Since we recently blocked accidental serialization of AR models, we are getting a 500 error in some cases with thumbnails. We can fix this by serializing the thumbnail, previously we just returned a raw OptimizedImage object.
Thumbnails are now attached to the serializer in core, therefore we no longer need to use add_to_serializer within the chat plugin to use thumbnails within chat message uploads.
This patch improves the custom `array` type available in contracts.
It’s now able to split strings on `|` on top of `,`, and to be more
consistent, it also tries to cast the resulting items to integers.
This is extracted from https://github.com/discourse/discourse/pull/29129.
In some chat specs, we provide an array as a value for group lists like
`chat_allowed_groups`, which is wrong. This results in a value like
`"1|2|[3]"` instead of `"1|2|3"`.
Currently, `ChatSDK.create` restricts what parameters can be provided to
the underlying service. This prevents the `discourse-ai` plugin from
using it in one of its specs.
This patch allows extra parameters to be provided.
In certain locales like English (GB), If a user posted 2 subsequent messages, the first would have a date displayed in 24 hour format, while the second message would be shown in 12 hour format (when hovering the message).
This change forces both messages to display in 12 hour format, the first message showing the am/pm, and the second showing the smaller version without am/pm.
This commit replaces all uppy-related mixins with standalone classes. The main entrypoint is now lib/uppy/uppy-upload.js, which has a list of its config options listed at the top of the file. Functionality & logic is completely unchanged.
The uppy-upload mixin is replaced with a backwards-compatibility shim, which will allow us to migrate to the new pattern incrementally.
Constants should always be only assigned once. The logical OR assignment
of a constant is a relic of the past before we used zeitwerk for
autoloading and had bugs where a file could be loaded twice resulting in
constant redefinition warnings.
The `id` column of `notifications` table and `notification_id` columns
of the other tables have been migrated to bigint in previous commits
(for example, 799a45a).
In order to run the migrations with zero downtime, the data had to be
copied to new columns and swapped, but the old columns have been kept
to allow for rollback. They are no longer needed now.
When chat channels are deleted, some users may be able to click the thread before it gets removed from the UI. This leads to a 500 error causing log noise. We can use the safe navigational operator to prevent calling chatable when the channel is not found (due to deleted_at constraint in query).
Support threads in DMs and group chats so members can keep their conversations organized.
This change adds a new toggle switch for threads within the Chat Channel Settings screen. For new direct message channels threading is enabled by default.
We have made a decision to exclude direct message threads from the My Threads screen for now.
If a plugin's JS fails to load for some reason, most commonly
ad blockers, the entire admin interface would break. This is because
we are adding links to the admin routes for plugins that define
them in the sidebar.
We have a fix for this already in the plugin list which shows a warning
to the admin. This fix just prevents the broken link from rendering
in the sidebar if the route is not valid.
This helps uncover issues with bigint columns that are joined with int
columns. It also introduces a temporary API for plugins to migrate int
columns to bigint in test environment to make tests pass.
This commit fixes an issue where the following happens:
1. The user opens a page where an alternative sidebar panel is displayed like /admin or other page where a plugin is displaying an alternative sidebar like the `docs-categories` plugin
2. Clicking the chat icon in the header and opening the drawer, or if you just minimize chat into drawer after it opens full-screen
3. The alternative sidebar panel is lost and reverted to the main panel.
There have been too many flaky tests as a result of leaking state in
Redis so it is easier to resolve them by ensuring we flush Redis'
database.
Locally on my machine, calling `Discourse.redis.flushdb` takes around
0.1ms which means this change will have very little impact on test
runtimes.
The script `send_chat_message` when used with the `post_created_edited` trigger now accepts `{{post_quote}}` as placeholder for the value of `message`.
This is made possible by a new method in `utils`. Usage:
```ruby
placeholders["foo"] = utils.build_quote(post)
```
On the chat channel settings page, we want to show a single Send push notifications setting instead of the current Desktop notifications and Mobile push notifications settings.
