Skip to content
This repository was archived by the owner on Sep 9, 2026. It is now read-only.

Commit 6f805cd

Browse files
JohannesMessnerJohannesMessner
authored
Debug ci (#1002)
* feat: native len for milvus Signed-off-by: Johannes Messner <messnerjo@gmail.com> * fix: make implementing len non-optional Signed-off-by: Johannes Messner <messnerjo@gmail.com> * ci: trigger the ci for debugging purposes Signed-off-by: Johannes Messner <messnerjo@gmail.com> * ci: only run oldproto tests Signed-off-by: Johannes Messner <messnerjo@gmail.com> * test: add some context mngrs Signed-off-by: Johannes Messner <messnerjo@gmail.com> * test: remove some tests Signed-off-by: Johannes Messner <messnerjo@gmail.com> * test: try to find minimal working changes Signed-off-by: Johannes Messner <messnerjo@gmail.com> * ci: fix success all tests Signed-off-by: Johannes Messner <messnerjo@gmail.com> Signed-off-by: Johannes Messner <messnerjo@gmail.com>
1 parent e812ec9 commit 6f805cd

1 file changed

Lines changed: 80 additions & 64 deletions

File tree

‎tests/unit/array/mixins/oldproto/test_eval_class.py‎

Lines changed: 80 additions & 64 deletions
Original file line numberDiff line numberDiff line change
@@ -42,7 +42,8 @@ def test_eval_mixin_perfect_match(metric_fn, kwargs, storage, config, start_stor
4242
da1 = DocumentArray.empty(10)
4343
da1.embeddings = np.random.random([10, 256])
4444
da1_index = DocumentArray(da1, storage=storage, config=config)
45-
da1.match(da1_index, exclude_self=True)
45+
with da1_index:
46+
da1.match(da1_index, exclude_self=True)
4647
r = da1.evaluate(ground_truth=da1, metrics=[metric_fn], strict=False, **kwargs)[
4748
metric_fn
4849
]
@@ -80,7 +81,8 @@ def test_eval_mixin_perfect_match_multiple_metrics(storage, config, start_storag
8081
da1 = DocumentArray.empty(10)
8182
da1.embeddings = np.random.random([10, 256])
8283
da1_index = DocumentArray(da1, storage=storage, config=config)
83-
da1.match(da1_index, exclude_self=True)
84+
with da1_index:
85+
da1.match(da1_index, exclude_self=True)
8486
r = da1.evaluate(ground_truth=da1, metrics=metric_fns, strict=False, **kwargs)
8587
for metric_fn in metric_fns:
8688
assert metric_fn in r
@@ -123,7 +125,8 @@ def test_eval_mixin_perfect_match_labeled(
123125
d.tags = {'label': 'A'}
124126
da1.embeddings = np.random.random([10, 256])
125127
da1_index = DocumentArray(da1, storage=storage, config=config)
126-
da1.match(da1_index, exclude_self=True)
128+
with da1_index:
129+
da1.match(da1_index, exclude_self=True)
127130
r = da1.evaluate(metrics=[metric_fn], **kwargs)[metric_fn]
128131
assert isinstance(r, float)
129132
assert r == 1.0
@@ -166,7 +169,8 @@ def test_eval_mixin_zero_labeled(storage, config, metric_fn, start_storage, kwar
166169
for d in da2:
167170
d.tags = {'label': 'B'}
168171
da1_index = DocumentArray(da2, storage=storage, config=config)
169-
da1.match(da1_index, exclude_self=True)
172+
with da1_index:
173+
da1.match(da1_index, exclude_self=True)
170174
r = da1.evaluate([metric_fn], **kwargs)[metric_fn]
171175
assert isinstance(r, float)
172176
assert r == 0.0
@@ -264,9 +268,10 @@ def test_eval_mixin_zero_match(storage, config, metric_fn, start_storage, kwargs
264268
da2 = copy.deepcopy(da1)
265269
da2.embeddings = np.random.random([10, 256])
266270
da2_index = DocumentArray(da2, storage=storage, config=config)
267-
da2.match(da2_index, exclude_self=True)
271+
with da2_index:
272+
da2.match(da2_index, exclude_self=True)
268273

