This class is used heavily basically everywhere, moving it to Utils
wouldn't make much sense. Also with this change, we can move
StatusValue to MediaWiki\Status as well.
Bug: T321882
Depends-On: I5f89ecf27ce1471a74f31c6018806461781213c3
Change-Id: I04c1dcf5129df437589149f0f3e284974d7c98fa
Convert this in HTMLFormField::validate to prevent callers from
receiving unexpected Status instances that will be identified as
an error result, even if the Status is good.
Bug: T338677
Change-Id: If2208cca92c5c496eef73f25363221bc490d9c88
Some of these classes are really huge, up to 2000 lines and more. I
hope this makes the source code a little more readable.
Change-Id: I3d4b2a042a34c14e6ea0ea30ea31ca53448d8d59
While this is not really a bottleneck, I found this does quite a lot
of unnecessary things that are not used in this particular context.
All we want to know is if one or more help messages exist. We don't
need the actual, parsed help messages in this context.
Note this patch changes the code in a way that is not 100% the same as
before. Let's say there is only 1 message, and it's disabled. Since
the possibly expensive exists() check is delayed shouldInfuseOOUI()
might return true in rare edge-case situations where it returned
false before. I would like to argue this is worth it. We just trade
one performance optimization for another.
Change-Id: Ib4550e954853355f361249b13ec877478de9c789
Currently it is documented that mParent in HTMLFormField may be null.
This can happen if the form element is constructed manually via
new, instead of the normal way via HTMLForm methods.
As it stands, much of the code assumes that mParent is always set
despite the documentation. Lets mark creating form fields without
parent set as deprecated. The current situation seems like a
recipe for bugs, and after the deprecation period this would allow
us to simplify some of the HTMLFormField code.
Bug: T326456
Change-Id: Ica0740049f0a3e8ec764903c5b71825e4d628a3f
Depends-On: I15a39605e3eec8a5c265c4a331039fa906eda036
When a form identifier is set, HTMLForm::prepareForm() initializes
mFieldData as an empty array. When resolving hide-if conditions based on
other fields, HTMLFormField looks up that field's value in the parent
form's mFieldData and expects its key to be set, but it isn't because
mFieldData is deliberately empty. That causes an E_NOTICE for an
undefined index.
Resolve this by having HTMLFormField::getNearestFieldValue() fall back
to the field's default value when the field's actual value isn't found
in mFieldData.
Change-Id: I3a12f6aec1e8a6da3cd3a29195d37c12ebf08b16
null is the “default default” in getDefault() (if the method has neither
been overridden nor $this->mDefault been set), so it can appear in the
field data. Don’t pass it to the validator (which may not expect a null
value) if the param is supposed to be required.
Bug: T319219
Change-Id: Ib6bb119b841ffcbb5e7fb30014355dc90ba9b46e
Callbacks for validation-callback definitions are given the submitted
field value, which was documented as being string or array, but
actually can include int, null, etc.
This fixes the docblock, and also updates
DefaultPreferencesFactory::validateSignature() which was assuming
a string and not expecting null. (This didn't matter before PHP 8.1.)
Bug: T318307
Change-Id: Ia9096d610bf377334bbeab9021a8ade9be62edd5
Various variables are left from ealier refactor are now unused
and can be removed to make the code easier to read
Change-Id: Id51770af1f08e85c7e7a02234a2cd2ab5b47ee7a
* Move the reset preferences link from the bottom of the preferences
form to the "basic information" section.
* Change the link from "destructive" red to default styling, since the
action requires confirmation so there is no reason to make the user be
afraid of clicking it by accident.
* Add a checkbox to the /reset form, for triple confirmation.
* Add a cancel button to the /reset form, to take the user back to
safety.
* Allow checkboxes to be "required" by fixing a detail in
HTMLFormField::validate(). The UI is not pretty, but it works.
Bug: T226325
Change-Id: I116d5275ba1a5beaaa44b32b8eff5824e94b437a
It gets wrapped into a Widget in ::getOOUI
Found by phan strict checks (used by tagfilter and cloner)
Change-Id: Ie1fd15fc2a66fcb39fa2e2f63761d9683029e395
This reverts commit a745a87216.
Reason for revert: logspam on use cases out side the scope of HTMLForm
Bug: T302512
Change-Id: Idcf1c7db89b4456308d57915ffd0dcc41ee577ad
The functions returning null or the class property is set explict null
Found by phan strict checks
Change-Id: I4a271093fb6526564d8083a08249c64cb21f2453
After its logic is refactored, we have more faith to use isDisabled() in these cases, which matches the original semantics of isHidden().
