gmail_attachment_download

gmail_attachment_download_with_regex.py

#python3 /custom-scripts/gmail_attachment_download.py --download_dir /opt/gmail_attachment_download/ --target_subject ".*spending.*" --start_date "16-May-2024" --end_date "18-May-2024"
#python3 /custom-scripts/gmail_attachment_download.py --download_dir /opt/gmail_attachment_download/ --target_subject ".*spending.*"
#python3 /custom-scripts/gmail_attachment_download.py -h

import imaplib
import email
import os
import argparse
import re
from datetime import datetime, timedelta

# Function to download attachments
def download_attachments(download_dir, target_subject_pattern, start_date, end_date):
    print("Connecting to Gmail server...")
    # IMAP configuration
    imap_host = 'imap.gmail.com'
    imap_port = 993

    # You need to provide your own credentials here
    username = ''                                               
    app_password = ''


    # Connect to the server
    mail = imaplib.IMAP4_SSL(imap_host, imap_port)
    mail.login(username, app_password)

    # Select the mailbox
    mail.select('inbox')

    print("Searching for emails between", start_date, "and", end_date)

    # Search for emails between start_date and end_date
    result, data = mail.search(None, '(SINCE "{}") (BEFORE "{}")'.format(start_date, end_date))

    for num in data[0].split():
        # Fetch the email
        result, email_data = mail.fetch(num, '(RFC822)')
        raw_email = email_data[0][1]
        msg = email.message_from_bytes(raw_email)

        # Get the email subject
        subject = msg.get('Subject')

        # Check if the email subject matches the target subject pattern using regex
        if re.search(target_subject_pattern, subject):
            print(f"Found email with subject matching '{target_subject_pattern}'.")

            # Check if the email has attachments
            for part in msg.walk():
                if part.get_content_maintype() == 'multipart':
                    continue
                if part.get('Content-Disposition'):
                    filename = part.get_filename()
                    if filename:
                        # Extract attachment timestamp from email headers
                        timestamp = email.utils.parsedate_to_datetime(msg['Date'])
                        if timestamp:
                            timestamp_str = timestamp.strftime('%Y-%m-%dT%H%M%S')
                            # Download attachment
                            filepath = os.path.join(download_dir, f"{filename}_{timestamp_str}.csv")  # Include original timestamp in filename
                            print("Downloading attachment to:", filepath)
                            with open(filepath, 'wb') as f:
                                f.write(part.get_payload(decode=True))
                                print(f"Attachment '{filename}' downloaded successfully to {filepath}")
                            # Set modification time of the file
                            os.utime(filepath, (timestamp.timestamp(), timestamp.timestamp()))
    
    # Logout and close connection
    mail.logout()
    print("Script finished.")

def main():
    parser = argparse.ArgumentParser(description="Download attachments from Gmail.")
    parser.add_argument("--download_dir", required=True, help="Directory to download attachments")
    parser.add_argument("--target_subject_pattern", required=True, help="Regular expression pattern to match the target email subject")
    parser.add_argument("--start_date", default=(datetime.today() - timedelta(days=1)).strftime('%d-%b-%Y'),
                        help="Start date for email search (default: yesterday) in format 'DD-MMM-YYYY'")
    parser.add_argument("--end_date", default=(datetime.today() + timedelta(days=1)).strftime('%d-%b-%Y'),
                        help="End date for email search (default: tomorrow) in format 'DD-MMM-YYYY'")

    args = parser.parse_args()

    # Validate required arguments
    if not args.download_dir or not args.target_subject_pattern:
        parser.print_help()
        parser.exit(1)

    download_attachments(args.download_dir, args.target_subject_pattern, args.start_date, args.end_date)

if __name__ == "__main__":
    main()

gmail_attachment_download_without_regex.py

#python3 /custom-scripts/gmail_attachment_download.py --download_dir /opt/gmail_attachment_download/ --target_subject "spending" --start_date "16-May-2024" --end_date "18-May-2024"
#python3 /custom-scripts/gmail_attachment_download.py --download_dir /opt/gmail_attachment_download/ --target_subject "spending"
#python3 /custom-scripts/gmail_attachment_download.py -h

import imaplib
import email
import os
import argparse
from datetime import datetime, timedelta

# Function to download attachments
def download_attachments(download_dir, target_subject, start_date, end_date):
    print("Connecting to Gmail server...")
    # IMAP configuration
    imap_host = 'imap.gmail.com'
    imap_port = 993

    # You need to provide your own credentials here
    username = ''                                               
    app_password = ''


    # Connect to the server
    mail = imaplib.IMAP4_SSL(imap_host, imap_port)
    mail.login(username, app_password)

    # Select the mailbox
    mail.select('inbox')

    print("Searching for emails between", start_date, "and", end_date)

    # Search for emails between start_date and end_date
    result, data = mail.search(None, '(SINCE "{}") (BEFORE "{}")'.format(start_date, end_date))

    for num in data[0].split():
        # Fetch the email
        result, email_data = mail.fetch(num, '(RFC822)')
        raw_email = email_data[0][1]
        msg = email.message_from_bytes(raw_email)

        # Get the email subject
        subject = msg.get('Subject')

        # Check if the email subject matches the target subject
        if subject == target_subject:
            print(f"Found email with subject '{target_subject}'.")

            # Check if the email has attachments
            for part in msg.walk():
                if part.get_content_maintype() == 'multipart':
                    continue
                if part.get('Content-Disposition'):
                    filename = part.get_filename()
                    if filename:
                        # Extract attachment timestamp from email headers
                        timestamp = email.utils.parsedate_to_datetime(msg['Date'])
                        if timestamp:
                            timestamp_str = timestamp.strftime('%Y-%m-%dT%H%M%S')
                            # Download attachment
                            filepath = os.path.join(download_dir, f"{filename}_{timestamp_str}.csv")  # Include original timestamp in filename
                            print("Downloading attachment to:", filepath)
                            with open(filepath, 'wb') as f:
                                f.write(part.get_payload(decode=True))
                                print(f"Attachment '{filename}' downloaded successfully to {filepath}")
                            # Set modification time of the file
                            os.utime(filepath, (timestamp.timestamp(), timestamp.timestamp()))
    
    # Logout and close connection
    mail.logout()
    print("Script finished.")

def main():
    parser = argparse.ArgumentParser(description="Download attachments from Gmail.")
    parser.add_argument("--download_dir", required=True, help="Directory to download attachments")
    parser.add_argument("--target_subject", required=True, help="Target email subject")
    parser.add_argument("--start_date", default=(datetime.today() - timedelta(days=1)).strftime('%d-%b-%Y'),
                        help="Start date for email search (default: yesterday) in format 'DD-MMM-YYYY'")
    parser.add_argument("--end_date", default=(datetime.today() + timedelta(days=1)).strftime('%d-%b-%Y'),
                        help="End date for email search (default: tomorrow) in format 'DD-MMM-YYYY'")

    args = parser.parse_args()

    # Validate required arguments
    if not args.download_dir or not args.target_subject:
        parser.print_help()
        parser.exit(1)

    download_attachments(args.download_dir, args.target_subject, args.start_date, args.end_date)

if __name__ == "__main__":
    main()