|
| 1 | +from typing import Optional, Type, Union |
| 2 | + |
| 3 | +import luigi |
| 4 | +import pytest |
| 5 | + |
| 6 | +import gokart |
| 7 | +from gokart.in_memory import InMemoryCacheRepository, InMemoryTarget |
| 8 | +from gokart.target import SingleFileTarget |
| 9 | + |
| 10 | + |
| 11 | +class DummyTask(gokart.TaskOnKart): |
| 12 | + task_namespace = __name__ |
| 13 | + param: str = luigi.Parameter() |
| 14 | + |
| 15 | + def run(self): |
| 16 | + self.dump(self.param) |
| 17 | + |
| 18 | + |
| 19 | +class DummyTaskWithDependencies(gokart.TaskOnKart): |
| 20 | + task_namespace = __name__ |
| 21 | + task: list[gokart.TaskOnKart[str]] = gokart.ListTaskInstanceParameter() |
| 22 | + |
| 23 | + def run(self): |
| 24 | + result = ','.join(self.load()) |
| 25 | + self.dump(result) |
| 26 | + |
| 27 | + |
| 28 | +class DumpIntTask(gokart.TaskOnKart[int]): |
| 29 | + task_namespace = __name__ |
| 30 | + value: int = luigi.IntParameter() |
| 31 | + |
| 32 | + def run(self): |
| 33 | + self.dump(self.value) |
| 34 | + |
| 35 | + |
| 36 | +class AddTask(gokart.TaskOnKart[Union[int, float]]): |
| 37 | + a: gokart.TaskOnKart[int] = gokart.TaskInstanceParameter() |
| 38 | + b: gokart.TaskOnKart[int] = gokart.TaskInstanceParameter() |
| 39 | + |
| 40 | + def requires(self): |
| 41 | + return dict(a=self.a, b=self.b) |
| 42 | + |
| 43 | + def run(self): |
| 44 | + a = self.load(self.a) |
| 45 | + b = self.load(self.b) |
| 46 | + self.dump(a + b) |
| 47 | + |
| 48 | + |
| 49 | +class TestTaskOnKartWithCache: |
| 50 | + @pytest.fixture(autouse=True) |
| 51 | + def clear_repository(slef): |
| 52 | + InMemoryCacheRepository().clear() |
| 53 | + |
| 54 | + @pytest.mark.parametrize('data_key', ['sample_key', None]) |
| 55 | + @pytest.mark.parametrize('use_unique_id', [True, False]) |
| 56 | + def test_key_identity(self, data_key: Optional[str], use_unique_id: bool): |
| 57 | + task = DummyTask(param='param') |
| 58 | + ext = '.pkl' |
| 59 | + relative_file_path = data_key + ext if data_key else None |
| 60 | + target = task.make_target(relative_file_path=relative_file_path, use_unique_id=use_unique_id) |
| 61 | + cached_target = task.make_cache_target(data_key=data_key, use_unique_id=use_unique_id) |
| 62 | + |
| 63 | + target_path = target.path().removeprefix(task.workspace_directory).removesuffix(ext).strip('/') |
| 64 | + assert cached_target.path() == target_path |
| 65 | + |
| 66 | + def test_make_cached_target(self): |
| 67 | + task = DummyTask(param='param') |
| 68 | + target = task.make_cache_target() |
| 69 | + assert isinstance(target, InMemoryTarget) |
| 70 | + |
| 71 | + @pytest.mark.parametrize(['cache_in_memory_by_default', 'target_type'], [[True, InMemoryTarget], [False, SingleFileTarget]]) |
| 72 | + def test_make_default_target(self, cache_in_memory_by_default: bool, target_type: Type[gokart.TaskOnKart]): |
| 73 | + task = DummyTask(param='param', cache_in_memory_by_default=cache_in_memory_by_default) |
| 74 | + target = task.output() |
| 75 | + assert isinstance(target, target_type) |
| 76 | + |
| 77 | + def test_complete_with_cache_in_memory_flag(self, tmpdir): |
| 78 | + task = DummyTask(param='param', cache_in_memory_by_default=True, workspace_directory=tmpdir) |
| 79 | + assert not task.complete() |
| 80 | + file_target = task.make_target() |
| 81 | + file_target.dump('data') |
| 82 | + assert not task.complete() |
| 83 | + cache_target = task.make_cache_target() |
| 84 | + cache_target.dump('data') |
| 85 | + assert task.complete() |
| 86 | + |
| 87 | + def test_complete_without_cache_in_memory_flag(self, tmpdir): |
| 88 | + task = DummyTask(param='param', workspace_directory=tmpdir) |
| 89 | + assert not task.complete() |
| 90 | + cache_target = task.make_cache_target() |
| 91 | + cache_target.dump('data') |
| 92 | + assert not task.complete() |
| 93 | + file_target = task.make_target() |
| 94 | + file_target.dump('data') |
| 95 | + assert task.complete() |
| 96 | + |
| 97 | + def test_dump_with_cache_in_memory_flag(self, tmpdir): |
| 98 | + task = DummyTask(param='param', cache_in_memory_by_default=True, workspace_directory=tmpdir) |
| 99 | + file_target = task.make_target() |
| 100 | + cache_target = task.make_cache_target() |
| 101 | + task.dump('data') |
| 102 | + assert not file_target.exists() |
| 103 | + assert cache_target.exists() |
| 104 | + |
| 105 | + def test_dump_without_cache_in_memory_flag(self, tmpdir): |
| 106 | + task = DummyTask(param='param', workspace_directory=tmpdir) |
| 107 | + file_target = task.make_target() |
| 108 | + cache_target = task.make_cache_target() |
| 109 | + task.dump('data') |
| 110 | + assert file_target.exists() |
| 111 | + assert not cache_target.exists() |
| 112 | + |
| 113 | + def test_gokart_build(self): |
| 114 | + task = AddTask( |
| 115 | + a=DumpIntTask(value=2, cache_in_memory_by_default=True), b=DumpIntTask(value=3, cache_in_memory_by_default=True), cache_in_memory_by_default=True |
| 116 | + ) |
| 117 | + output = gokart.build(task, reset_register=False) |
| 118 | + assert output == 5 |
0 commit comments