def test_tagging_and_lemmatization_and_ner(self, build_sentences_mock):
options = self._default_options(tagged=True, lemmatized=True, ner=True)
sentence = ["Hej", "mitt", "namn", "är"]
suc_tags = ["IN", "PS|NEU|SIN|DEF", "NN|NEU|SIN|IND|NOM", "VB|PRS|AKT"]
ud_tags = [
"INTJ|_",
"DET|Definite=Def|Gender=Neut|Number=Sing|Poss=Yes",
"NOUN|Case=Nom|Definite=Ind|Gender=Neut|Number=Sing",
"AUX|Mood=Ind|Tense=Pres|VerbForm=Fin|Voice=Act",
]
lemmas = ["hej", "min", "namn", "vara"]
ner_tags = ["O", "O", "O", "O"]
suc_tagger = MagicMock()
suc_tagger.tag.return_value = suc_tags
ud_tagger = MagicMock()
ud_tagger.tag.return_value = ud_tags
lemmatizer = MagicMock()
lemmatizer.predict.side_effect = lemmas
suc_ne_tagger = MagicMock()
suc_ne_tagger.tag.return_value = ner_tags
models = self._default_models(
suc_tagger=suc_tagger,
ud_tagger=ud_tagger,
lemmatizer=lemmatizer,
suc_ne_tagger=suc_ne_tagger,
)
self.assertEqual(
run_tagging_and_lemmatization(options, sentence, models),
(lemmas, ud_tags, suc_tags, ner_tags)
)
def test_tagging_and_lemmatization_and_ner(self, build_sentences_mock):
options = self._default_options(tagged=True, lemmatized=True, ner=True)
sentence = ["Hej", "mitt", "namn", "är"]
suc_tags = ["IN", "PS|NEU|SIN|DEF", "NN|NEU|SIN|IND|NOM", "VB|PRS|AKT"]
ud_tags = [
"INTJ|_",
"DET|Definite=Def|Gender=Neut|Number=Sing|Poss=Yes",
"NOUN|Case=Nom|Definite=Ind|Gender=Neut|Number=Sing",
"AUX|Mood=Ind|Tense=Pres|VerbForm=Fin|Voice=Act",
]
lemmas = ["hej", "min", "namn", "vara"]
ner_tags = ["O", "O", "O", "O"]
suc_tagger = MagicMock()
suc_tagger.tag.return_value = suc_tags
ud_tagger = MagicMock()
ud_tagger.tag.return_value = ud_tags
lemmatizer = MagicMock()
lemmatizer.predict.side_effect = lemmas
suc_ne_tagger = MagicMock()
suc_ne_tagger.tag.return_value = ner_tags
models = self._default_models(
suc_tagger=suc_tagger,
ud_tagger=ud_tagger,
lemmatizer=lemmatizer,
suc_ne_tagger=suc_ne_tagger,
)
self.assertEqual(
run_tagging_and_lemmatization(options, sentence, models),
(lemmas, ud_tags, suc_tags, ner_tags))
def test_tagging(self, build_sentences_mock):
options = self._default_options(tagged=True)
sentence = ["Hej", "mitt", "namn", "är"]
tags = ["IN", "PS|NEU|SIN|DEF", "NN|NEU|SIN|IND|NOM", "VB|PRS|AKT"]
tagger = MagicMock()
tagger.tag.return_value = tags
models = self._default_models(suc_tagger=tagger)
self.assertEqual(
run_tagging_and_lemmatization(options, sentence, models),
([], [],
['IN', 'PS|NEU|SIN|DEF', 'NN|NEU|SIN|IND|NOM', 'VB|PRS|AKT'], []))
def test_tagging(self, build_sentences_mock):
options = self._default_options(tagged=True)
sentence = ["Hej", "mitt", "namn", "är"]
tags = ["IN", "PS|NEU|SIN|DEF", "NN|NEU|SIN|IND|NOM", "VB|PRS|AKT"]
tagger = MagicMock()
tagger.tag.return_value = tags
models = self._default_models(suc_tagger=tagger)
self.assertEqual(
run_tagging_and_lemmatization(options, sentence, models),
(
[],
[],
['IN', 'PS|NEU|SIN|DEF', 'NN|NEU|SIN|IND|NOM', 'VB|PRS|AKT'],
[]
)
)
