-
Notifications
You must be signed in to change notification settings - Fork 237
Expand file tree
/
Copy pathtest_array_stacked.py
More file actions
680 lines (470 loc) · 17.6 KB
/
test_array_stacked.py
File metadata and controls
680 lines (470 loc) · 17.6 KB
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
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
from typing import Dict, Optional, Union
import numpy as np
import pytest
import torch
from pydantic import parse_obj_as
from docarray import BaseDoc, DocList
from docarray.array import DocVec
from docarray.documents import ImageDoc
from docarray.exceptions.exceptions import UnusableObjectError
from docarray.typing import AnyEmbedding, AnyTensor, NdArray, TorchTensor
@pytest.fixture()
def batch():
class ImageDoc(BaseDoc):
tensor: TorchTensor[3, 224, 224]
batch = DocVec[ImageDoc](
[ImageDoc(tensor=torch.zeros(3, 224, 224)) for _ in range(10)]
)
return batch
@pytest.fixture()
def nested_batch():
class ImageDoc(BaseDoc):
tensor: TorchTensor[3, 224, 224]
class MMdoc(BaseDoc):
img: DocList[ImageDoc]
batch = DocList[MMdoc](
[
MMdoc(
img=DocList[ImageDoc](
[ImageDoc(tensor=torch.zeros(3, 224, 224)) for _ in range(10)]
)
)
for _ in range(10)
]
)
return batch.to_doc_vec()
def test_create_from_list_docs():
list_ = [ImageDoc(tensor=torch.zeros(3, 224, 224)) for _ in range(10)]
da_stacked = DocVec[ImageDoc](docs=list_, tensor_type=TorchTensor)
assert len(da_stacked) == 10
assert da_stacked.tensor.shape == tuple([10, 3, 224, 224])
def test_len(batch):
assert len(batch) == 10
def test_create_from_None():
with pytest.raises(ValueError):
DocVec[ImageDoc]([])
def test_getitem(batch):
for i in range(len(batch)):
assert (batch[i].tensor == torch.zeros(3, 224, 224)).all()
def test_iterator(batch):
for doc in batch:
assert (doc.tensor == torch.zeros(3, 224, 224)).all()
def test_stack_setter():
class ImageDoc(BaseDoc):
tensor: TorchTensor[3, 224, 224]
batch = DocList[ImageDoc](
[ImageDoc(tensor=torch.zeros(3, 224, 224)) for _ in range(10)]
)
batch = batch.to_doc_vec()
batch.tensor = torch.ones(10, 3, 224, 224)
assert (batch.tensor == torch.ones(10, 3, 224, 224)).all()
for i, doc in enumerate(batch):
assert (doc.tensor == batch.tensor[i]).all()
def test_stack_setter_np():
class ImageDoc(BaseDoc):
tensor: NdArray[3, 224, 224]
batch = DocList[ImageDoc](
[ImageDoc(tensor=np.zeros((3, 224, 224))) for _ in range(10)]
)
batch = batch.to_doc_vec()
batch.tensor = np.ones((10, 3, 224, 224))
assert (batch.tensor == np.ones((10, 3, 224, 224))).all()
for i, doc in enumerate(batch):
assert (doc.tensor == batch.tensor[i]).all()
def test_stack_optional(batch):
assert (
batch._storage.tensor_columns['tensor'] == torch.zeros(10, 3, 224, 224)
).all()
assert (batch.tensor == torch.zeros(10, 3, 224, 224)).all()
def test_stack_numpy():
class ImageDoc(BaseDoc):
tensor: NdArray[3, 224, 224]
batch = DocList[ImageDoc](
[ImageDoc(tensor=np.zeros((3, 224, 224))) for _ in range(10)]
)
batch = batch.to_doc_vec()
assert (
batch._storage.tensor_columns['tensor'] == np.zeros((10, 3, 224, 224))
).all()
assert (batch.tensor == np.zeros((10, 3, 224, 224))).all()
assert (
batch.tensor.ctypes.data == batch._storage.tensor_columns['tensor'].ctypes.data
