File size: 1,595 Bytes
13065f0 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 | import unittest
import numpy as np
import torch
from hikka_forge import ForgeItem, ForgeVector
from hikka_forge.api import Forge2Vec
class PublicApiTests(unittest.TestCase):
def test_legacy_field_names_are_normalized(self):
item = ForgeItem.from_value({
"en_title": "Frieren",
"original_title": "Sousou no Frieren",
"alternate_names": ["Frieren at the Funeral"],
"en_description": "A fantasy journey.",
"type": "anime",
})
self.assertEqual(item.title, "Frieren")
self.assertEqual(item.native_title, "Sousou no Frieren")
self.assertEqual(item.synonyms, ["Frieren at the Funeral"])
def test_vector_arithmetic_preserves_shape(self):
a = ForgeVector(np.ones(256, dtype=np.float32))
b = ForgeVector(np.full(256, 2.0, dtype=np.float32))
result = a - b + 0.5 * b
self.assertEqual(result.shape, (256,))
np.testing.assert_allclose(result.numpy(), np.zeros(256), atol=1e-6)
def test_normalized_vector_has_unit_norm(self):
vector = ForgeVector(np.arange(1, 257, dtype=np.float32)).normalized()
self.assertAlmostEqual(float(np.linalg.norm(vector.numpy())), 1.0, places=6)
def test_tensor_poster_is_resized_and_normalized(self):
poster = torch.full((480, 320, 3), 255, dtype=torch.uint8)
pixels = Forge2Vec._poster_tensor(poster)
self.assertEqual(tuple(pixels.shape), (3, 224, 224))
self.assertTrue(torch.allclose(pixels, torch.ones_like(pixels)))
if __name__ == "__main__":
unittest.main()
|