Tim Düsterhus [Wed, 18 Aug 2021 09:35:20 +0000 (11:35 +0200)]
Merge pull request #4464 from WoltLab/session-id-unpack
Fix unpacking of the sessionId
Tim Düsterhus [Wed, 18 Aug 2021 07:43:22 +0000 (09:43 +0200)]
Add safety check for unpacked session cookie data
Tim Düsterhus [Wed, 18 Aug 2021 07:44:07 +0000 (09:44 +0200)]
Fix unpacking of the sessionId
As documented by PHP's reference documentation:
> The "a" code now retains trailing NULL bytes.
> The "A" code now strips all trailing ASCII whitespace (spaces, tabs,
> newlines, carriage returns, and NULL bytes).
Previously, with the 'A' code, sessionIds ending in ASCII whitespace would be
incorrectly unpacked, missing their trailing bytes. This ultimately resulted in
the session not being found and the user being logged out.
Five of the 256 possible characters exhibited this bug, making this fail in
roughly 2% of the cases.
However this likely was not noticable by the typical user. Once they have a
non-affected sessionId, this Id is not going to change. What the user might've
noticed is a login not working, despite showing a success message, because they
sessionId change after a successful login handed out an affected sessionId. But
then the user would likely try again, succeeding this time and writing off the
incident as a fluke.
Test script to reproduce the issue:
<?php
for ($i = 0; $i <= 255; $i++) {
$string = "foo".chr($i);
$packed = \pack(
'CA4',
1,
$string
);
$unpacked1 = \unpack('Cversion/A4string', $packed);
$unpacked2 = \unpack('Cversion/a4string', $packed);
if ($unpacked1['string'] !== $string) {
echo "$i: unpacked1\n";
}
if ($unpacked2['string'] !== $string) {
echo "$i: unpacked2\n";
}
}
Tim Düsterhus [Wed, 18 Aug 2021 07:04:18 +0000 (09:04 +0200)]
Fix informal phrase in de.xml
Marcel Werk [Mon, 16 Aug 2021 13:08:54 +0000 (15:08 +0200)]
Mark as read doesn't removed the badge from submenu entries
Marcel Werk [Mon, 16 Aug 2021 12:56:30 +0000 (14:56 +0200)]
Missing informal variant
Marcel Werk [Mon, 16 Aug 2021 12:52:59 +0000 (14:52 +0200)]
Typo
Tim Düsterhus [Mon, 16 Aug 2021 10:53:46 +0000 (12:53 +0200)]
Whitelist `abs` in enterprise mode
Tim Düsterhus [Mon, 16 Aug 2021 09:02:42 +0000 (11:02 +0200)]
Merge pull request #4463 from WoltLab/smiley-unicode-title
Fix Unicode in smiley titles
Tim Düsterhus [Mon, 16 Aug 2021 07:09:10 +0000 (09:09 +0200)]
Fix Unicode in smiley titles
see
5d0bf3ec233f62c6a5a68629e32b7eaa8c9d1dd3
see #4156
Tim Düsterhus [Sun, 15 Aug 2021 12:33:39 +0000 (14:33 +0200)]
Fix localization of PAGE_TITLE in __multifactorTotpSecretField.tpl
Tim Düsterhus [Fri, 13 Aug 2021 10:31:32 +0000 (12:31 +0200)]
Add `required` attributes to acptemplates/login
Marcel Werk [Fri, 13 Aug 2021 09:35:44 +0000 (11:35 +0200)]
Show article labels in recent activities
Tim Düsterhus [Fri, 13 Aug 2021 08:45:58 +0000 (10:45 +0200)]
Fix PHP 8.1 compatibility in DatabaseObject::getDatabaseTableIndexName()
Tim Düsterhus [Fri, 13 Aug 2021 07:20:30 +0000 (09:20 +0200)]
Merge branch '5.3' into 5.4
Tim Düsterhus [Fri, 13 Aug 2021 07:19:27 +0000 (09:19 +0200)]
Merge branch '5.2' into 5.3
Tim Düsterhus [Fri, 13 Aug 2021 07:18:36 +0000 (09:18 +0200)]
