|
| 1 | +from __future__ import annotations |
| 2 | + |
| 3 | +from typing import TYPE_CHECKING |
| 4 | + |
| 5 | +import pytest |
| 6 | + |
| 7 | +from crawlee.storage_clients._file_system._dataset_client import FileSystemDatasetClient |
| 8 | +from crawlee.storage_clients._file_system._key_value_store_client import FileSystemKeyValueStoreClient |
| 9 | +from crawlee.storage_clients._file_system._storage_client import FileSystemStorageClient |
| 10 | + |
| 11 | +if TYPE_CHECKING: |
| 12 | + from pathlib import Path |
| 13 | + |
| 14 | +pytestmark = pytest.mark.only |
| 15 | + |
| 16 | + |
| 17 | +@pytest.fixture |
| 18 | +async def client() -> FileSystemStorageClient: |
| 19 | + return FileSystemStorageClient() |
| 20 | + |
| 21 | + |
| 22 | +async def test_open_dataset_client(client: FileSystemStorageClient, tmp_path: Path) -> None: |
| 23 | + """Test that open_dataset_client creates a dataset client with correct type and properties.""" |
| 24 | + dataset_client = await client.open_dataset_client(name='test-dataset', storage_dir=tmp_path) |
| 25 | + |
| 26 | + # Verify correct client type and properties |
| 27 | + assert isinstance(dataset_client, FileSystemDatasetClient) |
| 28 | + assert dataset_client.name == 'test-dataset' |
| 29 | + |
| 30 | + # Verify directory structure was created |
| 31 | + assert dataset_client.path_to_dataset.exists() |
| 32 | + |
| 33 | + |
| 34 | +async def test_dataset_client_purge_on_start(client: FileSystemStorageClient, tmp_path: Path) -> None: |
| 35 | + """Test that purge_on_start=True clears existing data in the dataset.""" |
| 36 | + # Create dataset and add data |
| 37 | + dataset_client1 = await client.open_dataset_client( |
| 38 | + name='test-purge-dataset', |
| 39 | + storage_dir=tmp_path, |
| 40 | + purge_on_start=True, |
| 41 | + ) |
| 42 | + await dataset_client1.push_data({'item': 'initial data'}) |
| 43 | + |
| 44 | + # Verify data was added |
| 45 | + items = await dataset_client1.get_data() |
| 46 | + assert len(items.items) == 1 |
| 47 | + |
| 48 | + # Reopen |
| 49 | + dataset_client2 = await client.open_dataset_client( |
| 50 | + name='test-purge-dataset', |
| 51 | + storage_dir=tmp_path, |
| 52 | + purge_on_start=True, |
| 53 | + ) |
| 54 | + |
| 55 | + # Verify data was purged |
| 56 | + items = await dataset_client2.get_data() |
| 57 | + assert len(items.items) == 0 |
| 58 | + |
| 59 | + |
| 60 | +async def test_dataset_client_no_purge_on_start(client: FileSystemStorageClient, tmp_path: Path) -> None: |
| 61 | + """Test that purge_on_start=False keeps existing data in the dataset.""" |
| 62 | + # Create dataset and add data |
| 63 | + dataset_client1 = await client.open_dataset_client( |
| 64 | + name='test-no-purge-dataset', |
| 65 | + storage_dir=tmp_path, |
| 66 | + purge_on_start=False, |
| 67 | + ) |
| 68 | + await dataset_client1.push_data({'item': 'preserved data'}) |
| 69 | + |
| 70 | + # Reopen |
| 71 | + dataset_client2 = await client.open_dataset_client( |
| 72 | + name='test-no-purge-dataset', |
| 73 | + storage_dir=tmp_path, |
| 74 | + purge_on_start=False, |
| 75 | + ) |
| 76 | + |
| 77 | + # Verify data was preserved |
| 78 | + items = await dataset_client2.get_data() |
| 79 | + assert len(items.items) == 1 |
| 80 | + assert items.items[0]['item'] == 'preserved data' |
| 81 | + |
| 82 | + |
| 83 | +async def test_open_kvs_client(client: FileSystemStorageClient, tmp_path: Path) -> None: |
| 84 | + """Test that open_key_value_store_client creates a KVS client with correct type and properties.""" |
| 85 | + kvs_client = await client.open_key_value_store_client(name='test-kvs', storage_dir=tmp_path) |
| 86 | + |
| 87 | + # Verify correct client type and properties |
| 88 | + assert isinstance(kvs_client, FileSystemKeyValueStoreClient) |
| 89 | + assert kvs_client.name == 'test-kvs' |
| 90 | + |
| 91 | + # Verify directory structure was created |
| 92 | + assert kvs_client.path_to_kvs.exists() |
| 93 | + |
| 94 | + |
| 95 | +async def test_kvs_client_purge_on_start(client: FileSystemStorageClient, tmp_path: Path) -> None: |
| 96 | + """Test that purge_on_start=True clears existing data in the key-value store.""" |
| 97 | + # Create KVS and add data |
| 98 | + kvs_client1 = await client.open_key_value_store_client( |
| 99 | + name='test-purge-kvs', |
| 100 | + storage_dir=tmp_path, |
| 101 | + purge_on_start=True, |
| 102 | + ) |
| 103 | + await kvs_client1.set_value(key='test-key', value='initial value') |
| 104 | + |
| 105 | + # Verify value was set |
| 106 | + record = await kvs_client1.get_value(key='test-key') |
| 107 | + assert record is not None |
| 108 | + assert record.value == 'initial value' |
| 109 | + |
| 110 | + # Reopen |
| 111 | + kvs_client2 = await client.open_key_value_store_client( |
| 112 | + name='test-purge-kvs', |
| 113 | + storage_dir=tmp_path, |
| 114 | + purge_on_start=True, |
| 115 | + ) |
| 116 | + |
| 117 | + # Verify value was purged |
| 118 | + record = await kvs_client2.get_value(key='test-key') |
| 119 | + assert record is None |
| 120 | + |
| 121 | + |
| 122 | +async def test_kvs_client_no_purge_on_start(client: FileSystemStorageClient, tmp_path: Path) -> None: |
| 123 | + """Test that purge_on_start=False keeps existing data in the key-value store.""" |
| 124 | + # Create KVS and add data |
| 125 | + kvs_client1 = await client.open_key_value_store_client( |
| 126 | + name='test-no-purge-kvs', |
| 127 | + storage_dir=tmp_path, |
| 128 | + purge_on_start=False, |
| 129 | + ) |
| 130 | + await kvs_client1.set_value(key='test-key', value='preserved value') |
| 131 | + |
| 132 | + # Reopen |
| 133 | + kvs_client2 = await client.open_key_value_store_client( |
| 134 | + name='test-no-purge-kvs', |
| 135 | + storage_dir=tmp_path, |
| 136 | + purge_on_start=False, |
| 137 | + ) |
| 138 | + |
| 139 | + # Verify value was preserved |
| 140 | + record = await kvs_client2.get_value(key='test-key') |
| 141 | + assert record is not None |
| 142 | + assert record.value == 'preserved value' |
0 commit comments