def test_get_time_buckets_misaligned():
# Test for regression on bug when querying over x buckets for a timeframe
# (end - start) of < x buckets (i.e. end of B0 to start of B2)
start = DatalakeRecord.TIME_BUCKET_SIZE_IN_MS * 4 / 5
end = DatalakeRecord.TIME_BUCKET_SIZE_IN_MS * 11 / 5
buckets = DatalakeRecord.get_time_buckets(start, end)
assert buckets == [0, 1, 2]
def test_get_time_buckets_misaligned():
# Test for regression on bug when querying over x buckets for a timeframe
# (end - start) of < x buckets (i.e. end of B0 to start of B2)
start = DatalakeRecord.TIME_BUCKET_SIZE_IN_MS * 4 / 5
end = DatalakeRecord.TIME_BUCKET_SIZE_IN_MS * 11 / 5
buckets = DatalakeRecord.get_time_buckets(start, end)
assert buckets == [0, 1, 2]
def query_by_time(self, start, end, what, where=None, cursor=None):
results = []
buckets = DatalakeRecord.get_time_buckets(start, end)
if cursor:
current_bucket = cursor['current_time_bucket']
i = buckets.index(current_bucket)
buckets = buckets[i:]
for b in buckets:
cursor = self._query_time_bucket(b, results, start, end, what,
where, cursor)
return QueryResults(results, cursor)
def query_by_time(self, start, end, what, where=None, cursor=None):
results = []
buckets = DatalakeRecord.get_time_buckets(start, end)
if cursor:
current_bucket = cursor['current_time_bucket']
i = buckets.index(current_bucket)
buckets = buckets[i:]
for b in buckets:
cursor = self._query_time_bucket(b, results, start, end, what,
where, cursor)
if cursor and \
cursor.current_time_bucket and \
cursor.current_time_bucket > buckets[-1]:
# this is a corner case. It means that the next query would take us
# into the next bucket, but the next bucket is beyond the time of
# interest. Just clear the cursor in this case.
cursor = None
return QueryResults(results, cursor)
