I have an ASP MVC 3 application and in my Model I have implemented IValidatableObject.
When my controller posts for a create or edit, I obviously only want to save the model if it is valid.
I see many blogs and posts and answers that say something like
if(!ModelState.IsValid)
{
return View();
}
My question. Why is it that ModelState.IsValid is always true in a unit test on the Controller?
Example:
[Test]
public void InValidModelsAreNotAdded()
{
var invalidModel = new MyModel() { SomeField = "some data", SomeOtherField = "" };
var result = _controller.Submit(invalidModel);
_repository.AssertWasNotCalled(r => r.Add(Arg.Is.Anything));
}
Model code:
public class MyModel : IValidatableObject
{
public string SomeField { get; set; }
public string SomeOtherField { get; set; }
public IEnumerable Validate(ValidationContext validationContext)
{
if(string.IsNullOrWhiteSpace(SomeOtherField))
{
yield return
new ValidationResult("Oops invalid.", new[] {"SomeOtherField"});
}
}
}
The AssertWasNotCalled always fails this test.
I stepped through the test and noticed that the ModelState.IsValid is true for this test. It is as if the IValidatableObject.Validate is not being invoked. It seems to work when I run the project, but thats not much of a way to test drive an application.
Also, I realize I could use the [Required] attribute for my example, but my real code has much more complex validation to it.
Thoughts?
It's true because you haven't called anything which sets it false.
This normally happens during binding, but since you just pass the model directly in the test you skip that altogether.
If you're trying to test validation, do that directly. If you're trying to test the error path in your controller, your test's arrange can call _controller.ModelState.AddModelError( //...
Well, insted of simulate the model binding behavior you can do that:
public class YourController : Controller
{
//some code
public ViewResult someAction(Model model)
{
try
{
ValidateModel(model);
}
catch
{
// deal with errors
}
}
//some code
}
ValidateModel with "try catch" blocks are much more readable for me. But you can still use "if" blocks with the method TryValidateModel
Hope that helps!!
Related
I am trying to call a WebAPI method from Angular 5 like this:
selectClaims(state: DataSourceRequestState):Observable<DataResult>
{
return this.http.get<GridDataResult>(`${this.apiUrl}/SelectClaims?${toDataSourceRequestString(state)}`);
}
Which calls the API method as expected. The API method is:
[Route("SelectClaims")]
[HttpGet]
public IHttpActionResult SelectClaims([FromUri][DataSourceRequest]DataSourceRequest ClaimsRequest)
{
if(ClaimsRequest == null)
ClaimsRequest=new DataSourceRequest { Page=1, PageSize=20 };
var result = _db.Claims.ToDataSourceResult(ClaimsRequest, c => { c.SortHistory(); return c; });
return Ok(result);
}
The trouble is that ClaimsRequest only de-serializes Page and PageSize correctly. Filters and Sorts don't come through:
Fiddler tells me that the URL from Angular is:
GET /api/v1/Claims/SelectClaims?filter=id~eq~2&page=1&sort=firstName-asc&pageSize=20 HTTP/1.1, but in the controller both filter and sort are null.
If I create a URL through Swagger like: 'http://localhost:50223/api/v1/Claims/SelectClaims?ClaimsRequest.page=1&ClaimsRequest.pageSize=11&ClaimsRequest.sorts=firstName-desc' I do see a sort array in the API method, but the "Member" field is null.
Any attempt to add a filter through Swagger like 'http://localhost:50223/api/v1/Claims/SelectClaims?ClaimsRequest.page=1&ClaimsRequest.pageSize=11&ClaimsRequest.filters=id~eq~2' results in a "Cannot create an instance of an interface." error.
The state is a DataSourceRequestState in the angular component from a Kendo Grid for Angular.
I have simulated this in a simple test program and everything works fine there. The only difference in my test program is that the API controller targets .Net Core and the real system targets .Net 4.6.1.
Do I have to de-serialize manually in .Net 4.6.1 for some reason, or is something else going on here?