Change-Id: I1aece7c691307ac8bb3770e6dc4b8e2877e4d65c
This is a follow-up of change Ic414e7bb7933c8c5c58a277ac1c5c3aaf8c36119.
I noticed that HTMLForm would add a hidden field of edit token for forms
that require to be posted...
And I also noticed that there is a crazy use case (9ca8394943/includes/specials/SpecialTags.php (320))
that would break with the loose check (I'm also going to fix it by I5050311c37030a64daaa25d05e2223485ed86108).
So that change should be undone, but keep the POST check here to avoid
arbitrary token in a GET form.
Add more comments since there are some preconditions in another file and
avoid future regression.
(And I hope this is the last time I fix this test...)
Change-Id: Ib02cd6b4a45e3820c4378fe5b0c7fc61fe1251e7
This can help check fields default to true or have 'invert' enabled to
load correct data, because the 'false' values of checkboxes can never
be included in the request of normal use cases. Also better to other
use cases in HTMLCheckMatrix and HTMLMultiSelectField.
This partially reverted I5589ba7383587afdd9307c79e88849dacee02706.
I get trapped by the test case... The default value does need an invert
to check for hide-if and disable-if, like what
HTMLCheckField::getInputOOUI() do for display.
Depends-On: Icfa062eada75c50cd2c8bc5db2930602d80e9ae7
Change-Id: Ic414e7bb7933c8c5c58a277ac1c5c3aaf8c36119
Old codes relied on some assumptions, so:
* When a "key in form descriptor" is provided, the 'name' param can't be set otherwise the js wouldn't be able to find it.
* When a "name for submission" (with 'wp' prefix or same as the 'name' param) is provided, it works on the client-side but something might be broken on the server-side.
Since the documented usage is to use "key in form descriptor" and most use case is fine, the use of "name for submission" is explicitly disallowed here.
Use cases simply with the 'wp' prefix would still keep working on both sides though.
Depends-On: I27fd8fa9643d611b37e3f47e77b698245814d539
Change-Id: I9a42417a6161f42181badd8cdbec81ba85dc62f6
Don't catch and discard exceptions from the RequestTimeout library,
except when the exception is properly handled and the code seems to be
trying to wrap things up.
In most cases the exception is rethrown. Ideally it should instead be
done by narrowing the catch, and this was feasible in a few cases. But
sometimes the exception being caught is an instance of the base class
(notably DateTime::__construct()). Often Exception is the root of the
hierarchy of exceptions being thrown and so is the obvious catch-all.
Notes on specific callers:
* In the case of ResourceLoader::respond(), exceptions were caught for API
correctness, but processing continued. I added an outer try block for
timeout handling so that termination would be more prompt.
* In LCStoreCDB the Exception being caught was Cdb\Exception not
\Exception. I added an alias to avoid confusion.
* In ImageGallery I added a special exception class.
* In Message::__toString() the rationale for catching disappears
in PHP 7.4.0+, so I added a PHP version check.
* In PoolCounterRedis, let the shutdown function do its thing, but
rethrow the exception for logging.
Change-Id: I4c3770b9efc76a1ce42ed9f59329c36de04d657c
Values loaded from the default wouldn't be inverted, there is no need to copy the conditions to here, we can just simply bypass them.
Fortunately, nothing can be affected in a normal use case, since all value of fields would be set to server.
Thanks to the newly added tests, which helped me to realize this problem.
Change-Id: I5589ba7383587afdd9307c79e88849dacee02706
Follow-up changes will use cond-state params in other functions, split this out to make sure it's validated.
Change-Id: Icf358794b11a8f986fbd02c8f1b15ea9d1ef4d15
I believe there is a mistake, since DefaultPreferencesFactory::cleanSignature() uses it as field data, which is the only filter applied in the core.
Change-Id: Ic7aa509a3e5fd3a3c717259d83d5bf0a26d3556a
Data of disabled fields wouldn't be sent to the server, which needs to load data from default.
Bug: T298614
Bug: T298819
Change-Id: I58f9df384df8ecc5ebae8cac68ec2251351bc984
There is only one use case of cond-state classes passed to the client-side, and it can be done more natively.
Change-Id: Ib6835bb334bd65c6176f0a5c3881b20265901af9
Array addition in PHP does not renumber keys, so it can't be used to get
the union of two lists.
Bug: T297975
Change-Id: Ic9932b23e32c704b14fc9fe234c35a48b59767c5
