from odps.df.tools.lib import HyperLogLog
from odps.tests.core import TestBase
class Test(TestBase):
def testHLL(self):
hll = HyperLogLog(0.05)
buf = hll.buffer()
for i in xrange(10000):
hll(buf, str(i))
self.assertAlmostEqual(hll.getvalue(buf) / float(10000), 1, delta=0.1)
for i in xrange(100000, 200000):
hll(buf, str(i))
self.assertAlmostEqual(hll.getvalue(buf) / 110000, 1, delta=0.2)
buf2 = hll.buffer()
for i in xrange(10000):
hll(buf2, str(i))
hll.merge(buf, buf2)
self.assertAlmostEqual(hll.getvalue(buf) / 110000, 1, delta=0.2)
if __name__ == '__main__':
unittest.main()
self.assertTrue(expr._sort_fields[0]._ascending)
self.assertFalse(expr._sort_fields[1]._ascending)
expr = self.expr.map_reduce(mapper,
reducer,
group='name',
sort=['rating', 'id'],
ascending=[False, True])
self.assertEqual(expr.schema.names, ['name', 'rating'])
self.assertEqual(len(expr._sort_fields), 3)
self.assertTrue(expr._sort_fields[0]._ascending)
self.assertFalse(expr._sort_fields[1]._ascending)
self.assertTrue(expr._sort_fields[2]._ascending)
expr = self.expr.map_reduce(mapper,
reducer,
group='name',
sort=['rating', 'id'],
ascending=False)
self.assertEqual(expr.schema.names, ['name', 'rating'])
self.assertEqual(len(expr._sort_fields), 3)
self.assertTrue(expr._sort_fields[0]._ascending)
self.assertFalse(expr._sort_fields[1]._ascending)
self.assertFalse(expr._sort_fields[2]._ascending)
if __name__ == '__main__':
unittest.main()