- Display reason for validation error when logging in via an authenticator
- Fix email validation handling for 'Discourse SSO', and add a spec
Previously, validation errors (e.g. blocked or already-taken emails) would raise a generic error with no useful information.
When calculating whether the attached uploads went over the SiteSetting.email_total_attachment_size_limit_kb.kilobytes limit, we were using the original_upload for the calculation instead of the actually attached_upload, which will be smaller in most cases because it can be an optimized image.
At the moment, when filtering by group, the directory will unconditionally return the current user at the top of the list. This is quite unexpected, given that the user is deliberately trying to filter the list. This commit makes sure the 'include current user' logic only triggers for unfiltered directories
This adds a new min_trust_level_to_allow_ignore site setting that enables admins to control the point at which a user is allowed to ignore other users.
After the search term is parsed for advanced search filters, the term may
become empty. Later, the same term will be passed to Discourse.route_for
which will raise an ArgumentError.
> URI(nil)
ArgumentError: bad argument (expected URI object or URI string)
* FEATURE: display error if Oneboxing fails due to HTTP error
- display warning if onebox URL is unresolvable
- display warning if attributes are missing
* FEATURE: Use new Instagram oEmbed endpoint if access token is configured
Instagram requires an Access Token to access their oEmbed endpoint. The requirements (from https://developers.facebook.com/docs/instagram/oembed/) are as follows:
- a Facebook Developer account, which you can create at developers.facebook.com
- a registered Facebook app
- the oEmbed Product added to the app
- an Access Token
- The Facebook app must be in Live Mode
The generated Access Token, once added to SiteSetting.facebook_app_access_token, will be passed to onebox. Onebox can then use this token to access the oEmbed endpoint to generate a onebox for Instagram.
* DEV: update user agent string
* DEV: don’t do HEAD requests against news.yahoo.com
* DEV: Bump onebox version from 2.1.5 to 2.1.6
* DEV: Avoid re-reading templates
* DEV: Tweaks to onebox mustache templates
* DEV: simplified error message for missing onebox data
* Apply suggestions from code review
Co-authored-by: Gerhard Schlager <mail@gerhard-schlager.at>
Allowing the editing of remote themes has been something Discourse has advised against for some time. This commit removes the ability to edit or upload files to remote themes from Admin > Customize to enforce the recommended practice.
* FIX: Store Reviewable's force_review as a boolean.
Using the `force_review` flag raises the score to hit the minimum visibility threshold. This strategy turned out to be ineffective on sites with a high number of flags, where these values could rapidly fluctuate.
This change adds a `force_review` column on the reviewables table and modifies the `Reviewable#list_for` method to show these items when passing the `status: :pending` option, even if the score is not high enough. ReviewableQueuedPosts and ReviewableUsers are always created using this option.
CookedPostProcessor replaces all large images with their optimized
versions, but for GIF images the optimized version is limited to first
frame only. This caused animations it cooked posts to require a click
to show up the lightbox and start playing.
Now that we have dark logo settings in core, we can relatively easily ensure that static pages (such as the 404 page) use a logo that is appropriate for the given light or dark color scheme.
Animated emojis were converted to static images. This commit moves the
responsability on site admins to optimize their animated emojis before
uploading them (gifsicle is no longer used).
Plugin client.en.yml and server.en.yml can now be client/server-(1-100).en.yml. 1 is the lowest priority, and 100 is the highest priority. This allows plugins to set their priority higher than other plugins, so that they can override eachothers' translations.
It used to simply say "title is invalid" without giving any hint what
the problem could be. This commit adds different errors messages for
all caps titles, low entropy titles or titles with very long words.
We updated version of moment and moment-timezone as our current versions are outdated making Discourse Dates broken on places where timezone had updates, like here in Brazil.
This also update highlightJS to the latest version and corrected a test that relied on a no longer supported locale in
moment.
Previously, `/u/by-external/{id}` would only work for 'Discourse SSO' systems. This commit adds a new 'provider' parameter to the URL: `/u/by-external/{provider}/{id}`
This is compatible with all auth methods which have migrated to the 'ManagedAuthenticator' pattern. That includes all core providers, and also popular plugins such as discourse-oauth2-basic and discourse-openid-connect.
The new route is admin-only, since some authenticators use sensitive information like email addresses as the external id.
This commit adds an additional find_user_by_email hook to ManagedAuthenticator so that GitHub login can continue to support secondary email addresses
The github_user_infos table will be dropped in a follow-up commit.
This is the last core authenticator to be migrated to ManagedAuthenticator 🎉
PostDestroyer should accept the option to permanently destroy post from the database. In addition, when the first post is destroyed it destroys the whole topic.
Currently, that feature is limited to private messages and creator of the post. It will be used by discourse-encrypt to explode encrypted private messages.
When embedding secure images which have been oneboxed, we checked to see if the image's parent's parent had the class onebox-body. This was not always effective as if the image does not get resized/optimized then it does not have the aspect-image div wrapping it. This would cause the image to embed in the email but be huge.
This PR changes the check to see if any of the image's ancestors have the class onebox-body, or if the image has the onebox-avatar class to account for variations in HTML structure.
Per Google, sites are encouraged to upgrade from Universal Analytics v3 `analytics.js` to v4 `gtag.js` for Google Analytics tracking. We're giving admins the option to stay on the v3 API or migrate to v4. Admins can change the implementation they're using via the `ga_version` site setting. Eventually Google will deprecate v3, but our implementation gives admins the choice on what to use for now.
We chose this implementation to make the change less error prone, as many site admins are using custom events via the v3 UA API. With the site stetting defaulted to `v3_analytics`, site analytics won't break until the admin is ready to make the migration.
Additionally, in the v4 implementation, we do not enable automatic pageview tracking (on by default in the v4 API). Instead we rely on Discourse's page change API to report pageviews on transition to avoid double-tracking.
It used to simply say "not allowed" without giving any hint what the
problem could be. This commit refactors the code and tries to improve
readability.
There are issues around displaying images on published pages when secure media is enabled. This PR temporarily makes it appear as if published pages are enabled if secure media is also enabled.
* FEATURE - allow category group moderators to delete topics
* Allow individual posts to be deleted
* DEV - refactor for new `can_moderate_topic?` method
Previous to this change our anonymous rate limits acted as a throttle.
New implementation means we now also consider rate limited requests towards
the limit.
This means that if an anonymous user is hammering the server it will not be
able to get any requests through until it subsides with traffic.
When the linked topic is created we'll not hardcode the topic title and
let onebox work its magic instead so that the title can be updated
automatically.
Users could be silenced or suspended by two staff members at the same time and
would not be aware of it. This commit shows an error message if another penalty
has been applied.
- IgnoredUser records should all now have an expiring_at value. This commit enforces that in the DB, and fixes any corrupt rows
- Changes to the ignored user list are now handled by the `/u/{username}/notification_level` endpoint. This allows setting expiration dates on the ignore. This commit removes the old logic for saving a list of usernames in the user preferences.