)
def test_stack(batch):
assert (
batch._storage.tensor_columns['tensor'] == torch.zeros(10, 3, 224, 224)
).all()
assert (batch.tensor == torch.zeros(10, 3, 224, 224)).all()
assert batch._storage.tensor_columns['tensor'].data_ptr() == batch.tensor.data_ptr()
for doc, tensor in zip(batch, batch.tensor):
assert doc.tensor.data_ptr() == tensor.data_ptr()
for i in range(len(batch)):
assert batch[i].tensor.data_ptr() == batch.tensor[i].data_ptr()
def test_stack_mod_nested_document():
class ImageDoc(BaseDoc):
tensor: TorchTensor[3, 224, 224]
class MMdoc(BaseDoc):
img: ImageDoc
batch = DocList[MMdoc](
[MMdoc(img=ImageDoc(tensor=torch.zeros(3, 224, 224))) for _ in range(10)]
)
batch = batch.to_doc_vec()
assert (
batch._storage.doc_columns['img']._storage.tensor_columns['tensor']
== torch.zeros(10, 3, 224, 224)
).all()
assert (batch.img.tensor == torch.zeros(10, 3, 224, 224)).all()
assert (
batch._storage.doc_columns['img']._storage.tensor_columns['tensor'].data_ptr()
== batch.img.tensor.data_ptr()
)
def test_stack_nested_DocArray(nested_batch):
for i in range(len(nested_batch)):
assert (
nested_batch[i].img._storage.tensor_columns['tensor']
== torch.zeros(10, 3, 224, 224)
).all()
assert (nested_batch[i].img.tensor == torch.zeros(10, 3, 224, 224)).all()
assert (
nested_batch[i].img._storage.tensor_columns['tensor'].data_ptr()
== nested_batch[i].img.tensor.data_ptr()
)
def test_convert_to_da(batch):
class ImageDoc(BaseDoc):
tensor: TorchTensor[3, 224, 224]
batch = DocList[ImageDoc](
[ImageDoc(tensor=torch.zeros(3, 224, 224)) for _ in range(10)]
)
batch = batch.to_doc_vec()
da = batch.to_doc_list()
for doc in da:
assert (doc.tensor == torch.zeros(3, 224, 224)).all()
def test_unstack_nested_document():
class ImageDoc(BaseDoc):
tensor: TorchTensor[3, 224, 224]
class MMdoc(BaseDoc):
img: ImageDoc
batch = DocList[MMdoc](
[MMdoc(img=ImageDoc(tensor=torch.zeros(3, 224, 224))) for _ in range(10)]
)
batch = batch.to_doc_vec()
da = batch.to_doc_list()
for doc in da:
assert (doc.img.tensor == torch.zeros(3, 224, 224)).all()
def test_unstack_nested_DocArray(nested_batch):
batch = nested_batch.to_doc_list()
for i in range(len(batch)):
assert isinstance(batch[i].img, DocList)
for doc in batch[i].img:
assert (doc.tensor == torch.zeros(3, 224, 224)).all()
def test_stack_call():
class ImageDoc(BaseDoc):
tensor: TorchTensor[3, 224, 224]
da = DocList[ImageDoc](
[ImageDoc(tensor=torch.zeros(3, 224, 224)) for _ in range(10)]
)
da = da.to_doc_vec()
assert len(da) == 10
assert da.tensor.shape == (10, 3, 224, 224)
def test_stack_union():
class ImageDoc(BaseDoc):
tensor: Union[NdArray[3, 224, 224], TorchTensor[3, 224, 224]]
batch = DocList[ImageDoc](
[ImageDoc(tensor=np.zeros((3, 224, 224))) for _ in range(10)]
)
batch[3].tensor = np.zeros((3, 224, 224))
# union fields aren't actually doc_vec
# just checking that there is no error
batch.to_doc_vec()
@pytest.mark.parametrize(
'tensor_type,tensor',
[(TorchTensor, torch.zeros(3, 224, 224)), (NdArray, np.zeros((3, 224, 224)))],
)
def test_any_tensor_with_torch(tensor_type, tensor):
class ImageDoc(BaseDoc):
tensor: AnyTensor
da = DocVec[ImageDoc](
[ImageDoc(tensor=tensor) for _ in range(10)],
tensor_type=tensor_type,
)
for i in range(len(da)):
