mirror of
https://github.com/explosion/spaCy.git
synced 2024-12-28 02:46:35 +03:00
1f23c615d7
* Add implementation of batching + backwards compatibility fixes. Tests indicate issue with batch disambiguation for custom singular entity lookups. * Fix tests. Add distinction w.r.t. batch size. * Remove redundant and add new comments. * Adjust comments. Fix variable naming in EL prediction. * Fix mypy errors. * Remove KB entity type config option. Change return types of candidate retrieval functions to Iterable from Iterator. Fix various other issues. * Update spacy/pipeline/entity_linker.py Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com> * Update spacy/pipeline/entity_linker.py Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com> * Update spacy/kb_base.pyx Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com> * Update spacy/kb_base.pyx Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com> * Update spacy/pipeline/entity_linker.py Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com> * Add error messages to NotImplementedErrors. Remove redundant comment. * Fix imports. * Remove redundant comments. * Rename KnowledgeBase to InMemoryLookupKB and BaseKnowledgeBase to KnowledgeBase. * Fix tests. * Update spacy/errors.py Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Move KB into subdirectory. * Adjust imports after KB move to dedicated subdirectory. * Fix config imports. * Move Candidate + retrieval functions to separate module. Fix other, small issues. * Fix docstrings and error message w.r.t. class names. Fix typing for candidate retrieval functions. * Update spacy/kb/kb_in_memory.pyx Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/ml/models/entity_linker.py Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Fix typing. * Change typing of mentions to be Span instead of Union[Span, str]. * Update docs. * Update EntityLinker and _architecture docs. * Update website/docs/api/entitylinker.md Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com> * Adjust message for E1046. * Re-add section for Candidate in kb.md, add reference to dedicated page. * Update docs and docstrings. * Re-add section + reference for KnowledgeBase.get_alias_candidates() in docs. * Update spacy/kb/candidate.pyx * Update spacy/kb/kb_in_memory.pyx * Update spacy/pipeline/legacy/entity_linker.py * Remove canididate.md. Remove mistakenly added config snippet in entity_linker.py. Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com> Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
146 lines
4.1 KiB
Python
146 lines
4.1 KiB
Python
import warnings
|
|
from unittest import TestCase
|
|
import pytest
|
|
import srsly
|
|
from numpy import zeros
|
|
from spacy.kb.kb_in_memory import InMemoryLookupKB, Writer
|
|
from spacy.vectors import Vectors
|
|
from spacy.language import Language
|
|
from spacy.pipeline import TrainablePipe
|
|
from spacy.vocab import Vocab
|
|
|
|
from ..util import make_tempdir
|
|
|
|
|
|
def nlp():
|
|
return Language()
|
|
|
|
|
|
def vectors():
|
|
data = zeros((3, 1), dtype="f")
|
|
keys = ["cat", "dog", "rat"]
|
|
return Vectors(data=data, keys=keys)
|
|
|
|
|
|
def custom_pipe():
|
|
# create dummy pipe partially implementing interface -- only want to test to_disk
|
|
class SerializableDummy:
|
|
def __init__(self, **cfg):
|
|
if cfg:
|
|
self.cfg = cfg
|
|
else:
|
|
self.cfg = None
|
|
super(SerializableDummy, self).__init__()
|
|
|
|
def to_bytes(self, exclude=tuple(), disable=None, **kwargs):
|
|
return srsly.msgpack_dumps({"dummy": srsly.json_dumps(None)})
|
|
|
|
def from_bytes(self, bytes_data, exclude):
|
|
return self
|
|
|
|
def to_disk(self, path, exclude=tuple(), **kwargs):
|
|
pass
|
|
|
|
def from_disk(self, path, exclude=tuple(), **kwargs):
|
|
return self
|
|
|
|
class MyPipe(TrainablePipe):
|
|
def __init__(self, vocab, model=True, **cfg):
|
|
if cfg:
|
|
self.cfg = cfg
|
|
else:
|
|
self.cfg = None
|
|
self.model = SerializableDummy()
|
|
self.vocab = vocab
|
|
|
|
return MyPipe(Vocab())
|
|
|
|
|
|
def tagger():
|
|
nlp = Language()
|
|
tagger = nlp.add_pipe("tagger")
|
|
# need to add model for two reasons:
|
|
# 1. no model leads to error in serialization,
|
|
# 2. the affected line is the one for model serialization
|
|
tagger.add_label("A")
|
|
nlp.initialize()
|
|
return tagger
|
|
|
|
|
|
def entity_linker():
|
|
nlp = Language()
|
|
|
|
def create_kb(vocab):
|
|
kb = InMemoryLookupKB(vocab, entity_vector_length=1)
|
|
kb.add_entity("test", 0.0, zeros((1, 1), dtype="f"))
|
|
return kb
|
|
|
|
entity_linker = nlp.add_pipe("entity_linker")
|
|
entity_linker.set_kb(create_kb)
|
|
# need to add model for two reasons:
|
|
# 1. no model leads to error in serialization,
|
|
# 2. the affected line is the one for model serialization
|
|
nlp.initialize()
|
|
return entity_linker
|
|
|
|
|
|
objects_to_test = (
|
|
[nlp(), vectors(), custom_pipe(), tagger(), entity_linker()],
|
|
["nlp", "vectors", "custom_pipe", "tagger", "entity_linker"],
|
|
)
|
|
|
|
|
|
def write_obj_and_catch_warnings(obj):
|
|
with make_tempdir() as d:
|
|
with warnings.catch_warnings(record=True) as warnings_list:
|
|
warnings.filterwarnings("always", category=ResourceWarning)
|
|
obj.to_disk(d)
|
|
# in python3.5 it seems that deprecation warnings are not filtered by filterwarnings
|
|
return list(filter(lambda x: isinstance(x, ResourceWarning), warnings_list))
|
|
|
|
|
|
@pytest.mark.parametrize("obj", objects_to_test[0], ids=objects_to_test[1])
|
|
def test_to_disk_resource_warning(obj):
|
|
warnings_list = write_obj_and_catch_warnings(obj)
|
|
assert len(warnings_list) == 0
|
|
|
|
|
|
def test_writer_with_path_py35():
|
|
writer = None
|
|
with make_tempdir() as d:
|
|
path = d / "test"
|
|
try:
|
|
writer = Writer(path)
|
|
except Exception as e:
|
|
pytest.fail(str(e))
|
|
finally:
|
|
if writer:
|
|
writer.close()
|
|
|
|
|
|
def test_save_and_load_knowledge_base():
|
|
nlp = Language()
|
|
kb = InMemoryLookupKB(nlp.vocab, entity_vector_length=1)
|
|
with make_tempdir() as d:
|
|
path = d / "kb"
|
|
try:
|
|
kb.to_disk(path)
|
|
except Exception as e:
|
|
pytest.fail(str(e))
|
|
|
|
try:
|
|
kb_loaded = InMemoryLookupKB(nlp.vocab, entity_vector_length=1)
|
|
kb_loaded.from_disk(path)
|
|
except Exception as e:
|
|
pytest.fail(str(e))
|
|
|
|
|
|
class TestToDiskResourceWarningUnittest(TestCase):
|
|
def test_resource_warning(self):
|
|
scenarios = zip(*objects_to_test)
|
|
|
|
for scenario in scenarios:
|
|
with self.subTest(msg=scenario[1]):
|
|
warnings_list = write_obj_and_catch_warnings(scenario[0])
|
|
self.assertEqual(len(warnings_list), 0)
|