServerGroup/tracking.log-20140228.gz',
'FakeServerGroup/tracking.log-20140318.gz',
'FakeServerGroup/tracking.log-20140319-1395256622.gz',
'FakeServerGroup/tracking.log-20140401-1395254574.gz',
])
def test_expanded_interval(self):
task = PathSelectionByDateIntervalTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[r'.*?FakeServerGroup/tracking.log-(?P<date>\d{8}).*\.gz'],
expand_interval=datetime.timedelta(1),
)
self.assert_only_matched(task, [
'FakeServerGroup/tracking.log-20140228.gz',
'FakeServerGroup/tracking.log-20140318.gz',
'FakeServerGroup/tracking.log-20140319-1395256622.gz',
'FakeServerGroup/tracking.log-20140401-1395254574.gz',
])
def test_multiple_filtering_of_urls(self):
task = PathSelectionByDateIntervalTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[
r'.*?FakeServerGroup/tracking.log-(?P<date>\d{8}).*\.gz',
r'.*?FakeEdgeServerGroup/tracking.log-(?P<date>\d{8}).*\.gz',
r'.*tracking_\d{3,5}\.log\.gz$',
],
expand_interval=datetime.timedelta(0),
)
self.assert_only_matched(task, [
'FakeServerGroup/tracking.log-20140318.gz',
'FakeServerGroup/tracking.log-20140319-1395256622.gz',
'FakeEdgeServerGroup/tracking.log-20140324-1395670621.gz',
'FakeOldServerGroup3/tracking_14602.log.gz',
])
def test_multiple_filtering_of_urls(self):
task = PathSelectionByDateIntervalTask(
source=self.SOURCE,
interval=Month.parse('2014-03'),
pattern=[
r'.*?FakeServerGroup/tracking.log-(?P<date>\d{8}).*\.gz',
r'.*?FakeEdgeServerGroup/tracking.log-(?P<date>\d{8}).*\.gz',
r'.*tracking_\d{3,5}\.log\.gz$',
],
expand_interval=datetime.timedelta(0),
)
self.assert_only_matched(task, [
'FakeServerGroup/tracking.log-20140318.gz',
'FakeServerGroup/tracking.log-20140319-1395256622.gz',
'FakeEdgeServerGroup/tracking.log-20140324-1395670621.gz',
'FakeOldServerGroup3/tracking_14602.log.gz',
])
def test_pattern_from_config(self):
task = PathSelectionByDateIntervalTask(
interval=Month.parse('2014-03')
)
self.assertEquals(task.pattern, ('foobar',))
def test_pattern_override(self):
task = PathSelectionByDateIntervalTask(
interval=Month.parse('2014-03'),
pattern=['baz']
)
self.assertEquals(task.pattern, ('baz',))
def test_pattern_from_config(self):
task = PathSelectionByDateIntervalTask(
interval=Month.parse('2014-03')
)
self.assertEquals(task.pattern, ('foobar',))
def test_pattern_from_config(self):
task = EventLogSelectionTask(interval=Month.parse('2014-03'))
self.assertEquals(task.pattern, ('foobar', ))
def test_pattern_override(self):
task = EventLogSelectionTask(interval=Month.parse('2014-03'),
pattern=['baz'])
self.assertEquals(task.pattern, ('baz', ))
def test_default_pattern(self):
task = PathSelectionByDateIntervalTask(interval=Month.parse('2014-03'))
self.assertEquals(task.pattern, (
r'.*tracking.log-(?P<date>\d{8}).*\.gz',
r'.*tracking.notalog-(?P<date>\d{8}).*\.gz',
))
def test_default_source(self):
task = PathSelectionByDateIntervalTask(interval=Month.parse('2014-03'))
self.assertEquals(task.source, ('s3://fake/input/', 's3://fake/input2/'))
def test_pattern_from_config(self):
task = EventLogSelectionTask(
interval=Month.parse('2014-03')
)
self.assertEquals(task.pattern, 'foobar')
def test_pattern_override(self):
task = EventLogSelectionTask(
interval=Month.parse('2014-03'),
pattern='baz'
)
self.assertEquals(task.pattern, 'baz')
def test_default_source(self):
task = PathSelectionByDateIntervalTask(interval=Month.parse('2014-03'))
self.assertEquals(task.source, ('s3://fake/input/', 's3://fake/input2/'))
def test_default_pattern(self):
task = PathSelectionByDateIntervalTask(interval=Month.parse('2014-03'))
self.assertEquals(task.pattern, (
r'.*tracking.log-(?P<date>\\d{8}).*\\.gz',
r'.*tracking.notalog-(?P<date>\\d{8}).*\\.gz',
))
def test_pattern_override(self):
task = PathSelectionByDateIntervalTask(
interval=Month.parse('2014-03'),
pattern=['baz']
)
self.assertEquals(task.pattern, ('baz',))
