What I have is a model which has one of it's attributes dynamic. This dynamic attribute holds one of about 50 different objects. This model is send to a view that dynamic creates the page based on which object is used. This is working perfectly ... the issue is the postback. When the model posts back the modelbinder is not able to bind the dynamic attribute. I was expecting this and thought I would be able to handle it but nothing that I tried works appart from making an action for EACH different objects.
Model
public class VM_List
{
public Config.CIType CIType { get; set; }
public dynamic SearchData { get; set; }
//Lots of static fields
}
This works
public ActionResult List_Person(VM_List Model, VM_Person_List SearchData)
{
Model.SearchData = SearchData;
//Stuff
}
public ActionResult List_Car(VM_List Model, VM_Car_List SearchData)
{
Model.SearchData = SearchData;
//Stuff
}
But what I want is a single action
public ActionResult List(VM_List Model)
{
//Stuff
}
I have tried things like
public ActionResult List(VM_List Model)
{
switch (Model.CIType)
{
case Config.CIType.Person:
UpdateModel((VM_Person_List)Model.SearchData);
break;
default:
SearchData = null;
break;
}
//Stuff
}
and a Custom modelbinder
CIType CIType = (CIType)bindingContext.ValueProvider.GetValue("CIType").ConvertTo(typeof(CIType));
switch (CIType)
{
case Config.CIType.Person:
SearchData = (VM_Person_List)bindingContext.ValueProvider.GetValue("SearchData").ConvertTo(typeof(VM_Person_List));
break;
default:
SearchData = null;
break;
}
but I can't get either to work. Any ideas?
After trying many different things I finally found a way that works.
Action:
public ActionResult List(VM_List Model)
{
//If the defaultmodelbinder fails SearchData will be an object
if(Model.SearchData.GetType() == typeof(object))
{
//Get SearchData as a Dictionary
Dictionary<string, string> DSearchData = Request.QueryString.AllKeys.Where(k => k.StartsWith("SearchData.")).ToDictionary(k => k.Substring(11), k => Request.QueryString[k]);
switch (Model.CIType)
{
case Config.CIType.Person:
Model.SearchData = new VM_Person_List(DSearchData);
break;
case Config.CIType.Car:
Model.SearchData = new VM_Car_List(DSearchData);
break;
}
//Rest of action
//..
}
and for each object make a constructor that accepts a dictionary
public VM_Car_List(Dictionary<string, string> DSearchData)
{
this.Make = Convert.ToInt32(DSearchData["Make"]);
this.Model = Convert.ToInt32(DSearchData["Model"]);
this.Year = Convert.ToInt32(DSearchData["Year"]);
// ETC
}
Related
I'm trying to create form from such model:
class NewContractorModel
{
//...
public PhotoModel photos { get; set; }
//...
}
class PhotoModel
{
public List<Photo> f { get; set; }
}
From controller I do some manipulation (actually I removed some photos from the collection) on the model object and put them into the view page using this:
return new View("SomeView", model);
I've tried to create inputs (lets say hidden inputs) for each Photo.
for (int i = 0; i < Model.photos.f.Count; ++i)
{
#Html.HiddenFor(m => m.photos.f[i].Uri)
#Html.HiddenFor(m => m.photos.f[i].ThumbnailUri)
#Html.HiddenFor(m => m.photos.f[i].SmallThumbnailUri)
#Html.TextBoxFor(m => m.photos.f[i].Description, new { placeholder = "Dodaj opis" })
}
But as I noticed that this doesnt work because it dismiss all of model modifications (it still stores all Photos in List despite the fact that I've removed them in Controler method).
Then I tried this code:
for (int i = 0; i < Model.photos.f.Count; ++i)
{
Photo photo = Model.photos.f[i];
<input id="photos_f_#{#i}__Uri" name="photos.f[#{#i}].Uri" type="hidden" value="#photo.Uri"/>
<input id="photos_f_#{#i}__ThumbnailUri" name="photos.f[#{#i}].ThumbnailUri" type="hidden" value="#photo.ThumbnailUri"/>
<input id="photos_f_#{#i}__SmallThumbnailUri" name="photos.f[#{#i}].SmallThumbnailUri" type="hidden" value="#photo.SmallThumbnailUri"/>
<input id="photos_f_#{#i}__Description" name="photos.f[#{#i}].Description" placeholder="Dodaj opis" type="text" value="#photo.Description"/>
}
...and this time IT WORKS!
