|
| 1 | +from __future__ import annotations |
| 2 | + |
| 3 | +from typing import TYPE_CHECKING, Any |
| 4 | + |
| 5 | +from datacustomcode.client import Client |
| 6 | +from datacustomcode.config import ( |
| 7 | + AccessLayerObjectConfig, |
| 8 | + ClientConfig, |
| 9 | + SparkConfig, |
| 10 | + SparkProviderConfig, |
| 11 | +) |
| 12 | +from datacustomcode.io.reader.base import BaseDataCloudReader |
| 13 | +from datacustomcode.io.writer.base import BaseDataCloudWriter, WriteMode |
| 14 | +from datacustomcode.spark.base import BaseSparkSessionProvider |
| 15 | + |
| 16 | +if TYPE_CHECKING: |
| 17 | + from pyspark.sql import DataFrame as PySparkDataFrame |
| 18 | + |
| 19 | + |
| 20 | +class _Sentinel: |
| 21 | + pass |
| 22 | + |
| 23 | + |
| 24 | +SENTINEL_SPARK = _Sentinel() |
| 25 | + |
| 26 | + |
| 27 | +class MockReader(BaseDataCloudReader): |
| 28 | + CONFIG_NAME = "MockReader" |
| 29 | + last_spark: Any | None = None |
| 30 | + |
| 31 | + def __init__(self, spark): |
| 32 | + super().__init__(spark) |
| 33 | + MockReader.last_spark = spark |
| 34 | + |
| 35 | + def read_dlo(self, name: str): # type: ignore[override] |
| 36 | + raise NotImplementedError |
| 37 | + |
| 38 | + def read_dmo(self, name: str): # type: ignore[override] |
| 39 | + raise NotImplementedError |
| 40 | + |
| 41 | + |
| 42 | +class MockWriter(BaseDataCloudWriter): |
| 43 | + CONFIG_NAME = "MockWriter" |
| 44 | + last_spark: Any | None = None |
| 45 | + |
| 46 | + def __init__(self, spark): |
| 47 | + super().__init__(spark) |
| 48 | + MockWriter.last_spark = spark |
| 49 | + |
| 50 | + def write_to_dlo( |
| 51 | + self, name: str, dataframe: PySparkDataFrame, write_mode: WriteMode |
| 52 | + ) -> None: # type: ignore[override] |
| 53 | + raise NotImplementedError |
| 54 | + |
| 55 | + def write_to_dmo( |
| 56 | + self, name: str, dataframe: PySparkDataFrame, write_mode: WriteMode |
| 57 | + ) -> None: # type: ignore[override] |
| 58 | + raise NotImplementedError |
| 59 | + |
| 60 | + |
| 61 | +class FakeProvider(BaseSparkSessionProvider): |
| 62 | + CONFIG_NAME = "FakeProvider" |
| 63 | + |
| 64 | + def get_session(self, spark_config: SparkConfig): # type: ignore[override] |
| 65 | + return SENTINEL_SPARK |
| 66 | + |
| 67 | + |
| 68 | +def _reset_singleton(): |
| 69 | + # Reset Client singleton between tests |
| 70 | + Client._instance = None # type: ignore[attr-defined] |
| 71 | + |
| 72 | + |
| 73 | +def test_client_uses_provider_from_config(monkeypatch): |
| 74 | + _reset_singleton() |
| 75 | + |
| 76 | + cfg = ClientConfig( |
| 77 | + reader_config=AccessLayerObjectConfig( |
| 78 | + type_config_name=MockReader.CONFIG_NAME, options={} |
| 79 | + ), |
| 80 | + writer_config=AccessLayerObjectConfig( |
| 81 | + type_config_name=MockWriter.CONFIG_NAME, options={} |
| 82 | + ), |
| 83 | + spark_config=SparkConfig(app_name="test-app", master=None, options={}), |
| 84 | + spark_provider_config=SparkProviderConfig( |
| 85 | + type_config_name=FakeProvider.CONFIG_NAME, options={} |
| 86 | + ), |
| 87 | + ) |
| 88 | + |
| 89 | + from datacustomcode.config import config as global_config |
| 90 | + |
| 91 | + global_config.update(cfg) |
| 92 | + |
| 93 | + Client() |
| 94 | + assert MockReader.last_spark is SENTINEL_SPARK |
| 95 | + assert MockWriter.last_spark is SENTINEL_SPARK |
| 96 | + |
| 97 | + |
| 98 | +class ExplicitProvider(BaseSparkSessionProvider): |
| 99 | + CONFIG_NAME = "ExplicitProvider" |
| 100 | + |
| 101 | + def get_session(self, spark_config: SparkConfig): # type: ignore[override] |
| 102 | + return SENTINEL_SPARK |
| 103 | + |
| 104 | + |
| 105 | +def test_client_explicit_provider_overrides_config(monkeypatch): |
| 106 | + _reset_singleton() |
| 107 | + |
| 108 | + cfg = ClientConfig( |
| 109 | + reader_config=AccessLayerObjectConfig( |
| 110 | + type_config_name=MockReader.CONFIG_NAME, options={} |
| 111 | + ), |
| 112 | + writer_config=AccessLayerObjectConfig( |
| 113 | + type_config_name=MockWriter.CONFIG_NAME, options={} |
| 114 | + ), |
| 115 | + spark_config=SparkConfig(app_name="test-app", master=None, options={}), |
| 116 | + spark_provider_config=None, |
| 117 | + ) |
| 118 | + |
| 119 | + from datacustomcode.config import config as global_config |
| 120 | + |
| 121 | + global_config.update(cfg) |
| 122 | + |
| 123 | + provider = ExplicitProvider() |
| 124 | + Client(spark_provider=provider) |
| 125 | + assert MockReader.last_spark is SENTINEL_SPARK |
| 126 | + assert MockWriter.last_spark is SENTINEL_SPARK |
0 commit comments