For existing users, use the Mobile push notifications setting value for the new Send push notifications setting.
While using `OpenStruct` is nice, it’s generally not a very good idea as
it usually leads to performance problems.
The `OpenStruct` source code even says basically to avoid it.
Since the context object is crucial in our services, this patch replaces
`OpenStruct` with a custom implementation instead.
This commit attempts to improve the mobile experience for
admin page header and subheader by automatically collapsing
all action buttons in these components into a DMenu when viewing
mobile.
This is done by using different "list" wrapper components and a
DMenu trigger and a DropdownMenu on mobile only, and uses has-block
to determine whether to render the DMenu trigger at all.
This also removes the `PluginOutlet` in `AdminPluginConfigPage`, it
was too inflexible for this `DropdownMenu` case, and since the `:actions`
were always rendering we couldn't rely on `has-block`. A new plugin API,
`registerPluginHeaderActionComponent`, has been introduced instead to
replace it.
Makes channel_id and is_direct_message_channel consistent across desktop notifications, which also removes the need to lookup the channel from Chat Notification Manager.
Currently in services, the `contract` step is only used to define where
the contract will be called in the execution flow. Then, a `Contract`
class has to be defined with validations in it.
This patch allows the `contract` step to take a block containing
validations, attributes, etc. directly. No need to then open a
`Contract` class later in the service.
It also has a nice side effect, as it’s now easy to define multiples
contracts inside the same service. Before, we had the `class_name:`
option, but it wasn’t really useful as you had to redefine a complete
new contract class.
Now, when using a name for the contract other than `default`, a new
contract will be created automatically using the provided name.
Example:
```ruby
contract(:user) do
attribute :user_id, :integer
validates :user_id, presence: true
end
```
This will create a `UserContract` class and use it, also putting the
resulting contract in `context[:user_contract]`.
When trying to create a new automation based on a scriptable that has "force_triggerable" enable, it would break because of a typo in the code.
This fixes the typo and add a spec to ensure this code path is tested.
In 61c1d35f17 I added a
PluginOutlet to AdminPluginConfigPage. This was intended to be
used as a way to render actions buttons inside the header of
a plugin that has a custom admin UI page. This worked, but
since the outlet was generically named, as soon as one plugin
used it the button would show on all plugins.
This fixes the immediate issue by naming the outlet based
on the plugin, then having each plugin specify their own
outlet to render into. There may be a nicer way to do this,
but for now this stops the bleeding.
With the current implementation, a service step can be written as:
```ruby
def my_step(a_default_value: 2)
…
end
```
That’s a pattern we want to avoid as default values (if needed) should
be probably defined in a contract.
This patch makes a service raise an exception if a default value is
encountered.
We are seeing the following error on CI:
```
Errno::EADDRNOTAVAIL:
Failed to open TCP connection to localhost:31337 (Cannot assign requested address - connect(2) for "localhost" port 31337)
```
This is because the job being enqueued makes a HTTP request to the Rails
server but Capybara doesn't start the Rails server until a session is
actually required. Therefore, we need to enqueue the job after we are
sure that a session has been started.
This will help to enforce a consistent pattern for creating service
actions.
This patch also namespaces actions and policies, making everything
related to a service available directly in
`app/services/<concept-name>`, making things more consistent at that
level too.
This commit introduces a way to fetch the "serialized voters" for
multiple polls.
* Use a single query to fetch voters for all types of polls
* Refactor to introduce all_serialized_voters
* Cache serialized voters
Under certain conditions, a recurring automation can end up in a state with no pending automation records, causing it to not execute again until manually triggered.
