breaker

import os
import re
path = os.path.dirname(os.path.realpath(__file__))
file = path + '/zhwiktionary-latest-pages-articles-multistream.xml'
# file = path + '/test.xml'
folder = ""
everything = ""
title = ""
isPage = False
newFile = ""
gotTitle = False
num = 1
with open(file, encoding="utf-8") as f:
    everything = f.readlines()
    # print(everything)
# print(line)
for line in everything:
    if "<page>" in line.strip('\n'):
        isPage = True
    if "</page>" in line.strip('\n'):
        isPage = False
        gotTitle = False
        # print(everything)
        # print(title)

    # reset everything

    # everything = ""

    if isPage:
        if gotTitle:
            newFile.write(line)
        if '<title>' in line.strip('\n'):
            title = line[line.find("<title>")+7:line.find("</title>")]
            title = re.findall(r'[\u4e00-\u9fd5]+', title)
            # print(title)
            if len(title) > 0:
                name = "c" + str(num)
                num += 1
                newFile = open(os.path.dirname(os.path.realpath(__file__)) + "/output/" + name + ".txt", "a+")
                newFile.write(line)
                gotTitle = True
            else:
                gotTitle = False
print(num)

# print(file)