Automatic emoji conversion occurs in username strings
Categories
(Thunderbird :: Instant Messaging, defect)
Tracking
(Not tracked)
People
(Reporter: sopwerdna, Unassigned)
References
(Blocks 1 open bug)
Details
Attachments
(1 file)
17.25 KB,
image/png
|
Details |
Steps to reproduce:
Join any matrix room with the chat functionality (I encountered this in #matrix:matrix.org). Let some other user join with a homeserver starting with 's' (so that the username string ends up as @user:something.com
).
Actual results:
The characters :s
in the other user's join message are interpreted and replaced with the emoji U+1F615 (confused face) and so the displayed username is unreadable. I am attaching an example of the behavior with a test user from the above public chat room (user was @admin:synapse-performance-test.lab.element.dev).
Expected results:
Automatic emoji conversion should probably be skipped for usernames/room join/leave messages, and system messages in general.
Updated•3 years ago
|
Comment 2•3 years ago
|
||
This is a duplicate of bug 954572.
Hi Patrick,
I would argue this is definitely not a duplicate. Reading #954572, the issue is that the regex for automatic emoji conversions is a bit too enthusiastic and should be tuned to avoid word boundaries. This bug report is that automatic emoji conversion is occurring in system messages (i.e. join/leave messages, usernames, and server addresses) which should always be presented verbatim by Thunderbird and not modified. I would propose reopening this and investigating why these emoji conversions are happening in these sorts of places in the UI. I can try to track this down if someone can point me roughly to the right part of the codebase (although I'm not at all familiar with Thunderbird's codebase, I would like to try to assist).
Thanks!
RE: the previous question by Wayne Mery, I haven't had a chance to check, but I didn't think 91 had matrix support at all?
Comment 4•3 years ago
|
||
Sorry for the delay on responding.
(In reply to sopwerdna from comment #3)
I would argue this is definitely not a duplicate. Reading #954572, the issue is that the regex for automatic emoji conversions is a bit too enthusiastic and should be tuned to avoid word boundaries.
I think fixing bug 954572 would fix this as well.
This bug report is that automatic emoji conversion is occurring in system messages (i.e. join/leave messages, usernames, and server addresses) which should always be presented verbatim by Thunderbird and not modified.
I don't agree with this -- I don't see system messages as special that they might not have emoji in them. (E.g. a part message or subject could easily have plaintext emojis in them.)
It occurring in usernames does seem a bit different, and maybe not covered by bug 954572. We might be able to do something similar to bug 953653 to fix this case.
I would propose reopening this and investigating why these emoji conversions are happening in these sorts of places in the UI. I can try to track this down if someone can point me roughly to the right part of the codebase (although I'm not at all familiar with Thunderbird's codebase, I would like to try to assist).
The issue is the same root cause as bug 954572 -- the regex for finding text emoji is overly greedy.
RE: the previous question by Wayne Mery, I haven't had a chance to check, but I didn't think 91 had matrix support at all?
This isn't specific to Matrix, other protocols would exhibit the same behavior, e.g. an IRC usernames of foo^^
would match.
I do somewhat wonder if we even need to do plaintext emoji identification in incoming messages anymore now that everything has settled on Unicode, which includes fairly rich emoji in it.
Description
•