What:
Add hook that runs after a save attempt is made in ApiVisualEditorEdit.
The hook receives the same data available in ApiVisualEditorEdit, and
implementations of the hook can modify the API response.
Also introduce templated
parameters (https://www.mediawiki.org/wiki/API:Templated_parameters) in
the API parameters; this allows plugins to pass arbitrary data along
with their request using e.g. plugins=linkrecommendation&data-linkrecommendation=foo
Add ServiceWiring files, a PHP namespace, and a HookRunner class to
support the above changes.
Why:
VE plugins may wish to send additional data when saving an edit and take
action based on that data on the server-side. See for example the
AddLink plugin in I7a052f8e which sends annotation data, and then uses
the new hook to perform a database operation.
Change-Id: I392691475fbdcec766acbd832600e82efcb5bfe8
Otherwise, the global context is used (RequestContext::getMain()),
which is undesirable when you're building a rubegoldbergian
contraption and we're already inside an internal action API request
with a fake context.
Change-Id: I66e79e641eda185f7af2561d3655c92cba762135
Applying the changes from the MediaWiki core patch for T120883
(Ife272a0eb1f3322bc8eb30ca803bd21801acba3e) to our duplicated
code implementing the same functionality.
Bug: T270453
Change-Id: I1b2de322aa0c69eb6d3b3ffadaed3fbaa3a58bca
User::setOption() is deprecated and should be replaced with UserOptionsManager::setOption()
Bug: T277818
Change-Id: I7395a8620064dcbe019b92feba17d9c9996ffbb7
Remove using of User::isIP since this method will be hard-deprecated. Now it is soft-deprecated
Bug: T275602
Change-Id: Ia625be523706d1e24649f7aa15679491f9598b7f
The failure was:
includes/VisualEditorDataModule.php:37 PhanTypeMismatchArgument Argument 2 ($pretty) is ResourceLoader::inDebugMode() of type int but \FormatJson::encode() takes bool|false|string defined at ../../includes/json/FormatJson.php:115
includes/VisualEditorDataModule.php:41 PhanTypeMismatchArgument Argument 2 ($pretty) is ResourceLoader::inDebugMode() of type int but \FormatJson::encode() takes bool|false|string defined at ../../includes/json/FormatJson.php:115
Probably triggered by changes to MediaWiki in commit
Ieaf04e0c289646dd5d5b027b4f1f8278167b2d57.
Change-Id: Iab139c4dd69c9a16c9f4f9e2fd47ec76a9003152
If someone has edited before the switch (2016-ish), never edited since
then, and try to edit again now, it's probably no longer helpful to
show them a popup dialog about a change that happened 5 years ago.
The immediate motivation for this, though, is T273189. We discovered a
bug with the preferences that, among other issues, caused this dialog
to not be shown in some cases where it should have been, and fixing
the bug now would cause it to be shown the on the next edit attempt.
If someone has edited before the switch (2016-ish), *and* edited since
then, and try to edit again now, it's definitely not helpful at all to
show them a popup dialog about a change that happened 5 years ago.
Bug: T273189
Change-Id: I4b5a3d8dbdf1c853eb39fcfc85a9fe87a4db0f21
We have two preferences used for enabling/disabling VE:
'visualeditor-enable' and 'visualeditor-betatempdisable'.
(And 'visualeditor-autodisable', which sometimes overrides
them both, but it's not relevant here.)
The user can only set 'visualeditor-enable' when VE is Beta Feature,
and they can only set 'visualeditor-betatempdisable' when it is not.
However, when deciding if VE should be loaded, we always checked both
of the preferences. This gives incorrect results when the preference
that is not supposed to be settable actually exists, due to being set
in GlobalPreferences on another wiki where it's settable.
A similar issue could occur if VE configuration is changed from Beta
Feature to a normal preference (or the other way around), and the
preference values for the previous configuration still persist.
Bug: T271434
Change-Id: I7399b3c516f762429050a662ac85d1d392392323
Depending on the config, we would sometimes define both the
'visualeditor-enable' and the 'visualeditor-betatempdisable'
preferences, even though they contradict each other.
Bug: T273188
Change-Id: I6ac7e4580bf232a5198ad8842f814bc381131e00
* VisualEditorNewAccountEnableProportion
* VisualEditorAutoAccountEnable
Seemingly never used on Wikimedia wikis, and untested since 2016.
