Selectively remove text/html from multipart/alternative structures in email

rmhtml.py

from email import message_from_bytes
from email.iterators import typed_subpart_iterator, body_line_iterator
from email.policy import default


def nontrivial(textpart):
    """
    Check that this part contains at least three lines of text.
    """
    for idx, _ in enumerate(body_line_iterator(textpart)):
        if idx == 2:
            return True
    return False

def scrubbed(message):
    """
    Examine message; if it has a multipart/alternative structure
    with text/plain and text/html parts, and nontrivial() is True
    for the text/plain part, remove the text/html part and return
    True. Otherwise, return False. The message is modified in-place
    if the conditions are met.
    """
    text = None
    html = None
    for container in typed_subpart_iterator(
            message, maintype='multipart', subtype='alternative'):
        text = None
        html = None
        for part in typed_subpart_iterator(container):
            ctype = part.get_content_type()
            if ctype == 'text/plain' and nontrivial(part):
                text = part
            elif ctype == 'text/html':
                html = part
            if text and html:
                # FIXME: meddles with the internals;
                # calls get_content_type anew
                for idx, payload in enumerate(container._payload):
                    if payload.get_content_type() == 'text/html':
                        container._payload.pop(idx)
                        break
                return True
    return False

def killhtml_maybe(filename):
    """
    Read email from filename; replace with a scrubbed version
    if scrubbed() succeeds.
    """
    with open(filename, 'rb') as handle:
        message = message_from_bytes(handle.read(), policy=default)
    if scrubbed(message):
        with open(filename, 'wb') as handle:
            handle.write(message.as_bytes())


def main():
    import sys
    for filename in sys.argv[1:]:
        killhtml_maybe(filename)


if __name__ == '__main__':
    main()

Inspired by https://superuser.com/questions/1624144/bulk-delete-text-html-parts-of-email-messages

Unfortunately, when we rewrite a message, the new contents are rendered from Python's internal representation, which ends up rewriting a lot of the structures so you can't e.g. easily diff against the input message and see what changed.

https://stackoverflow.com/a/20250820/874188 has a healthier approach to removing attachments but I could not quickly get it to work like I wanted. I suspect I didn't try hard enough.