Merge pull request #4459 from SoftCreatR/bugfix/custom-errors
Unify error handling in several templates
Sascha Greuel [Thu, 12 Aug 2021 16:08:27 +0000 (18:08 +0200)]
Unified error handling in several templates
Tim Düsterhus [Thu, 12 Aug 2021 15:22:34 +0000 (17:22 +0200)]
Slightly improve phrasing in de.xml
Tim Düsterhus [Thu, 12 Aug 2021 08:09:17 +0000 (10:09 +0200)]
Merge branch '5.3' into 5.4
Tim Düsterhus [Thu, 12 Aug 2021 08:07:08 +0000 (10:07 +0200)]
Add missing call to ->loadVariables() before downloading Google Font during style import
Joshua Rüsweg [Thu, 12 Aug 2021 07:23:30 +0000 (09:23 +0200)]
Rename `Ui/Object/Action/Toogle` to `Toggle` (#4450)
Tim Düsterhus [Wed, 11 Aug 2021 13:31:13 +0000 (15:31 +0200)]
Merge pull request #4456 from WoltLab/style-edit-description
Prevent StyleEditForm from creating phrases with empty name
Tim Düsterhus [Wed, 11 Aug 2021 12:22:10 +0000 (14:22 +0200)]
Delete the empty phrase when updating to 5.4.5
Tim Düsterhus [Wed, 11 Aug 2021 12:19:53 +0000 (14:19 +0200)]
Validate the languageVariable in I18nHandler::save()
Tim Düsterhus [Wed, 11 Aug 2021 12:17:49 +0000 (14:17 +0200)]
Add LanguageEditor::validateItemName()
Tim Düsterhus [Wed, 11 Aug 2021 12:12:42 +0000 (14:12 +0200)]
Store deterministic language variable for style description in StyleEditForm
If the `styleDescription` was empty (as it might be for the default style) this
attempted to store the description in the language item with empty name.
Marcel Werk [Wed, 11 Aug 2021 11:26:02 +0000 (13:26 +0200)]
Missing information who reported a content, if report was made by a guest
Tim Düsterhus [Wed, 11 Aug 2021 09:56:40 +0000 (11:56 +0200)]
Merge branch '5.3' into 5.4
Tim Düsterhus [Wed, 11 Aug 2021 09:56:10 +0000 (11:56 +0200)]
Merge branch '5.2' into 5.3
Tim Düsterhus [Wed, 11 Aug 2021 09:37:23 +0000 (11:37 +0200)]
Merge pull request #4453 from WoltLab/iformbutton-is-available
Check whether an IFormButton is available before rendering
Tim Düsterhus [Wed, 11 Aug 2021 07:54:12 +0000 (09:54 +0200)]
Check whether an IFormButton is available before rendering
WoltLab [Tue, 10 Aug 2021 13:30:36 +0000 (13:30 +0000)]
Updating minified JavaScript files
Alexander Ebert [Tue, 10 Aug 2021 13:26:04 +0000 (15:26 +0200)]
Release 5.4.4
Tim Düsterhus [Tue, 10 Aug 2021 12:36:32 +0000 (14:36 +0200)]
Tim Düsterhus [Tue, 10 Aug 2021 12:34:12 +0000 (14:34 +0200)]
mutec [Tue, 10 Aug 2021 10:06:27 +0000 (12:06 +0200)]
use the sessions language id instead of the user's default in `I18nDatabaseObjectList`
`WCF::getUser()->languageID` doesn't return a valid language id as integer in every case. Additionally we (normally) want to use a specific language or the session's language - which might be different from the user's default in some moments.
see #4407
Alexander Ebert [Tue, 10 Aug 2021 09:05:07 +0000 (11:05 +0200)]
Release 5.4.3
WoltLab [Mon, 9 Aug 2021 16:39:16 +0000 (16:39 +0000)]
Updating minified JavaScript files
Alexander Ebert [Mon, 9 Aug 2021 16:36:13 +0000 (18:36 +0200)]
Workaround for unfreezing the screen in iOS Safari
The change to the CSS class and the properties cause a recalculation that could sometimes interfere with the page scrolling.