Can anyone explain me what is the difference between those two parts of code?
I've tried to swich this code more than ten times and it always work the same so it's not my fault. ;)
I think that there is a bug in HtmlHelper methods but is there any walk-around ? I'd like to use helpers methods instead of raw html.
EDIT:
This is simplified controller class.
public class MyController
{
private NewContractorModel _model = null;
protected override void OnActionExecuting(ActionExecutingContext filterContext)
{
_model = SerializationUtility.Deserialize(Request.Form["Data"]) as NewContractorModel;
if (_model == null)
_model = TempData["Data"] as NewContractorModel;
if (_model == null)
_model = new NewContractorModel() as NewContractorModel;
TryUpdateModel(_model);
}
protected override void OnResultExecuted(ResultExecutedContext filterContext)
{
if (filterContext.Result is RedirectToRouteResult)
TempData["Data"] = _model;
}
private bool CheckModel(object model)
{
Type type = model.GetType();
PropertyInfo[] properties = type.GetProperties();
foreach (PropertyInfo p in properties)
{
object[] attr = p.GetCustomAttributes(true);
foreach (object a in attr)
{
if (a is ValidationAttribute)
{
object value = p.GetValue(model, null);
if (!((ValidationAttribute)a).IsValid(value))
return false;
}
}
}
return true;
}
protected ActionResult SelectPage(string delPhoto)
{
if (!CheckModel(_model))
{
// Do some action
}
//.....
foreach (ZAY.Database.Photo p in _model.photos.f)
{
if (p.Uri == Request["delPhoto"])
{
_model.photos.f.Remove(p);
break;
}
}
//.....
return View("SomeView", _model);
}
}
I noticed that inside lambdas the model looks just like after TryUpdateModel call (before modifications). If I don't use lambdas the model is modified... :/
And also my Photo class (generated from EntityFramework - so there is nothing interesting) and also simplified:
public class Photo : EntityObject
{
[Required]
public string Uri { get; set; }
[Required]
public string ThumbnailUri { get; set; }
[Required]
public string SmallThumbnailUri { get; set; }
public string Description { get; set; }
}
I'm sorry that I'm writing only such small snippets but the whole code is more complicated - there is only the most interesting part of it.
This is the answer to my problem:
http://blogs.msdn.com/b/simonince/archive/2010/05/05/asp-net-mvc-s-html-helpers-render-the-wrong-value.aspx
I wonder why it is not mentioned in documentation... :/
From your description, I don't really understand what's going wrong in your first sample. But you certainly have a problem with the scope of the loop variable i.
Since the expression m => m.photos.f[i] involves closures, it will be evaluated at a later time, at a time when the for loop has already finished. The expression captures the variable i (and not the value of the variable i). When it is eventually evaluated, it finds the value Model.photos.f.Count in the variable i. So all hidden fields and textboxes will use the same invalid value of i.
Your second code sample avoids this problem by using a local variable within the for loop.
I need some help with this one....
I have this simple model:
public class Candidat
{
public string LoginEmail { get; set; }
[Required]
[DataType(DataType.Text)]
[Display(Name = "Prénom")]
public string FirstName { get; set; }
[Required]
[DataType(DataType.Text)]
[Display(Name = "Nom")]
public string LastName { get; set; }
}
I also have a controller like this:
[Authorize]
public ActionResult Index(Candidat model)
{
if (model.LoginEmail == null)
{
model = null;
using (var db = new rhDB())
{
MembershipUser user = Membership.GetUser();
if (user != null)
{
model = (from m in db.Candidates where m.LoginEmail == user.Email select m).SingleOrDefault();
}
if (model == null)
{
model = new Candidat();
model.LoginEmail = user.Email;
model.Email = user.Email;
}
}
}
return View("MyProfileCandidate", model);
}
As you can see, I check if the user as an existing record in the database. If not, I create a new instance of the model and set some default values... Then, I pass it to an EditView.
