|
| 1 | +import itertools |
| 2 | + |
| 3 | +from parameterized import parameterized |
| 4 | +from torch.testing._internal.common_utils import TestCase |
| 5 | +from torchdata.nodes import Header |
| 6 | +from torchdata.nodes.adapters import IterableWrapper |
| 7 | + |
| 8 | +from .utils import MockSource, run_test_save_load_state, StatefulRangeNode |
| 9 | + |
| 10 | + |
| 11 | +class TestHeader(TestCase): |
| 12 | + def test_header_basic(self) -> None: |
| 13 | + # Test with a simple range |
| 14 | + source = IterableWrapper(range(10)) |
| 15 | + node = Header(source, n=5) |
| 16 | + |
| 17 | + results = list(node) |
| 18 | + self.assertEqual(results, [0, 1, 2, 3, 4]) |
| 19 | + |
| 20 | + # Verify counter |
| 21 | + self.assertEqual(node._num_yielded, 5) |
| 22 | + |
| 23 | + # Test with n larger than source |
| 24 | + source = IterableWrapper(range(3)) |
| 25 | + node = Header(source, n=10) |
| 26 | + |
| 27 | + results = list(node) |
| 28 | + self.assertEqual(results, [0, 1, 2]) |
| 29 | + |
| 30 | + # Verify counter with n larger than source |
| 31 | + self.assertEqual(node._num_yielded, 3) |
| 32 | + |
| 33 | + # Test with n=0 (should yield nothing) |
| 34 | + source = IterableWrapper(range(10)) |
| 35 | + node = Header(source, n=0) |
| 36 | + |
| 37 | + results = list(node) |
| 38 | + self.assertEqual(results, []) |
| 39 | + |
| 40 | + # Verify counter with n=0 |
| 41 | + self.assertEqual(node._num_yielded, 0) |
| 42 | + |
| 43 | + def test_header_with_mock_source(self) -> None: |
| 44 | + num_samples = 20 |
| 45 | + source = MockSource(num_samples=num_samples) |
| 46 | + node = Header(source, n=7) # Limit to first 7 items |
| 47 | + |
| 48 | + # Test multi epoch |
| 49 | + for epoch in range(2): |
| 50 | + node.reset() |
| 51 | + results = list(node) |
| 52 | + self.assertEqual(len(results), 7) |
| 53 | + |
| 54 | + # Verify counter after each epoch |
| 55 | + self.assertEqual(node._num_yielded, 7) |
| 56 | + |
| 57 | + for i, result in enumerate(results): |
| 58 | + expected_step = i |
| 59 | + self.assertEqual(result["step"], expected_step) |
| 60 | + self.assertEqual(result["test_tensor"].item(), expected_step) |
| 61 | + self.assertEqual(result["test_str"], f"str_{expected_step}") |
| 62 | + |
| 63 | + def test_header_empty_source(self) -> None: |
| 64 | + source = IterableWrapper([]) |
| 65 | + node = Header(source, n=5) |
| 66 | + |
| 67 | + results = list(node) |
| 68 | + self.assertEqual(results, []) |
| 69 | + |
| 70 | + # Verify counter with empty source |
| 71 | + self.assertEqual(node._num_yielded, 0) |
| 72 | + |
| 73 | + @parameterized.expand(itertools.product([0, 3, 7])) |
| 74 | + def test_save_load_state(self, midpoint: int) -> None: |
| 75 | + n = 50 |
| 76 | + source = StatefulRangeNode(n=n) |
| 77 | + node = Header(source, n=20) # Limit to first 20 items |
| 78 | + run_test_save_load_state(self, node, midpoint) |
| 79 | + |
| 80 | + def test_header_reset_state(self) -> None: |
| 81 | + source = IterableWrapper(range(10)) |
| 82 | + node = Header(source, n=5) |
| 83 | + |
| 84 | + # Consume first two items |
| 85 | + self.assertEqual(next(node), 0) |
| 86 | + self.assertEqual(next(node), 1) |
| 87 | + |
| 88 | + # Check counter after consuming two items |
| 89 | + self.assertEqual(node._num_yielded, 2) |
| 90 | + |
| 91 | + # Get state and reset |
| 92 | + state = node.state_dict() |
| 93 | + node.reset(state) |
| 94 | + |
| 95 | + # Counter should be preserved after reset with state |
| 96 | + self.assertEqual(node._num_yielded, 2) |
| 97 | + |
| 98 | + # Should continue from where we left off |
| 99 | + self.assertEqual(next(node), 2) |
| 100 | + self.assertEqual(next(node), 3) |
| 101 | + self.assertEqual(next(node), 4) |
| 102 | + |
| 103 | + # Counter should be updated after consuming more items |
| 104 | + self.assertEqual(node._num_yielded, 5) |
| 105 | + |
| 106 | + # Should raise StopIteration after all items are consumed |
| 107 | + with self.assertRaises(StopIteration): |
| 108 | + next(node) |
| 109 | + |
| 110 | + def test_counter_reset(self) -> None: |
| 111 | + # Test that counter is properly reset |
| 112 | + source = IterableWrapper(range(10)) |
| 113 | + node = Header(source, n=5) |
| 114 | + |
| 115 | + # Consume all items |
| 116 | + list(node) |
| 117 | + |
| 118 | + # Verify counter after first pass |
| 119 | + self.assertEqual(node._num_yielded, 5) |
| 120 | + |
| 121 | + # Reset without state |
| 122 | + node.reset() |
| 123 | + |
| 124 | + # Counter should be reset to 0 |
| 125 | + self.assertEqual(node._num_yielded, 0) |
| 126 | + |
| 127 | + # Consume some items |
| 128 | + next(node) # 0 |
| 129 | + next(node) # 1 |
| 130 | + |
| 131 | + # Verify counter after partial consumption |
| 132 | + self.assertEqual(node._num_yielded, 2) |
| 133 | + |
| 134 | + def test_invalid_input(self) -> None: |
| 135 | + # Test with negative n |
| 136 | + source = IterableWrapper(range(10)) |
| 137 | + with self.assertRaises(ValueError): |
| 138 | + Header(source, n=-1) |
0 commit comments