This commit adds `parse_migration_status`, which takes in the string
output of `showmigrations` and parse it into key-value pair of installed
apps and a list of its migration status.
This is a prep commit to rework the check migrations function of
import/export which will parse the output of `showmigrations` to write
the `migration_status.json` file.
This consolidates the list of stale migration to
`lib/migration_status.py` as `STALE_MIGRATIONS`.
This is a prep work to make the migration status tool at
`migration_status.py` be able to clean its output of these migrations
too.
Currently if for what ever reason one decided to change how
`migration_status.json` is written, the check migrations tests in
`test_import_export.py` will happily just use the old and potentially
stale migration status test fixtures in
`fixtures/applied_migrations_fixtures` against other stale fixtures and
run the check migrations tests in a bubble.
This adds an assertion in `verify_migration_status_json` that makes sure
all the migration status fixtures we use in the tests resembles the
actual `migration_status.json` file our export tool will write.
in `get_migrations_status`, we clean up the printed output of any ANSI
codes used to format the output. Currently the regex only cleans up bold
ANSI escape code (\x1b[1m) and style reset code (\x1b[0m). So it won't
be able to clean up basic ANSI escape codes such as "\x1b\31;1m" which
is used to format `showmigrations` output for apps with no migrations.
e.g, "\x1b\31;1m (no migrations)"
This commit updates the regex to catch a wider range of basic ANSI
codes.
The `get_migration_status` command calls `connections.close_all()` when
its done and it was previously only called when we need to rebuild the
dev or test database and when running the `get_migration_status`
command.
This commit moves the `connections.close_all()` call out of the function
and into `test_fixtures.py` directly, making sure it will only be called
when we are rebuilding the dev/test database. This is a prep work to
refactor the check migration function of import/export later on which
plans to use `get_migration_status`.
This moves `get_migration_status` to its own file in
zerver/lib/migration_status.py. This is a prep work to refactor the
check migration function of import/export later on.
Some of the imports are moved into `get_migration_status` because we're
planning to share this file with `check-database-compatibility` which is
also called when one does `production-upgrade`, so we'd want to avoid
doing file-wide import on certain types of modules because it will fail
under that scenario.
In `test_fixtures.py`, `get_migration_status` is imported within
`Database.what_to_do_with_migrations` so that it is called after
`cov.start()` in `test-backend`. This is to avoid wierd interaction with
coverage, see more details in #33063.
Fixes#33063.
1. Fetching from the `/users.list` endpoint is supposed to use
pagination. Slack will return at most 1000 results in a single
request. This means that our Slack import system hasn't worked
properly for workspaces with more than 1000 users. Users after the
first 1000 would be considered by our tool as mirror dummies and thus
created with is_active=False,is_mirror_dummy=True.
Ref https://api.slack.com/methods/users.list
2. Workspaces with a lot of users, and therefore requiring the use of
paginated requests to fetch them all, might also get us to run into
Slack's rate limits, since we'll be doing repeating requests to the
endpoint.
Therefore, the API fetch needs to also handle rate limiting errors
correctly.
Per, https://api.slack.com/apis/rate-limits#headers, we can just read
the retry-after header from the rsponse and wait the indicated number
of seconds before repeating the requests. This is an easy approach to
implement, so that's what we go with here.
Fixes#32199
We only need a log in button since that will take users to
"/accounts/go" if we are on a non-realm specific URL.
"/accounts/go" already has link to go to "Find accounts" page.
Zulip production suite / ${{ matrix.name }} (zulip/ci:bookworm, --test-custom-db, Debian 12 production install with custom db name and user, bookworm) (push) Has been cancelled
Zulip production suite / ${{ matrix.name }} (zulip/ci:jammy, , Ubuntu 22.04 production install and PostgreSQL upgrade with pgroonga, jammy) (push) Has been cancelled
The original mocking example now uses time_machine, so I slimmed
down a lot of its comments, and then I created another
mocking example so that we still touch on mocking in
terms of mock.patch.
The new method reinforces the pattern of testing both the
sad path and happy path inside the same test.
This commit is a part of the work to support empty
string as a topic name.
Previously, empty string was not a valid topic name.
Adds `allow_empty_topic_name` boolean parameter to
`GET /users/me/{stream_id}/topics` endpoint to decide
whether the topic names in the fetched `topics` array
can be empty strings.
If False, the topic names in the fetched response will
have the value of `realm_empty_topic_display_name` field
in `POST /register` response replacing "".
Fixes part of #23291.
Fixes#32706.
A user with permission to invite users should be able to subscribe users
to any of the default streams whether they have the permission to do so
or not for each of those default streams or not. This should only happen
in the invite code path, and not the subscribe code path.
This commit also adds the ability to pick and chose default streams if
you do not have the permission to subscribe to any other channels.
Before this, if you did not have the permission to subscribe any other
channels, only the checkbox to subscribe to all the default streams at
once was available to you.