269-
r = da1.evaluate(ground_truth=da2, metrics=[metric_fn], **kwargs)[metric_fn]
274+
r = da1.evaluate(ground_truth=da2, metrics=[metric_fn], **kwargs)[metric_fn]
270275
assert isinstance(r, float)
271276
assert r == 1.0
272277
for d in da1:
@@ -337,17 +342,20 @@ def test_same_hash_same_len_fun_should_work(storage, config, start_storage):
337342
da1 = DocumentArray.empty(10)
338343
da1.embeddings = np.random.random([10, 3])
339344
da1_index = DocumentArray(da1, storage=storage, config=config)
340-
da1.match(da1_index)
345+
with da1_index:
346+
da1.match(da1_index)
341347
da2 = DocumentArray.empty(10)
342348
da2.embeddings = np.random.random([10, 3])
343349
da2_index = DocumentArray(da1, storage=storage, config=config)
344-
da2.match(da2_index)
345-
with pytest.raises(ValueError):
346-
da1.evaluate(ground_truth=da2, metrics=['precision_at_k'])
347-
for d1, d2 in zip(da1, da2):
348-
d1.id = d2.id
350+
with da2_index:
351+
da2.match(da2_index)
352+
with da1_index, da2_index:
353+
with pytest.raises(ValueError):
354+
da1.evaluate(ground_truth=da2, metrics=['precision_at_k'])
355+
for d1, d2 in zip(da1, da2):
356+
d1.id = d2.id
349357

350-
da1.evaluate(ground_truth=da2, metrics=['precision_at_k'])
358+
da1.evaluate(ground_truth=da2, metrics=['precision_at_k'])
351359

352360

353361
@pytest.mark.parametrize(
@@ -368,7 +376,8 @@ def test_adding_noise(storage, config, start_storage):
368376

369377
da.embeddings = np.random.random([10, 3])
370378
da_index = DocumentArray(da, storage=storage, config=config)
371-
da.match(da_index, exclude_self=True)
379+
with da_index:
380+
da.match(da_index, exclude_self=True)
372381

373382
da2 = copy.deepcopy(da)
374383

@@ -410,17 +419,18 @@ def test_adding_noise(storage, config, start_storage):
410419
def test_diff_match_len_in_gd(storage, config, metric_fn, start_storage, kwargs):
411420
da1 = DocumentArray.empty(10)
412421
da1.embeddings = np.random.random([10, 128])
413-
da1_index = DocumentArray(da1, storage=storage, config=config)
422+
# da1_index = DocumentArray(da1, storage=storage, config=config)
414423
da1.match(da1, exclude_self=True)
415424

416425
da2 = copy.deepcopy(da1)
417426
da2.embeddings = np.random.random([10, 128])
418427
da2_index = DocumentArray(da2, storage=storage, config=config)
419-
da2.match(da2_index, exclude_self=True)
420-
# pop some matches from first document
421-
da2[0].matches.pop(8)
428+
with da2_index:
429+
da2.match(da2_index, exclude_self=True)
430+
# pop some matches from first document
431+
da2[0].matches.pop(8)
422432

423-
r = da1.evaluate(ground_truth=da2, metrics=[metric_fn], **kwargs)[metric_fn]
433+
r = da1.evaluate(ground_truth=da2, metrics=[metric_fn], **kwargs)[metric_fn]
424434
assert isinstance(r, float)
425435
np.testing.assert_allclose(r, 1.0, rtol=1e-2) #
426436
for d in da1:
@@ -486,7 +496,8 @@ def test_useless_groundtruth_warning_should_raise(storage, config, start_storage
486496
d.tags = {'label': 'A'}
487497
da1.embeddings = np.random.random([10, 256])
488498
da1_index = DocumentArray(da1, storage=storage, config=config)
489-
da1.match(da1_index, exclude_self=True)
499+
with da1_index:
500+
da1.match(da1_index, exclude_self=True)
490501
da2 = DocumentArray.empty(10)
491502
with pytest.warns(UserWarning):
492503
da1.evaluate(ground_truth=da2, metrics=['precision_at_k'])
@@ -518,13 +529,14 @@ def test_embed_and_evaluate_single_da(storage, config, start_storage):
518529
dummy_embed_function(gt)
519530
gt.match(gt, limit=3)
520531

521-
res = queries_da.embed_and_evaluate(
522-
ground_truth=gt,
523-
metrics=['precision_at_k', 'reciprocal_rank'],
524-
embed_funcs=dummy_embed_function,
525-
match_batch_size=1,
526-
limit=3,
527-
)
532+
with queries_da:
533+
res = queries_da.embed_and_evaluate(
534+
ground_truth=gt,
535+
metrics=['precision_at_k', 'reciprocal_rank'],
536+
embed_funcs=dummy_embed_function,
537+
match_batch_size=1,
538+
limit=3,
539+
)
528540
assert all([v == 1.0 for v in res.values()])
529541

