有道词典导出的xml生词本文件提取出单词

作者: 再也不会用心爱
来源: 51数据库
2022-08-12

import xml.sax
import os,sys

class WordsHandler(xml.sax.ContentHandler):
    def __init__(self):
        self.CurrantTag = ''
        self.Word = ''
        self.Trans = ''
        self.wordfile = open('words.txt', 'w')
        
    def startElement(self, tag, attributes):
        self.CurrantTag = tag

    def endElement(self, tag):
        if self.CurrantTag == 'word':
            self.wordfile.write(self.Word.encode("utf-8") + '\n')
            print "Word : ", self.Word
        elif self.CurrantTag == 'trans':
            print "Trans : ", self.Trans
        self.CurrantTag = ''
    
    def characters(self, contant):
        if self.CurrantTag == 'word':
            self.Word = contant
        elif self.CurrantTag == 'trans':
            self.Trans = contant

    def endDocument(self):
        self.wordfile.close()
    
if __name__ == '__main__':
    parser = xml.sax.make_parser()
    parser.setFeature(xml.sax.handler.feature_namespaces, 0)
    
    handler = WordsHandler()
    parser.setContentHandler(handler)

    parser.parse("words.xml")

推荐阅读