Auto-import new mail to DEVONthink

E-mail archiving in DT is great and works very well. One thing I was missing though was the ability to automatically import my new mail. I have a simple archiving strategy: delete nothing, archive everything (this might not be everyone’s preference!)

I have a simple rule in Fastmail to add a “Year/2022” label to every incoming mail (meaning that basically everything I’ve received is labeled). With a Python script I simply check hourly for new mail in the 2022 mailbox + in my Sent folder. It also makes use of the "Extract attachment" script so all my incoming attachments are automatically searchable in DT.

For me this approach goes a long way in making DT my ‘single source of thruth’ for any document I have received in my e-mail, downloaded, clipped etc. I just search something in my DT databases and can almost immediately find it and e.g. send it to someone else or use it in a document I’m building.

2022-04-08 CleanShot CleanShot 13.29.49

Notes / bugs:
I’ve run into some weird situations looking like a race condition where DT tries to execute the smart rule (or tries to execute it too soon) while the database is not (fully) opened. Which is why I’ve included a simple database check in the Applescript below. Normally smart rules should only run when the database is fully opened.

We should be able to call a smart rule from within a smart rule, but that behavior is currently not working in 3.8.3 which is why I’m using a bit of a workaround: including the script and calling performsmartrule(theRecords) directly. This should be possible in 3.8.4 again. As a workaround the import tags are used to determine which records to process for replacing attachments.

Applescript: Auto import new mail

use AppleScript version "2.4" -- Yosemite (10.10) or later
use scripting additions

property pythonCmd : "/Users/mdbraber/.pyenv/shims/python3"
property pythonScriptPath : "/Users/mdbraber/iCloud Drive/Coding/Devonthink import mail/"
property pythonScriptName : "import-mail-to-dt.py"
property importCmd : pythonCmd & " " & quoted form of (pythonScriptPath & pythonScriptName)

-- 
property replaceScript : load script POSIX file "/Users/mdbraber/Library/Application Scripts/com.devon-technologies.think3/Smart Rules/Replace mail attachments.scpt"

on run
	tell application id "DNtp"
		my performSmartRule(selection as list)
	end tell
end run

on performSmartRule(theRecords)
	tell application id "DNtp"
		try
			-- Check if the database is open, because sometimes when starting
			-- DT it isn't open yet while the smart rule already runs (which is
			-- weird but might have to do with the opening taking longer?)
			set theDatabase to database "Mail"
			set theOutput to do shell script importCmd
			
			-- This is a workaround because 'perform smart rule' doesn't work
			-- from smart rules in DT 3.8.3 so we're including our script and 
			-- using tagged records. Ugly, but it works for now. This should
			-- be fixed in DT 3.8.4
			set theRecords to search "tags:import scope:\"Mail\""
			-- This script replaces attachments and also strips the import tag
			-- (for now)
			replaceScript's performSmartRule(theRecords)
			
		on error error_message number error_number
			if error_number is not -128 then log message error_message
		end try
	end tell
end performSmartRule

Python: import-mail-to-dt.py

#!/usr/bin/env python3
import imaplib
import email
from email.header import decode_header
from email.header import make_header
import uuid
import subprocess
import sys
import os
import logging
import keyring
import re
import datetime
import traceback

# login username
username = "your@email.com"
# login password (using keyring to save credentials)
# set password by issueing the command `keyring.set_password("import-mail-to-dt", username, "s3cr3tpa$sw0rd"
password = keyring.get_password("import-mail-to-dt", username)
imap_server = "imap.myprovider.com"
imap_port = 993
# get current year (my mailboxes are year-based)
year = datetime.datetime.now().date().strftime("%Y")
# DT database name
database = "Mail"
# group in DT database
group = "/" + year
# current path of this script
script_path = os.path.dirname(__file__)

logging.basicConfig(level=logging.INFO, format='%(asctime)s %(levelname)s %(message)s')

try:
	mail = imaplib.IMAP4_SSL(imap_server,imap_port)
	mail.login(username,password)

	count = 0
	for mailbox in [year, "Sent"]:
		# I use year-based archive mailboxes
		if mailbox == year:
			mailbox_name = "Year/" + year
		else:
			mailbox_name = mailbox