Forcing the scrolling into the next iteration of the event loop solves this issue by separating both actions. Since the scrolling does depend on the actions in the lines above it, this is reasonably safe to do.
joshuaruesweg [Mon, 9 Aug 2021 13:22:31 +0000 (15:22 +0200)]
Fix sorting status column in emailLogList.tpl
Alexander Ebert [Mon, 9 Aug 2021 12:00:00 +0000 (14:00 +0200)]
Skip XML that have been unchanged for 5.3 -> 5.4
Alexander Ebert [Mon, 9 Aug 2021 10:53:43 +0000 (12:53 +0200)]
Merge pull request #4445 from WoltLab/template-sandbox-foreachvars
Sandbox `foreachVars` in templates
Marcel Werk [Mon, 9 Aug 2021 09:35:07 +0000 (11:35 +0200)]
Merge pull request #4446 from WoltLab/mfa-setup-allocation-failed
Show nicer error message if allocating the MFA setup fails
Tim Düsterhus [Mon, 9 Aug 2021 09:27:51 +0000 (11:27 +0200)]
Show nicer error message if allocating the MFA setup fails
`Setup::allocateSetUpId` can deadlock if the form is submitted twice at the
same time. This error should not be normally seen by the user. If they do they
will be directed to "Try again" and then see that MFA is active, because one of
the requests succeeded. They will also receive the info mail letting them know
where to regenerate their backup codes if necessary.
Tim Düsterhus [Mon, 9 Aug 2021 09:07:27 +0000 (11:07 +0200)]
Add RejectEverythingFormField if email MFA is already active
We must not allow submitting the form in this case, because we will trigger the
assertion otherwise.
Tim Düsterhus [Mon, 9 Aug 2021 09:06:39 +0000 (11:06 +0200)]
Fix assertion in EmailMultifactorMethod::processManagementForm()
Alexander Ebert [Sun, 8 Aug 2021 09:29:26 +0000 (11:29 +0200)]
Sandbox `foreachVars` in templates
Nesting the same template inside a `foreach` loop that is also accessed inside the nested call will overwrite the values from the outer template due to identical identifiers being used.
The sandbox did not protected `$this->foreachVars` despite being stateful.
See #4431
Fixes #4444
Marcel Werk [Sat, 7 Aug 2021 09:42:38 +0000 (11:42 +0200)]
Search for not filled user option fields was not possible
Alexander Ebert [Sat, 7 Aug 2021 08:48:11 +0000 (10:48 +0200)]
Merge pull request #4442 from SoftCreatR/css-optimizations
Prefixed CSS First
Sascha Greuel [Fri, 6 Aug 2021 18:42:21 +0000 (20:42 +0200)]
Prefixed CSS First
Ensured vendor-prefixed versions of a CSS property are listed before the standardized, unprefixed version.
When multiple versions of the same CSS property are specified, the last supported one will be used due to how browsers handle fallback values. This means the order matters when using both vendor-prefixed and unprefixed versions of the same property. Specifically, the unprefixed version must be listed last to ensure standardized behavior takes precedence.
Alexander Ebert [Fri, 6 Aug 2021 16:15:55 +0000 (18:15 +0200)]
Removing inline formatting yielded empty elements and broke the selection
See https://community.woltlab.com/thread/291496-fettmarkierung-bei-einem-einzelnen-wort-kann-nicht-mehr-entfernt-werden/
WoltLab [Fri, 6 Aug 2021 12:52:45 +0000 (12:52 +0000)]
Updating minified JavaScript files
Tim Düsterhus [Fri, 6 Aug 2021 12:44:42 +0000 (14:44 +0200)]
Merge pull request #4440 from WoltLab/sourcemap-proxy
Add proxy_sourcemap.php
Alexander Ebert [Fri, 6 Aug 2021 12:41:10 +0000 (14:41 +0200)]
Merge branch '5.3' into 5.4
Alexander Ebert [Fri, 6 Aug 2021 12:41:00 +0000 (14:41 +0200)]
Prevent submitting an empty message to the preview
Tim Düsterhus [Fri, 6 Aug 2021 12:34:52 +0000 (14:34 +0200)]
Merge pull request #4441 from WoltLab/acp-logout-to-reauth
Redirect the user to ACP reauthentication instead of frontend after logout
Tim Düsterhus [Fri, 6 Aug 2021 08:04:23 +0000 (10:04 +0200)]
Add proxy_sourcemap.php
Tim Düsterhus [Fri, 6 Aug 2021 11:43:36 +0000 (13:43 +0200)]
Redirect the user to ACP reauthentication instead of frontend after logout
Tim Düsterhus [Fri, 6 Aug 2021 11:34:59 +0000 (13:34 +0200)]
Revert "Update Open Sans"
This reverts Open Sans back to the non-variable version, as the variable one
appears to render pretty badly on Windows.
