def test_computes_trending(self, db_request, with_purges):
projects = [
ProjectFactory.create(zscore=1 if not i else None) for i in range(3)
]
results = iter(
[
Row((projects[1].normalized_name, 2), {"project": 0, "zscore": 1}),
Row((projects[2].normalized_name, -1), {"project": 0, "zscore": 1}),
]
)
query = pretend.stub(result=pretend.call_recorder(lambda *a, **kw: results))
bigquery = pretend.stub(query=pretend.call_recorder(lambda q: query))
cacher = pretend.stub(purge=pretend.call_recorder(lambda keys: None))
def find_service(iface=None, name=None):
if iface is None and name == "gcloud.bigquery":
return bigquery
if with_purges and issubclass(iface, IOriginCache):
return cacher
raise LookupError
db_request.find_service = find_service
db_request.registry.settings = {
"warehouse.trending_table": "example.pypi.downloads*"
}
compute_trending(db_request)
assert bigquery.query.calls == [
pretend.call(
""" SELECT project,
IF(
STDDEV(downloads) > 0,
(todays_downloads - AVG(downloads))/STDDEV(downloads),
NULL
) as zscore
FROM (
SELECT project,
date,
downloads,
FIRST_VALUE(downloads) OVER (
PARTITION BY project
ORDER BY DATE DESC
ROWS BETWEEN UNBOUNDED PRECEDING
AND UNBOUNDED FOLLOWING
) as todays_downloads
FROM (
SELECT file.project as project,
DATE(timestamp) AS date,
COUNT(*) as downloads
FROM `example.pypi.downloads*`
WHERE _TABLE_SUFFIX BETWEEN
FORMAT_DATE(
"%Y%m%d",
DATE_ADD(CURRENT_DATE(), INTERVAL -31 day))
AND
FORMAT_DATE(
"%Y%m%d",
DATE_ADD(CURRENT_DATE(), INTERVAL -1 day))
GROUP BY file.project, date
)
)
GROUP BY project, todays_downloads
HAVING SUM(downloads) >= 5000
ORDER BY zscore DESC
"""
)
]
assert query.result.calls == [pretend.call()]
assert cacher.purge.calls == (
[pretend.call(["trending"])] if with_purges else []
)
results = dict(db_request.db.query(Project.name, Project.zscore).all())
assert results == {
projects[0].name: None,
projects[1].name: 2,
projects[2].name: -1,
}
def test_computes_trending(self, db_request, with_purges):
projects = [
ProjectFactory.create(zscore=1 if not i else None)
for i in range(3)
]
results = iter([
Row((projects[1].normalized_name, 2), {
"project": 0,
"zscore": 1
}),
Row((projects[2].normalized_name, -1), {
"project": 0,
"zscore": 1
}),
])
query = pretend.stub(
result=pretend.call_recorder(lambda *a, **kw: results))
bigquery = pretend.stub(query=pretend.call_recorder(lambda q: query))
cacher = pretend.stub(purge=pretend.call_recorder(lambda keys: None))
def find_service(iface=None, name=None):
if iface is None and name == "gcloud.bigquery":
return bigquery
if with_purges and issubclass(iface, IOriginCache):
return cacher
raise LookupError
db_request.find_service = find_service
db_request.registry.settings = {
"warehouse.trending_table": "example.pypi.downloads*"
}
compute_trending(db_request)
assert bigquery.query.calls == [
pretend.call(""" SELECT project,
IF(
STDDEV(downloads) > 0,
(todays_downloads - AVG(downloads))/STDDEV(downloads),
NULL
) as zscore
FROM (
SELECT project,
date,
downloads,
FIRST_VALUE(downloads) OVER (
PARTITION BY project
ORDER BY DATE DESC
ROWS BETWEEN UNBOUNDED PRECEDING
AND UNBOUNDED FOLLOWING
) as todays_downloads
FROM (
SELECT file.project as project,
DATE(timestamp) AS date,
COUNT(*) as downloads
FROM `example.pypi.downloads*`
WHERE _TABLE_SUFFIX BETWEEN
FORMAT_DATE(
"%Y%m%d",
DATE_ADD(CURRENT_DATE(), INTERVAL -31 day))
AND
FORMAT_DATE(
"%Y%m%d",
DATE_ADD(CURRENT_DATE(), INTERVAL -1 day))
GROUP BY file.project, date
)
)
GROUP BY project, todays_downloads
HAVING SUM(downloads) >= 5000
ORDER BY zscore DESC
""")
]
assert query.result.calls == [pretend.call()]
assert cacher.purge.calls == ([pretend.call(["trending"])]
if with_purges else [])
results = dict(db_request.db.query(Project.name, Project.zscore).all())
assert results == {
projects[0].name: None,
projects[1].name: 2,
projects[2].name: -1,
}