def test_add_all_resource_filter_mixed(self):
date1 = datetime(2019, 1, 1)
resource_counts1 = {
u'resource1': 10002,
u'cresource2': 51,
}
date2 = datetime(2020, 1, 1)
resource_counts2 = {
u'resource1': 29,
u'cresource2': 38290,
}
monthly_stats = MonthlyStats(resource_id=u'resource1')
monthly_stats.add_all(date1, resource_counts1)
monthly_stats.add_all(date2, resource_counts2)
assert monthly_stats.stats[u'1/2019'][u'research'][u'records'] == 10002
assert monthly_stats.stats[u'1/2019'][u'research'][
u'download_events'] == 1
assert monthly_stats.stats[u'1/2019'][u'collections'][u'records'] == 0
assert monthly_stats.stats[u'1/2019'][u'collections'][
u'download_events'] == 0
assert monthly_stats.stats[u'1/2020'][u'research'][u'records'] == 29
assert monthly_stats.stats[u'1/2020'][u'research'][
u'download_events'] == 1
assert monthly_stats.stats[u'1/2020'][u'collections'][u'records'] == 0
assert monthly_stats.stats[u'1/2020'][u'collections'][
u'download_events'] == 0
def test_update_from_gbif(self):
monthly_stats = MonthlyStats()
monthly_stats.update_from_gbif(10, 2018, 83, 4)
monthly_stats.update_from_gbif(10, 2018, 18, 5)
assert monthly_stats.stats[u'10/2018'][u'gbif'][
u'download_events'] == 4 + 5
assert monthly_stats.stats[u'10/2018'][u'gbif'][u'records'] == 83 + 18
def test_add(self):
# check that add just calls add_all
monthly_stats = MonthlyStats()
monthly_stats.add_all = MagicMock()
date = MagicMock()
resource_id = MagicMock()
count = MagicMock()
monthly_stats.add(date, resource_id, count)
assert monthly_stats.add_all.called
assert monthly_stats.add_all.call_args == call(date,
{resource_id: count})
def test_update_from_gbif_with_filters(self):
monthly_stats = MonthlyStats(month=10, year=2018)
monthly_stats.update_from_gbif(10, 2019, 2389, 223)
monthly_stats.update_from_gbif(4, 2012, 100, 28)
monthly_stats.update_from_gbif(10, 2018, 8344, 40)
monthly_stats.update_from_gbif(10, 2018, 40, 1)
assert monthly_stats.stats[u'10/2018'][u'gbif'][
u'download_events'] == 40 + 1
assert monthly_stats.stats[u'10/2018'][u'gbif'][
u'records'] == 8344 + 40
assert u'10/2019' not in monthly_stats.stats
assert u'4/2012' not in monthly_stats.stats
def test_month_year_format(self):
