python 分别使用Python中自带的xml.etree.ElementTree和第三方LXML来生成一个按首字母归类的城市列表XML文件。

Posted 2021-05-10

#! /usr/bin/env python
# coding=utf-8
__author__ = 'Kingson.Zhou'

import requests
import string
from pyquery import PyQuery as pq
import xml.etree.ElementTree as ET
from lxml import etree as ET

# Get city list from MeiTuan Website
def create_city():
    city_request = requests.get("http://www.meituan.com/index/changecity/initiative")
    city_content = city_request.content
    p = pq(city_content)
    city = []
    for i in string.uppercase:
        filter = '#city-' + i
        city_list = p('li').filter(filter).text()
        if city_list != None:
            city.append(city_list.split())    
    return city

# Create a xml file by xml.etree.ElementTree
def building_xml():
    city = create_city()
    root = ET.Element("plist")
    root.set("version", "1.0")
    array = ET.SubElement(root, "array")
    for i in range(len(city)):
        dict = ET.SubElement(array, "dict")
        key = ET.SubElement(dict, "key")
        key.text = "Title"
        string = ET.SubElement(dict, "string")
        string.text = city[i][0]
        key = ET.SubElement(dict, "key")
        key.text = "Cities"
        array1 = ET.SubElement(dict, "array")
        for n in range(1,len(city[i])):         
            dict = ET.SubElement(array1, "dict")
            key = ET.SubElement(dict, "key")
            key.text = "Title"
            string = ET.SubElement(dict, "string")
            string.text = city[i][n]
    tree = ET.ElementTree(root)
    tree.write("cities.xml", encoding="UTF-8")
    return True

# Pretty a xml file
def pretty_xmlfile():
    parser = ET.XMLParser(remove_blank_text=False, resolve_entities=True, strip_cdata=True)
    xmlfile = ET.parse("cities.xml",parser)
    pretty_xml = ET.tostring(xmlfile, encoding = 'UTF-8', xml_declaration = True, pretty_print = True,doctype='<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">')
    file = open("cities.xml","w")
    file.writelines(pretty_xml)
    file.close()

# Create a xml file by lxml
def create_xml():
    city = create_city()  
    plist = ET.Element('plist')
    plist.attrib['version']= '1.0'
    array = ET.SubElement(plist, 'array')
    for i in range(len(city)):      
        dict = ET.SubElement(array, 'dict')
        key = ET.SubElement(dict, 'key')
        key.text = "Title" 
        string = ET.SubElement(dict, "string")
        string.text = city[i][0]
        key = ET.SubElement(dict, "key")
        key.text = "Cities"
        array1 = ET.SubElement(dict, 'array')
        for n in range(1,len(city[i])):
            dict = ET.SubElement(array1, "dict")
            key = ET.SubElement(dict, "key")
            key.text = "Title"
            string = ET.SubElement(dict, "string")
            string.text = city[i][n]      
    print ET.tostring(plist, pretty_print=True, xml_declaration=True, encoding='UTF-8', doctype='<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">')    
    # write to file:  
    tree = ET.ElementTree(plist)  
    tree.write('cities.xml', pretty_print=True, xml_declaration=True, encoding='UTF-8') 
    return True

if __name__ == '__main__':
    building_xml()
#    pretty_xmlfile()
#   create_xml()