- Many specs were calling `IgnoredUser.create`. This commit changes them to use `Fabricate(:ignored_user)` for consistency
A site owner attempting to use both the email_subject site setting and translation overrides for normal post notification
email subjects would find themselves frusturated at the lack of template argument parity.
Make all the variables available for translation overrides by adding the subject variables to the custom interpolation keys list and applying them.
Reported at https://meta.discourse.org/t/customize-subject-format-for-standard-emails/20801/47?u=riking
This commit adds a site setting `auto_close_topics_create_linked_topic`
which when enabled works in conjunction with `auto_close_topics_post_count`
setting and creates a new linked topic for the topic just closed.
The auto-created new topic contains a link for all the previous topics
and the topic titles are appended with `(Part {n})`.
The setting is enabled by default.
We had an issue where onebox thumbnail was too large and thus was optimized, and we are using the image URLs in post to redact and re-embed, based on the sha1 in the URL. Optimized image URLs have extra stuff on the end like _99x99 so we were not parsing out the sha1 correctly. Another issue I found was for posts that have giant images, the original was being used to embed in the email and thus would basically never get included because it is huge.
For example the URL 787b17ea61_2_690x335.jpeg was not parsed correctly; we would end up with 787b17ea6140f4f022eb7f1509a692f2873cfe35_2_690x335.jpeg as the sha1 which would not find the image to re-embed that was already attached to the email.
This fix will use the first optimized image of the detected upload when we are redacting and then re-embedding to make sure we are not sending giant things in email. Also, I detect if it is a onebox thumbnail or the site icon and force appropriate sizes and styles.
This reverts commit e3de45359f.
We need to improve out strategy by adding a cache breaker with this change ... some assets on CDNs and clients may have incorrect CORS headers which can cause stuff to break.
There is a site setting reply_by_email_enabled which when combined with reply_by_email_address creates a Reply-To header in emails in the format "test+%{reply_key}@test.com" along with a PostReplyKey record, so when replying Discourse knows where to route the reply.
However this conflicts with the IMAP implementation. Since we are sending the email for a group via SMTP and from their actual email account, we want all replys to go to that email account as well so the IMAP sync job can pick them up and put them in the correct place. So if the group has IMAP enabled and configured, then the reply-to header will be correct.
This PR also makes a further fix to 64b0b50 by using the correct recipient user for the PostReplyKey record. If the post user is used we encounter this error:
if destination.user_id != user.id && !forwarded_reply_key?(destination, user)
raise ReplyUserNotMatchingError, "post_reply_key.user_id => #{destination.user_id.inspect}, user.id => #{user.id.inspect}"
end
This is because the user above is found from the from_address, but the destination which is the PostReplyKey is made by the post.user, which will be different people.
* PERF: we don't need to use a huge image to test thumbnails
Generating images with 5000x5000 dimensions is an expensive operation.
Using smaller images reduce the time of model spec from 11s to 3s and integration spec from 6s to 2s.
Documenting a few more endpoints so that our api docs can be
automatically generated. Made a couple other minor changes, like
including the "OK" example for our default success response.
Resending an invite moved the expire date in the future, but did not
invalidate it. For example, if an invite was sent to an email,
invalidated and then resent, it would still be left invalidated.
* FEATURE - Add SiteSettings to control JPEG image quality
`recompress_original_jpg_quality` - the maximum quality of a newly
uploaded file.
`image_preview_jpg_quality` - the maximum quality of OptimizedImages
* FEATURE: allow category group moderators to edit posts
If the `enable_category_group_moderation` SiteSetting is enabled, posts should be editable by those belonging to the appropraite groups.
Trying to include this attribute when topic_user is nil causes an error when visiting a topic as anon. Additionally, we don't display the slow mode banner for these users.
`max-width: 50%; max-height: 400px;` is a good fallback, however, if width and height are given and are smaller than fallback - we should persist that smaller size.
Our Email::Sender class accepts an optional user argument, which is used to create a PostReplyKey record when present. This record is used to sub out the %{reply_key} placeholder in the Reply-To mail header, so if we do not pass in the user we get a broken Reply-To header.
This is especially problematic in the IMAP group SMTP situation, because these emails go to customers that we are replying to, and when they reply to us the email bounces! This fixes the issue by passing user to the Email::Sender when sending a group_smtp email but there is still more to do in another PR.
This Email::Sender optional user is a bit of a footgun IMO, especially because most of the time we use it there is a user we can source. I would like to do another PR for this after this one to make the parameter not optional, so we don't end up with these reply issues down the line again.
This consolidates logic used to match routes in ApiKey, UserApiKey and DefaultCurrentUserProvider. This reduces duplicated logic, and will allow UserApiKeysScope to easily re-use the parameter matching logic from ApiKeyScope
Adds a new slow mode for topics that are heating up. Users will have to wait for a period of time before being able to post again.
We store this interval inside the topics table and track the last time a user posted using the last_posted_at datetime in the TopicUser relation.
It pauses Sidekiq, clears Redis (namespaced to the current site), clears Sidekiq jobs for the current site, restores the database and unpauses Sidekiq. Previously it stayed paused until the end of the restore.
Redis is cleared because we don't want any old data lying around (e.g. old Sidekiq jobs). Most data in Redis is prefixed with the name of the multisite, but Sidekiq jobs in a multisite are all stored in the same keys. So, deleting those jobs requires a little bit more logic.
Dependency on gifsicle, allow_animated_avatars and allow_animated_thumbnails
site settings were all removed. Animated GIF images are still allowed, but
the generated optimized images are no longer animated for those (which were
used for avatars and thumbnails).
The added 'animated' is populated by extracting information using FastImage.
This field was used to selectively reoptimize old animations. This process
happens in the background.
Now that we have support for user-selectable color schemes, it makes sense
to simplify seeding and theme updates in the wizard.
We now:
- seed only one theme, named "Default" (previously "Light")
- seed a user-selectable Dark color scheme
- rename the "Themes" wizard step to "Colors"
- update the default theme's color scheme if a default is set
(a new theme is created if there is no default)
When posts or topics are deleted we don't want to immediately delete associated bookmarks, so we have a grace period to recover them and their reminders if the post or topic is un-deleted. This PR adds a task to the Weekly scheduled job to go and delete bookmarks attached to posts or topics deleted > 3 days ago.
Per Google, sites are encouraged to upgrade from `analytics.js` to `gtag.js` for Google Analytics tracking. This commit updates core Discourse to use the new `gtag.js` API Google is asking sites to use. This API has feature parity with `analytics.js` but does not use trackers.
Creates a BabelHelper builder using a default list of plugins, to ensure the transpiled code is always using the same plugins instead of differents plugins in different cases.
When the server gets overloaded and lots of requests start queuing server
will attempt to shed load by returning 429 errors on background requests.
The client can flag a request as background by setting the header:
`Discourse-Background` to `true`
Out-of-the-box we shed load when the queue time goes above 0.5 seconds.