This reverts commit
33c8866a790e9e6bc46358f6e82df76d2e32f56e.
Tim Düsterhus [Fri, 6 Aug 2021 10:31:18 +0000 (12:31 +0200)]
Merge branch '5.3' into 5.4
Tim Düsterhus [Fri, 6 Aug 2021 10:26:36 +0000 (12:26 +0200)]
Merge branch '5.2' into 5.3
Joshua Rüsweg [Fri, 6 Aug 2021 09:22:18 +0000 (11:22 +0200)]
Merge pull request #4439 from xopez/5.2
Fixes Bings Sitemap Howto
Marcel Werk [Thu, 5 Aug 2021 15:56:02 +0000 (17:56 +0200)]
Reactions were not displayed in article list when sorted by title
Marcel Werk [Thu, 5 Aug 2021 15:48:20 +0000 (17:48 +0200)]
Mark as read doesn't removed the badge in the mobile menu
xopez [Thu, 5 Aug 2021 14:55:45 +0000 (16:55 +0200)]
Fixes Bings Sitemap Howto
This includes the correct links if the Bing Webmaster Tools help & how-to for sitemaps.
Tim Düsterhus [Thu, 5 Aug 2021 13:44:44 +0000 (15:44 +0200)]
Merge pull request #4438 from WoltLab/5.4-fix-missing-salts
Fix having an incorrect parameter if a hash does not contains a salt
joshuaruesweg [Thu, 5 Aug 2021 12:43:29 +0000 (14:43 +0200)]
Fix having an incorrect parameter if a hash does not contains a salt
Fixes #4416
Alexander Ebert [Thu, 5 Aug 2021 11:39:03 +0000 (13:39 +0200)]
Reset the avatar cache to force the use of the WebP variants
Fixes #4424
joshuaruesweg [Thu, 5 Aug 2021 11:04:46 +0000 (13:04 +0200)]
Delete reaction type icon on delete action
Tim Düsterhus [Thu, 5 Aug 2021 09:40:52 +0000 (11:40 +0200)]
Merge branch '5.2' into 5.3
Tim Düsterhus [Thu, 5 Aug 2021 09:40:15 +0000 (11:40 +0200)]
Run SCSS prettier
Tim Düsterhus [Thu, 5 Aug 2021 07:16:18 +0000 (09:16 +0200)]
Fix typo in watchedArticleList.tpl
Tim Düsterhus [Wed, 4 Aug 2021 13:32:36 +0000 (15:32 +0200)]
Merge branch '5.3' into 5.4
Marcel Werk [Wed, 4 Aug 2021 13:27:50 +0000 (15:27 +0200)]
Some external links in the admin panel did not open in a new window
Alexander Ebert [Wed, 4 Aug 2021 12:23:34 +0000 (14:23 +0200)]
Trim the preview height using traditional overflow
`-webkit-line-clamp` is broken in Firefox and Safari once complex HTML is used.
