Return of the missing Yahoo HTML email stripping bug

Contribute Bug

AI-generated summary

The discussion revolves around the recurrence of a bug where Yahoo HTML email signatures are not being stripped from notification replies in Discourse. downey initially reports the issue, noting it now includes extra quoted email content. codinghorror assigns it to zogstrip, who deploys a fix via commit 46b6c55. However, downey confirms the bug persists, sharing a screenshot. zogstrip requests the raw email, which downey provides. zogstrip notes the system works as intended but admits challenges in identifying user signatures and line breaks. wesochuck asks if a preferred signature format exists for easier stripping, suggesting a delimiter like -------------------- before signatures. They also reference Mailgun’s open-source parsing library as a potential solution. sam clarifies that Discourse’s library, email_reply_trimmer, is a custom re-implementation inspired by Mailgun Talon and others, and believes the issue is now resolved.

1

Continuing the discussion from HTML email signature not being stripped out of notification reply:

https://meta.discourse.org/t/html-email-signature-not-being-stripped-out-of-notification-reply/21351?u=downey

It’s back, although now with extra email quoting inclusion as well. :slightly_smiling:

2

Yes this is for @zogstrip

3

Anyone who wants the original message source via Mess feel free to ping me.

1 me gusta
4

This should now be fixed for good :wink:

https://github.com/discourse/discourse/commit/46b6c55197a41e13c163f10e00138d78d6e79820

5 Me gusta
5

Or not… :frowning:

Pasted image769×274 67.5 KB

6

Can you send me the raw email?

7

Of course; check your messages. :wink:

1 me gusta
8

Looks like it’s working as expected here. (Except for the fact that we can’t properly identify user signatures and new lines, which are really hard problems)

9

Is there a preferred format for email signatures that is easily strippable by Discourse? If so we, could let our members know the format.

1 me gusta
10

Looks like this is optimized for stripping out Mobile signatures:

Our biggest offenders come from Desktop clients, for example:

Pasted image461×590 126 KB

In our case, we could gently encourage our users to proceed their signatures with something like: --------------------<line break> if that is acceptable.

Has this option already been considered: http://blog.mailgun.com/open-sourcing-our-email-signature-parsing-library/ ?

11

Hemos reescrito nuestra biblioteca a lo largo de los años, está inspirada en MailGun Talon y otras bibliotecas.

Creo que esto está solucionado.

1 me gusta