The only request we shed at the moment is the request to load up a new post
when someone posts to a topic.
We can extend this as we go with a more general pattern on the client.
Previous to this change, rate limiting would "break" the post stream which
would make suggested topics vanish and users would have to scroll the page
to see more posts in the topic.
Server needs this protection for cases where tons of clients are navigated
to a topic and a new post is made. This can lead to a self inflicted denial
of service if enough clients are viewing the topic.
Due to the internal security design of Discourse it is hard for a large
number of clients to share a channel where we would pass the full post body
via the message bus.
It also renames (and deprecates) triggerNewPostInStream to triggerNewPostsInStream
This allows us to load a batch of new posts cleanly, so the controller can
keep track of a backlog
Co-authored-by: Joffrey JAFFEUX <j.jaffeux@gmail.com>
A follow-up to a follow-up. (6932a373a3 and 572da7a57b)
Our `discourse_test` Docker image uses git 2.20.1 released on Dec 15, 2018. It does not support `git branch --show-current`. (it was added in 2.22.0)
Previously, any errors in those files would e.g. blow up the update process in docker_manager.
Now it prints out an error and proceeds as if there was no compatibility file.
Includes:
* DEV: Extract setup_git_repo
* DEV: Use `Dir.mktmpdir`
* DEV: Default to `main` branch (The latest versions of git already do this, so to avoid problems do this by default)
* FIX: second factor cannot be enabled if SSO is enabled
If `enable_sso` setting is enabled then admin should not be able to
enable `enforce_second_factor` setting as that will lock users out.
Co-authored-by: Robin Ward <robin.ward@gmail.com>
DEV: Replace instances of Discourse.base_uri with Discourse.base_path
This is clearer because the base_uri is actually just a path prefix. This continues the work started in 555f467.
See https://meta.discourse.org/t/email-address-change-confirmation-email-not-sent-but-every-other-notification-emails-are/165358
In short: with disable emails set to non-staff, email address change confirmation emails (those sent to the new address) are not sent for staff or admin members.
This was happening because we were looking up the staff user with the to_address of the email, but the to address was the new email address because we are sending a confirm email change email, and thus the user could not be found. We didn't need to do this anyway because we are passing the user into the Email::Sender class anyway.
* PERF: avoid lookbehinds when indexing search
Previously we used a `EmailCook.url_regexp` this regex used lookbehinds
Unfortunately certain strings could lead to pathological behavior causing
CPU to skyrocket and regex replace to take a very very long time.
EmailCook still needs a fix, but it is less urgent cause it already splits
to single lines. That said we will correct that as well in a seperate PR.
New implementation is far more naive and relies on the extra spaces search
indexer inserts.
When that site setting is enabled, the category counts (new/unread)
include the subcategory definition topics, but the topics aren't included
in the list. This fixes that discrepancy.
Previously, Jobs::EnqueueDigestEmails would enqueue a digest job for every user, even if there are no topics to send. The digest job would exit, no email would send, and last_emailed_at would not change. 30 minutes later, Jobs::EnqueueDigestEmails would run again and re-enqueue jobs for the same users.
120fa8ad introduced a temporary mitigation for this issue, by randomly selecting a subset of those users each time.
This commit adds a new `digest_attempted_at` column to the `user_stats` table. This column is updated every time a digest job completes for a user. Using this, we can avoid scheduling digest jobs for the same user every 30 minutes. This also removes the random user selection in 120fa8ad, and instead prioritizes users who had digests attempted the longest time ago.
To avoid blocking the sidekiq queue a limit of 10,000 digests per 30 minutes
is introduced.
This acts as a safety measure that makes sure we don't keep pouring oil on
a fire.
On multisites it is recommended to set the number way lower so sites do not
dominate the backlog. A reasonable default for multisites may be 100-500.
This can be controlled with the environment var
DISCOURSE_MAX_DIGESTS_ENQUEUED_PER_30_MINS_PER_SITE
See https://meta.discourse.org/t/changing-a-users-email/164512 for additional context.
Previously when an admin user changed a user's email we assumed that they would need a password reset too because they likely did not have access to their account. This proved to be incorrect, as there are other reasons a user needs admin to change their email. This PR:
* Changes the admin change email for user flow so the user is sent an email to confirm the change
* We now record who the email change request was requested by
* If the requested by user is admin and not the user we note this in the email sent to the user
* We also make the confirm change email route open to anonymous users, so it can be clicked by the user even if they do not have access to their account. If there is a logged in user we make sure the confirmation matches the current user.
* FEATURE: Export the entire user profile as json, not just bio/website
* FEATURE: Add session log information to user export
Even though the columns are named 'auth_token' etc, the content is not actually usable to log into the forum with. Despite all that, it is still truncated for export, to avoid any 'token hash cracking' situations.
Allows site administrators to pick different fonts for headings in the wizard and in their site settings. Also correctly displays the header logos in wizard previews.
When enable_personal_messages was disabled, moderators could not see
the private messages for the "moderators" group. The link was displayed
on the client side, but the checks on the server side did not allow it.
Previously this site setting `embed unlisted` defaulted to false and
empty topics would be generated for embed, but those topics tend to take
up a lot of room on the topic lists.
This new default creates invisible topics by default until they receive
their first reply.
Previously, moving a category into another one, that already had a child category of that name (but with a non-conflicting slug) would cause a 500 error:
```
# PG::UniqueViolation:
# ERROR: duplicate key value violates unique constraint "unique_index_categories_on_name"
# DETAIL: Key (COALESCE(parent_category_id, '-1'::integer), name)=(5662, Amazing Category 0) already exists.
```
It now returns 422, and shows the same message as when you're renaming a category: "Category Name has already been taken".
UploadRecovery only worked on missing Upload records. Now it also works with existing ones that have an invalid_etag status.
The specs (first that test the S3 path) are a bit of stub-a-palooza, but that's how much this class interacts with the the outside world. 🤷♂️
Upload.secure_media_url? raised an exceptions when the URL was invalid,
which was a issue in some situations where secure media URLs must be
removed.
For example, sending digests used PrettyText.strip_secure_media,
which used Upload.secure_media_url? to replace secure media with
placeholders. If the URL was invalid, then an exception would be raised
and left unhandled.
Now instead in UrlHelper.rails_route_from_url we return nil if there is something wrong with the URL.
Co-authored-by: Bianca Nenciu <nenciu.bianca@gmail.com>
Error messages for exceeded rate limits and invalid parameters always used the English locale instead of the default locale or the current user's locale.
The download link on the lightbox for images was not downloading the image if the upload was marked secure, because the code in the upload controller route was not respecting the dl=1 param for force download.
This PR fixes this so the download link works for secure images as well as regular ligthboxed images.
See https://meta.discourse.org/t/changing-a-users-email/164512 for context.
When admin changes an email for a user, we were incorrectly sending the password reset email to the user's old address. Also the new email does not come into effect until the reset password process is done, so this PR adds some notes to the admin to make this clearer.