It should be a POST not a GET. Something like this:
return this.http.post<GridDataResult>(`${this.apiUrl}/SelectClaims`, toDataSourceRequestString(state)});
I needed it to be a GET (URL) so i created a new object
public class GridParamaterBinder
{
public int Page { get; set; }
public int PageSize { get; set; }
public string Filter { get; set; }
public string Sort { get; set; }
public DataSourceRequest ToDataSourceRequest(IConfigurationProvider mapper, Func<string, string> OverDefaultParamaterMapping)
{
DataSourceRequest result = new DataSourceRequest();
result.Page = Page;
result.PageSize = PageSize;
result.Sorts = GridDescriptorSerializer.Deserialize<SortDescriptor>(Sort);
result.Filters = FilterDescriptorFactory.Create(Filter);
return result;
}
}
and used it instead of the Telerik effort.
in API I Bind it like so
public virtual DataSourceResult Get([FromUri]GridParamaterBinder request)
And then used it like
DataSourceResult results = query.ToDataSourceResult(request.ToDataSourceRequest(), r => (r)));
Thanks #KevDevMan for your solution. I found this example,
then I changed my API controller like this and it worked like a charm :
[HttpGet, Route("for-kendo-grid")]
public DataSourceResult GetProducts([System.Web.Http.ModelBinding.ModelBinder(typeof(WebApiDataSourceRequestModelBinder))] DataSourceRequest request)
explanation here
The website that I'm working on is heavily depending on ajax/json and knockout.js.
I would like to have a lot of my Controllers return view-tailored 'json objects', without wrapping them in a JsonResult when returning the method.
This would mean I could easily composite multiple calls into one parent object, but still be able to call the Actions separately too.
Simplified example:
public object Main(int groupId)
{
var viewModel = new
{
Persons = Employees(groupId),
Messages = AllMessages()
};
return viewModel;
}
public object Employees(int groupId)
{
return DatabaseContext.Employees.Where(e => e.GroupId == groupId).ToList();
}
public object AllMessages()
{
return DatabaseContext.Messages.ToList();
}
I was hoping I could capture the returned object in OnActionExecuted and at that point wrap the whole result up in a final JsonResult.
The result is already converted to a string and captured in a ContentResult though.
Any ideas? :) Thanks,
A good approach on this is to create helper methods for your entity calls. Or if you have those methods already somewhere, they can actually serve as the helper methods. In that manner you can return a list of strongly-typed Messages and Employees as well as returning your desired parent object. You can then have individual controller methods that returns json objects. In addition, you can extend the parent viewmodel to return additional fields.
The Parent ViewModel
public class ParentModel {
public Employee Persons {get;set;}
public Message Messages {get;set;}
}
The Helper Methods
The beauty of using helper methods similar to what is defined here is that you can apply a few more logic to your query, and more, and you don't have to change anything in your controller methods.
public ParentModel GetMain(int groupId)
{
var viewModel = new ParentModel
{
Persons = Employees(groupId),
Messages = AllMessages()
};
return viewModel;
}
public IEnumerable<Employee> Employees(int groupId)
{
return DatabaseContext.Employees.Where(e => e.GroupId == groupId).ToList();
}
public IEnumerable<Message> AllMessages()
{
return DatabaseContext.Messages.ToList();
}
The Controller Methods
public ActionResult GetParent(int groupId){
return Json(helperinstance.GetMain());
}
public ActionResult GetEmployees(int groupId){
return Json(helperinstance.Employees());
}
public ActionResult GetMessages(int groupId){
return Json(helperinstance.AllMessages());
}
Thanks for the answer. I'm not going for the solution of von v. because I like to keep the boilerplate as small as possible.
In the end I am trying out the following approach. It seems to work pretty well for now, but I still have to test it in real production.
If anyone has some (security) concerns with this, I'm happy to hear them in the comments.
