| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
| |
This check was done when generating complete nicklist, but not when
updating a single user. Deleted users are still part of the member list
of a channel, if they don't leave the channel first. This caused a
deleted user to be inserted into the nicklist when processing a joined
channel message or when receiving a presence change for that user.
When investigating this, I saw that we still receive presence change
events for deleted users if we subscribe to them. We might want to
filter out the deleted users when generating the list of users to
subscribe to.
|
|\
| |
| | |
Include members set and team in SlackThreadChannel
|
| | |
|
| |
| |
| | |
Allows autocompletion of nicks to work (instead of error from `nick_completion_cb`)
|
|\ \
| | |
| | | |
Fix quit callback function
|
|/ /
| |
| |
| |
| | |
The function was renamed at some point without renaming the hook signal,
so weechat would print an error when exiting.
|
|\ \
| | |
| | | |
Temporary fix user presence subscription crashing big teams
|
|/ / |
|
|\ \
| | |
| | | |
SlackUser: use profile.display_name if available
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Since recent changes in Slack, usernames can no longer be set directly, and
are instead inferred from the email address. This is not particularly useful,
so the recommendation is to instead use display_name. display_name may have
spaces in it, however, which confuses tab completion.
With this change, wee-slack uses display name for both display and completion,
after first stripping all whitespace from it. Ref linking (via linkify_text)
ensures that @mentions continue to work properly even when this whitespace
removal means that the name being @mentioned locally doesn't match the "real"
username.
|
|\ \ \
| |/ /
|/| | |
Restore OAuth support
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
|\ \ \
| | | |
| | | | |
Fix presence subscription now mandatory to get state changes
|
|/ / /
| | |
| | |
| | | |
https://api.slack.com/docs/presence-and-status#batching
|
|\ \ \
| | | |
| | | | |
Suppress SSLWantReadError errors
|
|/ / /
| | |
| | |
| | |
| | | |
This exception is expected to some degree and does not affect the functioning
of wee-slack. With this change, it will no longer appear in the log.
|
|\ \ \
| |_|/
|/| | |
Handle memberlist updates in channelsinfo message
|
|/ /
| |
| |
| |
| |
| |
| | |
In large Slacks, channel membership isn't included in the initial channel
info, but in channelsinfo messages sent after connection. This fixes an
issue where most users would not ever show up in the nicklist when
connecting to such slacks.
|
|\ \
| |/
|/| |
Add completion of custom emojis
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
It's simpler to just reload the whole list instead of adding and
removing specific emojis, especially since we would have to remove
aliases for an emoji when the emoji is removed (and we don't store if an
emoji is an alias or not).
Since this probably doesn't happen that often, it shouldn't be a problem
to do it this way.
|
| |
| |
| |
| |
| |
| |
| | |
After the last commit, the EMOJI variable is accessed in the
initialization of SlackTeam. This made some tests fail since they don't
run the main code, so it would not be defined. By setting EMOJI to an
empty list in the top, we ensure that this is not a problem.
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This loads the list of custom emojis for each team after wee-slack has
connected to the team, and adds them to the emoji completion list for
that team.
The list of custom emojis is only loaded if the list of the standard
emojis has been loaded, since it would be a bit strange to only have
completions of the custom emojis.
|
| | |
|
| |
| |
| |
| |
| | |
This prevents an error that EMOJI is not defined when trying to tab
complete after the emoji list could not be loaded.
|
| |
| |
| |
| |
| |
| |
| |
| | |
By printing the exception instead of sys.exc_info, we get more useful
error messages, including the path to the file it is trying to load if
it can't find it.
Fixes #431
|
| |
| |
| |
| |
| | |
Emoji completion works fine for me without it, so this doesn't seem to
be necessary anymore.
|
|\ \
| | |
| | | |
feat: Remove Slacks auto prefixing of url protocols
|
| | | |
|
| | |
| | |
| | |
| | |
| | | |
Instead of each of the callers of unfurl_refs having to get the config
values, get them in unfurl_refs if they are not set by the caller.
|
| | |
| | |
| | |
| | |
| | |
| | | |
This changes the feature from the previous commit so it's also possible
to only display the url when the text and the url (without the protocol)
in a link is equal.
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
When you write multiple words separated by dots, or email adresses,
Slack will turn those into links. Wee-slack will display this as
`text (url)`, where url is just the text with http:// or mailto:
prefixed. This makes the text hard to read, and might not even make
sense (e.g. when writing server names or user@hostname).
When you receive such an url, the alt text of the url will be the
original text. This means that we can check if the url is just the alt
text with a protocol prefixed. If it is, we should just print the alt
text and ignore the url.
I guess some people may want the urls, so they are able to click the
links. And since it removes information coming from slack, I made it
into an option instead of making it the default.
|
|\ \ \
| | | |
| | | | |
Use the timestamp of the last message in mark_read
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
All of the times we used sorted_message_keys, we were accessing some of
the last elements in the list. Since OrderedDict is doubly linked, using
reversed(keys) and next/islice is much more efficient than converting it
to a list.
Because of this the method now returns a generator in reverse order. It
is renamed to main_message_keys_reversed to make it clear that it isn't
returning thread messages, and that it is now reversed.
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
By using an OrderedDict, we can sort the messages when we store them, so
we don't have to sort them each time we access them with
sorted_message_keys.
If we are guaranteed that the message we store is always the newest
message, we can even drop the sorting, as OrderedDicts are sorted in
insertion order. However, I'm not sure if we might store an older
message at some point, so to be safe, we sort the messages each time we
store one.
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
When sending a mark read message to slack, use the timestamp of the last
message in the channel, instead of the current timestamp. This should
prevent potentially marking messages as read, if they arrive after
mark_read was called and the clock of the slack server and the machine
wee-slack runs on is not exactly in sync.
According to issue #307 messages might also be marked as read if they
arrive right after you switch buffer, and you have a slow internet
connection. Though, since the time used in mark_read is set
synchronously when the buffer is changed, I'm not sure if that can be
the case. Either way, I think this patch should fix the issue.
If there are no messages in the channel, fall back to using the current
time as timestamp.
Fixes #307
|
|\ \ \
| | | |
| | | | |
Attachment tests and minor fixes
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The title and from_url fields of attachments doesn't seem to be escaped,
so the check if they already are in the text didn't work if the link
contained e.g. an ampersand.
|
| | | | |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
This is a follow-up of commit 0345ba1. That prevented rendering the link
if it already was in the text. This prevents rendering both the
title_link and from_url if they are equal, but not in the text. In this
case it only renders the title_link.
|
| |/ / |
|
|\ \ \
| | | |
| | | | |
Add ability to open multiparty DMs with /query and /slack talk
|
| | | | |
|
| | | |
| | | |
| | | |
| | | |
| | | | |
conversations.open doesn't return `unread_count_display` for MPDMs for
some reason (only DMs), so use mpim.open instead.
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
- Fix incorrect parameter name for MPDMs: `users` instead of `user_ids`
- Accept comma-separated nicks instead of space-separated to match
weechat's `/query` syntax
- Check user IDs instead of usernames to find channels
- Use `SLACK_API_TRANSLATOR` instead of hard coding API methods
- Update documentation
|
|/ / /
| | |
| | |
| | |
| | | |
`command_talk` can now accept multiple arguments, which allows commands
like `/query user1 user2` to open a MPDM with user1 and user2.
|