Bug: T273177
Change-Id: Ic132ef2091399fd223626ae307830f278e72ec01
This name is consistent with T91820.
Bug: T91820
Bug: T259685
Depends-On: I0c4ec63bb26641b237c92dbd3bc5367811ca0675
Followup-to: 3561167493
Change-Id: I4c4c5f83ad56a198d08095d629a6ba86ce9dc1a4
A couple of things make this quite a simple change:
* Most messages are generated from a single message key
* $title->getEditNotices() already returns an array
keyed in the same way, we just currently flatten it.
* The VE client already flattens the array immediately
in ArticleTarget#parseMetadata, so no change required.
* PHP preserves the order in which keys are added, not
that notice order is particularly important.
This partially undoes c824dc383a,
which changed the format from a mixed array to an indexed array.
Bug: T272188
Change-Id: I646667fe2513e371e0c8270761553253d1abe2b6
isRegistered is part of the slick UserIdentity interface, i.e.
it's the more "canonical" form. This change makes it a bit
easier to move away from using the huge (4000+ LOC) User class
everywhere, in favor of the UserIdentity interface, where
possible.
This patch is meant as a small step towards this goal. I tried
to replace some usages of User type hints already, but prefer
to go in small, incremental steps.
Change-Id: I827b83a5304b1975437d5fd5083f2877dba6f6d8
Although wikitext is (expected to be) in Unicode Normalization Form C,
the output HTML may not be, due to the presence of explicit entities in
the wikitext representing non-NFC codepoints.
Bug: T266140
Depends-On: I2e78e660ba1867744e34eda7d00ea527ec016b71
Change-Id: I0d34c9a01f1132c2616ed3392ea40d8b73e15325
The hook should work now.
I tested it manually and preloading with extensions works fine.
Preloading the default text for system messages is fixed as well.
Bug: T266404
Change-Id: Icf40112bd5fabf24452f9b807b388399c7c920f0
This patch starts using watchlist related values from ApiEditPage
results instead of updating the "watch link" based on whether the
checkbox was selected or not at the time of saving the article.
This change does not depend on T261030 and can be merged without it
but T261030 needs to be fixed or temporarily watched items will not
display the right tooltip when hovering the "watch link" or star icon.
Bug: T260434
Change-Id: I2c844223620d7d28f36a0cd8ae3dee4b0c8ae5bf
We'll update this before 1.36 is released, unless we get Parsoid
integrated "properly" before then.
Change-Id: I92d8555b1f5dc121c5f596b5cb6d59414280388f
Parsoid always enables `<ref>` processing, but our Cite extension
implementation tries to fetch $wgCiteResponsiveReferences, which won't
be set if the Cite extension is not installed.
Change-Id: Idde8af07e5bf40983b2ec878ebf70aabb522a800
(cherry picked from commit 0ca4ae6908b626d34f8445d9048342378d0e3c23)
Currently we always register VE as a Beta Feature, and then
expect users to use $wgHiddenPrefs to hide it, hackily.
Also, set this new preference to false so that 3rd party
wikis don't show the BF by default.
Bug: T254349
Change-Id: I92fe3d44bb4d762ca7b1bc693b7d2e74367c84ec
getDefaultMessageText() returns false when the message doesn't exist.
It was incorrectly treated as a string by this code.
Follow-up to 3ec5e04a37.
Change-Id: Ib94a30697bd20133e46cbd31c14caf5f0c4169dd
Set the 'forwardCookies' flag if the wiki is private (not everyone is
allowed 'read' access).
Bug: T260201
Change-Id: I0b958e8b75c04e4a27f50f91276be221a5b1404d
The output is assumed to be the same as you would get if you
fetched the page again, so this output-modifying hook needs to run.
Bug: T258980
Change-Id: I015ac183a0c25dafb9b95c577edd4ef59c112d43
Previously, when the editsection link of the section edit links had
a class defined (either by the skin or through an extension), the
mw-editsection-visualeditor class would not be added, as the + operator
on arrays does not merge existing entries.
The side effect of the mw-editsection-visualeditor class missing is
that the initialization code in ve.init.mw.DesktopArticleTarget.init
would add an additional link to the section edit links, resulting in
three links shown when VisualEditorUseSingleEditTab is false.