We use the `RRule` gem to calculate the next execution date and time for recurring automations. The gem takes the interval, frequency, start date, and a time range, and returns all dates/times within this range that meet the recurrence rule. For example:
```ruby
RRule::Rule
.new("FREQ=DAILY;INTERVAL=1", dtstart: Time.parse("2023-01-01 07:30:00 UTC"))
.between(Time.zone.now, Time.zone.now + 2.days)
# => [Sat, 14 Sep 2024 07:30:00.000000000 UTC +00:00, Sun, 15 Sep 2024 07:30:00.000000000 UTC +00:00]
```
However, if the time component of the first point provided to `.between()` is slightly ahead of the start date (e.g., `dtstart`), the first date/time returned by `RRule` can fall outside the specified range by the same subsecond amount. For instance:
```ruby
RRule::Rule
.new("FREQ=DAILY;INTERVAL=1", dtstart: Time.parse("2023-01-01 07:30:00 UTC"))
.between(Time.parse("2023-01-01 07:30:00.999 UTC"), Time.parse("2023-01-03 07:30:00 UTC"))
.first
# => Sun, 01 Jan 2023 07:30:00.000000000 UTC +00:00
```
Here, the start date/time given to `.between()` is 999 milliseconds after 07:30:00, but the first date returned is exactly 07:30:00 without the 999 milliseconds. This causes the next recurring date to fall into the past if the automation executes within a subsecond of the start time, leading to the automation stalling.
I'm not sure why `RRule` does this, but it seems intentional judging by the source of the `.between()` method:
b9911b7147/lib/rrule/rule.rb (L28-L32)
This commit fixes the issue by selecting the first date ahead of the current time from the list returned by `RRule`, rather than the first date directly.
Internal topic: t/138045.
This change adds full names to direct message channel titles when the following conditions are met:
- SiteSetting.enable_names = true
- SiteSetting.display_name_on_posts = true
- SiteSetting.prioritize_username_in_ux = false
If a user's full name is blank, it will fallback to their username in both 1-1 channels and Group DM channels.
- fetch models inside services
- validate `user_id` in contracts
- use policy objects
- extract more logic to actions
- write specs for services and action
`track_sql_queries` only returned queries that were executed by
ActiveRecord. All queries executed through DB.exec, DB.query and others
were not returned.
We're seeing a problem where some recurring automations end up in a state where they don't have any `pending_automations` records scheduled which effectively makes the recurring automation dead. We need to add some debugging to figure out what might be causing this problem.
Internal topic: t/138045.
These fields are often used when serializing topics which may contain
multiple polls. On average, serializing a poll took 2+N queries where N
is the number of options. This change reduces the number of queries to
3, one for each field (Poll#voters_count, PollOption#voters_count and
Poll#has_voted?).
On occasion we get an error popup on desktop due to the channel not being found.
This change means that we only check the cached channels in ChatChannelsManager for the matching channel id, but we skip doing manual lookup which results in ajax popup when it fails.
This commit converts the current chat plugin UI into the
new "show plugin" UI already followed by AI and Gamification.
In the process, I also:
* Made a dedicated /new route to create new webhooks
* Converted the webhook form to FormKit
* Made some fixes and improvements to the `AdminPluginConfigPage`, `AdminPageHeader`,
and `AdminPageSubheader` generic components, so more plugins can
adopt the UI guidelines too. This includes adding a header outlet so plugins
can add action buttons to the plugin show page header.
* Fixes the submit button loading state for FormKit (by Joffrey)
---------
Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com>
This upgrade is designed to be fully backwards-compatible. Any icon names which have changed will be automatically remapped to the new name. For now, this will happen silently. In future, once core & official themes/plugins have been updated, we will start raising deprecation errors to help theme/plugin authors update their code.
Extracted from https://github.com/discourse/discourse/pull/28715
Announcement at https://meta.discourse.org/t/were-upgrading-our-icons-to-font-awesome-6/325349
Co-authored-by: awesomerobot <kris.aubuchon@discourse.org>
⚠️ This commit is a revert of a revert due to a migration which was causing `{}` metadata to be transformed into `{"value": [null]}`. The new migration shouldn't cause this and will also clean the existing errors, there shouldn't be any data loss given the affected fields where not containing actual data. We might want to stop storing these empty fields in the future.
