Dumps all emails via IMAP

dump_emails.py

#!/usr/bin/env python3
import argparse
import functools
import imaplib
import subprocess
import sys
import time
import zipfile

print_err = functools.partial(print, file=sys.stderr, flush=True)


def process_mailbox(M, mailbox, zip_file):
    print_err(f"process {mailbox!r}")
    rv, messages = M.search(None, "ALL")
    if rv != "OK":
        print_err("messages not found")
        return

    # print_err(messages)
    message_nums = messages[0].decode().split()
    print_err("total messages:", len(message_nums))
    for num in message_nums:
        time.sleep(0.1)
        rv, data = M.fetch(num, "(RFC822)")
        if rv != "OK":
            print_err("can't fetch", num)
            continue

        filename = f"./{mailbox}/{num}.eml"
        print_err(f"add to archive: {filename!r}")
        zip_file.writestr(filename, data[0][1])


# Мне было лень разбираться как декодировать UTF-7 этот
def decode_mailbox(s):
    try:
        return subprocess.check_output(
            "tr '&' + | iconv -f UTF-7 -t UTF-8",
            shell=True,
            input=s.encode(),
        ).decode()
    except:  # noqa: E722
        return s


def quote_if_needed(s):
    return f'"{s}"' if any(c.isspace() for c in s) else s


def main():
    parser = argparse.ArgumentParser(description="Dumps all emails via IMAP")
    parser.add_argument("-H", "--host", required=True)
    parser.add_argument("-p", "--port", type=int, default=993)
    parser.add_argument("-u", "--username", required=True)
    parser.add_argument("-P", "--password", required=True)
    parser.add_argument("-o", "--output", type=argparse.FileType("wb"), default="-")
    args = parser.parse_args()

    M = imaplib.IMAP4_SSL(args.host, args.port)
    M.login(args.username, args.password)

    # ('OK', [b'(\\HasNoChildren \\Inbox) "/" "INBOX"', b'(\\HasNoChildren \\Sent) "/" "Sent Messages"', b'(\\HasNoChildren \\Drafts) "/" "Drafts"', b'(\\HasNoChildren \\Trash) "/" "Deleted Messages"', b'(\\HasNoChildren \\Junk) "/" "Junk"', b'(\\HasNoChildren) "/" "&sLSsjMT0ulTHfNVo-"', b'(\\HasNoChildren) "/" "&zK2tbAC3rLDIHA-"', b'(\\HasNoChildren) "/" "&znTTmA-"', b'(\\HasNoChildren) "/" "SNS"', b'(\\HasNoChildren) "/" "&1QS4XLqowVg-"'])
    rv, mailboxes = M.list()
    assert rv == "OK"

    mailboxes = [item.decode().split('"')[-2] for item in mailboxes]
    print_err(mailboxes)

    with zipfile.ZipFile(
        args.output, "w", compression=zipfile.ZIP_DEFLATED
    ) as zip_file:
        for mailbox in mailboxes:
            # Бросит ошибку если имя содержит пробелы
            # Вроде декодированное имя выбрать нельзя
            rv, _ = M.select(quote_if_needed(mailbox))
            if mailbox.startswith("&"):
                mailbox = decode_mailbox(mailbox)
            if rv != "OK":
                print_err(f"can't select mailbox: {mailbox}")
                continue
            try:
                process_mailbox(M, mailbox, zip_file)
            finally:
                M.close()

    M.logout()


if __name__ == "__main__":
    main()