def test_it_fill_dictionaries_with_treetagger(self):
"""
Tests the class method of the KeywordExtractor _fill_dictionaries_with_treetagger.
Passes an Italian sentence to this method and checks the content of dictionaries and sets filled by it.
"""
kw_extractor = KeywordExtractor(
os.path.join(self.script_folder, "test"), "small-mixed2.txt",
self.output_folder)
kw_extractor._fill_dictionaries_with_treetagger(
'Ieri il primario di Casa Basaglia Lorenzo Toresini è andato in pensione.',
1.0, ('il', 'di', 'essere', 'casa', 'in'))
self.assertEqual(
kw_extractor.noun_lemma_dict, {
'primario': 1.0,
'Basaglia': 1.0,
'Lorenzo': 1.0,
'Toresini': 1.0,
'pensione': 1.0,
'andare': 1.0
})
self.assertEqual(
kw_extractor.token_to_lemma_dict_original_case, {
'Basaglia': 'basaglia',
'Casa': 'casa',
'Ieri': 'ieri',
'Lorenzo': 'lorenzo',
'Toresini': 'toresini',
'andato': 'andare',
'di': 'di',
'il': 'il',
'in': 'in',
'pensione': 'pensione',
'primario': 'primario',
'è': 'essere'
})
self.assertEqual(
kw_extractor.lemma_token_to_POS, {
'Basaglia': {'NOM'},
'Casa': {'NPR'},
'Ieri': {'ADV'},
'Lorenzo': {'NPR'},
'Toresini': {'NOM'},
'andare': {'VER:pper'},
'andato': {'VER:pper'},
'di': {'PRE'},
'essere': {'VER:pres'},
'il': {'DET:def'},
'in': {'PRE'},
'pensione': {'NOM'},
'primario': {'NOM'},
'è': {'VER:pres'}
})
self.assertEqual(kw_extractor.title_noun_lemmas_dict, {})
self.assertEqual(kw_extractor.tree_taggers_proper_nouns,
{'Basaglia', 'Lorenzo', 'Toresini'})
def test_de_fill_dictionaries_with_treetagger(self):
"""
Tests the class method of the KeywordExtractor _fill_dictionaries_with_treetagger.
Passes a German sentence to this method and checks the content of dictionaries and sets filled by it.
"""
kw_extractor = KeywordExtractor(
os.path.join(self.script_folder, "test"), "small-mixed.txt",
self.output_folder)
kw_extractor._fill_dictionaries_with_treetagger(
'Antworten auf diese Fragen gab es aus dem Passeiertal bereits einige.',
1.0, ('auf', 'dem', 'es', 'aus'))
self.assertEqual(kw_extractor.noun_lemma_dict, {
'Antwort': 1.0,
'Frage': 1.0,
'Passeiertal': 1.0,
'geben': 1.0
})
self.assertEqual(
kw_extractor.token_to_lemma_dict_original_case, {
'Fragen': 'frage',
'bereits': 'bereits',
'es': 'es',
'aus': 'aus',
'gab': 'geben',
'auf': 'auf',
'Passeiertal': 'passeiertal',
'Antworten': 'antwort',
'diese': 'dies',
'einige': 'einige',
'dem': 'die'
})
self.assertEqual(
kw_extractor.lemma_token_to_POS, {
'Frage': {'NN'},
'die': {'ART'},
'bereits': {'ADV'},
'einige': {'PIS'},
'Antworten': {'NN'},
'auf': {'APPR'},
'diese': {'PDAT'},
'dem': {'ART'},
'Fragen': {'NN'},
'geben': {'VVFIN'},
'Antwort': {'NN'},
'aus': {'APPR'},
'gab': {'VVFIN'},
'Passeiertal': {'NN'},
'dies': {'PDAT'},
'es': {'PPER'}
})
self.assertEqual(kw_extractor.title_noun_lemmas_dict, {})
self.assertEqual(kw_extractor.tree_taggers_proper_nouns, set())