def setUp(self):
QueueDatabaseTest.setUp(self)
# registration for all the editors below
self.r = r = 20150101000000
# exactly 30 days after registration
self.m = m = 20150131000000
self.r_plus_30 = format_pretty_date(d(self.m))
self.create_test_cohort(
editor_count=5,
revisions_per_editor=8,
revision_timestamps=[
# this one will make 5 edits within 30 days of self.r_plus_30
[r + 1, r + 2, r + 3, r + 4, r + 5, m + 6, m + 7, m + 8],
# this one will make 3 edits within 30 days of self.r_plus_30
[r + 1, r + 2, r + 3, m + 4, m + 5, m + 6, m + 7, m + 8],
# this one will make 8 edits within 30 days of self.r_plus_30
[r + 1, r + 2, r + 3, r + 4, r + 5, r + 6, r + 7, r + 8],
# this one will make 0 edits within 30 days of self.r_plus_30
[m + 1, m + 2, m + 3, m + 4, m + 5, m + 6, m + 7, m + 8],
# this one will make the 5th edit right on self.r_plus_30
[r + 1, r + 2, r + 3, r + 4, m + 0, m + 6, m + 7, m + 8],
],
user_registrations=r,
revision_lengths=10
)
def normalize_datetime_slices(self, results_by_user, submetrics):
"""
Starting from a sparse set of timeseries results, fill in default values
for the specified list of sub-metrics. Also make sure the chronological
first timeseries slice is >= self.start_date.
If self.timeseries is NONE, this is a simple identity function.
Parameters
results_by_user : dictionary of submetrics dictionaries by user
submetrics : list of tuples of the form (label, index, default)
Returns
the results, filled in with default values
"""
if self.timeseries.data == TimeseriesChoices.NONE:
return results_by_user
slice_delta = self.get_delta_from_choice()
timeseries_slices = OrderedDict()
start_slice_key = format_pretty_date(self.start_date.data)
timeseries_slices[start_slice_key] = None
first_slice = self.get_first_slice()
first_slice_key = format_pretty_date(first_slice)
slice_to_default = first_slice
while slice_to_default < self.end_date.data:
date_key = format_pretty_date(slice_to_default)
timeseries_slices[date_key] = None
slice_to_default += slice_delta
for user_id, user_submetrics in results_by_user.iteritems():
for label, i, default in submetrics:
if not label or not user_submetrics or label not in user_submetrics:
continue
defaults = timeseries_slices.copy()
defaults.update(user_submetrics[label])
for k, v in defaults.iteritems():
if not v:
defaults[k] = default
if start_slice_key != first_slice_key:
# coerce the first datetime slice to be self.start_date
defaults[start_slice_key] = defaults.pop(first_slice_key)
user_submetrics[label] = defaults
return results_by_user
def get_date_from_tuple(self, row_tuple, start_index, stop_index):
"""
Suppose you have a tuple like this:
([data], [data], ... , year, month, day, [data], [data])
Then this function will parse out the year, month, day, and hour
into a date string. Anything beyond year, month, day is optional.
"""
date_pieces = row_tuple[start_index:stop_index]
year, month, day, hour = 1970, 1, 1, 0
if len(date_pieces) > 0:
year = date_pieces[0]
if len(date_pieces) > 1:
month = date_pieces[1]
if len(date_pieces) > 2:
day = date_pieces[2]
if len(date_pieces) > 3:
hour = date_pieces[3]
return format_pretty_date(datetime(year, month, day, hour))
def test_parse_pretty_date(self):
date = datetime.datetime(2012, 2, 3, 4, 5)
assert_equal(date, parse_pretty_date(format_pretty_date(date)))
def test_parse_pretty_date(self):
date = datetime(2012, 2, 3, 4, 5)
assert_equal(date, parse_pretty_date(format_pretty_date(date)))