assert (da[i].tensor == tensor).all()
assert 'tensor' in da._storage.tensor_columns.keys()
assert isinstance(da._storage.tensor_columns['tensor'], tensor_type)
def test_any_tensor_with_optional():
tensor = torch.zeros(3, 224, 224)
class ImageDoc(BaseDoc):
tensor: Optional[AnyTensor] = None
class TopDoc(BaseDoc):
img: ImageDoc
da = DocVec[TopDoc](
[TopDoc(img=ImageDoc(tensor=tensor)) for _ in range(10)],
tensor_type=TorchTensor,
)
for i in range(len(da)):
assert (da.img[i].tensor == tensor).all()
assert 'tensor' in da.img._storage.tensor_columns.keys()
assert isinstance(da.img._storage.tensor_columns['tensor'], TorchTensor)
def test_dict_stack():
class MyDoc(BaseDoc):
my_dict: Dict[str, int]
da = DocVec[MyDoc]([MyDoc(my_dict={'a': 1, 'b': 2}) for _ in range(10)])
da.my_dict
def test_get_from_slice_stacked():
class Doc(BaseDoc):
text: str
tensor: NdArray
N = 10
da = DocVec[Doc](
[Doc(text=f'hello{i}', tensor=np.zeros((3, 224, 224))) for i in range(N)]
)
da_sliced = da[0:10:2]
assert isinstance(da_sliced, DocVec)
tensors = da_sliced.tensor
assert tensors.shape == (5, 3, 224, 224)
texts = da_sliced.text
assert len(texts) == 5
for i, text in enumerate(texts):
assert text == f'hello{i * 2}'
def test_stack_embedding():
class MyDoc(BaseDoc):
embedding: AnyEmbedding
da = DocVec[MyDoc]([MyDoc(embedding=np.zeros(10)) for _ in range(10)])
assert 'embedding' in da._storage.tensor_columns.keys()
assert (da.embedding == np.zeros((10, 10))).all()
@pytest.mark.parametrize('tensor_backend', [TorchTensor, NdArray])
def test_stack_none(tensor_backend):
class MyDoc(BaseDoc):
tensor: Optional[AnyTensor] = None
da = DocVec[MyDoc](
[MyDoc(tensor=None) for _ in range(10)], tensor_type=tensor_backend
)
assert 'tensor' in da._storage.tensor_columns.keys()
def test_to_device():
da = DocVec[ImageDoc]([ImageDoc(tensor=torch.zeros(3, 5))], tensor_type=TorchTensor)
assert da.tensor.device == torch.device('cpu')
da.to('meta')
assert da.tensor.device == torch.device('meta')
def test_to_device_with_nested_da():
class Video(BaseDoc):
images: DocVec[ImageDoc]
da_image = DocVec[ImageDoc](
[ImageDoc(tensor=torch.zeros(3, 5))], tensor_type=TorchTensor
)
da = DocVec[Video]([Video(images=da_image)])
assert da.images[0].tensor.device == torch.device('cpu')
da.to('meta')
assert da.images[0].tensor.device == torch.device('meta')
def test_to_device_nested():
class MyDoc(BaseDoc):
tensor: TorchTensor
docs: ImageDoc
da = DocVec[MyDoc](
[MyDoc(tensor=torch.zeros(3, 5), docs=ImageDoc(tensor=torch.zeros(3, 5)))],
tensor_type=TorchTensor,
)
assert da.tensor.device == torch.device('cpu')
assert da.docs.tensor.device == torch.device('cpu')
da.to('meta')
assert da.tensor.device == torch.device('meta')
assert da.docs.tensor.device == torch.device('meta')
def test_to_device_numpy():
da = DocVec[ImageDoc]([ImageDoc(tensor=np.zeros((3, 5)))], tensor_type=NdArray)
with pytest.raises(NotImplementedError):
da.to('meta')
def test_keep_dtype_torch():
class MyDoc(BaseDoc):
tensor: TorchTensor
da = DocList[MyDoc](
[MyDoc(tensor=torch.zeros([2, 4], dtype=torch.int32)) for _ in range(3)]
)
assert da[0].tensor.dtype == torch.int32
da = da.to_doc_vec()
assert da[0].tensor.dtype == torch.int32
assert da.tensor.dtype == torch.int32