Change-Id: I4b25c63884fa367fb0e44c61323a19a71fb6b9d8
* Remove postHTML/postData separation, not very useful.
* Pass $oldid instead of full params object.
* Return full repsonse like other methods, instead of just body.
Change-Id: I0fe301475adeb0fd5952809c3f74f0cb19f89cfe
This copies the Parsoid extension code into includes/VEParsoid
to allow a "one extension, zero-configuration" install of
VisualEditor for MW's LTS release. The Parsoid code has been
re-namespaced (`VEParsoid` instead of `MWParsoid`) to avoid
autoloader conflicts if you actually install Parsoid as an
extension (as we do in Wikimedia production). Similarly, we
arrange that the ServiceWiring and RestRoutes configurations
are skipped unless running in zeroconf mode, to avoid
conflicts with the Parsoid extension.
This import matches Parsoid commit b30f223.
Bug: T248343
Change-Id: Ic63ce40f59c4be8f4fdc5f9ac17798353fc86866
This allows Special:ApiSandbox to display multi-line text boxes for
them, which makes testing the API easier.
Change-Id: I10541a8e9033d81740033da80d842f58d1d3e0de
Calling the linter is very slow as the result is not
cached. Extensions needing this (DiscussionTools) should
just call the Linter extension API directly.
Bug: T253799
Change-Id: I994b52ca70c29a32900741a36087f10144396720
The current default for Message::__toString() is to parse, so this
keeps that behaviour unchanged.
Change-Id: I528a60c1d9c0f8c1596a15e06a764200a2b2565f
This allows it to respect the proposed limit on HTTP request timeout.
Bug: T245170
Depends-On: I8252f6c854b98059f4916d5460ea71cf4b580149
Change-Id: I1c3d96720709253ad15bb8528cdd132571de2e4e
Goals:
* Allow other extensions to reuse these methods (maybe upstream them
to MediaWiki core later)
* Allow ApiVisualEditorEdit to extend ApiEditPage. We'll be able to
reuse its definitions for API parameters instead of duplicating
them, and we won't have to pass around unrecognized parameters.
Bug: T252573
Change-Id: If5c8d95560cbb078ae4980f4a912cbaeafe53d3e
We go through all this trouble to pass the config to the API modules,
and then we don't use it at all (we removed the uses recently in
ce094c72d and d85d30f9b).
If we end up needing the config there again, we can just get it by using
MediaWikiServices::getInstance()->getConfigFactory()->makeConfig( 'visualeditor' )
anywhere we want, like we do all over the place in VisualEditorHooks.
Change-Id: I9d254a9946f0d24783baf68c409b10291a8fd1b3
While we pretend that the ConfirmEdit CAPTCHA support is added by
ve.init.mw.CaptchaSaveErrorHandler in the ConfirmEdit extension,
we still have a bunch of code here required for it to work.
This commit removes some of it, no longer needed after
I6605017fd31a4f96c529dd0beb69e9f4433cebc1.
Depends-On: I6605017fd31a4f96c529dd0beb69e9f4433cebc1
Change-Id: I41e032fd754927b7ea6cfb767eb9f21b522ccacd
The client-side code sends the empty string as the value of this
parameter when the edit should be marked as minor, and doesn't set it
at all when it shouldn't (this is the normal convention for boolean
parameters). However, the API code here was incorrectly handling the
empty string, and not marking the edit as minor as a result.
This was revealed by 50883dd7fe. Prior
to that change the original 'minor' parameter was forwarded to the
action=edit call if it was provided, overriding the 'notminor'
parameter, which was effectively added unconditionally.
Bug: T248257
Change-Id: I37fd73851d94906d79943692fb9136da03ea95fa
Our client-side code doesn't use them, and they should not be needed.
ApiVisualEditor still returns those, and they are needed there; this
method used to be shared by both classes until 25afae342a.
ApiVisualEditorEdit still has 'basetimestamp' and 'starttimestamp'
input parameters.
Change-Id: I7108468df2ff20f05bd141cb617e4e0f90c027ce
This can lead to problems if the names in the VE API are
still used inthe Edit API, as previously happened with
`watch`/`watchlist`.
Explicitly list the parameters from VE API that were
previously getting passed to the Edit API automatically.