		uid_filename = os.path.join(script_path, ".uid." + mailbox)
		uid_file = open(uid_filename, "r")
		uid_line = uid_file.readline().split(" ")
		
		uidnext_prev = int(uid_line[0])
		uidvalidity_prev = int(uid_line[1])

		status = mail.status(mailbox_name, '(UIDNEXT UIDVALIDITY)')
		if status[0] == "OK":
			uidnext_now = int(re.search(r"UIDNEXT (\d*)",str(status[1][0].decode())).group(1))
			uidvalidity_now = int(re.search(r"UIDVALIDITY (\d*)",str(status[1][0].decode())).group(1))

			if not (uidvalidity_prev == uidvalidity_now):
				exit("UIDVALIDITY of " + mailbox_name + " has changed - stopping")
		else:
			exit("Can't get mailbox status of " + mailbox_name)

		if uidnext_now > uidnext_prev:
			mail.select(mailbox_name,True)
			type, data = mail.search(None, 'UID ' + str(uidnext_prev) + ':*')
			id_list = data[0].split()
			for num in id_list:
				resp, data = mail.fetch(num, '(RFC822)')
				raw_email = data[0][1]
				raw_email_string = raw_email.decode('utf-8')
				msg = email.message_from_string(raw_email_string)
				filename = os.path.join(script_path, str(uuid.uuid4()) + ".eml")
				file = open(filename, "w")
				a = file.write(str(msg))
				file.close()
				
				if os.path.exists(filename):

					script = '''
					tell application id "DNtp"
						if it is running
							set theGroup to create location "{group}" in database "{database}"
							set theRecord to import POSIX path of "{filename}" to theGroup
							set theMetadata to meta data of theRecord
							set theSubject to |kMDItemSubject| of theMetadata
							set name of theRecord to theSubject
							set unread of theRecord to false
							set modification date of theRecord to creation date of theRecord
							-- This does not work in DT 3.8.3 so we use a workaround by
							-- adding an 'import tag' (see calling AppleScript for details)
							--perform smart rule "Mail: replace attachments" record theRecord
							set tags of theRecord to {{"import"}}
						else
							error "DEVONthink not running" number -10000
						end if
					end tell
					'''

					proc = subprocess.Popen(['osascript', '-'], stdin=subprocess.PIPE, stdout=subprocess.PIPE)
					proc_input = bytes(script.format(group=group,database=database,filename=filename),'utf-8')
					proc.communicate(input=proc_input)

					if not proc.returncode == 0:
						exit("Something went wrong importing the e-mail - is DT running?")	
			

					subject = str(make_header(decode_header(msg['subject']))).replace("\\n", "")
					logging.info("Downloading %s",subject)
					count = count + 1
					
					os.remove(filename)

			uid_file = open(uid_filename, "w")
			a = uid_file.write(str(uidnext_now) + " " + str(uidvalidity_now))
			uid_file.close()
			mail.close()

	if count > 0:
		logging.info("Downloaded %s messages",count)

	mail.logout()


except Exception as e:
	traceback.print_exc()
	print (str(e))
1 Like

Very interesting. What I don’t see - what client are you important from? I’ve to come out of Outlook.

Thanks!

No client - it’s importing all e-mail directly via IMAP

OK interesting, thanks.

Yes, it might still be relevant as my current way of working is using a local imap server (running in a docker container) that I push the mails over to. My way of importing currently is by grabbing the files from the imap server’s directory, doing some parsing to get the subject into the filename, then pushing them over into my inbox for archival. It works very very stable since over 2 years or so, but of course having DTP actually kind of playing IMAP client would be even much nicer.

Thanks, I’ll have a look

M

If it ain’t broke don’t fix it :slight_smile:

I saw some your code on Github. If you’re actually only temporarily using files to import e-mail into DT you can also use Applescript to simply do set theSubject to |kMDItemSubject| of (meta data of theRecord) after importing (which is what I’m using when importing the .eml files). This would also give you the added benefit of not having to deal with filesystem/filename limitations in your subjects.

Very well spotted. The one issue I do have is file names: For identical subjects, I just keep appending dots, which I later parse out; the other is special characters / foreign alphabets. Those just then tend to not move over at all - no big deal, but the only reason I occasionally even use the default import function of DTP.

So great, thanks for that recommendation, I’ll try that out.

Best,

M