As promised, a little bit late, due to urgent work and holidays. So, this is a very simple milter, written in Python, that do the following when a new message arrive.
1. Get the recipient uid "UID" from the email address. 2. Get the list of address books in the database for the user with uid "UID" 3. Search the sender email address in all the user's address books 4. If found at least one, add one header "X-AddressBook" which contains a list of all the address books found, with a prefixed syntax: The header contains a list written for instance "SOGo:Personnal, SOGo: Professional", it should be, in theory, possible to add other address books, like RoundCube or an LDAP directory, with the same header. There are some limitations, of course, but it is still functional. - Once active, sieve filters should be simpler, no need to have a long list of from email address, just create an address book, and a put your contacts in it. - You can create a different automatic responder for each address book. - Because I am using this on Debian Stretch for now, I am using the available version of python-milter, which uses python2. When I will port this to buster, I will use the python3-milter package. I tried to made it compatible with version 3 as much as I could. - This is freshly new, and has not been tested in production yet. I am working on it. - You can test the behaviour using an SQLite database. Because my first idea was to use a global sqlite database synchronised with multiple address books, it was supposed to contain not the email address but a hash (sha256). - I am not a Python expert, and perhaps there are some errors in the code, but it should be readable. I had to disable a few pylint warnings, as I could not solve them at all, especially those related to the inherited class. - This is the first time I am writing a milter script, be indulgent. The code: https://github.com/progmaticltd/sogo-milters/blob/master/milter-abook/milter-abook.py I initially wrote this for HomeBox¹, I may add some documentation for the deployment if needed. Kind regards, André PS: I do not understand the subtleties of the SOGo address book schema in the database, but it looks like a very weird implementation without any SQL principle in mind. I think this is probably now too late to rewrite the database schema to something convenient. PPS: If anyone knows an RFC that specify a header name that contains the source address book of a message sender, I am happy to change the header name. Otherwise, maybe I may submit an RFC draft. -- [email protected] https://inverse.ca/sogo/lists