Change-Id: I8fe705b178af82d8285067909168d32cfb9a0421
The code for setting 'watchlist' in the EditAPI request
was completely broken as it always evaluated to 'unwatch'.
Instead pass through 'watchlist' directly from the client
where it must be set to 'watch' or 'unwatch'.
Bug: T245579
Change-Id: Ia5a2bb76ef35a685b39bcc0c4727796acd0f510d
It turns out anonymous users can't apply change tags, so change
I2c1d0f8d69bc03e5c1877c790247e165f160e966 broke editing for them.
Bug: T242184
Change-Id: I7c27e4d9995428e213a980819810f235fdfe9435
In the core MediaWiki editor, the detailed messages about page
protection are only shown when the user is allowed to edit the page;
otherwise, a generic permission error is shown. Do the same here.
Change-Id: Ia0ca52b9bf556354218b2aa91f141b429b0c5880
This opens up the API so that other tools can use it without being
forced to tag those edits as being from VE.
Also, document that tags is a working parameter that can be passed
through to the edit API.
Bug: T242184
Change-Id: I2c1d0f8d69bc03e5c1877c790247e165f160e966
Replace Title::userCan with PermissionManager::userCan.
Replace Title::getUserPermissionsErrors with
PermissionManager::getPermissionErrors.
Change-Id: I1afec4ba62185c3cd555a10ae35cef01b7194221
In many places we check whether VE is available before doing things
(init.isVisualAvailable). This variable includes checks for whether
the page is wikitext, etc. Now it will also include checking whether
VE is enabled in user preferences.
In almost all places where init.isVisualAvailable is used, we were
already also checking if VE is enabled, so this doesn't affect the
behavior. But notably, we didn't do it when showing the option to
switch to VE in the welcome dialog and in the toolbar, causing T243723.
Changing init.isVisualAvailable this way makes it consistent with
init.isWikitextAvailable, which has always included a checking whether
NWE is enabled in user preferences.
Bug: T243723
Change-Id: Ie174bc3f16bceb29cb155b9223e0acef70167fd6
* Query permission errors for 'edit' and 'create' in the same way,
and consistently with how MediaWiki core EditPage does it.
* Prefer using Title::getUserPermissionsErrors() instead of custom
code for user blocks. The result is the same, except for an extra
line like "You do not have permission to edit this page, for the
following reason:".
Note that this loses the 'type' => 'block' property on each notice,
which was previously used to track when a block notice was shown.
This was however removed in 96de1353d3,
and could probably be re-implemented by using the root 'blockinfo'
property anyway.
* When Title::getUserPermissionsErrors() returns multiple messages
(for example, you're blocked *and* the page is protected),
display them all in a list instead of only the first one, using
OutputPage::formatPermissionsErrorMessage().
That method returns wikitext, which we have to parse ourselves. This
is a bit silly, but I found this approach in SpecialChangeContentModel
in MediaWiki core, so it should be fine.
Change-Id: Ifaf95d8aab836e45665b1fbdf98dd1980a867d8c
The message ends with "The latest log entry is provided below for
reference:", but we were not providing it. Use the same code as
'protectedpagewarning' and 'semiprotectedpagewarning', which behave
the same.
Change-Id: Ibe5463aa3d93cd1d6d6e3c0b9da82bfa2c813f86
The following sniffs are failing and were disabled:
* MediaWiki.Commenting.FunctionComment.MissingDocumentationPrivate
* MediaWiki.Commenting.FunctionComment.MissingParamName
* MediaWiki.Commenting.FunctionComment.MissingParamTag
Change-Id: If59d450236a7d1afa3c6a05536abea819535f984
I'm not sure if this qualifies as an actual fix. I don't really know
what this code is supposed to do. Maybe the bad array access is a hint
for a completely different error somewhere else. All this patch really
does is silencing the bad array access.
This line of code was introduced with I1b35b28 in May already. Why does
it start failing now?
Bug: T240961
Change-Id: I5ec2dc453e5d8db5d2f9e49058eda001ed021d3d
I'm trying to track down possible values of 'result' for client-side
error handling and this was confusing.
Change-Id: I325249f7c57936c9c11a1d3dc9166b1ca3737c39
Follow-up to 57ad605dc4.
