Extract attachements from .eml email downloaded file from GMail, email client, etc.

email_extract_attachment.py

#!/usr/bin/env python3

"""
Use Python stdlib "email" to extract attachments from an .eml file
as obtained from an email client or Download from Gmail etc.

based on https://gist.github.com/urschrei/5258588
"""

from pathlib import Path
import email
import email.policy
import email.message
import argparse
import logging


def extract_payload(att: email.message.EmailMessage, extract_dir: Path) -> Path | None:

    try:
        outfn = att.get_filename()
    except AttributeError:
        logging.error(f"Could not get attachment filename")
        return None

    if not outfn:
        logging.error(f"Attachment has no filename")
        return None

    outpath = extract_dir / outfn
    with outpath.open("wb") as of:
        try:
            if pd := att.get_payload(decode=True):
                of.write(pd)  # type: ignore
                return outpath
        except TypeError:
            logging.error(f"Couldn't get payload for {outfn}")

    return None


def extract_attachments(fn: Path, extract_dir: Path) -> None:

    count = 0

    with fn.open("r") as f:
        msg = email.message_from_file(f, policy=email.policy.default)
        for att in msg.iter_attachments():
            if outpath := extract_payload(att, extract_dir):
                count += 1
                print(f"Extracted {outpath}")

        if count == 0:
            raise ValueError(f"No attachment found for file {f.name}")


if __name__ == "__main__":
    p = argparse.ArgumentParser(description="extract attachments from .eml file")
    p.add_argument("file", help=".eml file to extract attachments from")
    p.add_argument(
        "-extract_dir", help="directory to extract attachment to", default="."
    )
    P = p.parse_args()

    eml = Path(P.file).expanduser()

    extract_attachments(eml, Path(P.extract_dir).expanduser().resolve(strict=True))