For the stream pill typeahead, we don't show streams that the user
cannot subscribe other users to. For more details, see
https://chat.zulip.org/#narrow/channel/101-design/topic/can.20subscribe.20other.20users.20permission.20invite
The function to check relevant permissions does so for multiple streams
at once to save us database query counts. Doing it one by one for every
stream would become very expensive.
We've also added `insufficient_permission_streams` to the filter
functions return type for streams for which the current user does not
have permission to subscribe other users.
We remove `invite_to_stream_policy` from the backend wherever applicable
except deleting the field. We have just ported the existing behaviour of
`invite_to_stream_policy` to `can_add_subscribers_group` except one
change. We have added an explicit exception for admins to have this
permission whether they are part of this group or not. The reason for
this is we are adding `stream.can_add_susbcribers_group` in the future
which will grant all admins permission to subscribe other users to a
channel given they have access to a channel. So it makes sense that we
add this exception to the realm level property also.
See https://chat.zulip.org/#narrow/channel/101-design/topic/Can.20subscribe.20other.20users.20on.20user.20profile/near/2039825
While `can_subscribe_other_users` property will make sense for the
current permissions structure where the ability to add subscribers to
channels is dictated with a realm level setting. In the future, we are
adding a channel level `can_add_subscribers_group`, and having a
property called `can_subscribe_other_users` in state_data will be
confusing since the permission to add subscribers will vary channel to
channel.
We have not removed user.can_subscribe_other_users, that will be better
removed when we add the channel level setting.
See more discussion at
https://chat.zulip.org/#narrow/channel/378-api-design/topic/invite_to_stream_policy.20deprecation/near/2039787
b4fedaa765 introduced
this helper, and I assume that the weird loop over
zip made sense at the time.
The assertEqual approach on the whole
set gives nice messages in modern Python.
This new test verifies that the server will
canonicalize the America/Montreal timezone to
America/Toronto. In general, we canonicalize via
the standard Python libraries for cities that
are aliases.
This is essentially a failing test that we can improve
by addressing #32934. It is also a useful test to measure
the actual impact of any improvements (with some more
instrumentation).
The use of assertCountEqual preceded
my changes to use a set instead of a list.
Now assertEqual is more clear and concise.
I should have noticed this in 36a6f0c547
but I caught it now.
stream_topic variable is needed only when updating content
so we set the field inside is_content_edited block.
Also added a comment clarifying about why we use orig_stream
for stream ID.
We already have realm_id from message object passed to
do_update_message so there is no need to pass "None" to
update_message_cache for direct messages.
Previously, if "None" was passed to update_message_cache,
realm_id was eventually computed from stream if it was
a stream message and from Message object for direct messages.
But we always passed a value which is not "None" for stream
messages, and we can be sure that the message.realm will
always be the realm in which stream is present to which the
message was sent.
This was used before `alt="{{ integration.display_name }} logo"` was set
in the common template `integrations/index.html` for all integration
logos. It is no longer necessary.
The `hubot-scripts` account has been taken over.
While calls to the current set of URLs work as they are redirected to
the new links, this commit updates the link to refer to the right
GitHub account.
The query in question runs in a loop when you have
multiple group mentions in a message. We can at
least make it slim.
This is my way to address #32934. It doesn't undo
the O(N) behavior, but N here is usually 1 or 2.
Zulip production suite / ${{ matrix.name }} (zulip/ci:bookworm, --test-custom-db, Debian 12 production install with custom db name and user, bookworm) (push) Has been cancelled
Zulip production suite / ${{ matrix.name }} (zulip/ci:jammy, , Ubuntu 22.04 production install and PostgreSQL upgrade with pgroonga, jammy) (push) Has been cancelled
We now allow changing description and all the permission settings
for deactivated groups as well, as there is no need to restrict
it and makes handling UI for deactivated groups easier.
Description for "PATCH /user_groups/{user_group_id}" endpoint
only mentioned updating name and description, but permission
settings are also updated using the same endpoint.
The bug we're fixing here leaks information by returning an "invalid
subdomain" error when an attempt is made to log in to user@example.com
on a subdomain X when user@example.com does not exist on X, but does
on another subdomain Y.
This allows an attacker to determine that a certain email address has an
account on the server.
Instead, this should just return a regular authentication error.
71406ac767 switched the IMAGE_BOMB_TOTAL_PIXELS cutoff for what
images we preview to include the number of frames in the calculation.
While accurate to the implementation (thumbnailing a 1k-frame animation is
prohibitive, even a small resolutions), this was a behaviour change
from without thumbnailing -- animated gifs did not display inline at
all anymore.
Switch to thumbnailing as many frames as we can fit into a pixel-based
animated thumbnailing threshold, with a minimum of three (to be able
to convey that the image is actually animated). Smaller-resolution
images will hence get more frames in their preview. This also allows
the standard animate-on-hover or always-animate behaviour to be true
to their configurations, without confusing edge cases.
Fixes: #32609.
This commit makes changes to the edit message endpoint
to disallow resolving empty string topic.
It also removes the resolve topic button in the web client
from topic popover and message header for empty string topic.