530542

@@ -601,15 +613,16 @@ def test_embed_and_evaluate_two_das(storage, config, sample_size, start_storage)
601613
dummy_embed_function(gt_index)
602614
gt_queries.match(gt_index, limit=3)
603615

604-
res = queries_da.embed_and_evaluate(
605-
ground_truth=gt_queries,
606-
index_data=index_da,
607-
metrics=['precision_at_k', 'reciprocal_rank'],
608-
embed_funcs=dummy_embed_function,
609-
match_batch_size=1,
610-
limit=3,
611-
query_sample_size=sample_size,
612-
)
616+
with index_da:
617+
res = queries_da.embed_and_evaluate(
618+
ground_truth=gt_queries,
619+
index_data=index_da,
620+
metrics=['precision_at_k', 'reciprocal_rank'],
621+
embed_funcs=dummy_embed_function,
622+
match_batch_size=1,
623+
limit=3,
624+
query_sample_size=sample_size,
625+
)
613626
assert all([v == 1.0 for v in res.values()])
614627

615628

@@ -681,25 +694,26 @@ def emb_func(da):
681694
da1 = DocumentArray([Document(text=str(i), tags={label_tag: i}) for i in range(3)])
682695
da2 = DocumentArray(da1, storage=storage, config=config, copy=True)
683696

684-
if (
685-
use_index
686-
): # query and index da are distinct # (different embeddings are generated)
687-
res = da1.embed_and_evaluate(
688-
index_data=da2,
689-
metrics=metric_fns,
690-
embed_funcs=emb_func,
691-
match_batch_size=1,
692-
limit=3,
693-
label_tag=label_tag,
694-
)
695-
else: # query and index are the same (embeddings of both das are equal)
696-
res = da2.embed_and_evaluate(
697-
metrics=metric_fns,
698-
embed_funcs=emb_func,
699-
match_batch_size=1,
700-
limit=3,
701-
label_tag=label_tag,
702-
)
697+
with da2:
698+
if (
699+
use_index
700+
): # query and index da are distinct # (different embeddings are generated)
701+
res = da1.embed_and_evaluate(
702+
index_data=da2,
703+
metrics=metric_fns,
704+
embed_funcs=emb_func,
705+
match_batch_size=1,
706+
limit=3,
707+
label_tag=label_tag,
708+
)
709+
else: # query and index are the same (embeddings of both das are equal)
710+
res = da2.embed_and_evaluate(
711+
metrics=metric_fns,
712+
embed_funcs=emb_func,
713+
match_batch_size=1,
714+
limit=3,
715+
label_tag=label_tag,
716+
)
703717
for key in metric_fns:
704718
assert key in res
705719
assert abs(res[key] - expected[key]) < 1e-4
@@ -799,9 +813,10 @@ def test_embed_and_evaluate_with_embed_model(
799813
[Document(text=f'some text {i}', tags={'label': str(i)}) for i in range(5)]
800814
)
801815
da = DocumentArray(da, storage=storage, config=config)
802-
res = da.embed_and_evaluate(
803-
metrics=['precision_at_k'], embed_models=model, collate_fns=collate_fn
804-
)
816+
with da:
817+
res = da.embed_and_evaluate(
818+
metrics=['precision_at_k'], embed_models=model, collate_fns=collate_fn
819+
)
805820
assert res
806821
assert res['precision_at_k'] == 0.2
807822

@@ -873,12 +888,13 @@ def emb_func(da):
873888
)
874889
da2 = DocumentArray(da1, storage=storage, config=config, copy=True)
875890

876-
res = da1.embed_and_evaluate(
877-
index_data=da2,
878-
metrics=metric_fns,
879-
embed_funcs=emb_func,
880-
query_sample_size=sample_size,
881-
)
891+
with da2:
892+
res = da1.embed_and_evaluate(
893+
index_data=da2,
894+
metrics=metric_fns,
895+
embed_funcs=emb_func,
896+
query_sample_size=sample_size,
897+
)
882898
expected_size = (
883899
sample_size if sample_size and (sample_size < len(da1)) else len(da1)
884900
)

0 commit comments

Comments
 (0)