Extracted commonly used spec helpers into spec/support/uploads_helpers.rb, removed unused stubs and let definitions. Makes it easier to write new S3-related specs without copy and pasting setup steps from other specs.
The NewUserOfTheMonth badge is part of the Badges::GettingStarted group. This group is skipped in BadgeGranter if the user skips the new user tips. However, the NewUserOfTheMonth badge granter job does not account for this. Instead, it notifies the user they've received the badge even if they did not.
This commit introduces a simple fix to allow granting of this badge even to users who skipped the new user tips.
This allows administrators to stop automatic redirect to an external authenticator. It only takes effect when there is a single authentication method, and the site is login_required
This is a little bit of refactoring. Core Discourse should have default promotion message for TL2.
In addition, when the Discobot plugin is enabled, the user is invited to advanced training
Since 9e4ed03, moderators can view groups with visibility level set to "Group owners, members and moderators".
This fixes an issue where moderators can see the group in /g but then get a 404 when clicking on individual groups.
typographer can change " to ” leading to breakages in parser
At least codify this. Longer term we want to re-prioritize typographer so
it always runs after bbcode parsing.
Previously attributes such as `[test a='a"a' b="a'a"]` were not correctly
handled.
This amends the regex parser to ensure it correctly parses attributes
without breaking incorrectly on the first nested quote
To check if a post contains any embedded media, we look if the "image_sizes" attribute is present in the new post manager arguments.
We want to see one boxed links, but we only store the raw content of the post. To work around this, I extracted the onebox logic from the composer editor into a module.
With secure media and the UploadSecurity class, we need a nice way for plugins to register custom upload types that should be considered public and never secure.
We are making the changes from the PR #10563 the default behaviour. Now, if secure media is enabled, secure images will be embedded in emails by default instead of redacting them and displaying a message. This will be a nicer overall experience by default, and for forums that want to be super strict with redaction this setting can always be disabled.
Groups page was loading fields that are only used on the group show
page, so move those fields to the GroupShowSerializer.
Also only fetch the default category and tag notifications once.
This is intended for use by plugins which are building their own topic lists, and want to include PMs alongside regular topics (e.g. discourse-assign). It does not get used directly in core.
Overall stubs lead to long term instability, this helps stabilize them.
Also fixed flaky spec around plugin hooks.
It was relying you `.posts` on system_user which could be loaded already
and invalid. Instead we now load it by hand.
This PR removes the user reminder topic timers, because that system has been supplanted and improved by bookmark reminders. The option is removed from the UI and all existing user reminder topic timers are migrated to bookmark reminders.
Migration does this:
* Get all topic_timers with status_type 5 (reminders)
* Gets all bookmarks where the user ID and topic ID match
* Loops through the found topic timers
* If there is no bookmark for the OP of the topic, then we just create a bookmark with a reminder
* If there is a bookmark for the OP of the topic and it does **not** have a reminder set, then just
update it with the topic timer reminder
* If there is a bookmark for the OP of the topic with a reminder then just discard the topic timer
* Cancels all outstanding user reminder topic timers
* **Trashes (not deletes) all user reminder topic timers**
Notes:
* For now I have left the user reminder topic timer job class in place; this is so the jobs can be cancelled in the migration. It and the specs will be deleted in the next PR.
* At a later date I will write a migration to delete all trashed user topic timers. They are not deleted here in case there are data issues and they need to be recovered.
* A future PR will change the UI of the topic timer modal to make it look more like the bookmark modal.
Editing a post didn't update the `post_uploads` right away. Instead it relied on the `CookedPostProcessor`. This can lead to an inconsistent state if uploads are added or removed during an edit and, for some reason, the `ProcessPost` job doesn't run (successfully). This inconsistency leads to missing uploads, because the newly added uploads appear to be unused and will be deleted by the `CleanUpUploads` job. In addition to that, uploads, which got removed during the edit, appear to be still in use and won't be deleted by the background job.
This commit ensures that the `post_uploads` are updated during the edit without relying on a background job.
In some cases Discourse admins may opt for sessions not to persist when a
browser is closed.
This is particularly useful in healthcare and education settings where
computers are shared among multiple workers.
By default `persistent_sessions` site setting is enabled, to opt out you
must disable the site setting.
Currently, if a group's visibility is set to "Group owners, members" then the mods can't view those group pages. The same rule is applied for members visibility setting too.
This reverts commit 7fc7090. And fixed the spec test fails.
Moderators should not be able to see `UserSerializer#group_users` and `UserSerializer#second_factor_enabled` of other users.
Impact of leaking this is low because the information leaked is not
exploitable.
Currently, if a group's visibility is set to "Group owners, members" then the mods can't view those group pages. The same rule is applied for members visibility setting too.
After restoring a backup it takes up to 48 hours for uploads stored on S3 to appear in the S3 inventory. This change prevents alerts about missing uploads by preventing the EnsureS3UploadsExistence job from running in the first 48 hours after a restore. During the restore it deletes the count of missing uploads from the PluginStore, so that an alert isn't triggered by an old number.
We must guarantee that "rel=noopener" was set if "target=_blank" is present, which is not always the case for trusted users. Also, if the link contains the "nofollow" attribute, it has to have the "ugc" attribute as well.
If a user always read all group messages, we will never update the
`first_pm_unread_at` column since the previous query will not return the
group_user. Instead, we should update `first_pm_unread_at` to the
current timestamp if the user has read everything.
Follow-up to 9b75d95fc6
In c6ceda8c, a bug was introduced where an admin searching for his own
private messages will actually end up searching through all private
messages on the site.
Follow-up to c6ceda8c4e
This PR introduces a few important changes to secure media redaction in emails. First of all, two new site settings have been introduced:
* `secure_media_allow_embed_images_in_emails`: If enabled we will embed secure images in emails instead of redacting them.
* `secure_media_max_email_embed_image_size_kb`: The cap to the size of the secure image we will embed, defaulting to 1mb, so the email does not become too big. Max is 10mb. Works in tandem with `email_total_attachment_size_limit_kb`.
`Email::Sender` will now attach images to the email based on these settings. The sender will also call `inline_secure_images` in `Email::Styles` after secure media is redacted and attachments are added to replace redaction messages with attached images. I went with attachment and `cid` URLs because base64 image support is _still_ flaky in email clients.
All redaction of secure media is now handled in `Email::Styles` and calls out to `PrettyText.strip_secure_media` to do the actual stripping and replacing with placeholders. `app/mailers/group_smtp_mailer.rb` and `app/mailers/user_notifications.rb` no longer do any stripping because they are earlier in the pipeline than `Email::Styles`.
Finally the redaction notice has been restyled and includes a link to the media that the user can click, which will show it to them if they have the necessary permissions.
- Lets child components extend color definitions
- Includes default theme color definitions
- Fails gracefully on color stylesheet SCSS errors
- Includes theme variables when extending colors
There is a request spec that was ignored with the `xit` flag almost a
year ago and every time you generate the api docs with
```
rake rswag:specs:swaggerize
```
it shows the output of this pending test and I guess I finally got sick
of looking at it, so here is a fix for it.
