分詞程式碼

# -*- coding:utf-8 -*-
   
#簡單的支援中文的正向最大匹配的機械分詞
   
import string
__dict = {}
   
def load_dict(dict_file='words.dic'):
    #載入詞庫，把詞庫載入成一個key為首字元，value為相關詞的列表的字典
   
    words = [line.split() for line in open(dict_file)]
    
    for word in words:
        
        first_char = word[0][0]
        __dict
 
.setdefault(first_char, [])
        __dict[first_char].append(word[0])
      
    #按詞的長度倒序排列
    for first_char, twords in __dict.items():
        __dict[first_char] = sorted(twords, key=lambda x:len(x), reverse=True)
   
def __match_ascii(i, input):
    #返回連續的英文字母，數字，符號, 對英文,字母,符號不處理
    result = ''
 

    for i in range(i, len(input)):
        if  input[i] in string.printable: # and input[i] not in string.whitespace: #string.ascii_letters or input[i] in string.digits: 
            result += input[i]
        else:
            break
    
    return result.strip()
   
   
def __match_word(first_char
 
, i , input):
    #根據當前位置進行分詞，ascii的直接讀取連續字元，中文的讀取詞庫
   
    if not __dict.get(first_char):
        try:
            if first_char in string.printable: #string.ascii_letters or first_char in string.digits:

                return __match_ascii(i, input)
        except:
            print('except:',first_char,chr(first_char))
        return first_char
   
    words = __dict[first_char]
    for word in words:
        if input[i:i+len(word)] == word:
            return word
   
    return first_char
   
def tokenize(input):
    #對input進行分詞
   
    if not input: return []
   
    tokens = []
    i = 0
    while i < len(input):
        first_char = input[i]
        matched_word = __match_word(first_char, i, input)
        tokens.append(matched_word)
        i += len(matched_word)
   
    return tokens
   
   
if __name__ == '__main__':
    def get_test_text():
        import requests
        url = "http://www.zhb.gov.cn/xxgk/gzdt/201703/t20170321_408538.shtml"
        #url="http://mil.news.sina.com.cn/2016-12-30/doc-ifxzczff3445251.shtml"
        #text = requests.get(url).content
        text = requests.get(url,'utf8').content
        #return text.decode('gbk')
        #print(text.decode('utf8'))
        return text.decode('utf8')
   
    def load_dict_test():
        load_dict()
        i=0;
        for first_char, words in __dict.items():
            print('%d. %s:%s' % (i,first_char, ' '.join(words)))
            i=i+1
            if i>10:
                break
            
   
    def tokenize_test(text):
        load_dict()
        tokens = tokenize(text)
        for token in tokens:
            print(token)
            
    #load_dict_test()
    tokenize_test('美麗的花園裡有各種各樣的小動物')
    tokenize_test('他購買了一盒Rosetta Stone品牌的SHA-PA型號24/6的訂書釘，總價￥24.3元.')
    tokenize_test('1949年10月1日，毛主席站在天安門城樓上莊嚴宣佈：中華人民共和國中央人民政府成立了！');
    tokenize_test('A Happy New Yeear and a Merry Christmas