To achieve it, this commit does the following:
- create a new `groups field`, ideally we would have reused the existing group field, but many automations now have the expectation that this field will return a group id and not an array of group ids, which makes it a dangerous change
- alter the code in `post_created_edited` to use this new groups field and change the logic to use an array
- migrate the existing group fields post_created_edited automations to change name from `restricted_group` to `restricted_groups`, the component from `group` to `groups` and the metadata from `{"value": integer}` to `{"value": [integer]}`
<!-- NOTE: All pull requests should have tests (rspec in Ruby, qunit in JavaScript). If your code does not include test coverage, please include an explanation of why it was omitted. -->
Every time a desktop chat sound plays, there should be some visual cue as to why the sound was played in the first place.
This change follows the chat indicator preference:
- All New Messages - a blue dot is shown for all messages, so we attempt to play a sound every time
- Direct Messages, Mentions and Watched Threads - a green dot is shown for all urgent messages, so we attempt to play a sound for urgent chat notifications
- Only Mentions - only play chat sounds when user is mentioned
- Never - we never play chat sounds, as user wouldn’t know why the sound was played
* FEATURE: Add user to topic_tags_changed event
Add user to topic_tags_changed event context
Update automation plugin with new arguments in event
Update tests for new arguments
relates to https://github.com/discourse/discourse-chat-integration/pull/214
* DEV: change variable name for better readability
changed `tags` to be payload and used `values_at` to get the values of the keys
To achieve this this commit does the following:
- create a new `groups field, ideally we would have reused the existing group field, but many automations now have the expectation that this field will return a group id and not an array of group ids, which makes it a dangerous change
- alter the code in `post_created_edited` to use this new groups field and change the logic to use an array
- migrate the existing group fields post_created_edited automations to change name from `restricted_group` to `restricted_groups`, the component from `group` to `groups` and the metadata from `{"value": integer}` to `{"value": [integer]}`
Chat toggle relies on using a height: auto to collapse drawer, however in channel threads we should be only displaying threads when drawer is expanded. Displaying threads conditionally based on drawer toggle status fixes this.
This change increases the visibility of unread channels to make them stand out more in drawer mode (desktop).
When a channel is unread:
- it floats to the top;
- when multiple channels are unread, they are sorted alphabetically (equal to how it’s done on mobile)
- the unread indicator blue dot moves to directly right of the channel name
A previous commit had broken this codepath, this commit ensures it's fixed and is adding a test. It's not testing the copy/paste behavior as fairly complex to test.
This patch removes the `with_service` helper from the code base.
Instead, we can pass a block with actions directly to the `.call` method
of a service.
This simplifies how to use services:
- use `.call` without a block to run the service and get its result
object.
- use `.call` with a block of actions to run the service and execute
arbitrary code depending on the service outcome.
It also means a service is now “self-contained” and can be used anywhere
without having to include a helper or whatever.
We need to start printing deprecation notices when the `show_in_ui` argument is used because it works only for the old about page which will be removed soon. For the new about page, we've introduced a new API `addAboutPageActivity` which is more flexible than a true/false argument on the server side.
Internal topic: t/136551.
- Uses `helper.renderGlimmer` with GJS to render the `<Poll` component without any widgets
- Moves some logic into component, so that only `@post`, `@poll` and `@titleHTML` need to be passed into the component (no more 'attrs')
- Updates `modifyClass` calls to modern syntax
- Replaces observer in `Post` model with a native setter & tracked property
- Replaced Poll EmberObject instances with TrackedObject
- Updated component tests with new arguments
- Updated some tests to qunit-dom
- Fixed up core `repliesBelow` and `repliesAbove` logic to create post models properly. Previously it was passing 'transformed' versions of posts into the model, which doesn't make sense.
Currently, categories support designating only 1 group as a moderation group on the category. This commit removes the one group limitation and makes it possible to designate multiple groups as mods on a category.
Internal topic: t/124648.
A new setting attribute is used to define the areas (separated by `|`).
In addition, endpoint `/admin/config/site_settings.json` accepts new `filter_area` data.