Original Commit: d84c34ad75
This PR ensures that new bookmarks cannot be created for deleted posts and topics, and also makes sure that if a bookmark was created and then the topic deleted that the show topic page does not error from trying to retrieve the bookmark reminder at.
It is possible that a user could exist without an email, if so we should
not enqueue a job to download their gravatar.
This commit resolves this error that can occur:
```
Job exception: undefined method `email' for nil:NilClass
/var/www/discourse/app/models/user.rb:1204:in `email'
/var/www/discourse/app/jobs/regular/update_gravatar.rb:12:in `execute'
```
This commit also fixes the original spec which actually was wrong. The
job never enqueued in the original spec and so the gravatar was never
actually updated and the test was checking if the two values were the
same, but they were both null and never updated, so of course they were
the same!
A new test has also been added to make sure the gravatar job isn't
enqueued when a user's email is missing.
DEV: add plugin hooks for silence message parameters
Allows plugins to add, and update extra silence message params for custom
i18n vars
Allows plugins to override system messages via `message_title` and
`message_raw` parameters. We can later expose these params where necessary via event
hooks. Expose the parameter for the on user_silenced trigger.
When a category is removed from `auto_watch_category` we are removing
CategoryUser. However, there are still TopicUser with notification level
set to `watching` which was inherited from Category.
We should move them back to `regular` unless they were modified by a user.
* FEATURE: Use predictable filenames inside the user archive export
* FEATURE: Include badges in user archive export
* FEATURE: Add user_visits table to the user archive export
Previously in some cases the test suite could fail due to a bad entry in
redis from previous tests
This ensures the correct cache is expired when needed
Additionally improves performance of the redis check
"en_US" doesn't contain most of the translations, so it falls back to "en". But that behavior stopped translation overrides to work for pluralized strings in "en_US", because it relies on existing translations. This fixes it by looking up the existing translation in all fallback locales.
This is in preparation for improvements to the user archive export data.
Some refactors happened along the way, including calling the different _export methods 'components' of the zip file.
Additionally, make the test for post export much more comprehensive.
Copy sources:
app/jobs/regular/export_csv_file.rb
spec/jobs/export_csv_file_spec.rb
This commit adds a new site setting "allowed_onebox_iframes". By default, all onebox iframes are allowed. When the list of domains is restricted, Onebox will automatically skip engines which require those domains, and use a fallback engine.
Originally disabled in 0c0192e. Upload specs now use separate paths for each spec worker.
Fixes an issue in UploadRecovery#recover_from_local – it didn't take into account the testing infix (e.g. test_0) in the uploads/tombstone paths.
With the addition of `PostSearchData#private_message`, a partial
index consisting of only search data from regular posts can be created.
The partial index helps to speed up searches on large sites since PG
will not have to do an index scan on the entire search data index which
has shown to be a bottle neck.
These fields are required when using the UI and if `suspend_until`
params isn't used the user never is actually suspended so we should
require these fields when suspending a user.
This commit is addressing an issue where it is possible that there could
be multiple topic timer jobs running to close a topic or a weird race
condition state causing a topic that was just closed to be re-opened.
By removing the logic from the Topic Timer model into the Topic Timer
controller endpoint we isolate the code that is used for setting an
auto-open or an auto-close timer to just that functionality making the
topic timer background jobs safer if multiple are running.
Possibly in the future if we would like this logic back in the model a
refactor will be needed where we actually pass in the auto-close and
auto-open action instead of mixing it with the close and open
action that is currently being passed to the controller.
When someone wants to add > 1000 users at once they will hit a timeout.
Therefore, we should introduce limit and inform the user when limit is exceeded.
* ensure emails don't have spaces
* import banned users as suspended for 1k yrs
* upgrade users to TL2 if they have comments
* topic: import views, closed and pinned info
* import messages
* encode vanilla usernames for permalinks. Vanilla usernames can contain spaces and special characters.
* parse Vanilla's new rich body format
The filter noops if an incorrect username is passed. This filter is not
exposed as part of the UI but is only used when an admin transitions
from a search within a user's personal messages to the full page search.
Follow-up to 4b30799054.
Renamed from `private_messages` to `personal_messages` without
deprecation because the `private_messages` advanced search filter never
worked in the first place when it was implemented.
This also ensures that restoring a backup works when it was created with the wrong upload paths in the time between ab4c0a4970 (shortly after v2.6.0.beta1) and this fix.
This was likely introduced with the refactor to make ColorSchemeColor a database object. Add a test so this doesn't happen again.
Also test other basics of the WizardSerializer.
For some reason, the .as_json left Ruby objects in; I solved this with a round trip through JSON during the test.
Because we allow all the other flag types on a deleted post we should be
able to send a pm to the user letting them know why we deleted their
post.
Bug report:
https://meta.discourse.org/t/-/161156
Like "default watching" and "default tracking" categories option now the "regular" categories support is added. It will be useful for sites that are muted by default. The user option will be displayed only if `mute_all_categories_by_default` site setting is enabled.
* FIX: Unlike own posts on ownership transfer
If a user has liked a post that has passed the
`post_undo_action_window_mins` system setting window and you transfer ownership
of that post to that user you will be the owner of a post that you have
liked, but cannot unlike resulting in a weird UI behavior. This commit
fixes this issue.
The existing tests didn't check for the timeout window for unliking
posts so I added that in.
I couldn't find a good way to do this logic inside of the guardian class
so rather than duplicating behavior of the `PostActionDestroyer` class
inside of the `PostOwnerChanger` I decided to pass in a "bypass"
variable that could be used to check if the calling class is the
'post_owner_changer' and bypass the guardian instead. I went this route
because the guardian `can_delete_post_action` method has no way of
distinguishing how to allow a user to be able to unlike their own posts
after the timeout window but only on a post owner change.
* use an options hash instead
The rotp gem is currently pinned to version 5.1.0 and this will bump it
up to version 6.0.1.
Follow up to: 85d4370f79
because this issue we were waiting on is now closed:
https://github.com/mdp/rotp/issues/98
Because version 6 is now encoding the params I needed to update the
tests as well.
Enabling the moderators_manage_categories_and_groups site setting will allow moderator users to create/manage groups.
* show New Group form to moderators
* Allow moderators to update groups and read logs, where appropriate
* Rename site setting from create -> manage
* improved tests
* Migration should rename old log entries
* Log group changes, even if those changes mean you can no longer see the group
* Slight reshuffle
* RouteTo /g if they no longer have permissions to view group
Themes can now declare custom colors that get compiled in core's color definitions stylesheet, thus allowing themes to better support dark/light color schemes.
