2017-10-09 01:02:23 +03:00
|
|
|
import pytest
|
2020-02-18 17:38:18 +03:00
|
|
|
from thinc.api import Adam
|
2023-06-26 12:41:03 +03:00
|
|
|
|
2020-07-22 14:42:59 +03:00
|
|
|
from spacy import registry
|
2023-06-26 12:41:03 +03:00
|
|
|
from spacy.attrs import NORM
|
|
|
|
from spacy.pipeline import DependencyParser
|
2020-07-22 14:42:59 +03:00
|
|
|
from spacy.pipeline.dep_parser import DEFAULT_PARSER_MODEL
|
2018-07-25 00:38:44 +03:00
|
|
|
from spacy.tokens import Doc
|
2023-06-26 12:41:03 +03:00
|
|
|
from spacy.training import Example
|
|
|
|
from spacy.vocab import Vocab
|
2017-10-09 01:02:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
@pytest.fixture
|
|
|
|
def vocab():
|
|
|
|
return Vocab(lex_attr_getters={NORM: lambda s: s})
|
|
|
|
|
2018-07-25 00:38:44 +03:00
|
|
|
|
2020-09-08 23:44:25 +03:00
|
|
|
def _parser_example(parser):
|
|
|
|
doc = Doc(parser.vocab, words=["a", "b", "c", "d"])
|
|
|
|
gold = {"heads": [1, 1, 3, 3], "deps": ["right", "ROOT", "left", "ROOT"]}
|
|
|
|
return Example.from_dict(doc, gold)
|
|
|
|
|
|
|
|
|
2017-10-09 01:02:23 +03:00
|
|
|
@pytest.fixture
|
|
|
|
def parser(vocab):
|
2020-12-13 04:08:32 +03:00
|
|
|
vocab.strings.add("ROOT")
|
2020-07-25 16:01:15 +03:00
|
|
|
cfg = {"model": DEFAULT_PARSER_MODEL}
|
2020-09-27 23:21:31 +03:00
|
|
|
model = registry.resolve(cfg, validate=True)["model"]
|
2021-06-17 10:33:00 +03:00
|
|
|
parser = DependencyParser(vocab, model)
|
2018-11-27 03:09:36 +03:00
|
|
|
parser.cfg["token_vector_width"] = 4
|
|
|
|
parser.cfg["hidden_width"] = 32
|
|
|
|
# parser.add_label('right')
|
|
|
|
parser.add_label("left")
|
2020-09-29 13:05:38 +03:00
|
|
|
parser.initialize(lambda: [_parser_example(parser)])
|
2020-01-29 19:06:46 +03:00
|
|
|
sgd = Adam(0.001)
|
2017-10-09 01:02:23 +03:00
|
|
|
|
|
|
|
for i in range(10):
|
|
|
|
losses = {}
|
2018-11-27 03:09:36 +03:00
|
|
|
doc = Doc(vocab, words=["a", "b", "c", "d"])
|
2020-06-26 20:34:12 +03:00
|
|
|
example = Example.from_dict(
|
|
|
|
doc, {"heads": [1, 1, 3, 3], "deps": ["left", "ROOT", "left", "ROOT"]}
|
|
|
|
)
|
|
|
|
parser.update([example], sgd=sgd, losses=losses)
|
2017-10-09 01:02:23 +03:00
|
|
|
return parser
|
|
|
|
|
2018-07-25 00:38:44 +03:00
|
|
|
|
2017-10-09 01:02:23 +03:00
|
|
|
def test_no_sentences(parser):
|
2018-11-27 03:09:36 +03:00
|
|
|
doc = Doc(parser.vocab, words=["a", "b", "c", "d"])
|
2017-10-09 01:02:23 +03:00
|
|
|
doc = parser(doc)
|
2017-10-12 22:18:22 +03:00
|
|
|
assert len(list(doc.sents)) >= 1
|
2017-10-09 01:02:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
def test_sents_1(parser):
|
2018-11-27 03:09:36 +03:00
|
|
|
doc = Doc(parser.vocab, words=["a", "b", "c", "d"])
|
2017-10-09 01:02:23 +03:00
|
|
|
doc[2].sent_start = True
|
|
|
|
doc = parser(doc)
|
2017-10-09 01:29:37 +03:00
|
|
|
assert len(list(doc.sents)) >= 2
|
2018-11-27 03:09:36 +03:00
|
|
|
doc = Doc(parser.vocab, words=["a", "b", "c", "d"])
|
2017-10-09 01:02:23 +03:00
|
|
|
doc[1].sent_start = False
|
|
|
|
doc[2].sent_start = True
|
|
|
|
doc[3].sent_start = False
|
|
|
|
doc = parser(doc)
|
|
|
|
assert len(list(doc.sents)) == 2
|
|
|
|
|
|
|
|
|
|
|
|
def test_sents_1_2(parser):
|
2018-11-27 03:09:36 +03:00
|
|
|
doc = Doc(parser.vocab, words=["a", "b", "c", "d"])
|
2017-10-09 01:02:23 +03:00
|
|
|
doc[1].sent_start = True
|
|
|
|
doc[2].sent_start = True
|
|
|
|
doc = parser(doc)
|
2017-11-03 16:36:08 +03:00
|
|
|
assert len(list(doc.sents)) >= 3
|
2017-10-09 01:02:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
def test_sents_1_3(parser):
|
2018-11-27 03:09:36 +03:00
|
|
|
doc = Doc(parser.vocab, words=["a", "b", "c", "d"])
|
2020-12-13 04:08:32 +03:00
|
|
|
doc[0].is_sent_start = True
|
|
|
|
doc[1].is_sent_start = True
|
|
|
|
doc[2].is_sent_start = None
|
|
|
|
doc[3].is_sent_start = True
|
2017-10-09 01:02:23 +03:00
|
|
|
doc = parser(doc)
|
2017-10-11 09:38:34 +03:00
|
|
|
assert len(list(doc.sents)) >= 3
|
2018-11-27 03:09:36 +03:00
|
|
|
doc = Doc(parser.vocab, words=["a", "b", "c", "d"])
|
2020-12-13 04:08:32 +03:00
|
|
|
doc[0].is_sent_start = True
|
|
|
|
doc[1].is_sent_start = True
|
|
|
|
doc[2].is_sent_start = False
|
|
|
|
doc[3].is_sent_start = True
|
2017-10-09 01:02:23 +03:00
|
|
|
doc = parser(doc)
|
|
|
|
assert len(list(doc.sents)) == 3
|