// BaseController
protected override void OnActionExecuting(ActionExecutingContext filterContext)
{
var returnType = ((ReflectedActionDescriptor)filterContext.ActionDescriptor).MethodInfo.ReturnType;
// is the returnType not deriving from ActionResult? Automatically wrap it in a JsonResult
if ( !typeof(ActionResult).IsAssignableFrom(returnType) )
{
var result = filterContext.ActionDescriptor.Execute(filterContext, filterContext.ActionParameters);
filterContext.Result = Json( result );
}
}
Consider, for example's sake, the logic "A user may only edit or delete a comment that the user has authored".
My Controller Actions will repeat the logic of checking whether the currently logged in user can affect the comment. Example
[Authorize]
public ActionResult DeleteComment(int comment_id)
{
var comment = CommentsRepository.getCommentById(comment_id);
if(comment == null)
// Cannot find comment, return bad input
return new HttpStatusCodeResult(400);
if(comment.author != User.Identity.Name)
// User not allowed to delete this comment, return Forbidden
return new HttpStatusCodeResult(403);
// Error checking passed, continue with delete action
return new HttpStatusCodeResult(200);
}
Of course, I can bundle that logic up in a method so that I'm not copy / pasting that snippet; however, taking that code out of the controller and putting it in a ValidationAttribute keeps my Action smaller and easier to write tests for. Example
public class MustBeCommentAuthorAttribute : ValidationAttribute
{
// Import attribute for Dependency Injection
[Import]
ICommentRepository CommentRepository { get; set; }
protected override ValidationResult IsValid(object value, ValidationContext validationContext)
{
int comment_id = (int)value;
var comment = CommentsRepository.getCommentById(comment_id);
if(comment == null)
return new ValidationResult("No comment with that ID");
if(comment.author != HttpContext.Current.User.Identity.Name)
return new ValidationResult("Cannot edit this comment");
// No errors
return ValidationResult.Success;
}
}
public class DeleteCommentModel
{
[MustBeCommentAuthor]
public int comment_id { get; set; }
}
Is Model Validation the right tool for this job? I like taking that concern out of the controller Action; but in this case, it may complicate things further. This is especially true when you consider that this Action is part of a RESTful API and needs to return a different HTTP Status Code depending on the Validation errors in the ModelState.
Is there "best practice" in this case?
Personally, I think that it looks nice, but you are getting carried away with annotations. I think that this does not belong in your presentation layer and it should be handled by your service layer.
I would have something on the lines of:
[Authorize]
public ActionResult DeleteComment(int comment_id)
{
try
{
var result = CommentsService.GetComment(comment_id, Auth.Username);
// Show success to the user
}
catch(Exception e)
{
// Handle by displaying relevant message to the user
}
}
I am using a Remote validation attribute on my view model to validate a Bank Account that is specified for my Company:
ViewModel:
[Remote("CheckDefaultBank", "Company")]
public string DefaultBank
{
This in the controller I have:
[HttpGet]
public JsonResult CheckDefaultBank(string defaultBank)
{
bool result = BankExists(defaultBank);
return Json(result, JsonRequestBehavior.AllowGet);
}
That all works well. But, I have two other banks related to my company as well. However, when the remote validation js calls the action it uses a parameter mactching the field name of "DefaultBank"... so I use that as a parameter in my action.
Is there some attribute I can add in the view so that it will use a parameter of say "bankId" on the ajax get so I don't need an action for each field which are basically exactly the same?
The goal here is to eliminate now having to have this in my controller:
[HttpGet]
public JsonResult CheckRefundBank(string refundBank)
{
bool result = BankExists(defaultBank);
return Json(result, JsonRequestBehavior.AllowGet);
}
[HttpGet]
public JsonResult CheckPayrollBank(string payrollBank)
{
bool result = BankExists(defaultBank);
return Json(result, JsonRequestBehavior.AllowGet);
}
I was hoping I could do something like this in the view:
#Html.EditorFor(model => model.DefaultBank, new { data-validate-parameter: bankId })
This way I could just use the same action for all of the Bank entries like:
[HttpGet]
public JsonResult CheckValidBank(string bankId)
{
bool result = BankExists(bankId);
return Json(result, JsonRequestBehavior.AllowGet);
}
Possible?