See https://community.woltlab.com/thread/291373-rich-embeds-mit-liste-fehlerhaft/
Alexander Ebert [Wed, 4 Aug 2021 11:49:05 +0000 (13:49 +0200)]
Skip the user menu initialization if there are no items
Alexander Ebert [Wed, 4 Aug 2021 10:33:43 +0000 (12:33 +0200)]
Detect `<center>` as a block element
Tim Düsterhus [Wed, 4 Aug 2021 09:26:15 +0000 (11:26 +0200)]
Merge branch '5.2' into 5.3
Tim Düsterhus [Wed, 4 Aug 2021 09:25:25 +0000 (11:25 +0200)]
Merge pull request #4435 from WoltLab/php-ddl-diff
PHP DDL Fixes
Tim Düsterhus [Wed, 4 Aug 2021 08:57:06 +0000 (10:57 +0200)]
Take the array key into account when checking whether a column is up to date in DatabaseTableChangeProcessor
Previously updating a column that looks like this:
column VARCHAR(1) NOT NULL
to:
column VARCHAR(1) NULL
would not do anything.
Converted into the `getData()` representation of the PHP DDL API these would
look like:
[ 'default' => null
, 'notNull' => 1
, 'type' => 'VARCHAR'
, 'length' => 1
]
and
[ 'default' => null
, 'notNull' => 0
, 'type' => 'VARCHAR'
, 'length' => 1
]
respectively.
Now taking the diff of the first array against the second array (subtracting
the second from the first) will remove *both* 1 values, resulting in an
empty difference, thus believing both columns are identical.
Fix this issue by using `array_diff_assoc` which will also take the key into
account.
Tim Düsterhus [Wed, 4 Aug 2021 08:36:12 +0000 (10:36 +0200)]
Take the array key into account when checking whether a KEY is up to date in DatabaseTableChangeProcessor
Previously updating a (pretty contrived) KEY that looks like this:
[…] UNIQUE KEY someIndex (`UNIQUE`)
to:
[…] KEY someIndex (`UNIQUE`)
would not do anything.
Converted into the `getData()` representation of the PHP DDL API these would
look like:
[ 'columns' => 'UNIQUE'
, 'type' => 'UNIQUE'
]
and
[ 'columns' => 'UNIQUE'
, 'type' => null
]
respectively.
Now taking the diff of the first array against the second array (subtracting
the second from the first) will remove *both* 'UNIQUE' values, resulting in an
empty difference, thus believing both KEYs are identical.
Fix this issue by using `array_diff_assoc` which will also take the key into
account.
Alexander Ebert [Tue, 3 Aug 2021 15:17:12 +0000 (17:17 +0200)]
Incorrect handling of `null` values for metacode attributes
Tim Düsterhus [Tue, 3 Aug 2021 15:09:29 +0000 (17:09 +0200)]
Merge branch '5.2' into 5.3
Tim Düsterhus [Tue, 3 Aug 2021 15:08:35 +0000 (17:08 +0200)]
Merge pull request #4433 from WoltLab/php-ddl-diff
PHP DDL Fixes
Tim Düsterhus [Tue, 3 Aug 2021 14:33:43 +0000 (16:33 +0200)]
Take the array key into account when checking whether a FOREIGN KEY is up to date in DatabaseTableChangeProcessor
Previously updating a FOREIGN KEY that looks like this:
[…] FOREIGN KEY (fooID) REFERENCES wcf1_foo (fooID) ON DELETE SET NULL ON UPDATE CASCADE
to:
[…] FOREIGN KEY (fooID) REFERENCES wcf1_foo (fooID) ON DELETE CASCADE ON UPDATE SET NULL
would not do anything.
Converted into the `getData()` representation of the PHP DDL API these would
look identical when looking at the values only. Both span the same column and
reference the same column on the same table. The `ON …` actions are one
`CASCADE` and one `SET NULL` for both.
For this reason the diff is empty, believing that the FOREIGN KEY already
matches the expected configuration.
Fix this issue by using `array_diff_assoc` which will also take the key into
account.
Tim Düsterhus [Tue, 3 Aug 2021 14:13:58 +0000 (16:13 +0200)]
Take the array key into account when matching up FOREIGN KEYs in DatabaseTableChangeProcessor
Previously a FOREIGN KEY within the database that looks like this:
[…] FOREIGN KEY (someOtherUserID) REFERENCES wcf1_user (userID) […]
would match up a FOREIGN KEY definition like the following:
[…] FOREIGN KEY (userID) REFERENCES wcf1_user (userID) […]
Converted into the `getDiffData()` representation of the PHP DDL API these
would like:
[ 'columns' => 'someOtherUserID'
, 'referencedColumns' => 'userID'
, 'referencedTable' => 'wcf1_user'
]
and
[ 'columns' => 'userID'
, 'referencedColumns' => 'userID'
, 'referencedTable' => 'wcf1_user'
]
respectively.
