[feature-request] Add mailpit instance ID (MP_LABEL) to hook json.

[baiomys]

Sometimes it is necessary to make API callbacks from hook processing code, for example to delete messages with unknown recipients.
Client IP is not reliable enough to determine API URL for callbacks.

[axllent]

That makes sense. Currently the webhook is a 1:1 "copy" of a message summary (the same as the web UI has), so I don't think it's a good idea to clone / modify the object for every message just to add a new field because there could be 100+ messages per second (wasted CPU).

Would an additional HTTP header work with the webhook request? I'm thinking something like Mailpit-Label: blaah if the label is set - easy to implement, and no wasted CPU.

[baiomys]

Great! Header should be fine, tnx.

It would be nice also to somehow include info about absent html part.

Hook code generates

1. preview link for telegram bot, pointing via Caddy/Nginx to view/{ID}.html,
2. in some cases preview image of HTML content

mailpit return error code which MUST be handled on reverse proxy side, it would be better not to send link and do not call image generation API (which is really CPU hungry) at all in case there is no HTML.

[axllent]

Adding the header isn't an issue (and I will do this), however your other request to add an extra field is problematic because it introduces the same problem I explained earlier: The webhook receives the same "MessageSummary" that the browser's websocket connection receives - which does not include any HTML or Text (other than a "snippet" summary).

I am assuming your Telegram bot has logic built in? When it receives a webhook, can it not call the API and determine that itself?

• Incoming webhook has message ID XwxLXpPvasMibz9GbX5ZaB
• Telegram bot does a HTTP request to Caddy/Nginx proxy to <https://<server>/api/v1/message/XwxLXpPvasMibz9GbX5ZaB
• Telegram bot can decide what to do based on the value of HTML (or Text as a fallback) etc

[baiomys]

Is it possible to include in header summary info on size of all attachment parts? I.e. html: 0, txt: 105, attachment1: 200, attachment2: 300 and so on?

Hook will handle thousands of messages per minute, so implementing logic to track availability of HTML on HOOK side seems not practical/possible. To implement logic on BOT side it will be necessary to make it stateful instead of stateless and keep track of bot recipient messages instead of sending just one link.

Anyway, in current condition problem can be solved on proxy side, if I am asking for something too hard to implement just don't mention.
=)

P.S. One more question, are IDs really unique? It looks like old plain MD5 coded to base64. Do they include timestamp before encoding?

[axllent]

I think your request for all that information is too specific to your own needs, rather than beneficial to everyone and the project, so I won't add that information to the webhook. Sorry....

I am interested to hear what you are building that is going to be receiving tousands of webhook connections per minute? Obviously you have multiple Mailpit instances (instance ID in the webhook request), but what are you building (if you don't mind me asking)?

I will add the label to the header in the next day or two and include it in a new Mailpit release by the end of the weekend 👍

To answer your question, yes, the IDs should be really unique. It is definitely not MD5, it uses a shortened version of UUID which is case-sensitive (XwxLXpPvasMibz9GbX5ZaB != xwxlxppvasmibz9gbx5zab) - so if you are storing IDs then make sure they are case sensitive.

[baiomys]

I am interested to hear what you are building that is going to be receiving tousands of webhook connections per minute?

public SMTP to telegram gateway hosted on dozen of VPS worldwide

Thanks for your time and fantastic multipurpose application. =)

[axllent]

This feature has just been released in v1.21.6 👍