|
| 1 | +import unittest |
| 2 | + |
| 3 | +import pyarrow as pa |
| 4 | +import pyiceberg.types as types |
| 5 | +from pyiceberg.expressions import EqualTo |
| 6 | +from pyiceberg.schema import Schema |
| 7 | + |
| 8 | +from servc.svc.com.storage.iceberg import IceBerg |
| 9 | +from servc.svc.com.storage.lake import LakeTable, Medallion |
| 10 | + |
| 11 | +schema = pa.schema( |
| 12 | + [ |
| 13 | + ("date", pa.string()), |
| 14 | + ("some_int", pa.int64()), |
| 15 | + ] |
| 16 | +) |
| 17 | + |
| 18 | +mytable: LakeTable = { |
| 19 | + "name": "test", |
| 20 | + "partitions": ["date"], |
| 21 | + "medallion": Medallion.BRONZE, |
| 22 | + "schema": Schema( |
| 23 | + types.NestedField(field_id=1, name="date", type=types.StringType()), |
| 24 | + types.NestedField(field_id=2, name="some_int", type=types.IntegerType()), |
| 25 | + ), |
| 26 | +} |
| 27 | + |
| 28 | +config = { |
| 29 | + "database": "default", |
| 30 | + "catalog_name": "default", |
| 31 | + "catalog_properties": { |
| 32 | + "type": "sql", |
| 33 | + "uri": "sqlite:////tmp/pyiceberg.db", |
| 34 | + "echo": "true", |
| 35 | + "init_catalog_tables": "true", |
| 36 | + "warehouse": "file:///tmp/warehouse", |
| 37 | + }, |
| 38 | +} |
| 39 | + |
| 40 | + |
| 41 | +class TestLakeIceberg(unittest.TestCase): |
| 42 | + @classmethod |
| 43 | + def setUpClass(cls) -> None: |
| 44 | + cls.iceberg = IceBerg(config, mytable) |
| 45 | + |
| 46 | + def test_connect(self): |
| 47 | + self.iceberg._connect() |
| 48 | + self.assertTrue(self.iceberg.isOpen) |
| 49 | + |
| 50 | + def test_name(self): |
| 51 | + self.assertEqual(self.iceberg.tablename, "default.bronze_test") |
| 52 | + |
| 53 | + def test_insert(self): |
| 54 | + self.iceberg.overwrite([]) |
| 55 | + self.iceberg.insert([{"date": "2021-01-01", "some_int": 1}]) |
| 56 | + data = self.iceberg.read(["date"]).to_pylist() |
| 57 | + self.assertEqual(len(data), 1) |
| 58 | + self.assertEqual(data, [{"date": "2021-01-01"}]) |
| 59 | + |
| 60 | + def test_overwrite(self): |
| 61 | + self.iceberg.overwrite([]) |
| 62 | + self.iceberg.insert([{"date": "2021-01-01", "some_int": 1}]) |
| 63 | + self.iceberg.insert([{"date": "2021-01-02", "some_int": 1}]) |
| 64 | + self.iceberg.insert([{"date": "2021-01-02", "some_int": 3}]) |
| 65 | + |
| 66 | + data = self.iceberg.read(["date"]).to_pylist() |
| 67 | + self.assertEqual(len(data), 3) |
| 68 | + |
| 69 | + self.iceberg.overwrite([], {"date": ["2021-01-02"]}) |
| 70 | + data = self.iceberg.read(["date"]).to_pylist() |
| 71 | + self.assertEqual(len(data), 1) |
| 72 | + |
| 73 | + def test_reading_partitions(self): |
| 74 | + self.iceberg.overwrite([]) |
| 75 | + self.iceberg.insert([{"date": "2021-01-01", "some_int": 1}]) |
| 76 | + self.iceberg.insert([{"date": "2021-01-02", "some_int": 1}]) |
| 77 | + self.iceberg.insert([{"date": "2021-01-02", "some_int": 3}]) |
| 78 | + |
| 79 | + data = self.iceberg.read( |
| 80 | + ["date"], partitions={"date": ["2021-01-01"]} |
| 81 | + ).to_pylist() |
| 82 | + self.assertEqual(len(data), 1) |
| 83 | + |
| 84 | + data = self.iceberg.read( |
| 85 | + ["date"], partitions={"date": ["2021-01-02"]} |
| 86 | + ).to_pylist() |
| 87 | + self.assertEqual(len(data), 2) |
| 88 | + |
