python解析XML文档

来源：互联网发布：centos 开机启动脚本编辑：程序博客网时间：2024/05/16 01:52

XML文件

<?xml version="1.0" encoding="UTF-8"?>
<info>
   <intro>保存用户的信息</intro>
    <list id='001'>
        <head>auto_userone</head>
        <name>Jordy</name>
        <number>12345678</number>
        <age>20</age>
        <sex>男</sex>
        <hobby>上彀</hobby>
    </list>
    <list id='002'>
        <head>auto_usertwo</head>
        <name>功夫</name>
         <number>34443678</number>
         <age>18</age>
         <sex>男</sex>
         <hobby>功夫</hobby>
    </list>
</info>

用sax解析

from xml.dom import minidom

class praseXML:
    def __init__(self):
        self.dom = minidom.parse("data.xml")
        self.root = self.dom.documentElement
        self.elements = self.root.childNodes
        for element in self.elements:
            if element.nodeType is not element.TEXT_NODE:
                if element.hasAttribute('id'):
                    print element.nodeName,'-->',element.getAttribute('id')
                    nodes = element.childNodes
                    for node in nodes:
                        if node.nodeType is not node.TEXT_NODE:
                            print node.nodeName,'-->',node.firstChild.nodeValue

    def getTagByName(self,tagname):
        nodes = self.root.getElementsByTagName(tagname)
        return nodes

if __name__ == '__main__':
    obj = praseXML()
    nodes = obj.getTagByName('name')
    for node in nodes:
        print node.nodeName,'-->',node.firstChild.nodeValue

用etree解析

import xml.etree.ElementTree as ET

class parseXML:
    def __init__(self):
        self.filename = 'data.xml'
    def load_xml_file(self):
        root = ET.parse(self.filename).getroot()
        print root.tag
        intro = root.find('intro')
        print intro.tag,'-->',intro.text
        all_users = root.findall('list')
        usermap = {}
        count = 0
        for user in all_users:
            count += 1
            head = user.find('head')
            name = user.find('name')
            sex = user.find('sex')
            print sex.text
            usermap[head.tag+str(count)] = head.text
            usermap[name.tag+str(count)] = name.text
            usermap[sex.tag+str(count)] = sex.text
        return usermap
if __name__ == '__main__':
    obj = parseXML()
    usermap = obj.load_xml_file()
    print usermap