cleanHeaders.py 文件源码

def extractEmailBody(filename):
    with open(filename, 'r') as fin:
        mail = fin.readlines()
        mail = ''.join(mail)
        msg = Parser().parsestr(mail)

        metadata = {
            'To': msg['to'],
            'From': msg['from'],
            'Cc': msg['cc'],
            'Bcc': msg['bcc'],
            'Subject': msg['subject'],
            'Date': msg['date'],
            'References': msg['references']
        }
        return msg.get_payload(), metadata

# Removes everything after 'Original Message' or '<somebody> wrote:'
# TODO: Fix this for replies AFTER original messages