|
| 1 | +import unittest |
| 2 | +from datetime import datetime, timedelta, timezone |
| 3 | +from .stats import get_statistics |
| 4 | +from .types import Table, MaxValuePartition, PositionPartition |
| 5 | + |
| 6 | + |
| 7 | +def mkPPart(name, *pos): |
| 8 | + p = PositionPartition(name) |
| 9 | + for x in pos: |
| 10 | + p.add_position(x) |
| 11 | + return p |
| 12 | + |
| 13 | + |
| 14 | +ts = datetime(1949, 1, 12, tzinfo=timezone.utc) |
| 15 | + |
| 16 | + |
| 17 | +class TestStatistics(unittest.TestCase): |
| 18 | + def test_statistics_no_partitions(self): |
| 19 | + s = get_statistics(list(), ts, Table("no_parts")) |
| 20 | + self.assertEqual(s, {"partitions": 0}) |
| 21 | + |
| 22 | + def test_statistics_single_unnamed_partition(self): |
| 23 | + s = get_statistics([MaxValuePartition("p_start", 1)], ts, Table("single_part")) |
| 24 | + self.assertEqual(s, {"partitions": 1}) |
| 25 | + |
| 26 | + def test_statistics_single_partition(self): |
| 27 | + s = get_statistics( |
| 28 | + [MaxValuePartition("p_19480113", 1)], ts, Table("single_part") |
| 29 | + ) |
| 30 | + self.assertEqual( |
| 31 | + s, {"partitions": 1, "time_since_last_partition": timedelta(days=365)} |
| 32 | + ) |
| 33 | + |
| 34 | + def test_statistics_two_partitions(self): |
| 35 | + s = get_statistics( |
| 36 | + [mkPPart("p_19480101", 42), MaxValuePartition("p_19490101", 1)], |
| 37 | + ts, |
| 38 | + Table("two_parts"), |
| 39 | + ) |
| 40 | + self.assertEqual( |
| 41 | + s, |
| 42 | + { |
| 43 | + "partitions": 2, |
| 44 | + "time_since_last_partition": timedelta(days=11), |
| 45 | + "mean_partition_delta": timedelta(days=366), |
| 46 | + "max_partition_delta": timedelta(days=366), |
| 47 | + }, |
| 48 | + ) |
| 49 | + |
| 50 | + def test_statistics_weekly_partitions_year(self): |
| 51 | + parts = list() |
| 52 | + base = datetime(2020, 5, 20, tzinfo=timezone.utc) |
| 53 | + for w in range(0, 52): |
| 54 | + partName = f"p_{base + timedelta(weeks=w):%Y%m%d}" |
| 55 | + parts.append(mkPPart(partName, w * 1024)) |
| 56 | + parts.append(MaxValuePartition(f"p_{base + timedelta(weeks=52):%Y%m%d}", 1)) |
| 57 | + |
| 58 | + s = get_statistics( |
| 59 | + parts, base + timedelta(weeks=54), Table("weekly_partitions_year_retention") |
| 60 | + ) |
| 61 | + self.assertEqual( |
| 62 | + s, |
| 63 | + { |
| 64 | + "partitions": 53, |
| 65 | + "time_since_last_partition": timedelta(days=14), |
| 66 | + "mean_partition_delta": timedelta(days=7), |
| 67 | + "max_partition_delta": timedelta(days=7), |
| 68 | + }, |
| 69 | + ) |
0 commit comments