| 89 | + data = self.iceberg.read( |
| 90 | + ["date"], partitions={"date": ["2021-01-02", "2021-01-01"]} |
| 91 | + ).to_pylist() |
| 92 | + self.assertEqual(len(data), 3) |
| 93 | + |
| 94 | + data = self.iceberg.read( |
| 95 | + ["date"], |
| 96 | + options={"limit": 1}, |
| 97 | + ).to_pylist() |
| 98 | + self.assertEqual(len(data), 1) |
| 99 | + |
| 100 | + data = self.iceberg.read( |
| 101 | + ["date"], |
| 102 | + partitions={"date": ["2021-01-02"]}, |
| 103 | + options={"row_filter": EqualTo("some_int", 3)}, |
| 104 | + ).to_pylist() |
| 105 | + self.assertEqual(len(data), 1) |
| 106 | + |
| 107 | + data = self.iceberg.read( |
| 108 | + ["date"], |
| 109 | + partitions={"date": ["2021-01-02"], "some_int": [3]}, |
| 110 | + ).to_pylist() |
| 111 | + self.assertEqual(len(data), 1) |
| 112 | + |
| 113 | + def test_load_from_catalog(self): |
| 114 | + self.iceberg.insert([{"date": "2021-01-01", "some_int": 1}]) |
| 115 | + orig_data = self.iceberg.read(["date"]).to_pylist() |
| 116 | + |
| 117 | + iceberg = IceBerg(config, "default.bronze_test") |
| 118 | + iceberg._connect() |
| 119 | + self.assertTrue(iceberg.isOpen) |
| 120 | + |
| 121 | + data = iceberg.read(["date"]).to_pylist() |
| 122 | + self.assertEqual(orig_data, data) |
| 123 | + self.assertGreater(len(data), 0) |
| 124 | + |
| 125 | + def test_version_travel(self): |
| 126 | + self.iceberg.insert([{"date": "2021-01-01", "some_int": 1}]) |
| 127 | + orig_data = self.iceberg.read(["date"]).to_pylist() |
| 128 | + currentVersion = self.iceberg.getCurrentVersion() |
| 129 | + |
| 130 | + versions = self.iceberg.getVersions() |
| 131 | + self.assertGreater(len(versions), 0) |
| 132 | + self.assertIn(currentVersion, versions) |
| 133 | + |
| 134 | + self.iceberg.insert([{"date": "2021-01-02", "some_int": 1}]) |
| 135 | + new_version = self.iceberg.getCurrentVersion() |
| 136 | + self.assertNotEqual(currentVersion, new_version) |
| 137 | + |
| 138 | + data = self.iceberg.read(["date"], version=currentVersion).to_pylist() |
| 139 | + self.assertEqual(len(data), len(orig_data)) |
| 140 | + self.assertEqual(data, orig_data) |
| 141 | + |
| 142 | + def test_partitions(self): |
| 143 | + self.iceberg.overwrite([]) |
| 144 | + self.iceberg.insert([{"date": "2021-01-01", "some_int": 1}]) |
| 145 | + self.iceberg.insert([{"date": "2021-01-02", "some_int": 1}]) |
| 146 | + self.iceberg.insert([{"date": "2021-01-02", "some_int": 3}]) |
| 147 | + |
| 148 | + partitions = self.iceberg.getPartitions() |
| 149 | + self.assertEqual(list(partitions.keys()), ["date"]) |
| 150 | + self.assertEqual(len(partitions["date"]), 2) |
| 151 | + self.assertIn("2021-01-01", partitions["date"]) |
| 152 | + self.assertIn("2021-01-02", partitions["date"]) |
| 153 | + |
| 154 | + def test_schema(self): |
| 155 | + schema = self.iceberg.getSchema() |
| 156 | + self.assertIsInstance(schema, pa.Schema) |
| 157 | + self.assertEqual(len(schema.names), 2) |
| 158 | + self.assertEqual(schema.names, ["date", "some_int"]) |
| 159 | + |
| 160 | + def test_close(self): |
| 161 | + self.iceberg.close() |
| 162 | + self.iceberg.connect() |
| 163 | + self.iceberg.close() |
| 164 | + |
| 165 | + |
| 166 | +if __name__ == "__main__": |
| 167 | + unittest.main() |
0 commit comments