def __read_conllu_vocabs(file: Path, *args, **kwargs) -> Tuple[Vocab, Vocab]:
word_vocab, tag_vocab = Vocab(), Vocab()
with open(file, **kwargs) as f:
for line in f:
if not line.strip() or line.strip().startswith("#"):
continue
tok = CoNLLUToken(*line.strip().split("\t"))
word_vocab += tok.form
tag_vocab += tok.upos
return word_vocab, tag_vocab
def test_unknown_token_in_example_yields_unk_udposdataset():
word_vocab = Vocab()
tag_vocab = Vocab()
form = "Pierre"
entity_tag = "B-PER"
assert form not in word_vocab
tag_vocab += entity_tag
examples = [CoNLL2003Example([CoNLL2003Token(form, "", "", entity_tag)])]
ds = CoNLL2003NERDataset(examples, word_vocab, tag_vocab)
(token_vector, tag_vector) = ds[0]
assert word_vocab.unk_index in token_vector
assert tag_vocab.unk_index not in tag_vector
def test_unknown_token_and_tag_in_example_yields_unk():
word_vocab = Vocab()
tag_vocab = Vocab()
form = "Pierre"
entity_tag = "B-PER"
assert form not in word_vocab
assert entity_tag not in tag_vocab
examples = [CoNLL2003Example([CoNLL2003Token(form, "", "", entity_tag)])]
ds = CoNLL2003Dataset(examples, word_vocab, tag_vocab, lambda t: t.entity_tag)
(token_vector, tag_vector) = ds[0]
assert word_vocab.unk_index in token_vector
assert tag_vocab.unk_index in tag_vector
def test_uddataset_from_file(tmp_path, example_dataset, cls, func):
file = tmp_path / "tmp.txt"
word_vocab, tag_vocab = Vocab(), Vocab()
with open(file, "w", encoding="utf-8") as f:
f.write(example_dataset)
if func:
ds = cls.from_file(file, word_vocab, tag_vocab, func, encoding="utf-8")
else:
ds = cls.from_file(file, word_vocab, tag_vocab, encoding="utf-8")
assert len(ds) == 2
def test_unknown_token_and_tag_in_example_yields_unk():
word_vocab = Vocab()
tag_vocab = Vocab()
form = "Pierre"
pos = "NNP"
assert form not in word_vocab
assert pos not in tag_vocab
examples = [
CoNLLUExample([CoNLLUToken(1, form, "", pos, "", "", 0, "", "", "")])
]
ds = UDDataset(examples, word_vocab, tag_vocab, lambda t: t.upos)
(token_vector, tag_vector) = ds[0]
assert word_vocab.unk_index in token_vector
assert tag_vocab.unk_index in tag_vector
def test_unknown_token_in_example_yields_unk_udposdataset():
word_vocab = Vocab()
tag_vocab = Vocab()
form = "Pierre"
pos = "NNP"
assert form not in word_vocab
tag_vocab += pos
examples = [
CoNLLUExample([CoNLLUToken(1, form, "", pos, "", "", 0, "", "", "")])
]
ds = UDUPOSDataset(examples, word_vocab, tag_vocab)
(token_vector, tag_vector) = ds[0]
assert word_vocab.unk_index in token_vector
assert tag_vocab.unk_index not in tag_vector
def tag_vocab():
vocab = Vocab()
vocab += "ODD"
vocab += "EVEN"
return vocab
def word_vocab():
vocab = Vocab()
for i in range(11):
vocab += str(i)
return vocab
def test_uddataset_len_is_len_examples(examples):
word_vocab = Vocab()
tag_vocab = Vocab()
ds = UDDataset(examples, word_vocab, tag_vocab, lambda t: t.upos)
assert len(ds) == len(ds)
def tag_vocab():
tag_vocab_ = Vocab()
tag_vocab_ += "EVEN"
tag_vocab_ += "ODD"
return tag_vocab_
def word_vocab():
word_vocab_ = Vocab()
for i in range(11):
word_vocab_ += str(i)
return word_vocab_
def test_uddataset_len_is_len_examples(examples):
word_vocab = Vocab()
tag_vocab = Vocab()
ds = CoNLL2003NERDataset(examples, word_vocab, tag_vocab, lambda t: t.entity_tag)
assert len(ds) == len(ds)
def nonempty_vocab(example_embedding):
vocab = Vocab()
vocab += example_embedding[0].split()[0]
return vocab
def empty_vocab(example_embedding):
return Vocab()
