Download as
Browse Code »

Commit 4c653c2e87ba5fb2735ccbbf04d78a4d56c60dd1

Authored by Erickson Silva
1 parent 647ef3f2
Exists in devel

Corrige tratamento de palavras compostas

Showing 1 changed file with 7 additions and 3 deletions   Show diff stats
src/AplicaSinonimos.py
  Diff comments   View file @4c653c2
... ... @@ -7,6 +7,7 @@
7 7 #LAViD - Laboratório de Aplicações de Vídeo Digital
8 8  
9 9 import os
  10 +import re
10 11 import csv
11 12 import sys
12 13 from nltk.tree import Tree
... ... @@ -58,12 +59,15 @@ class AplicaSinonimos(object):
58 59 sentenca_corrigida = "_".join(lista).upper()
59 60 except:
60 61 sentenca_corrigida = "_".join([str(x[0]) for x in lista]).upper()
61   -
62 62 for p in palavras_compostas:
63   - if p in sentenca_corrigida:
64   - sentenca_corrigida = sentenca_corrigida.replace(p, p.replace("_", "#*#"))
  63 + for m in re.finditer(p, sentenca_corrigida):
  64 + first = "_" if m.start() == 0 else sentenca_corrigida[m.start()-1]
  65 + last = "_" if m.end() == len(sentenca_corrigida)-1 else sentenca_corrigida[m.end()]
  66 + if first == "_" and last == "_":
  67 + sentenca_corrigida = sentenca_corrigida.replace(p, p.replace("_", "#*#"))
65 68 return sentenca_corrigida.replace("_", " ").replace("#*#", "_")
66 69  
  70 +
67 71 def carregar_palavras_compostas(self):
68 72 path = self.localizar_arquivo_palavras_compostas()
69 73 return set(open(path).read().decode('utf-8').split())
... ...