add tsid channel information for complete BS support.

[rec10/rec10-git.git] / rec10 / trunk / src / n_gram.py

diff --git a/rec10/trunk/src/n_gram.py b/rec10/trunk/src/n_gram.py
index bfc72ee..f24a097 100644 (file)
--- a/rec10/trunk/src/n_gram.py
+++ b/rec10/trunk/src/n_gram.py
@@ -1,7 +1,7 @@
 #!/usr/bin/python
 # coding: UTF-8
 # Rec10 TS Recording Tools
-# Copyright (C) 2009 Yukikaze
+# Copyright (C) 2009-2011 Yukikaze
 
 import zenhan
 def bigram(str1, str2):
@@ -22,7 +22,7 @@ def bigram(str1, str2):
             tmp = x
     point = 0
     for x in gram:
-        i = find_gram(x, str2)
+        i = findGram(x, str2)
         if i > 0:
             i = 90 + 10 * i
         else:
@@ -30,13 +30,14 @@ 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)
@@ -50,18 +51,18 @@ def trigram(str1, str2):
     point = 0
     count = 0
     for x in gram:
-        i = find_gram(x, str2)
+        i = findGram(x, str2)
         if i > 0:
             i = 90 + 10 * i
             count=count + 1
         else:
             i = 0
         point = point + i
-    point =point + 20*count*count
+    point = point + 20*count*count
     point = point / len(str1) * 10
     point = point / len(str2) * 10
     return point
-def find_gram(gram, s):
+def findGram(gram, s):
     """
     s中にあらわれるgramの数を調べる
     """
@@ -71,4 +72,49 @@ def find_gram(gram, s):
         i = i + 1
         st = st[st.find(gram) + 1:]
     return i
+def getNounQuadGram(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.checkCharacterType(tmp1)>0:
+                        if (zenhan.checkCharacterType(tmp1)==zenhan.checkCharacterType(tmp2)):
+                            if (zenhan.checkCharacterType(tmp2)==zenhan.checkCharacterType(tmp3)):
+                                if (zenhan.checkCharacterType(tmp3)==zenhan.checkCharacterType(x)):
+                                    gram[tmps]=gram.get(tmps,0)+1
+                                    skipnum=0
+                                else:
+                                    skipnum=3
+                            else:
+                                if (zenhan.checkCharacterType(tmp3)==zenhan.checkCharacterType(x)):
+                                    skipnum=2
+                                else:
+                                    skipnum=3
+                        else:
+                            if (zenhan.checkCharacterType(tmp2)==zenhan.checkCharacterType(tmp3)):
+                                if (zenhan.checkCharacterType(tmp3)==zenhan.checkCharacterType(x)):
+                                    skipnum=0
+                                else:
+                                    skipnum=3
+                            else:
+                                if (zenhan.checkCharacterType(tmp3)==zenhan.checkCharacterType(x)):
+                                    skipnum=2
+                                else:
+                                    skipnum=3
+            tmp1 = tmp2
+            tmp2 = tmp3
+            tmp3 = x
+    return gram
 ""
\ No newline at end of file