# currently the month/year keys are not zero padded so we might as well test that that is
# still the case in case a developer gets refactor happy and changes it (I did this so maybe
# this test is for me...)
monthly_stats = MonthlyStats()
for month in range(1, 12):
date = datetime(2017, month, 1)
monthly_stats.add_all(date, resource_counts=dict(resource1=10))
month_year = u'{}/2017'.format(month)
assert monthly_stats.stats[month_year][u'research'][
u'records'] == 10
assert monthly_stats.stats[month_year][u'research'][
u'download_events'] == 1
def test_add_all_no_filters_mixed(self):
date = datetime(2020, 1, 1)
resource_counts = {
u'resource1': 10002,
u'cresource2': 51,
}
monthly_stats = MonthlyStats()
monthly_stats.add_all(date, resource_counts)
assert monthly_stats.stats[u'1/2020'][u'collections'][u'records'] == 51
assert monthly_stats.stats[u'1/2020'][u'collections'][
u'download_events'] == 1
assert monthly_stats.stats[u'1/2020'][u'research'][u'records'] == 10002
assert monthly_stats.stats[u'1/2020'][u'research'][
u'download_events'] == 1
def test_update_from_backfill_with_previous_data(self):
monthly_stats = MonthlyStats()
# this is a collection resource
monthly_stats.add(datetime(2018, 10, 1), u'resource1', 20)
# this isn't a collection resource
monthly_stats.add(datetime(2018, 10, 1), u'resource2', 15)
# update from the backfill
monthly_stats.update_from_backfill(
u'10', u'2018', {
u'collections': {
u'download_events': 3,
u'records': 10004,
},
u'research': {
u'download_events': 5829,
u'records': 32894932,
}
})
assert monthly_stats.stats[u'10/2018'][u'collections'][
u'download_events'] == 3 + 1
assert monthly_stats.stats[u'10/2018'][u'collections'][
u'records'] == 10004 + 20
assert monthly_stats.stats[u'10/2018'][u'research'][
u'download_events'] == 5829 + 1
assert monthly_stats.stats[u'10/2018'][u'research'][
u'records'] == 32894932 + 15
def test_add_all_complex_filter_mixed(self):
date = datetime(2019, 6, 1)
resource_counts = {
u'resource1': 10,
u'cresource2': 51,
u'resource2': 19,
}
monthly_stats = MonthlyStats(month=6,
year=2019,
resource_id=u'resource2')
monthly_stats.add_all(date, resource_counts)
assert monthly_stats.stats[u'6/2019'][u'research'][u'records'] == 0
assert monthly_stats.stats[u'6/2019'][u'research'][
u'download_events'] == 0
assert monthly_stats.stats[u'6/2019'][u'collections'][u'records'] == 19
assert monthly_stats.stats[u'6/2019'][u'collections'][
u'download_events'] == 1
def test_update_from_backfill_no_previous_data(self):
monthly_stats = MonthlyStats()
monthly_stats.update_from_backfill(
u'10', u'2018', {
u'collections': {
u'download_events': 102,
u'records': 1029
},
u'research': {
u'download_events': 4,
u'records': 902832
}
})
assert monthly_stats.stats[u'10/2018'][u'collections'][
u'download_events'] == 102
assert monthly_stats.stats[u'10/2018'][u'collections'][
u'records'] == 1029
assert monthly_stats.stats[u'10/2018'][u'research'][
u'download_events'] == 4
assert monthly_stats.stats[u'10/2018'][u'research'][
u'records'] == 902832
def test_update_from_backfill_with_filters(self):
monthly_stats = MonthlyStats(month=10, year=2018)
monthly_stats.update_from_backfill(
u'10', u'2018', {
u'collections': {
u'download_events': 3,
u'records': 10004,
},
u'research': {
u'download_events': 5829,
u'records': 32894932,
}
})
monthly_stats.update_from_backfill(
u'11', u'2018', {
u'collections': {
u'download_events': 3,
u'records': 10004,
},
u'research': {
u'download_events': 5829,
u'records': 32894932,
}
})
monthly_stats.update_from_backfill(
u'11', u'2017', {
u'collections': {
u'download_events': 3,
u'records': 10004,
},
u'research': {
u'download_events': 5829,
u'records': 32894932,
}
})
assert u'10/2018' in monthly_stats.stats
assert u'11/2018' not in monthly_stats.stats
assert u'11/2017' not in monthly_stats.stats
def test_as_dict(self):
monthly_stats = MonthlyStats()
monthly_stats.add(datetime(2017, 4, 1), u'resource1', 29)
monthly_stats.add(datetime(2017, 4, 20), u'resource1', 39)
monthly_stats.add(datetime(2013, 6, 18), u'resource2', 1000)
monthly_stats.add(datetime(2013, 1, 8), u'resource5', 55)
monthly_stats.add(datetime(2013, 11, 4), u'resource2', 1000)
monthly_stats.add(datetime(2020, 10, 10), u'resource4', 92)
stats = monthly_stats.as_dict()
assert isinstance(stats, OrderedDict)
assert list(stats.keys()) == [
u'1/2013', u'6/2013', u'11/2013', u'4/2017', u'10/2020'
]