def test_keep_dtype_np():
class MyDoc(BaseDoc):
tensor: NdArray
da = DocList[MyDoc](
[MyDoc(tensor=np.zeros([2, 4], dtype=np.int32)) for _ in range(3)]
)
assert da[0].tensor.dtype == np.int32
da = da.to_doc_vec()
assert da[0].tensor.dtype == np.int32
assert da.tensor.dtype == np.int32
def test_del_item(batch):
assert len(batch) == 10
assert batch.tensor.shape[0] == 10
with pytest.raises(NotImplementedError):
del batch[2]
def test_np_scalar():
class MyDoc(BaseDoc):
scalar: NdArray
da = DocList[MyDoc]([MyDoc(scalar=np.array(2.0)) for _ in range(3)])
assert all(doc.scalar.ndim == 0 for doc in da)
assert all(doc.scalar == 2.0 for doc in da)
stacked_da = da.to_doc_vec()
assert type(stacked_da.scalar) == NdArray
assert all(type(doc.scalar) == NdArray for doc in stacked_da)
assert all(doc.scalar.ndim == 1 for doc in stacked_da)
assert all(doc.scalar == 2.0 for doc in stacked_da)
# Make sure they share memory
stacked_da.scalar[0] = 3.0
assert stacked_da[0].scalar == 3.0
def test_torch_scalar():
class MyDoc(BaseDoc):
scalar: TorchTensor
da = DocList[MyDoc](
[MyDoc(scalar=torch.tensor(2.0)) for _ in range(3)],
)
assert all(doc.scalar.ndim == 0 for doc in da)
assert all(doc.scalar == 2.0 for doc in da)
stacked_da = da.to_doc_vec(tensor_type=TorchTensor)
assert type(stacked_da.scalar) == TorchTensor
assert all(type(doc.scalar) == TorchTensor for doc in stacked_da)
assert all(doc.scalar.ndim == 1 for doc in stacked_da) # TODO failing here
assert all(doc.scalar == 2.0 for doc in stacked_da)
stacked_da.scalar[0] = 3.0
assert stacked_da[0].scalar == 3.0
def test_np_nan():
class MyDoc(BaseDoc):
scalar: Optional[NdArray] = None
da = DocList[MyDoc]([MyDoc() for _ in range(3)])
assert all(doc.scalar is None for doc in da)
assert all(doc.scalar == doc.scalar for doc in da)
stacked_da = da.to_doc_vec()
assert stacked_da.scalar is None
assert all(doc.scalar is None for doc in stacked_da)
# Stacking them turns them into np.nan
def test_from_storage():
class ImageDoc(BaseDoc):
tensor: TorchTensor[3, 224, 224]
batch = DocVec[ImageDoc](
[ImageDoc(tensor=torch.zeros(3, 224, 224)) for _ in range(10)]
)
DocVec[ImageDoc].from_columns_storage(batch._storage)
def test_validate_from_da():
class ImageDoc(BaseDoc):
tensor: TorchTensor[3, 224, 224]
batch = DocList[ImageDoc](
[ImageDoc(tensor=torch.zeros(3, 224, 224)) for _ in range(10)]
)
da = parse_obj_as(DocVec[ImageDoc], batch)
assert isinstance(da, DocVec)
for d in da:
assert isinstance(d, ImageDoc)
def test_validation_column_tensor(batch):
batch.tensor = torch.zeros(10, 3, 224, 244)
assert isinstance(batch.tensor, TorchTensor)
def test_validation_column_tensor_fail(batch):
with pytest.raises(ValueError):
batch.tensor = ['hello'] * 10
with pytest.raises(ValueError):
batch.tensor = torch.zeros(11, 3, 224, 244)
@pytest.fixture()
def batch_nested_doc():
class Inner(BaseDoc):
hello: str
class Doc(BaseDoc):
inner: Inner
batch = DocVec[Doc]([Doc(inner=Inner(hello='hello')) for _ in range(10)])
return batch, Doc, Inner
def test_validation_column_doc(batch_nested_doc):
batch, Doc, Inner = batch_nested_doc
batch.inner = DocList[Inner]([Inner(hello='hello') for _ in range(10)])
assert isinstance(batch.inner, DocVec)
for d in batch.inner:
