Implement BODYSTRUCTURE to optimize fetching HTML/TEXT only nodes

[stevebauman]

Currently we're fetching entire bodies using withBody() on message queries which include attachments and text/html.

We can optimize what developers can retrieve from emails using a BODYSTRUCTURE command to only retrieve specific components (using withAttachments(), withHtml(), etc.).

[ev3rlost]

Any news on the progress of this feature?

Right now we need to use withBody(), but even with chunking - if a inbox has, lets say, over 2000 messages, it simply overloads - just to check if a message has attachments or not.

Highly anticipated feature!

[stevebauman]

Hi @ev3rlost,

Haven't started work on this yet, but if you'd like to submit a PR then I'd be happy to get it merged 👍

Otherwise I'll be starting work on it this upcoming weekend.

[stevebauman]

Also, in regards to this:

if a inbox has, lets say, over 2000 messages, it simply overloads - just to check if a message has attachments or not.

Are you using chunking, or retrieving all 2000 messages at once? Asking as this shouldn't happen with chunking, and using bodystructure won't actually fix this if you are retrieving all 2000 at once, since it will still retrieve all 2000 attachments, html/text bodies without chunking, and just reduce the size of the responses.

[ovidiu-vas]

I doubt this can be done reliably. BODYSTRUCTURE is for each message, not for the whole mailbox. IMAP doesn't guarantee that all messages in the same mailbox have the same BODYSTRUCTURE so it wouldn't be reliable to fetch all messages with body part 1 / 2 and automatically assume you're always getting the plain / html body parts.

Most email providers send the plain text and html in body part 1 and 2 - but because IMAP doesn't guarantee this, it's not a certainty that you will always get this.

I do believe it's valuable to add BODYSTRUCTURE in the MessasgeQuery with a function like withBodyStructure, and let developers analyze the structure and identify parts. Also, it would be useful to have a function in the Message class to get a body part by body part number. This would make the following possible:

• create query to fetch the messages with minimal data like id, headers, flags and bodyparts
• for each message, analyze the bodypart and only get the attachment(s) you need: for example extracting calendar invites or inline attachments.

This would give a developer the granularity to only download the attachments they need for a message.

[stevebauman]

Hey @ovidiu-vas, thanks for this valuable information, appreciate it! I agree with all your suggestions. Let me see what I can cook 🧑‍🍳