def test_transform_clients(generate_data):
submission_dates = generate_dates(start_ds)
snippets = [
# not new, default
{
"client_id": "0",
"profile_creation_date": 0,
},
# new, but duplicate
{
"client_id": "1"
},
{
"client_id": "1",
"timestamp": submission_dates["timestamp"] + 1,
},
# new, not default
{
"client_id": "2",
"is_default_browser": False
}
]
df = generate_data(snippets)
res = topline.transform(df, start_ds, "weekly")
assert res.count() == 1
row = res.first()
assert row.actives == 3
assert row.new_records == 2
assert row.default == 2
def test_transform_hours(generate_data):
snippets = [
{
"country": "US",
"subsession_length": topline.seconds_per_hour
},
{
"country": "CA",
"subsession_length": topline.seconds_per_hour
},
{
"subsession_length": 181 * topline.seconds_per_day
},
{
"subsession_length": -1 * topline.seconds_per_day
},
]
df = generate_data(snippets)
res = topline.transform(df, start_ds, "weekly")
assert res.count() == 2
assert res.groupBy().sum().first()["sum(hours)"] == 2.0
assert (res.groupBy("geo").sum().where(
F.col("geo") == "CA").first()["sum(hours)"]) == 1.0
def test_transform_searches_filters_incontent(generate_data):
snippets = [
{
'search_counts': [search_row("google", source="in-content")]
}, # no
{
'search_counts': [search_row("google", source="contextmenu")]
}, # yes
{
'search_counts': [search_row("google", source="abouthome")]
}, # yes
]
df = generate_data(snippets)
res = topline.transform(df, start_ds, "weekly")
assert res.groupBy().sum().first()["sum(google)"] == 2
def test_transform_searches(generate_data):
snippets = [
{
"search_counts": None
},
{
"search_counts": [search_row("google")]
},
{
"country": "CA",
"search_counts": [search_row("hooli"),
search_row("google")]
},
]
df = generate_data(snippets)
res = topline.transform(df, start_ds, "weekly")
assert res.count() == 2
assert res.groupBy().sum().first()["sum(google)"] == 2
assert (res.groupBy("geo").sum().where(
F.col("geo") == "CA").first()["sum(other)"]) == 1