|
a |
|
b/tests/pipelines/misc/test_sections.py |
|
|
1 |
from pytest import fixture |
|
|
2 |
|
|
|
3 |
from edsnlp.pipelines.misc.sections import Sections, patterns |
|
|
4 |
from edsnlp.utils.examples import parse_example |
|
|
5 |
|
|
|
6 |
sections_text = ( |
|
|
7 |
"Le patient est admis pour des douleurs dans le bras droit, " |
|
|
8 |
"mais n'a pas de problème de locomotion. " |
|
|
9 |
"Historique d'AVC dans la famille. pourrait être un cas de rhume.\n" |
|
|
10 |
"NBNbWbWbNbWbNBNbNbWbWbNBNbWbNbNbWbNBNbWbNbNBWbWbNbNbNBWbNb" |
|
|
11 |
"WbNbWBNbNbWbNbNBNbWbWbNbWBNbNbWbNBNbWbWbNb\n" |
|
|
12 |
"Pourrait être un cas de rhume.\n" |
|
|
13 |
"Motif :\n" |
|
|
14 |
"<ent section=motif>Douleurs</ent> dans le bras droit.\n" |
|
|
15 |
"Pas d'anomalie détectée." |
|
|
16 |
) |
|
|
17 |
|
|
|
18 |
empty_sections_text = """ |
|
|
19 |
Antécédents : |
|
|
20 |
Conclusion : |
|
|
21 |
<ent section=conclusion>Patient</ent> va mieux |
|
|
22 |
|
|
|
23 |
Au total: |
|
|
24 |
sortie du patient |
|
|
25 |
""" |
|
|
26 |
|
|
|
27 |
|
|
|
28 |
def test_section_detection(doc): |
|
|
29 |
assert doc.spans["sections"] |
|
|
30 |
|
|
|
31 |
|
|
|
32 |
@fixture |
|
|
33 |
def sections_factory(blank_nlp): |
|
|
34 |
|
|
|
35 |
default_config = dict( |
|
|
36 |
sections=patterns.sections, |
|
|
37 |
add_patterns=True, |
|
|
38 |
attr="NORM", |
|
|
39 |
ignore_excluded=True, |
|
|
40 |
) |
|
|
41 |
|
|
|
42 |
def factory(**kwargs): |
|
|
43 |
|
|
|
44 |
config = dict(**default_config) |
|
|
45 |
config.update(kwargs) |
|
|
46 |
|
|
|
47 |
return Sections( |
|
|
48 |
nlp=blank_nlp, |
|
|
49 |
**config, |
|
|
50 |
) |
|
|
51 |
|
|
|
52 |
return factory |
|
|
53 |
|
|
|
54 |
|
|
|
55 |
def test_sections(blank_nlp, sections_factory): |
|
|
56 |
|
|
|
57 |
blank_nlp.add_pipe("normalizer") |
|
|
58 |
|
|
|
59 |
sections = sections_factory() |
|
|
60 |
|
|
|
61 |
text, entities = parse_example(example=sections_text) |
|
|
62 |
|
|
|
63 |
doc = blank_nlp(text) |
|
|
64 |
doc.ents = [ |
|
|
65 |
doc.char_span(ent.start_char, ent.end_char, "generic") for ent in entities |
|
|
66 |
] |
|
|
67 |
|
|
|
68 |
doc = sections(doc) |
|
|
69 |
|
|
|
70 |
for entity, ent in zip(entities, doc.ents): |
|
|
71 |
|
|
|
72 |
for modifier in entity.modifiers: |
|
|
73 |
|
|
|
74 |
assert ( |
|
|
75 |
getattr(ent._, modifier.key) == modifier.value |
|
|
76 |
), f"{modifier.key} labels don't match." |
|
|
77 |
|
|
|
78 |
|
|
|
79 |
def test_empty_sections(blank_nlp, sections_factory): |
|
|
80 |
|
|
|
81 |
blank_nlp.add_pipe("normalizer") |
|
|
82 |
|
|
|
83 |
sections = sections_factory() |
|
|
84 |
|
|
|
85 |
text, entities = parse_example(example=empty_sections_text) |
|
|
86 |
|
|
|
87 |
doc = blank_nlp(text) |
|
|
88 |
doc.ents = [ |
|
|
89 |
doc.char_span(ent.start_char, ent.end_char, label="ent") for ent in entities |
|
|
90 |
] |
|
|
91 |
|
|
|
92 |
doc = sections(doc) |
|
|
93 |
|
|
|
94 |
for section in doc.spans["sections"]: |
|
|
95 |
for ent in section.ents: |
|
|
96 |
ent._.section = section.label_ |
|
|
97 |
|
|
|
98 |
for entity, ent in zip(entities, doc.ents): |
|
|
99 |
for modifier in entity.modifiers: |
|
|
100 |
assert ( |
|
|
101 |
getattr(ent._, modifier.key) == modifier.value |
|
|
102 |
), f"{modifier.key} labels don't match." |