* Add Tooltip component to common
Will be used to provide backwards compatibility when we switch to CSS tooltips.
All other methods of creating tooltips are deprecated and this component-based method should be used instead.
* Modify direct child instead of using container element
Instead of using a container to house the tooltip, we'll now modify the
first direct child of the Tooltip component.
The Tooltip component will ensure that:
- children are passed to it
- only one child is present
- that child is an actual HTML Element and not a text node, or similar
- that child is currently present in the DOM
Only after all of the above are satisfied, will the tooltip be created
on that element. We store a reference to the DOM node that the tooltip
should be created on, then use this to perform tooltip actions via
jQuery. If this element gets changes (e.g. the tooltip content is
updated to another element) then the tooltip will be recreated.
If any of the first 3 requirements are not satisfied, an error will
be thrown to alert the developer to their misuse of this component.
To make this work, we do need to overwrite the title attribute of
the element with the tooltip, but this is the only solution other than
specifying `title` as an option when making the tooltip, but this is
not accessible by screenreaders unless they simulate a hover on the
element.
* Add warning about component overwriting `title` attr
* Update previous uses of Tooltip component
Frontend extenders exist in a weird state of limbo, where they are technically defined, but aren't used or tested at all. In 114779bcfd, we shifted from passing `extension.extend` to `flattenDeep` to calling `flat` on `extension.extend`. If an extension doesn't define extenders (as is the case for most extensions), the change breaks the forum. All we do here is add a null check.
* Remove `lodash-es` dependency
* Replace `escapeRegExp` with home-made util
* Replace `throttle` with `throttle-debounce` library
* Use native browser methods for `deepFlatten`
We need a polyfill for iOS 11 and below. I think using a native method with this polyfill is better than having our own function instead, even if the bundle size is ~150B more.
* Save a few bytes in `escapeRegExp`
* Fix typo in comment
* Undo import re-organisation
* Use spread instead of slice
* Use smaller Array.flat polyfill from MDN
* Export new utils in `compat.js`
* Add focus ring mixin
These mixins allow us to restore default browser focus rings on elements which no longer have them.
* Add info about custom outline styles; use `#private` namespace and fix mixin name
I just learned that Less has namespaces! https://lesscss.org/features/#mixins-feature-namespaces
- revisions now use <asset>.<type>?v=<revision> instead of <asset>-<revision>.<type>- remove deprecated filename for revision method
- reconsider use of cache differentiator and implement something that
prevents recompiling css every single time
- allow force recompilation
* Commit initial WIP code
* Fix squashed grid on mobile
* Add pagination support; rename to userList
* Improve grid sizing
* Improve grid row shading
* Move EditUserModal to common
* Add link to profile page in grid
* Use Less styling vars
* Move EditUserModal translations to lib
* Add edit user button to grid
* Fix incorrect profile link priority
* Update profile link translation key
* Add priorities to other columns
* Add group badges to grid
* Add username to profile link tooltip
* Organise imports
* Use variable for header border bottom color
* Fix broken export
* Add total user count to API payload's metadata
* Add new metadata to ApiPayload type
* Implement correct page number
* Remove debug code
* Use function to get the total pages
This allows us to use the raw count elsewhere in the component (pssst... check the next commit!)
* Center profile link in column
* Add profile link header
* Show total users above table
* Use ItemList's itemName property for column data attributes
* Add user email column, hidden by default
This column is hidden by default using a placeholder email and blur filter. These are then removed when the visibility toggle is pressed.
This prevents any over-the-shoulder accidental data leakage, as emails are classed as PII under GDPR.
* Fix incorrect tooltip translation keys
* Add extra padding between email and visibility toggle button
* Prevent selection of blurred email
* Fix incorrect icon state for email toggle
* Update API response type to include metadata (for now)
* Increase number of users per page to 50
* Update compat files with new locations
* Format
* Add @deprecated notices for forum compat export
* Use AdminPayload for user count instead of supplying as REST API metadata
* Make nav look less squashed using bottom margin
* Suppress TS warning
* StyleCI fixes
* Fix TS error
* Update based on review comments
* Rename user list -> users
* Rename internal instances of user_list to users
* Fix formatting
* Use CSS custom properties for the table column count
* Use .Button--icon instead of custom style
* Make fake email more realistic length
* Add a11y attributes
* Use padding bottom instead of margin bottom for page spacing
* Make compatible with new CSS LoadingIndicator
I won't let it break here! :P
* Integrate profile link into username column
* Don't force columns to be 300px
This made the grid look very bloated and intimidating -- lets instead increase the padding between items and make it only the width it needs to be.
* Center edit user button in column
* Increase spacing between email and visibility toggle button
* Rename `statistics` to `modelStatistics` in Admin payload
This prevents any possible conflicts with core and `flarum/statistics`. We might want to consider migrating the stats extension to extend this object in the future.
* Update comments, fix TS error
* Various translation key changes
* Change gmail.com -> example.com
* Stretch 'edit user' button to entire cell size
* Update translations
* Is the YAML formatted right this time? 🙈
* Remove email placeholder
Fixes an issue where the table would jump if an email was unhidden that was longer than the placeholder.
* Re-order lib translations
* Clicking blurred email now unblurs
* Correct header class
* Improve edit user button centring
* Improve vertical row item centering
* Fix incorrect column length in aria attribute
* Use .Button--text!
* Update Loading Indicator
- Fix mistake in LoadingIndicator Less
- Middle align the loading indicator when inline
- Fix Loading Indicator not correctly accepting container class names
- Add inline and block attributes
* Fix loading indicator in composer
* Fix loading indicator on notification list
* Fix loading indicator on discussion page
* Fix loading indicator on button
* Update more uses of loading indicator
* Fix loading indicator in Search box
* Fix AvatarEditor loading spinner
* Set default spinner props
* Replace "tiny" with "small" in Less
* Improve spinner vertical centring in buttons
* Reduce size specificity
* Use single attribute for block/inline
* Use new display attribute
* Use classes for different sizes
* Use `display=block` by default
#b4738febaa2e9a34cea181b621016b50a145115d and #0fb3548f46719090579a5ef4e8cda26234d24943
both added/improved searching tests, but did so in separate files. As a result, the tests did not consider each other, and when both were merged, started failing. This commit combines the tests into one file that tests both order and search in titles.
* Fix searching titles in discussions
* Apply fixes from StyleCI
* Fix tests
* Distinct by discussion ID
* Replace distinct with groupBy
Co-authored-by: Alexander Skvortsov <askvortsov1@users.noreply.github.com>