Prior to this fix we could exit early if tags was `[]` as `tags && (tags & post.topic.tags.map(&:name)).empty?` would have returned true. This commit ensures it's not the case anymore and adds a test for it.
Co-Authored-By: Martin Brennan <mjrbrennan@gmail.com>
This change introduces a new thread notification level allowing users to get notified when someone replies to the thread.
Users who watch a thread will get a green notification on the chat icon and a user notification (blue). User notifications are consolidated based on thread id to prevent cluttering the original users notification area.
---------
Co-authored-by: Régis Hanol <regis@hanol.fr>
We're working to remove all decorators from object-literal-properties. This commit refactors all initializers which were using these decorators into classes, where decorators are allowed. Also adds cleanup logic to the local-dates initializer, which was previously missing.
This commit introduces a little bit of duplication
since the old plugin UIs not using the new plugin show
page look different from ones like AI and Gamification
which have been converted. We can use the new admin
header component on the plugins list, but for the other
pages we are manually rendering a breadcrumb trail and
the list of plugin tabs.
Over time as we convert more plugins to use the new UI
guidelines and show page we can get rid of this duplication.
Removes chat-specific changes from dfc947a97d, and adds `preventScroll: true` to prevent timing issues between the emoji picker being brought into the viewport and being focussed.
The following test is flakey. We don't care about the order because we check if the tags are being sent.
```
Failure/Error: measurement = Benchmark.measure { example.run }
expected: ["tag4", "tag5"]
got: ["tag5", "tag4"]
(compared using ==)
```
Create a topic can fail in many different ways and we don't want this to impact the rest of the application, the call will now be wrapped in a begin/rescue block and log an error if it fails.
Previous to this change there is no clean way to apply keyboard shortcuts
to things such as "add poll" and other hidden options in the toolbar
This allows shortcuts to be specified similar to how they are on the toolbar
Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com>
Now that `ActiveRecord` relations are properly handled in a `model`
step, putting this step back will allow the service to stops its
execution if there are no users to be removed without calling the pretty
big SQL query contained in the `CalculateMembershipsForRemoval` action.
When we use CTRL/CMD + K to search, on the server we run 4 queries:
- 1 for users
- 1 for groups
- 1 for category channels
- 1 for direct message channels
The server returns up to 10 results per type and the client concatenate all the results (in the order I described) and then takes the first 10 or so.
Let's say you've had 1:1s with john1, john2, and john3. Searching for “john” would return all the johns as “users”. It doesn’t make sense to also return them as 1:1 dms.
That’s what this fixes. When the search returns some users, we skip all 1:1s because we assume they will show up as the results of the “users” query.
We’ll always return matching direct messages with more than 2 users (aka. “group chats”). All 3 other queries (users, groups, and category channels) are unaffected.
Internal ref - t/136079
This patch allows using an AR relation as a model in services without
fetching associated records. It will just check if the relation is empty
or not. In the former case, the execution will stop at that point, as
expected.
This commit introduces a new frontend API to add custom items to the "Site activity" section in the new /about page. The new API is called `addAboutPageActivity` and it works along side the `register_stat` serve-side API which serializes the data that the frontend API consumes. More details of how the two APIs work together is in the JSDoc comment above the API function definition.
Internal topic: t/128545/9.
This commit improves the hilight-ing of mentions in posts and chat messages.
- `@here` and `@all` will generate a `<a class="mention --wide">`
- bots will generate a `<a class="mention --bot">`
- current user will generate a `<a class="mention --current">`
To achieve this change the following value transformer has been added: "mentions-class". It will be run in posts and chat messages after the mention is rendered.
A bug were bots were not considered in mentioned users has also been fixed as part of this PR.
This is a variation on bc3e8a9963cf9a64d114ec751c875025af169690, which was reverted due to issues on iOS. Safari's "in response to user action" check cannot follow the `runAfterFramePaint` chain of interaction -> requestAnimationFrame -> messageChannel, and so some sensitive browser APIs (e.g. clipboard, upload, etc.) were blocked.
