類似文章を比較してみる

ToDo

類似文章の比較するために作ったバギースクリプト。 mecabを使用。

mecab_test.py

# -*- coding: cp932 -*-
from ctypes import *

class Mecab_Test:
    def __init__(self):
        self.lib=cdll.LoadLibrary("C:/Program Files/MeCab/bin/libmecab.dll")
        getattr(self.lib, "<b style="color:black;background-color:#ffff66">mecab_new2</b>")
        self.lib.<b style="color:black;background-color:#ffff66">mecab_new2</b>.restype = c_void_p
        self.lib.<b style="color:black;background-color:#ffff66">mecab_new2</b>.argtypes = [c_char_p, c_char_p]
        getattr(self.lib, "mecab_sparse_tostr")
        self.lib.mecab_sparse_tostr.restype = c_char_p
        self.m = self.lib.<b style="color:black;background-color:#ffff66">mecab_new2</b>('-a', 'c://Progmecab/bin/libmecab.dll')

    def conv_mecab(self,str):
        return self.lib.mecab_sparse_tostr(self.m,str)

if __name__=="__main__":
    str='今日は元気だ。'
    m=Mecab_Test()
    print m.conv_mecab(str)
    str='明日は雨だろう。'
    print m.conv_mecab(str);

test.py

# -*- coding: cp932 -*-
import re,string,mecab_test
filename='sample6.txt'
inputfile=open(filename,'r')
lines=inputfile.readlines();
inputfile.close()

m_test=mecab_test.Mecab_Test()


class CharSet:
    def __init__(self):
        self.first_array=[]
        self.second_array=[]

    def drawArray(self):
        for line in self.second_array:
            print line
            
cs=CharSet()

def analizer_str(strs):
    str_array=string.split(strs,'\n')
    for basestr in str_array:
        if(basestr=='EOS'):
            cs.second_array.append(cs.first_array)
            cs.first_array=[]
            break
        (b1,b2)=string.split(basestr,'\t')
        if re.match(r'名詞',b2):
            b2=string.split(b2,',')
            if re.match('一般',b2[1]):
                cs.first_array.append(b1)
#                print b1
#            if re.search(r'一般',b2):
#                print 'str1 %s youso %s' % (b1,b2)

for line in lines:
    line=string.strip(line)
    strs=string.split(line,'。')
    for str in strs:
        str=str+'。'
        if str == '。':break
        analizer_str(m_test.conv_mecab(str))

cs.drawArray()