#!/usr/bin/python
# coding: UTF-8
# Rec10 TS Recording Tools
-# Copyright (C) 2009 Yukikaze
+# Copyright (C) 2009-2010 Yukikaze
import zenhan
def bigram(str1, str2):
point = point + i
return point
def trigram(str1, str2):
+ """
+ trigramによる単語の近さを計算します。
+ 単語文字列の長さも考慮にいれます。
+ """
str1 = zenhan.toHankaku(str1)
str2 = zenhan.toHankaku(str2)
str1 = str1.replace(" ", "")
str2 = str2.replace(" ", "")
- """
- trigramによる単語の近さを計算します。
- """
gram = []
if len(str1) < 3:
gram.append(str1)
i = i + 1
st = st[st.find(gram) + 1:]
return i
+def get_noun_quad_gram(s):
+ """
+ 。
+ """
+ str1 = zenhan.toHankaku(s)
+ gram={}
+ if len(str1) > 4:
+ tmp1 = str1[0]
+ tmp2 = str1[1]
+ tmp3 = str1[2]
+ skipnum=0
+ for x in str1[3:]:
+ if skipnum>0:
+ skipnum=skipnum-1
+ else:
+ tmps=tmp1 + tmp2 + tmp3 + x
+ if tmps.find(" ")<0:
+ if zenhan.check_Character_Type(tmp1)>0:
+ if (zenhan.check_Character_Type(tmp1)==zenhan.check_Character_Type(tmp2)):
+ if (zenhan.check_Character_Type(tmp2)==zenhan.check_Character_Type(tmp3)):
+ if (zenhan.check_Character_Type(tmp3)==zenhan.check_Character_Type(x)):
+ gram[tmps]=gram.get(tmps,0)+1
+ skipnum=0
+ else:
+ skipnum=3
+ else:
+ if (zenhan.check_Character_Type(tmp3)==zenhan.check_Character_Type(x)):
+ skipnum=2
+ else:
+ skipnum=3
+ else:
+ if (zenhan.check_Character_Type(tmp2)==zenhan.check_Character_Type(tmp3)):
+ if (zenhan.check_Character_Type(tmp3)==zenhan.check_Character_Type(x)):
+ skipnum=0
+ else:
+ skipnum=3
+ else:
+ if (zenhan.check_Character_Type(tmp3)==zenhan.check_Character_Type(x)):
+ skipnum=2
+ else:
+ skipnum=3
+ tmp1 = tmp2
+ tmp2 = tmp3
+ tmp3 = x
+ return gram
""
\ No newline at end of file