This commit is similar, but uses `next()` instead of `runAfterFramePaint()`. The result seems the same, but doesn't have the same issue on iOS.
The chat-emoji-picker change was required to resolve a test failure. The emoji picker has never closed-on-scroll on desktop, so there is no user-facing change in behavior.
This is a follow-up to 671f40ce07 and
ed11ee9d05.
While the optimisations in the previous commits were sound, it did not
resolve the memory bloat we were seeing. It turns out that we call
`.blank?` on the model's result if the model has not been marked
optional. The problem with this is that if the model returns an
ActiveRecord relation, calling `.blank?` on the relation basically loads
everything into memory.
Therefore, this commit removes `users` as a model in the since it really isn't
a model but just a relation.
This is a follow up to ed11ee9d05.
In `Chat::AutoRemove::HandleCategoryUpdated`, we are currently loading
the related users record in batches and then handing it off to
`Chat::Action::CalculateMembershipsForRemoval.call`. However, we are
still seeing memory spike as a result of this.
This commit eliminates the allocation of `User` ActiveRecord objects until
absolutely necessary. `Chat::Action::CalculateMembershipsForRemoval.call` has been
updated to accept an ActiveRecord relation instead which allows us to
avoid the ActiveRecord allocations.
This change delays the notify watching job to allow time for message to be marked as seen within chat.
Without a slight delay the job fires straight away and often means that messages are seen on screen but the user also receives a notification due to Chat::Notifier.user_has_seen_message? returning false.
This commit seeks to reduce the memory footprint of `Chat::AutoRemove::HandleCategoryUpdated.call`
by optimizing the
`Chat::AutoRemove::HandleCategoryUpdated#remove_users_without_channel_permission` method which was
loading all the ActiveRecord users objects into memory at once. This
change updates the method call to load the ActiveRecord user objects in
batches instead.
See: https://meta.discourse.org/t/cant-edit-topic-with-poll-bug-occurs/320845?u=merefield
When the Poll is set to "results ON_CLOSE", vote numbers for each option are only streamed to the browser when the vote is Closed. It is therefore not possible to render the Results.
The current issue is that when you refresh the page, for those that have voted the default view is results. For this type of poll this should NOT happen. The Results view in this mode should not be possible to see until closure, even for the Author.
Because the votes are not yet serialised when this kind of poll remains open, an attempt to display results causes a JavaScript exception and in any case does not make logical sense.
So the fix here is making sure the default view, for Polls that have results on close, is the voting view until the Poll is Closed.
I've added a test to cover this scenario.
Additionally, this requires a refresh of the page when the poll admin actions a Close to ensure the results are serialized in.
Initially, the poll results display a maximum of 25 voters per option. If the number of voters exceeds this limit, a button allows users to load additional pages of voters.
When this button is clicked, a method is called to retrieve the additional voter information. However, there was a bug where the local tracked object was not properly updated for ranked choice voters. This is due to the existence of two attributes per option: one indicating whether new data is currently loading, and the other containing the list of voters.
Instead of updating the entire option key with the voters list, the fix requires updating the voters attribute only.
This is a small but critical change. The pull request also includes a new test that significantly increases coverage, addressing this issue and beyond.
Adds a new statistics (hidden from the UI, but available via the API) that tracks daily participating users.
A user is considered as "participating" if they have
- Reacted to a post
- Replied to a topic
- Created a new topic
- Created a new PM
- Sent a chat message
- Reacted to a chat message
Internal ref - t/131013
chapoi
Loïc Guitaut
David Battersby
Régis Hanol
David Taylor
Joffrey JAFFEUX
Discourse Translator Bot
Alan Guo Xiang Tan
Jarek Radosz
Bianca Nenciu
Martin Brennan
Sérgio Saquetim
Gabriel Grubba
Jan Cernik
Penar Musaraj
Osama Sayegh
Kris
Robert
Krzysztof Kotlarek
Renato Atilio
Roman Rizzi
Jan Cernik
Sam