Pass parameters like when switching from wikitext editor:
* bodyOnly=false
* stash=true
Bug: T233320
Change-Id: Ied2d9a48e884e033ef9d41b2da9cfa3599784ae0
Previously we were not returning it, so when saving the edit, wikitext
syntax would not be preserved. This was probably not a big problem,
but I noticed it coming up in the logs for T233320.
Now making an edit starting with preloaded content behaves like
switching from wikitext to visual mode, rather than like starting the
edit in visual mode.
Similar to 679e777cfa.
Bug: T233320
Change-Id: Id1ee6877b103fa4274deec11b1b3cacbdcdae606
The configuration of Wikimedia wikis should be fixed.
This reverts commit 04407e9eb1
and also makes similar changes to new logging code added in
5d1a67757a.
Bug: T234564
Change-Id: Ic999b050e68b71f5a1737366e16f133e5a557307
Something is causing the 'ETag' headers produced by the "public"
RESTBase (queried directly from the client) to be mangled or lost.
My theory is that some proxy or browser extension is doing that.
When we detect a bad etag when fetching the page contents, discard
the result and try querying the "private" RESTBase via the MediaWiki
API (similar to what we do on private wikis, except there we talk
directly to Parsoid instead of RESTBase). After I463a84de63, that
returns the etag as part of the payload rather than HTTP headers,
and should pass unharmed through whatever is mangling the data.
Also compare and log the two etags.
Bug: T233320
Change-Id: I2ef0ca872597566f74b650aea71bf3f15747a6d7
For consistency, I guess. Also I need this in I2ef0ca8725.
Previously, when querying the HTML content of an existing page, we did
not return the 'etag', on the assumption that anyone who needs it will
instead query RESTBase directly.
Bug: T233320
Change-Id: I463a84de631598243893946ad1d060a9aa0b180e
The 'eye' icon is in the 'accessibility' pack, not 'alerts'.
Compare to the dependencies of 'ext.visualEditor.diffPage.init'.
Change-Id: Ie14ab6be756fd9e0bef59475466674a41273046f
No longer needed after Id36aa6bdb8f873fe7deb8123a7fc774103721c01,
which teaches SpamBlacklist to return its own error messages.
Depends-On: Id36aa6bdb8f873fe7deb8123a7fc774103721c01
Bug: T211443
Change-Id: I462c2002f9b596cbdfd7ce3673d4e362e5bd1aaf
Unnecessary suppression causes build failures. Probably fixed by the
EditPage.php changes in e2e543f7c2a98f40c9b43ba3989d0f6689f4cb67.
Change-Id: I2ac7e95886ce6bef2ba08e1614728caae7d26442
Context items can be created for specific template titles. Titles
are mapped to context items using an on-wiki message.
Bug: T211243
Change-Id: Icfc39e350452da238d0e0c17cb2305c60d9ca16a
New changes:
1a7460058 Remove ve.newMobileContext feature flag
Local changes:
* Remove ve.newMobileContext feature flag
Change-Id: Ia8def997b7cba4623866080752b06068d2118cc3
* Change the query in ve.init.mw.ArticleTargetLoader#requestParsoidData
so that in non-RESTBase mode with wikitext it still returns the
metadata required to initialize the editor, using the backend API
code added in I1b35b28e428a1f86d2e34d90ddbe73361ce14818. This fixes
the exception from T222312.
* Introduce new configuration option $wgVisualEditorAllowLossySwitching
to control this feature. It is enabled by default, fixing T214542.
We allow it to be disabled because switching in non-RESTBase mode may
cause "dirty diffs" (non-semantic changes to the wikitext), which are
undesirable on wikis where users carefully review all changes.
Bug: T214542
Bug: T222312
Change-Id: I58879cba5612002c70c24731306214d2577c2c52
In general action=edit could be bound to a wikitext-specific
edit link, but in the case of redlinks we can use the
preferred editor instead.
Bug: T223793
Change-Id: Ib0851e9e2ce441ae93311153801e2c3de0a2063d
We don't currently support it in NWE. It has a very different design
for previews that wouldn't really do what users expect. Let the old
editor handle this.
