(Attempt to) extract (everything) from EML files (specified in command args), and write them to (a well-named) output subdir

parseml.py

#!/usr/bin/env python

"""
2020 update:
- More iterators, fewer lists
- Python 3 compatible
- Processes files in parallel
(one thread per CPU, but that's not really how it works)

hacked to dump everything
"""

import os
import email
from multiprocessing import Pool
import mimetypes
import sys
import traceback

def printe(*args, **kwargs):
    if not kwargs.get("file"):
        kwargs["file"] = sys.stderr
    return print(*args, **kwargs)

def extract(filename):
    if not os.path.exists(filename):
        printe(f"File {filename} does not exist!")
        return 0, 0

    # ensure that an output dir exists
    od = "output_" + os.path.splitext(os.path.basename(filename))[0]
    os.path.exists(od) or os.makedirs(od)

    output_count = 0
    with open(filename, "r") as f:
        msg = email.message_from_file(f)

    for attachment in msg.walk():
        if attachment.is_multipart(): continue

        try:
            output_filename = attachment.get_filename() or hex(hash(attachment))
            if not output_filename:
                printe("Couldn't make a name for a part of %s. Skipping." % filename)
                continue

            output_filename = output_filename.replace("/", "_")

            if os.path.splitext(output_filename)[1] == "":
                ext = mimetypes.guess_extension(attachment.get_content_type())
                if ext:
                    output_filename = output_filename + ext
        except Exception:
            printe("Couldn't make a name for a part of %s. Skipping." % filename)
            traceback.print_exc()
            continue

        with open(os.path.join(od, output_filename), "wb") as of:
            payload = attachment.get_payload(decode=True)
            if not payload:
                payload = attachment.get_content().encode(attachment.get_content_charset())
            if not payload:
                printe("Couldn't get payload for %s" % output_filename)
                continue

            of.write(payload)
            output_count += 1

    if output_count > 0:
        print(f"{filename} -> {od}")
    else:
        printe("No attachment found for file %s!" % filename)

    return output_count > 0 and 1 or 0, output_count


if __name__ == "__main__":
    files = sys.argv[1:]
    if len(files) == 0:
        printe("No input files specified")
        exit(1)

    pool = Pool()
    res = pool.map(extract, files)
    pool.close()
    pool.join()

    done_count, out_file_count = (sum(i) for i in zip(*res))
    print("Done: Processed {} files with {} attachments.".format(done_count, out_file_count))
    if done_count < len(files): exit(1)