assert isinstance(d, Inner)
def test_validation_list_doc(batch_nested_doc):
batch, Doc, Inner = batch_nested_doc
batch.inner = [Inner(hello='hello') for _ in range(10)]
assert isinstance(batch.inner, DocVec)
for d in batch.inner:
assert isinstance(d, Inner)
def test_validation_col_doc_fail(batch_nested_doc):
batch, Doc, Inner = batch_nested_doc
with pytest.raises(ValueError):
batch.inner = ['hello'] * 10
with pytest.raises(ValueError):
batch.inner = DocList[Inner]([Inner(hello='hello') for _ in range(11)])
def test_doc_view_update(batch):
batch[0].tensor = 12 * torch.ones(3, 224, 224)
assert (batch.tensor[0] == 12 * torch.ones(3, 224, 224)).all()
def test_doc_view_nested(batch_nested_doc):
batch, Doc, Inner = batch_nested_doc
batch[0].inner = Inner(hello='world')
assert batch.inner[0].hello == 'world'
def test_type_error_no_doc_type():
with pytest.raises(TypeError):
DocVec([BaseDoc() for _ in range(10)])
def test_doc_view_dict(batch: DocVec[ImageDoc]):
doc_view = batch[0]
assert doc_view.is_view()
d = doc_view.dict()
assert d['tensor'].shape == (3, 224, 224)
assert d['id'] == doc_view.id
doc_view_two = batch[1]
assert doc_view_two.is_view()
d = doc_view_two.dict()
assert d['tensor'].shape == (3, 224, 224)
assert d['id'] == doc_view_two.id
def test_doc_vec_equality():
class Text(BaseDoc):
text: str
da = DocVec[Text]([Text(text='hello') for _ in range(10)])
da2 = DocList[Text]([Text(text='hello') for _ in range(10)])
assert da != da2
assert da == da2.to_doc_vec()
@pytest.mark.parametrize('tensor_type', [TorchTensor, NdArray])
def test_doc_vec_equality_tensor(tensor_type):
class Text(BaseDoc):
tens: tensor_type
da = DocVec[Text](
[Text(tens=[1, 2, 3, 4]) for _ in range(10)], tensor_type=tensor_type
)
da2 = DocVec[Text](
[Text(tens=[1, 2, 3, 4]) for _ in range(10)], tensor_type=tensor_type
)
assert da == da2
da2 = DocVec[Text](
[Text(tens=[1, 2, 3, 4, 5]) for _ in range(10)], tensor_type=tensor_type
)
assert da != da2
@pytest.mark.tensorflow
def test_doc_vec_equality_tf():
from docarray.typing import TensorFlowTensor
class Text(BaseDoc):
tens: TensorFlowTensor
da = DocVec[Text](
[Text(tens=[1, 2, 3, 4]) for _ in range(10)], tensor_type=TensorFlowTensor
)
da2 = DocVec[Text](
[Text(tens=[1, 2, 3, 4]) for _ in range(10)], tensor_type=TensorFlowTensor
)
assert da == da2
da2 = DocVec[Text](
[Text(tens=[1, 2, 3, 4, 5]) for _ in range(10)], tensor_type=TensorFlowTensor
)
assert da != da2
def test_doc_vec_nested(batch_nested_doc):
batch, Doc, Inner = batch_nested_doc
batch2 = DocVec[Doc]([Doc(inner=Inner(hello='hello')) for _ in range(10)])
assert batch == batch2
def test_doc_vec_tensor_type():
class ImageDoc(BaseDoc):
tensor: AnyTensor
da = DocVec[ImageDoc]([ImageDoc(tensor=np.zeros((3, 224, 224))) for _ in range(10)])
da2 = DocVec[ImageDoc](
[ImageDoc(tensor=torch.zeros(3, 224, 224)) for _ in range(10)],
tensor_type=TorchTensor,
)
assert da != da2
def teste_unusable_state_raises_exception():
from docarray import DocVec
from docarray.documents import ImageDoc
docs = DocVec[ImageDoc]([ImageDoc(url='http://url.com/foo.png') for _ in range(10)])
docs.to_doc_list()
with pytest.raises(UnusableObjectError):
docs.url
with pytest.raises(UnusableObjectError):
docs.url = 'hi'