Python:如何拆分WARC文件？

pip install warcio

from warcio.archiveiterator import ArchiveIterator

def split_warc_file(file_path):
    with open(file_path, 'rb') as warc_file:
        warc_iterator = ArchiveIterator(warc_file)

        current_warc_record = None
        current_warc_payload = None
        current_warc_filename = None

        for record in warc_iterator:
            if record.rec_type == 'warcinfo':
                continue

            if record.rec_type == 'response':
                if current_warc_record is not None:
                    current_warc_record.close()
                    current_warc_payload.close()

                current_warc_filename = f"{record.rec_headers.get_header('WARC-Target-URI').replace('/', '_')}.warc"
                current_warc_record = open(current_warc_filename, 'wb')
                current_warc_payload = record.content_stream()

            if current_warc_record is not None:
                current_warc_record.write(record.raw_stream.read())

        if current_warc_record is not None:
            current_warc_record.close()
            current_warc_payload.close()

split_warc_file('example.warc')