For example, if you need your theme to use tertiary for an element in a light color scheme and quaternary in a dark scheme, you can add the following SCSS to your theme's `color_definitions.scss` file:
```
:root {
--mytheme-tertiary-or-quaternary: #{dark-light-choose($tertiary, $quaternary)};
}
```
And then use the `--mytheme-tertiary-or-quaternary` variable as the color property of that element. You can also use this file to add color variables that use SCSS color transformation functions (lighten, darken, saturate, etc.) without compromising your theme's compatibility with different color schemes.
There is an fk to user_profile that can make destroying uploads fail
if they happen to be set as user profile.
This ensures we clear this information when destroying uploads.
There are more relationships, but this makes some more progress.
Convert all IMAP logging to write to a database table for easier inspection. These logs are cleaned up daily if they are > 5 days old.
Logs can easily be watched in dev by setting DISCOURSE_DEV_LOG_LEVEL=\"debug\" and running tail -f development.log | grep IMAP
It's possible that the original topic image is broken in some form, so
we shouldn't try and generate a topic thumbnail for it. The fix will
prevent the generate_topic_thumbnails job being enqueued every time the
topic is viewed.
For sites that are configured to mute some or all categories and tags
for users by default, groups can now be configured to set members'
notification level to normal from the group manage UI.
* FEATURE: don't notify about changed tags for a private message
Only staff members observing specific tag should receive a notification
* FIX: remove other category which is not used
* FIX: improved specs to ensure that revise was succesful
Previously we did an early return if either SiteSetting.tagging_enabled or SiteSetting.allow_staff_to_tag_pms was false when updating the email on the IMAP server -- however this also stopped us from archiving or deleting emails if either of these were disabled.
The category model already has a default value for `color` and
`text_color` so they don't need to be required via the API. The ember UI
already requires that colors be selected.
The name of the category also doesn't need to be required when updating
the category either because we are already passing in the id for the
category we want to change.
These changes improve the api experience because you no longer have to
lookup the category name, color, or text color before updating a single
category attribute. When creating a category the name is still required.
https://meta.discourse.org/t/-/132424/2
This fixes an issue where a non-default theme set to use the base color
scheme (i.e. the theme had an empty `color_scheme_id`) was loading the
default theme's color scheme instead.
Sometime parallel spec if failing with error:
```
NoMethodError:
undefined method `data' for nil:NilClass
# ./spec/models/topic_tracking_state_spec.rb:339:in `block (4 levels) in <main>'
```
I have a theory that it might be related to instance variables in before block
Adds functionality to reflect topic delete in Discourse to IMAP inbox (Gmail only for now) and reflecting Gmail deletes in Discourse.
Adding lots of tests, various refactors and code improvements.
When Discourse topic is destroyed in PostDestroyer mark the topic incoming email as imap_sync: true, and do the opposite when post is recovered.
The `/directory_items` route needs to have a .json url, but the rails
url helper `_path` doesn't return the format of the route.
I tried passing in a format options to `directory_items_path`. Which
works in the rails console
```
[8] pry(main)> directory_items_path(params.merge(:format => :json))
=> "/directory_items.json?page=1"
```
but when I added that some logic to the controller it comes out as
```
/directory_items?format=json&page=1
```
(which is actually how I expect it to work based on how you pass in the
format param). Anyways, because I couldn't figure out how to pass a
format to the `_path` helper I just used URI.parse to append `.json`
manually.
When we run the S3 inventory, mark uploads that exist as verified true, those that don't as verified false, and uploads not included in the check / not yet checked as verified nil.
The UI prevents users from trying to create tags on topics when they
don't have permission, but if you are trying to add tags to a topic via
the API and you don't have permission before this change it would
silently succeed in creating the topic, but it wouldn't have any tags.
Now a 422 error will be returned with an error message when trying to
create a topic with tags when tagging is disabled or you don't have
enough trust level to add tags to a topic.
Bug report: https://meta.discourse.org/t/-/70525/14
Normally, secure media urls are linked like `/secure-media-uploads/...`. In this case, uploads were already being linked correctly.
But sometimes (e.g. when pulling hotlinked onebox images) secure media is referenced with a full domain name (`//example.com/secure-media-uploads`). This commit ensures that those uploads are also linked correctly.
When a tab is open but left unattended for a while, the red, green, and blue
pills tend to go out of sync.
So whevener we open the notifications menu, we sync up the notification count
(eg. blue and green pills) with the server.
However, the reviewable count (eg. the red pill) is not a notification and
is located in the hamburger menu. This commit adds a new route on the server
side to retrieve the reviewable count for the current user and a ping
(refreshReviewableCount) from the client side to sync the reviewable count
whenever they open the hamburger menu.
REFACTOR: I also refactored the hamburger-menu widget code to prevent repetitive uses
of "this.".
PERF: I improved the performance of the 'notify_reviewable' job by doing only 1 query
to the database to retrieve all the pending reviewables and then tallying based on the
various rights.
Similar to `advanced_filter` I introduced `advanced_order`.
I needed a new option because default orders are evaluated after advanced_filter so I couldn't use it.
Also, that part is a little bit more generic
```
elsif word =~ /order:\w+/
@order = word.gsub('order:', '').to_sym
nil
```
After those changes, I can use them in plugins in this way:
```
Search.advanced_order(:votes) do |posts|
posts.reorder("COALESCE((SELECT dvvc.counter FROM discourse_voting_vote_counters dvvc WHERE dvvc.topic_id = subquery.topic_id), 0) DESC")
end
```
* FEATURE: set notification levels when added to a group
This feature allows admins and group owners to define default
category and tag tracking levels that will be applied to user
preferences automatically at the time when users are added to the
group. Users are free to change those preferences afterwards.
When removed from a group, the user's notification preferences aren't
changed.
In the near future, we will be swtiching to PG headlines to generate the
search blurb. As such, we need to replace audio and video links in the
raw data used for headline generation. This also means that we avoid
replacing links each time we need to generate the blurb.
Previously we would unconditionally keep all images downloaded via pull_hotlinked_images, even if they are later removed from the post. This commit removes that logic, and relies on the existing link_post_uploads process to pick up the downloaded images in `cooked`. Specs are added to ensure this is working correctly for regular hotlinked images, and for oneboxes.
This commit should cause no functional change
- Split into functions to avoid deep nesting
- Register custom field type, and remove manual json parse/serialize
- Recover from deleted upload records
Also adds a test to ensure pull_hotlinked_images redownloads secure images only once
The assign plugin is one of two situations where a post can be both a whisper and a small-action. Check the action_code field to filter out small-actions.
* Fixed an issue I introduced in the last PR where I am just archiving everything regardless of whether it is actually archived in Discourse man_facepalming
* Refactor group list_mailboxes IMAP code to use providers, add specs, and add provider code to get the correct prodivder
A first step to adding automatic dark mode color scheme switching. Adds a new SCSS file at `color_definitions.scss` that serves to output all SCSS color variables as CSS custom properties. And replaces all SCSS color variables with the new CSS custom properties throughout the stylesheets.
This is an alpha feature at this point, can only be enabled via console using the `default_dark_mode_color_scheme_id` site setting.
