- #4690
- move `ModelBindingMessageProvider` init from `DefaultBindingMetadataProvider` to `DefaultModelMetadata`
- in addition to avoiding error cases, this removes some boilerplate
- add specific errors to `BodyModelBinderProvider`, `CompilerCache`, `CompositeViewEngine`, `ModelBinderFactory`,
and `ObjectResultExecutor`
- `DefaultRazorViewEngineFileProviderAccessor.FileProvider` now a `NullFileProvider` in empty case
This change to ModelBinderFactory makes the caching much more aggressive,
by caching all non-root binders. There's some trickiness here around
making sure we have the right behavior when all providers return null. See
the tests and comments.
I also kept the change I made for a temporary workaround to use a
dictionary rather than a "stack" for cycle breaking. This seems like an
overall improvement in clarity.
- #4652
- previously ignored for top-level models
- `ModelBinderProviderContext.BindingInfo` is now never `null`
- similarly, use type metadata (as well as parameter info) for `ModelBindingContext.BinderModelName`
- previously ignored when overridden in `ControllerArgumentBinder`
This change simplifies a bunch of code and fits more in line with the
current design of model binding.
Now, a model binder only has to do anything if it was successful.
'return' is enough to indicate failure.
We want this change to avoid MVC eagerly reading the form. This is good
for general perf and also for scenarios where you want read the body
yourself (large file uploads).
We DO have scenarios where you want to configure the value providers
per-request or also to change the limits on the value providers (form) so
it's worth keeping these around on the context.
This change renames IPropertyBindingPredicateProvider to
IPropertyFilterProvider. The changes here are mostly renames of
parameters/variables from predicate -> propertyFilter. I did a
find+replace and left the term 'predicate' in some of the docs because it
refers to a predicate in the abstract sense.
This change also simplifies BindAttribute and removes support for type
activation.
This commit introduces application parts as a concept on MVC.
An application part is an abstraction that allows you to expose some
feature or corncern in a way that is decoupled from their underlying source.
Examples of this include types in an assembly, emdeded resources, files on
disk etc.
Application parts are configured during startup by adding or removing them from
the application part manager available as part of IMvcBuilder and IMvcCoreBuilder.
The application part manager provides the ability to populate features from the
list of available application parts by using a list of application feature providers.
Application feature providers are responsible for populating a given feature given a
list of application parts.
Examples of application providers can be a ControllerFeatureProvider
that goes through the list of application parts, sees which one of those parts exposes types,
determines which of those types are controller types, and adds them to a ControllerFeature
that holds a list of all the types that will be considered controllers in the application.
This change no longer suppresses validation for IFormFile and
IFormFileCollection model values. This will allow the use of [Required] on an
IFormFile model, or a custom attribute for validating IFormFileCollection.
These types already have ValidateChildren = false, so we don't recurse
into them.
This change separates model binding into IModelBinderProvider (decides
which binder to use) and IModelBinder (does binding). The
IModelBinderFactory is a new services with coordinates the creation of
model binders.
- #3482
- see new tests; many failed without fixes in the product code
- add support for binding `IFormFileCollection` properties
- make `FormFileModelBinder` / `HeaderModelBinder` collection handling consistent w/ `GenericModelBinder`++
- see also dupe bug #4129 which describes some of the prior inconsistencies
- add checks around creating collections and leaving non-top-level collections `null` (not empty)
- move smarts down to `ModelBindingHelper.GetCompatibleCollection<T>()` (was `ConvertValuesToCollectionType<T>()`)
- add `ModelBindingHelper.CanGetCompatibleCollection()`
- add fallback for cases like `public IEnumerable<T> Property { get; set; } = new T[0];`
- #4193
- allow `Exception`s while activating collections to propagate
- part of #4181
- `CollectionModelBinder` no longer creates an instance to check if it can create an instance
- not a complete fix since it still creates unnecessary intermediate lists
nits:
- correct a few existing test names since nothing is not the same as `ModelBindingResult.Failed()`
- remove a couple of unnecessary `return` statements
- correct stale "optimized" comments
- explicit `(string)`
- unrelated to #3482 except that I discovered the issue while investigating that issue
- tests previously set `BindingDetails.IsReadOnly` for a `Type` and that was ignored
- same for `DictionaryModelBinderTest`
This undoes a behavior change introduced in
7b18d1d3f1.
The intent was to have ClearValidationState do the right thing for a case
where a collection was bound to the empty prefix, and then used again with
TryUpdateModel.
This change was implemented by saying that a key like "[0].Foo" is a match
for the prefix of "Foo". This isn't really right, and it's only
interesting for the ClearValidationState case.
The problem is that we don't know what the keys look like for a
collection. We can assume that they start with [0] but that's not really a
guarantee, it's a guess.
This change fixes the behavior of StartsWithModel, and move the
responsibility for this case back into ClearValidationState.
This change also removes the call to ClearValidationState from
TryUpdateModel. If you need this behavior, then call ClearValidationState
manually. Trying to bind and then re-bind a model object isn't really what
we intend.
