Commit 408ce64883cb450ab0acbf9694e90e75d3952b46
1 parent
a5237893
Exists in
master
and in
1 other branch
Remove caractere especial 'marcador' ao classificar sentença.
Showing
1 changed file
with
2 additions
and
1 deletions
Show diff stats
src/ClassificaSentencas.py
| @@ -50,6 +50,7 @@ class ClassificaSentencas(object): | @@ -50,6 +50,7 @@ class ClassificaSentencas(object): | ||
| 50 | decodificada = regex2.sub('',regex.sub('"',s.replace("–", "-").replace("—", "-"))).decode("utf-8") | 50 | decodificada = regex2.sub('',regex.sub('"',s.replace("–", "-").replace("—", "-"))).decode("utf-8") |
| 51 | except: | 51 | except: |
| 52 | decodificada = s.decode("utf-8") | 52 | decodificada = s.decode("utf-8") |
| 53 | + | ||
| 53 | return Toqueniza.TOK_PORT.tokenize(decodificada) | 54 | return Toqueniza.TOK_PORT.tokenize(decodificada) |
| 54 | 55 | ||
| 55 | def obter_classificacao_morfologica(self): | 56 | def obter_classificacao_morfologica(self): |
| @@ -63,7 +64,7 @@ class ClassificaSentencas(object): | @@ -63,7 +64,7 @@ class ClassificaSentencas(object): | ||
| 63 | while (anotada[0][1] is None): | 64 | while (anotada[0][1] is None): |
| 64 | time.sleep(random.choice(sleep_times)) | 65 | time.sleep(random.choice(sleep_times)) |
| 65 | anotada = AnotaCorpus.anota_sentencas([s],etiquetador,"hunpos")[0] | 66 | anotada = AnotaCorpus.anota_sentencas([s],etiquetador,"hunpos")[0] |
| 66 | - regex = re.compile('[%s]' % re.escape('!"#&()*+,-./:;<=>?@[\]^_`{|}~')) | 67 | + regex = re.compile('[%s]' % re.escape(u'\u2022''!"#&()*+,-./:;<=>?@[\]^_`{|}~')) |
| 67 | tag_punctuation = [".",",","QT","("] | 68 | tag_punctuation = [".",",","QT","("] |
| 68 | anotada_corrigida = [] | 69 | anotada_corrigida = [] |
| 69 | for x in anotada: | 70 | for x in anotada: |