Bug: T195914
Change-Id: If0c0312347c212447bd8da7336c80bd4a1cb246a
There are various circumstances where the wgIsProbablyEditable check
gives incorrect results (hence the 'probably'):
* User is blocked (T111217)
* Page is protected from creation (T173763)
* Page is transcluded on a cascade-protected page (T217217)
Bug: T111217
Bug: T173763
Bug: T217217
Change-Id: I7df8909c31f29d2e7521bef8612c27cb61146a4d
Prior to 80bfbfc54b this worked by
accident, and with a number of bugs depending on your settings (see
T219457). It turns out that Wikipedia users have invented various
workflows that depended on this bug (mostly involving sandbox pages in
namespaces where VE is not enabled). Restore it as a supported
feature, and in a way that avoids the problems it previously caused.
Bug: T221892
Change-Id: I62714b6f2905efd1d1b34c7a13b9917cb6c609fc
Use an API 'parse' call with a sectionid to trigger
a full document expansion using replaceSectionAtRev.
Then send this off the RESTBase to convert to HTML
and statsh.
Ensure the etag is passed back to the API response.
Bug: T117716
Bug: T223023
Change-Id: I1b35b28e428a1f86d2e34d90ddbe73361ce14818
This is necessary to allow the unmitigated removal of
ApiQueryUserInfo::getBlockInfo to be fixed. See discussion on
I84ed21641c44b2f65.
Change-Id: I9f40666a31bd4af50762c197c2ce5bf089a5e68c
The method ApiQueryUserInfo::getBlockInfo() was removed in unannounced
breaking change in MW core: I84ed21641c44b2f65ebe1980b0893d1846db3b34.
Apparently we're supposed to use the method from ApiBlockInfoTrait now.
Bug: T209599
Change-Id: I7ab5492310980b1527c7329faf65655330b8bef0
RESTBase is changing the way it is storing HTML/data-parsoid renders. In
order to still support VE (and other editors) with quick access to HTML
and matching data-parsoid (needed later for transforming the modified
HTML to wikitext), VE now needs to let RESTBase know it intends to
perform a transformation call later on. In that case, RESTBase will make
sure to keep the matching data-aprsoid around for long enough.
Bug: T222639
Change-Id: I02672e29bd0f331794fd77d9e56f9cc6822d9b9e
Is replaced by MediaWikiServices::getInstance()->getConfigFactory().
Also, did a few minor cleanups where necessary such as; objects are
passed as references by default etc.
Change-Id: I42cd242ebdbc0b091a99e771289020d498bf2bba
Mobile builds its own block message from the block information,
rather than using parseAsBlock.
Bug: T209599
Change-Id: I8b200f258b50f7048aded22ec7ab81a49937c5a9
This is a clean up after collecting the necessary data related to
blocks and how often users see the block notices
See: https://phabricator.wikimedia.org/T189724
Bug: T214214
Change-Id: I532a0cd95009109ba25caa8dd31badd5c1900da7
This check currently requires LCStore, MessageCache, and (sometimes)
Database to involved to check whether the message and/or local override
exist.
Using `useDatabase(false)` should take away most of this cost
by no longer performing the multiple Memcached/Database roundtrips,
and leaving only a cheap in-process check on LCStore.
Bug: T221294
Change-Id: I6bf47cd84cdf9bfdd63bee0a613425bb79595e4f
The latter is being deprecated as of I74a9535918e because it was
almost never used intentionally. When module objects are created,
the appropriate context object is injected via setConfig. That is
the one the modules should use.
The context object has a reference to the ResourceLoader object
(although unsure why actually), but shouldn't be used for this
purpose as there could be a 1 to many relationships further down
towards modules.
Change-Id: Icab0f12141a46476618f984d4548a82fdae33275
For a while now, the 'ResourceLoaderRegisterModules' hook is
the last oppertunity to register modules. Therefore, the
isModuleRegistered() check covers everything it needs to.
In addition, at this point modifications to ResourceModules would
be ignored even if it did contain additional entries.
Change-Id: I77714fca0f561f5817a45dd3be5fd8d3ba42f969
These do not vary by user or page, and can thus be loaded asynchronously
via the startup module, rather than blocking rendering and fetching
of modules on all pages.
In a future change, it might be better to go a step further and bundle
these with a module so that they only load as-needed instead of still
on all page views, but this should be an improvement nonetheless.
Change-Id: Icae3712ac5546a90bc7ffd787b0f3285dff6a26f