Apparently latest.json and latest.rss are not routed to the same
controller methods. This change allows for any passed in query
parameters to actually be applied to the rss route.
This came in as a request on meta:
https://meta.discourse.org/t/-/155812/6
Currently, we only reset `email_digests`, `email_level` and `email_messages_level` when the user wants to unsubscribe from all email.
`mailing_list_mode` should be reset as well
Adds a imap_group_id column to IncomingEmail to deal with an issue where we were trying to update emails in the mailbox, calling IncomingEmail.where(imap_sync: true). However UID and UIDVALIDITY could be the same across accounts. So if group A used IMAP details for Gmail account A, and group B used IMAP details for Gmail account B, and both tried to sync changes to an email with UID of 3 (e.g. changing Labels), one account could affect the other. This even applied to Archiving!
Also in this PR:
* Fix error occurring if we do a uid_fetch and no emails are returned
* Allow for creating labels within the target mailbox (previously we would not do this, only use existing labels)
* Improve consistency for log messages
* Add specs for generic IMAP provider (Gmail specs still to come)
* Add custom archiving support for Gmail
* Only use Message-ID for uniqueness of IncomingEmail if it was generated by us
* Various refactors and improvements
* DEV: Show message when cannot invite user to PM
When inviting a user to a PM return a message that says, "Sorry, this
user can't be invited." if they have been muted or are not in a users
allowed pm users list.
* Minor refactor & improved some text
For the following conditions, the TopicUser.bookmarked column was not updated correctly:
* When a bookmark was auto-deleted because the reminder was sent
* When a bookmark was auto-deleted because the owner of the bookmark replied to the topic
This adds another migration to fix the out-of-sync column and also some refactors to BookmarkManager to allow for more of these delete cases. BookmarkManager is used instead of directly destroying the bookmark in PostCreator and BookmarkReminderNotificationHandler.
This changes PG text search to only match the given title against
lexemes that are formed from the title. Likewise, the given raw will
only be matched against lexemes that are formed from the post's raw.
If a user posted a topic and Akismet decided it was spam, the topic gets deleted and put into the review queue. If a category moderator for that category marked the post/topic as "Not Spam" the topic did not get recovered correctly because Guardian.new(@user).can_review_topic?(@post.topic) returned false incorrectly because the topic was deleted.
We do prefix matching in search so there is no need to inject the extra
terms.
Before:
```
"'discourse':10,11 'discourse.org':10,11 'org':10,11 'test':8A,10,11 'test.discourse.org':10,11 'titl':4A 'uncategor':9B"
```
After:
```
"'discourse.org':10,11 'org':10,11 'test':8A 'test.discourse.org':10,11 'titl':4A 'uncategor':9B"
```
There was a bug that even when `email_digest` was set to false but
`digest_after_minutes` was positive, we were not displaying correct
status.
In addition, the message is improved when the user is unsubscribed +
unsubscribe from all is hidden.
This reverts commit 84de643c04.
Users are using the search endpoint as public API even though it is
meant to be internal. Revert for now while we figure out the path
forward on providing a more stable API to end users.
This adds a special filter to topic lists that will filter to tracked and
watched categories.
To use it you can visit:
`https://sitename/?filter=tracked`
`https://sitename/unread?filter=tracked`
and so on
Note, we do not include explicitly tracked and watched topics **outside** of
the tracked categories and tags.
We can consider a `filter=all_tracked` to cover this edge case.
Before this commit if you were bulk removing group members and passed in
a user who wasn't currently a member of that group the whole request
would fail. This change will return a 200 response now listing the users
that were removed and those that were skipped.
To reproduce the initial issue here:
1. A user makes a post, which discourse-akismet marks as spam (I cheated and called `DiscourseAkismet::PostsBouncer.new.send(:mark_as_spam, post)` for this)
2. The post lands in the review queue
3. The category the topic is in has a `reviewable_by_group_id`
4. A user in that group goes and looks at the Review queue, decides the post is not spam, and clicks Not Spam
5. Weird stuff happens because the `PostDestroyer#recover` method didn't handle this (the user who clicked Not Spam was not the owner of the post and was not a staff member, so the post didn't get un-destroyed and post counts didn't get updated)
Now users who belong to a group who can review a category now have the ability to recover/delete posts fully.
This adds an option to "delete on owner reply" to bookmarks. If you select this option in the modal, then reply to the topic the bookmark is in, the bookmark will be deleted on reply.
This PR also changes the checkboxes for these additional bookmark options to an Integer column in the DB with a combobox to select the option you want.
The use cases are:
* Sometimes I will bookmark the topics to read it later. In this case we definitely don’t need to keep the bookmark after I replied to it.
* Sometimes I will read the topic in mobile and I will prefer to reply in PC later. Or I may have to do some research before reply. So I will bookmark it for reply later.
* FEATURE: Allow List for PMs
This feature adds a new user setting that is disabled by default that
allows them to specify a list of users that are allowed to send them
private messages. This way they don't have to maintain a large list of
users they don't want to here from and instead just list the people they
know they do want. Staff will still always be able to send messages to
the user.
* Update PR based on feedback
In 1bd8a075, a hidden site setting was added that causes Email::Styles
to treat its input as a complete document in all cases.
This commit enables that setting by default.
Some tests were removed that were broken by this change. They tested the
behaviour of applying email styles to empty strings. They weren't useful
because:
* Sending empty email is not something we ever intend to do,
* They were testing incidental behaviour - there are lots of
valid ways to process the empty string,
* Their intent wasn't clear from their descriptions,
It seems there was a discrepancy in that background images were attached
to the full slug category class: `category-:slug-:id` and our body class
only had `category-:slug`.
This fix adds support for both formats.
* Remove unneeded bookmark name index.
* Change bookmark search query to use post_search_data. This allows searching on topic title and post content
* Tweak the style/layout of the bookmark list so the search looks better and the whole page fits better on mobile.
* Added scopes UI
* Create scopes when creating a new API key
* Show scopes on the API key show route
* Apply scopes on API requests
* Extend scopes from plugins
* Add missing scopes. A mapping can be associated with multiple controller actions
* Only send scopes if the use global key option is disabled. Use the discourse plugin registry to add new scopes
* Add not null validations and index for api_key_id
* Annotate model
* DEV: Move default mappings to ApiKeyScope
* Remove unused attribute and improve UI for existing keys
* Support multiple parameters separated by a comma
It's possible through an import or other means to have images larger
than the current max allowed image size in the db.
If this happens the thumbnail generation job will keep running
indefinitely trying to download a new copy of the original but
discarding it because it is larger than the max_file_size eventually
causing this error
`Job exception: undefined method `path' for nil:NilClass`
because the newly downloaded image is now nil.
This fix stops the enqueuing of the `GenerateTopicThumbnails` job for
all images that happen to be larger than the max image size.
Considering document length in search introduced too much variance in
our search results such that it makes certain searches better but at the
same time made certain searches worst. Instead, we want to have a more
determistic way of ranking search so that it is easier to reason about
why a post is rank higher in search than another.
