parse_bwic.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
parse_bwic.py

example script showing how to parse a structured product BWIC embedded in an
email via the Empirasign Parser as a Service API endpoint.

To get set up with a Free Trial, contact us at:
info@empirasign.com
+1 646-837-8849

https://gist.github.com/empirasign/0835086dc08aec6e767a7bfc6f96c85c

example usage:
./parse_bwic.py hf_bwic.eml       # eml/rfc822 formated run
./parse_bwic.py reit_bwic.msg     # MS Outlook formated run

notes on email file formats
https://www.msoutlookware.com/difference/msg-vs-eml.html
https://www.w3.org/Protocols/rfc822/
https://blogs.msdn.microsoft.com/openspecification/2009/11/06/msg-file-format-part-1/
"""

import argparse
import base64
import json
import pprint

import requests

API_KEY = "MY_API_KEY"        # provided by Empirasign
API_SECRET = "MY_API_SECRET"  # provided by Empirasign
API_URL = 'https://api.empirasign.com/v1/parse-bwic/'


def main():
    """
    the main event
    """
    parser = argparse.ArgumentParser(description='parse BWIC email message in eml/msg format')
    parser.add_argument("infile", help="input msg/eml filename")
    parser.add_argument('-o', dest="outfile", help="output filename [not required]")
    args = parser.parse_args()

    infile = args.infile
    if not args.outfile:
        outfile = infile.rsplit(".", 1)[0] + ".json"
    elif not args.outfile.endswith(".json"):
        outfile = args.outfile + ".json"
    else:
        outfile = args.outfile

    post_data = {'api_key': API_KEY, 'api_secret': API_SECRET}
    if args.infile.lower().endswith(".msg"):
        post_data["msg"] = base64.b64encode(open(infile, "rb").read()).decode("ascii")
    else:
        post_data["rfc"] = open(infile).read()
    res = requests.post(API_URL, json=post_data, timeout=3.5)

    if res.status_code == 200:
        obj = res.json()
        pprint.pprint(obj)
        with open(outfile, "w") as f:
            f.write(json.dumps(obj, sort_keys=True, indent=2, ensure_ascii=False))
        print("\nresults file saved: {}".format(outfile))
    else:
        print("some sort of error occurred, http status code: {}".format(res.status_code))
        pprint.pprint(obj)


if __name__ == "__main__":
    main()