|
a |
|
b/tests/test_reducers.py |
|
|
1 |
import copy |
|
|
2 |
import pickle |
|
|
3 |
|
|
|
4 |
import dill |
|
|
5 |
import pytest |
|
|
6 |
from spacy.tokens import Doc |
|
|
7 |
|
|
|
8 |
import edsnlp |
|
|
9 |
|
|
|
10 |
if not hasattr(Doc, "custom_test_reducers"): |
|
|
11 |
Doc.set_extension("custom_test_reducers", default=None) |
|
|
12 |
|
|
|
13 |
|
|
|
14 |
@pytest.mark.parametrize("pickler_module", ["pickle", "dill"]) |
|
|
15 |
def test_reducers(pickler_module): |
|
|
16 |
pickler = pickle if pickler_module == "pickle" else dill |
|
|
17 |
doc = edsnlp.blank("eds")("This is a test.") |
|
|
18 |
doc.spans["test"] = [doc[0:1]] |
|
|
19 |
span = doc.spans["test"][0] |
|
|
20 |
token = doc[0] |
|
|
21 |
data = [ |
|
|
22 |
doc, # data[0] |
|
|
23 |
token, # data[1] |
|
|
24 |
token, # data[2] |
|
|
25 |
span, # data[3] |
|
|
26 |
span, # data[4] |
|
|
27 |
] |
|
|
28 |
doc._.custom_test_reducers = data |
|
|
29 |
doc_bytes = pickler.dumps(doc) |
|
|
30 |
|
|
|
31 |
doc_bis = pickler.loads(doc_bytes) |
|
|
32 |
data_bis = doc_bis._.custom_test_reducers |
|
|
33 |
assert data_bis[0] is doc_bis |
|
|
34 |
assert data_bis[1] == doc_bis[0] |
|
|
35 |
assert data_bis[1] is data_bis[2] |
|
|
36 |
assert data_bis[3] is data_bis[4] |
|
|
37 |
|
|
|
38 |
doc_ter = pickler.loads(doc_bytes) |
|
|
39 |
data_ter = doc_ter._.custom_test_reducers |
|
|
40 |
assert data_ter[0] is not data_bis[0] |
|
|
41 |
|
|
|
42 |
|
|
|
43 |
def test_deep_copy(): |
|
|
44 |
doc = edsnlp.blank("eds")("This is a test.") |
|
|
45 |
doc.spans["test"] = [doc[0:1]] |
|
|
46 |
span = doc.spans["test"][0] |
|
|
47 |
token = doc[0] |
|
|
48 |
data = [ |
|
|
49 |
doc, # data[0] |
|
|
50 |
token, # data[1] |
|
|
51 |
token, # data[2] |
|
|
52 |
span, # data[3] |
|
|
53 |
span, # data[4] |
|
|
54 |
] |
|
|
55 |
|
|
|
56 |
bis = copy.deepcopy(data) |
|
|
57 |
ter = copy.deepcopy(data) |
|
|
58 |
assert bis[1].doc is bis[0] |
|
|
59 |
assert bis[3].doc is bis[0] |
|
|
60 |
assert bis[1] is bis[2] |
|
|
61 |
assert bis[3] is bis[4] |
|
|
62 |
|
|
|
63 |
assert bis[0] is not doc |
|
|
64 |
assert bis[0] is not ter[0] |