* Broken InputFormatter into InputFormatter and TextInputFormatter
* Added an exception to ModelState inside ReadAsync on TextInputFormatter
when we can't find a valid encoding to read the body.
Closes#4007, Fixes#3622, Fixes#3621
Squashed commit of the following:
commit 4b3095671d945ae79baa4fc4ba22a2ce5ebe488e
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Mon Feb 1 15:05:58 2016 -0800
PR Feedback #4007
ModelBindingContext.ModelType. Need similar changes a couple of
other places in this class.
commit 7c45847d1d4c3eb02c48710bb4f100d4d1305385
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Mon Feb 1 15:01:13 2016 -0800
PR Feedback #4007
Please clean out the now-extra added usings for System.Diagnostics.
commit 13263fb29a0a12b472eab130eafe6760f6279262
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Mon Feb 1 14:58:52 2016 -0800
PR Feedback #4007
Confusing to read: How can overallResult be non-null here?
commit dfb52e1bd48397580eacf87be3541b159dbe79f1
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Mon Feb 1 14:47:28 2016 -0800
PR Feedback #4007
End these new guys with a period.
commit f4c745a0a805b960234cd4d810378bc40b8e5c8c
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Mon Feb 1 14:44:10 2016 -0800
Fixing unit tests
Also fixes rebase error in CancellationTokenModelBinder
commit f3874cc27388552e863d99d8e6631a2901f8e000
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Mon Feb 1 13:01:02 2016 -0800
PR Feedback #4007
Remove commented-out code.
commit 44ec1a5a3bd4fdeb4d1c4de9f0b402ff0fd34ffe
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Mon Feb 1 12:22:04 2016 -0800
PR Feedback #4007
Why now public?
commit 9aa45eda42ec9b11b7e6b73a14e887f1be8cb130
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Mon Feb 1 12:18:19 2016 -0800
PR Feedback #4007
Leave out .Test
commit 58d32d809797d25cc9b5b4c0516ae9e5992fa66e
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Fri Jan 29 15:41:14 2016 -0800
PR Feedback #4007
By making this async, we've undone an optimization here.
Notice BindModelAync now has a state machine.
commit f7503228b77803b786c19e06e60f618f6f5e15c5
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Fri Jan 29 15:30:54 2016 -0800
PR Feedback #4007
Don't need extra line
commit cf26f3e5125792d6af0ca2afda672e241fe0e164
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Fri Jan 29 15:28:23 2016 -0800
PR Feedback #4007
No longer need most of the using System.Diagnostics additions
commit e90c40c4e4315462efc9ade1585e2e8c085588a4
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Fri Jan 29 15:12:50 2016 -0800
PR Feedback #4007
Suggest Microsoft.AspNetCore.Mvc.Internal.
commit 5360599e9c09ab97b8136977ee0917fd88f51f76
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Fri Jan 29 14:57:28 2016 -0800
PR Feedback #4007
Remove this line.
commit ce63edba51721412684a54886109e6b2225c6c99
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Fri Jan 29 14:14:02 2016 -0800
PR Feedback #4007
incorrect mentions of DefaultModelBindingContext in doc comments
commit bbc738ca2304243839e8d68615ff4bbf14f3279c
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Fri Jan 29 13:59:50 2016 -0800
PR Feedback #4007
Could we convert most of these setups to avoid the
"async but not await" mess?
commit 0827e6917481e3fafb2905a3fa93cf90afb9be40
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Fri Jan 29 11:02:00 2016 -0800
PR Feedback #4007
remove the file
commit 3dd0d10d7f16c94df5f6d34ecb5f2efda62bccd8
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Thu Jan 28 13:34:31 2016 -0800
Revert "PR Feedback"
This reverts commit 589d3e65d406b127a0b833d86c05ad4a8c5172c8.
It seems the assignment of these properties is not appropriate
in both cases.
commit c8a97ada63fa16ee7df6c165031bf15b1c23e8a2
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Thu Jan 28 12:37:18 2016 -0800
PR Feedback
DefaultModelBindingContext should not be in Abstractions
commit 5e3b6a1486283f45867830693d4fd23cc967f06f
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Thu Jan 28 12:35:28 2016 -0800
PR Feedback
Is it still needed and, if so, where does the problem lie?
commit 44635a75af05034863f0d4b80b2eb857d43766b2
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Thu Jan 28 11:06:26 2016 -0800
PR Feedback
* this is either supported (shouldn't assert) or not supported (should throw). I don't feel like these asserts add much value
* this isn't something that should throw
commit 648975cad14ac5e3738ea454c490ca221f650155
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Thu Jan 28 10:54:28 2016 -0800
PR Feedback
remove the other static constructors for Task<ModelBindingResult>
they don't have a purpose anymore
commit d0e15211530668ec21406121342715b0d0f606a6
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Thu Jan 28 10:51:24 2016 -0800
PR Feedback
the reason for the separation was to put the implementation somewhere else
commit 2caf88df7a37664db829d660cbb5266cf6871f2c
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Thu Jan 28 10:48:25 2016 -0800
PR Feedback
* coding standards...