For just such a situation, I wrote a RemoteReusableAttribute, which may be helpful to you. Here is a link to it: Custom remote Validation in MVC 3
Since MVC uses the default model binder for this, just like a normal action method. You could take a FormsCollection as your parameter and lookup the value. However, I personally would find it much easier to just use several parameters to the function, unless you start having dozens of different parameters.
You could also write a custom model binder, that would translate the passed parameter to a generic one.
Consider encapsulating the logic, "BankExists" in this case into a ValidationAttribute (Data Annotations Validator). This allows other scenarios as well.
Then use a wrapper ActionResult like the one below, which lets you pass in any validator.
[HttpGet]
public ActionResult CheckRefundBank(string refundBank)
{
var validation = BankExistsAttribute();
return new RemoteValidationResult(validation, defaultBank);
}
Here is the code for the ActionResult that works generically with Validators.
using System.ComponentModel.DataAnnotations;
using System.Web.Mvc;
public class RemoteValidationResult : ActionResult
{
public RemoteValidationResult(ValidationAttribute validation, object value)
{
this.Validation = validation;
this.Value = value;
}
public ValidationAttribute Validation { get; set; }
public object Value { get; set; }
public override void ExecuteResult(ControllerContext context)
{
var json = new JsonResult();
json.JsonRequestBehavior = JsonRequestBehavior.AllowGet;
if (Validation.IsValid(Value))
{
json.Data = true;
}
else
{
json.Data = Validation.FormatErrorMessage(Value.ToString());
}
json.ExecuteResult(context);
}
}
As an extra enhancement consider creating a Controller Extension method to dry up your return call even more.
(couldn't think of a better title, sorry)
So I've got my layout page, on this page there is a searchbar + options. Choosing whatever, should take you through to the search page, with the results etc. Fairly standard. What I've done to get this working is to create a MasterModel class, with a SearchDataModel class member on it. This SearchDataModel contains the various parameters for the search (search term, what fields to search on etc).
I've then strongly typed my layout page to the MasterModel class, and using a Html.BeginForm... I've constructed the search form for it. However all the checkboxes relating to the fields aren't checked by default, even though the default value for all the fields is true (via a private getter/setter setup).
Yet when I submit the form to the SearchController, all the checkboxes are set to true. So I'm a bit confused as to why it knows they should be true, yet not set the checkboxes to be checked?
Putting breakpoints in key places seems to show that the model isn't insantiated on the get requests, only the post to the Search controller?
I may be going about this all wrong, so if so, pointers as to the right way always appreciated.
public class MasterModel {
public SearchDataModel SearchModel { get; set; }
}
public class SearchDataModel{
private bool _OnTags = true;
private bool _OnManufacturers = true;
private bool _OnCountries = true;
[Display(Name= "Tags")]
public bool OnTags {
get { return _OnTags; }
set { _OnTags = value; }
}
[Display(Name= "Manufacturers")]
public bool OnManufacturers {
get { return _OnManufacturers; }
set { _OnManufacturers = value; }
}
[Display(Name= "Countries")]
public bool OnCountries {
get { return _OnCountries; }
set { _OnCountries = value; }
}
[Required]
[Display(Name="Search Term:")]
public string SearchTerm { get; set; }
}
Then in the _layout page:
#Html.CheckBoxFor(m => m.SearchModel.OnTags, new { #class="ddlCheckbox", #id="inpCheckboxTag" })
#Html.LabelFor(m =>m.SearchModel.OnTags)
Make sure you return a MasterModel with initialized SearchModel from your views:
public ActionResult Index()
{
var model = new MasterModel
{
SearchModel = new SearchDataModel()
};
return View(model);
}
Another possibility to implement this functionality than strongly typing your master layout to a view model is yo use Html.Action as shown by Phil Haack in his blog post.