The long term plan to tackle repeated terms is to restrict the number of
positions for a given lexeme in our search index.
Follow up to d8c796bc4.
Note that his change increases query time by around 40% in the following
benchmark against `dev.discourse.org` but this is a tradeoff that has to be taken so that relevance
search is accurate.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number ORDER BY rank DESC, bumped_at DESC))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
18.040 (± 0.0%) i/s - 181.000 in 10.035241s
current aggregate search query with proper ranking
12.992 (± 0.0%) i/s - 130.000 in 10.007214s
Comparison:
current aggregate search query: 18.0 i/s
current aggregate search query with proper ranking: 13.0 i/s - 1.39x (± 0.00) slower
```
Indexing query strings in URLS produces inconsistent results in PG and
pollutes the search data for really little gain.
The following seems to work as expected...
```
discourse_development=# SELECT TO_TSVECTOR('https://www.discourse.org?test=2&test2=3');
to_tsvector
------------------------------------------------------
'2':3 '3':5 'test':2 'test2':4 'www.discourse.org':1
```
However, once a path is present
```
discourse_development=# SELECT TO_TSVECTOR('https://www.discourse.org/latest?test=2&test2=3');
to_tsvector
----------------------------------------------------------------------------------------------
'/latest?test=2&test2=3':3 'www.discourse.org':2 'www.discourse.org/latest?test=2&test2=3':1
```
The lexeme contains both the path and the query string.
```
discourse_development=# SELECT alias, lexemes FROM TS_DEBUG('www.discourse.org');
alias | lexemes
-------+---------------------
host | {www.discourse.org}
discourse_development=# SELECT TO_TSVECTOR('www.discourse.org');
to_tsvector
-----------------------
'www.discourse.org':1
```
Given the above lexeme, we will inject additional lexeme by splitting
the host on `.`. The actual tsvector stored will look something like
```
tsvector
---------------------------------------
'discourse':1 'discourse.org':1 'org':1 'www':1 'www.discourse.org':1
```
Previously, we would only take either the `MIN` or `MAX` for
`post_number` during aggregation meaning that the ranking is not
considered.
```
require 'benchmark/ips'
Benchmark.ips do |x|
x.config(time: 10, warmup: 2)
x.report("current aggregate search query") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT topics.id, min(posts.post_number) post_number FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted)) GROUP BY topics.id ORDER BY MAX((
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
) DESC, topics.bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.report("current aggregate search query with proper ranking") do
DB.exec <<~SQL
SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id" FROM "posts" JOIN (SELECT *, row_number() over() row_number FROM (SELECT subquery.topic_id id, (ARRAY_AGG(subquery.post_number))[1] post_number, MAX(subquery.rank) rank, MAX(subquery.bumped_at) bumped_at FROM (SELECT "posts"."id", "posts"."user_id", "posts"."topic_id", "posts"."post_number", "posts"."raw", "posts"."cooked", "posts"."created_at", "posts"."updated_at", "posts"."reply_to_post_number", "posts"."reply_count", "posts"."quote_count", "posts"."deleted_at", "posts"."off_topic_count", "posts"."like_count", "posts"."incoming_link_count", "posts"."bookmark_count", "posts"."score", "posts"."reads", "posts"."post_type", "posts"."sort_order", "posts"."last_editor_id", "posts"."hidden", "posts"."hidden_reason_id", "posts"."notify_moderators_count", "posts"."spam_count", "posts"."illegal_count", "posts"."inappropriate_count", "posts"."last_version_at", "posts"."user_deleted", "posts"."reply_to_user_id", "posts"."percent_rank", "posts"."notify_user_count", "posts"."like_score", "posts"."deleted_by_id", "posts"."edit_reason", "posts"."word_count", "posts"."version", "posts"."cook_method", "posts"."wiki", "posts"."baked_at", "posts"."baked_version", "posts"."hidden_at", "posts"."self_edits", "posts"."reply_quoted", "posts"."via_email", "posts"."raw_email", "posts"."public_version", "posts"."action_code", "posts"."locked_by_id", "posts"."image_upload_id", (
TS_RANK_CD(
post_search_data.search_data,
TO_TSQUERY('english', '''postgres'':*ABCD'),
1|32
) *
(
CASE categories.search_priority
WHEN 2
THEN 0.6
WHEN 3
THEN 0.8
WHEN 4
THEN 1.2
WHEN 5
THEN 1.4
ELSE
CASE WHEN topics.closed
THEN 0.9
ELSE 1
END
END
)
)
rank, topics.bumped_at bumped_at FROM "posts" INNER JOIN "post_search_data" ON "post_search_data"."post_id" = "posts"."id" INNER JOIN "topics" ON "topics"."id" = "posts"."topic_id" AND ("topics"."deleted_at" IS NULL) LEFT JOIN categories ON categories.id = topics.category_id WHERE ("posts"."deleted_at" IS NULL) AND "posts"."post_type" IN (1, 2, 3, 4) AND (topics.visible) AND (topics.archetype <> 'private_message') AND (post_search_data.search_data @@ TO_TSQUERY('english', '''postgres'':*ABCD')) AND (categories.id NOT IN (
SELECT categories.id WHERE categories.search_priority = 1
)
) AND ((categories.id IS NULL) OR (NOT categories.read_restricted))) subquery GROUP BY subquery.topic_id ORDER BY rank DESC, bumped_at DESC LIMIT 51 OFFSET 0) xxx) x ON x.id = posts.topic_id AND x.post_number = posts.post_number WHERE ("posts"."deleted_at" IS NULL) ORDER BY row_number;
SQL
end
x.compare!
end
```
```
Warming up --------------------------------------
current aggregate search query
1.000 i/100ms
current aggregate search query with proper ranking
1.000 i/100ms
Calculating -------------------------------------
current aggregate search query
17.726 (± 0.0%) i/s - 178.000 in 10.045107s
current aggregate search query with proper ranking
17.802 (± 0.0%) i/s - 178.000 in 10.002230s
Comparison:
current aggregate search query with proper ranking: 17.8 i/s
current aggregate search query: 17.7 i/s - 1.00x (± 0.00) slower
```
David Taylor
Martin Brennan
tshenry
Dan Ungureanu
Roman Rizzi
jbrw
Penar Musaraj
Kane York
Arpit Jalan
Krzysztof Kotlarek
Bianca Nenciu
Justin DiRose
Gerhard Schlager
Mark VanLandingham
Rafael dos Santos Silva
Guo Xiang Tan
Sam
Vinoth Kannan
Blake Erickson
Jahan Gagan
Angus McLeod
Gerhard Schlager
Joffrey JAFFEUX
Jarek Radosz
Arpit Jalan
Daniel Waterworth
Robin Ward
Vinoth Kannan
Neil Lalonde
Jeff Wong
Joshua Rosenfeld
Faizaan Gagan
siriwatknp
Rachel Carvalho
Régis Hanol
Martin Brennan
jbrw