* yea
* private readonly
commit 6c2c640611686a554a8c417a407fe169c45a690e
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Thu Jan 28 10:46:12 2016 -0800
PR Feedback
* disposable -> <see cref="ModelBindingContextDisposable"/>
* Fill in the blanks.
commit df07891c798627ac9c8fce6a7598d9386e979615
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Thu Jan 28 10:44:07 2016 -0800
PR Feedback
Value to assign to the ... for modelName and model as well.
commit 8e9ade7378577eddc0ba8e939deafe8d12781b8a
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Thu Jan 28 10:43:50 2016 -0800
PR Feedback
Should this just be an abstract class rather than an interface?
commit a66685e2470a84b0acbf4e797eff5010a1a066ab
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 23:47:44 2016 -0800
PR Feedback
Suggest resetting FallbackToEmptyPrefix for current context as part
of PushContext().
commit d85e66f59d24e0959652f1683cb94c4b513d75c0
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 23:34:42 2016 -0800
PR Feedback
Add a using for the Internal namespace.
commit bef09536f881d400f15a83b7788c828903679a38
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 23:27:30 2016 -0800
PR Feedback
needs lots of /// doc love and blank lines
commit 4173488112f2889d662d5a7c55f249aefa821c22
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 23:06:49 2016 -0800
PR Feedback
Temp file accidentally added. Oops.
commit 6790d6104d5bd72ce7eab270bfe9a360cacf74f2
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 23:04:40 2016 -0800
PR Feedback
please remove the else bits after returns
commit 8d077ee27109133d18c5cddfe3b9b6c6e9a40e0e
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 22:57:38 2016 -0800
PR Feedback
The code at one point matched the comment -- always returning
something other than NoResult. Likely the comment needs
to change.
commit 0ea6c2186212444f46c7a48394b86b14190a9026
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 22:50:02 2016 -0800
PR Feedback
Don't recall why this was virtual but why remove that?
commit 0b27f3e62df6836e6e8751967c47c6192c3eec9a
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 22:35:57 2016 -0800
PR Feedback
No need for this line; just asserted that Result is already null.
commit c85648bb680bdbdb33381cd1af91133d04d56592
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 22:34:26 2016 -0800
PR Feedback
Comment needs some love since NoResult is no more.
commit a606cf32386de9e4566a678b34a195cc0720211a
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 22:18:55 2016 -0800
PR Feedback
* Long enough to start wrapping this. Same for a few other properties
in this class.
* Why is this check debug-only?
* /// docs and some blank lines between members
commit 7efce5675a78350a4a7fde4b54fcb7886fe18252
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 20:29:50 2016 -0800
PR Feedback
Why not use the Task helper used elsewhere?
commit eb5fe6f95a463eb2f162f51649ffea57db87e286
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 20:28:31 2016 -0800
PR Feedback
Sort usings
commit df3bd00c4b59d2434e3a7be28e637d3c8008ae32
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 20:27:38 2016 -0800
PR Feedback
Dead code
commit fe7ec17fbb5209f2904f067ec1f91169ab05fb33
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 20:26:04 2016 -0800
PR Feedback
Putting internal back in place
commit 496f1f97ac48b2e95572dd6d463ced3366f8d36c
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 20:21:03 2016 -0800
PR Feedback
stylecop 13040: Move these to the top of the file.
commit cd003d0bca6516a907a0ae889d2e6130ff473b57
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Wed Jan 27 19:38:15 2016 -0800
Renames to asp.net core
commit e1cf523119084e35b70f11f67f64f651c39e1c8b
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Thu Jan 21 21:21:44 2016 -0800
Removing NoResult and fixing tests
commit 5af6223a4d60b6c4139c1fc62ebf297ab0e5454f
Author: Louis DeJardin <lodejard@microsoft.com>
Date: Thu Dec 17 11:49:08 2015 -0800
Reducing ModelBindingContext allocations
This is a fix for the specific case reported in #3743 and also a few
related cases where the object being validated does not come from user
input.
Because the CancellationToken is bound using the 'empty prefix',
suppressing validation causes the validation system to suppress ALL keys
in the MSD. This will wipe out validation info for other, unrelated model
data. The same can occur for [FromServices], IFormCollection, IFormFile,
and HttpRequestMessage (anythere when MSD key == null and
SuppressValidation == true).
The other change here is related to the [FromBody]. We don't want to
create an entry in the model state for a 'valid' model with the empty
prefix. This can cause the system to miss validation errors. If you end up
in a situation where there's an entry with the empty prefix for an invalid
state, you won't accidentally miss validation errors.
* Add an UnsupportedContentType to the ModelState dictionary when no formatter can read the body.
* Add a filter to the pipeline that searches for that specific exception and transforms the response into 415.
This change removes a layer of abstraction. These validators now just do
what they do now without creating a bunch of intermediate objects.
ModelClientValidationRule has been removed, and client validations
manipulate the attributes collection of a tag directly.
Kiran Challa
Doug Bunting
Ryan Brandenburg
Ryan Nowak
mnltejaswini
Sebastien Ros
jacalvar
Pranav K
Ajay Bhargav Baaskaran
Louis DeJardin
Derek Gray
N. Taylor Mullen