在Python中拆分大型XML文件

import xml.sax

class SplitLargeXMLHandler(xml.sax.ContentHandler):
    def __init__(self, file_prefix, max_size):
        self.file_prefix = file_prefix
        self.max_size = max_size
        self.file_count = 0
        self.element_count = 0
        self.current_file = None
        self.current_size = 0

    def startElement(self, name, attrs):
        if name == 'element':
            self.element_count += 1
            if self.current_size >= self.max_size:
                self.current_file.close()
                self.file_count += 1
                self.current_file = open(f'{self.file_prefix}_{self.file_count}.xml', 'w')
                self.current_size = 0
            self.current_file.write(f'<{name}>')

    def endElement(self, name):
        if name == 'element':
            self.current_file.write(f'</{name}>')
            self.current_size += 1

    def characters(self, content):
        self.current_file.write(content)

    def endDocument(self):
        self.current_file.close()

    def parse(self, file_path):
        xml.sax.parse(file_path, self)

handler = SplitLargeXMLHandler('output', 1000)
handler.parse('input.xml')