Now taking the diff of the second array against the first array (subtracting
the first from the second) will remove *both* 'userID' values, resulting in an
empty difference, thus believing both FOREIGN KEYs are identical
Fix this issue by using `array_diff_assoc` which will also take the key into
account.
Alexander Ebert [Tue, 3 Aug 2021 12:36:05 +0000 (14:36 +0200)]
Merge pull request #4430 from WoltLab/js-legacy-inheritance
Resolved side effects from previous inheritance approach
Tim Düsterhus [Tue, 3 Aug 2021 12:35:40 +0000 (14:35 +0200)]
Merge pull request #4429 from WoltLab/form-field-data-handler-es6
Support ES 6 default exports for form builder field JavaScript data handlers
Tim Düsterhus [Mon, 2 Aug 2021 15:05:36 +0000 (17:05 +0200)]
Support ES 6 default exports for form builder field JavaScript data handlers
For legacy exports the value loaded by `require()` (`FormBuilderField`) will be
whatever the module author returned within the module definition. ES 6 modules
on the other hand will be passed as an object containing the exported values,
with the default export residing in the `default` key.
tslib's `__importDefault` will transform a legacy export into a ES 6 module, by
putting the exported value into the `default` key of an freshly created object.
This allows us to handle both legacy as well as ES 6 default exports identically
by simply using the `default` value of the `FormBuilderField` variable.
Tim Düsterhus [Tue, 3 Aug 2021 10:39:34 +0000 (12:39 +0200)]
Merge branch '5.3' into 5.4
Marcel Werk [Tue, 3 Aug 2021 08:56:51 +0000 (10:56 +0200)]
Merge pull request #4431 from WoltLab/foreach-empty
Fix `{foreach}` loops when the to-be-iterated value is empty
Tim Düsterhus [Tue, 3 Aug 2021 07:55:35 +0000 (09:55 +0200)]
Fix `{foreach}` loops when the to-be-iterated value is empty
The restoring of the `item` value failed, because the necessary data in
`foreachVars` was only being filled when the `foreach()` loop was actually
entered. Move this saving of the old value up to ensure it always happens.
see
75ce18bc18904d1215c7d021ac0ac18c0a7a5d42
see #4425
Alexander Ebert [Mon, 2 Aug 2021 17:06:53 +0000 (19:06 +0200)]
Resolved side effects from previous inheritance approach
This is a follow-up for
db23f8af33398c4851d6ba36436592f406b35a0d which introduced a flawed change.
The iteration over the prototype chain caused the prototype itself being bound to an object on runtime, conflicting with other objects.
The root cause was that some parts of the inherited functions were still bound to `constructed`, which was attempted to be fixed by poking the prototype chain.
This new fix is a bit weird, unless one understands that the call to `Reflect.construct()` is a bit tricky because any bound call inside the constructor of `legacyClass` will be bound to `constructed`.
This change is more of a sledge hammer approach, but it works all cases that I tested, including those that were initially the cause for the previous fix as well as new issues caused by the fix.
Alexander Ebert [Sat, 31 Jul 2021 10:03:07 +0000 (12:03 +0200)]
Sneaky whitespace
Alexander Ebert [Sat, 31 Jul 2021 09:46:23 +0000 (11:46 +0200)]
Incorrect value type when using the legacy MySQL extension
Marcel Werk [Fri, 30 Jul 2021 21:21:07 +0000 (23:21 +0200)]
Improved a11y of content items
Marcel Werk [Fri, 30 Jul 2021 17:14:02 +0000 (19:14 +0200)]
Improved usability of footer links on mobile devices
Tim Düsterhus [Fri, 30 Jul 2021 12:51:19 +0000 (14:51 +0200)]
Merge branch '5.3' into 5.4
Tim Düsterhus [Fri, 30 Jul 2021 12:45:20 +0000 (14:45 +0200)]
Merge branch '5.2' into 5.3