The problem is that my view show with the error validation messages... For all required fields...
Of course, this is because those fields are empty and required... It seems like the view think I am posting back an invalid model... Is there a way to hide those validation message ?
Try clearing the model state if you intend to modify some of the values on your model in the POST action:
[Authorize]
public ActionResult Index(Candidat model)
{
if (model.LoginEmail == null)
{
model = null;
using (var db = new rhDB())
{
MembershipUser user = Membership.GetUser();
if (user != null)
{
ModelState.Clear();
model = (from m in db.Candidates where m.LoginEmail == user.Email select m).SingleOrDefault();
}
if (model == null)
{
ModelState.Clear();
model = new Candidat();
model.LoginEmail = user.Email;
model.Email = user.Email;
}
}
}
return View("MyProfileCandidate", model);
}
The reason for this is that HTML helpers will use model state values that were initially posted instead of those in the model. You could also clear individual fields in the model state: ModelState.Remove("FirstName");.
I had two fields some thing like phone number and mobile number. Some thing like..
[Required]
public string Phone { get; set; }
[Required]
public string Mobile{ get; set; }
But user can enter data in either one of it. One is mandatory. How to handle them i.e how to disable the required field validator for one field when user enter data in another field and viceversa. In which event i have to handle it in javascript and what are the scripts i need to add for this. Can anyone please help to find the solution...
One possibility is to write a custom validation attribute:
public class RequiredIfOtherFieldIsNullAttribute : ValidationAttribute, IClientValidatable
{
private readonly string _otherProperty;
public RequiredIfOtherFieldIsNullAttribute(string otherProperty)
{
_otherProperty = otherProperty;
}
protected override ValidationResult IsValid(object value, ValidationContext validationContext)
{
var property = validationContext.ObjectType.GetProperty(_otherProperty);
if (property == null)
{
return new ValidationResult(string.Format(
CultureInfo.CurrentCulture,
"Unknown property {0}",
new[] { _otherProperty }
));
}
var otherPropertyValue = property.GetValue(validationContext.ObjectInstance, null);
if (otherPropertyValue == null || otherPropertyValue as string == string.Empty)
{
if (value == null || value as string == string.Empty)
{
return new ValidationResult(string.Format(
CultureInfo.CurrentCulture,
FormatErrorMessage(validationContext.DisplayName),
new[] { _otherProperty }
));
}
}
return null;
}
public IEnumerable<ModelClientValidationRule> GetClientValidationRules(ModelMetadata metadata, ControllerContext context)
{
var rule = new ModelClientValidationRule
{
ErrorMessage = FormatErrorMessage(metadata.GetDisplayName()),
ValidationType = "requiredif",
};
rule.ValidationParameters.Add("other", _otherProperty);
yield return rule;
}
}
which you would apply to one of the properties of your view model:
public class MyViewModel
{
[RequiredIfOtherFieldIsNull("Mobile")]
public string Phone { get; set; }
public string Mobile { get; set; }
}
then you could have a controller:
public class HomeController : Controller
{
public ActionResult Index()
{
return View(new MyViewModel());
}
[HttpPost]
public ActionResult Index(MyViewModel model)
{
return View(model);
}
}
and finally a view in which you will register an adapter to wire the client side validation for this custom rule:
#model MyViewModel
<script src="#Url.Content("~/Scripts/jquery.validate.js")" type="text/javascript"></script>
<script src="#Url.Content("~/Scripts/jquery.validate.unobtrusive.js")" type="text/javascript"></script>
<script type="text/javascript">
jQuery.validator.unobtrusive.adapters.add(
'requiredif', ['other'], function (options) {
var getModelPrefix = function (fieldName) {
return fieldName.substr(0, fieldName.lastIndexOf('.') + 1);
}
var appendModelPrefix = function (value, prefix) {
if (value.indexOf('*.') === 0) {
value = value.replace('*.', prefix);
}
return value;
}
var prefix = getModelPrefix(options.element.name),
other = options.params.other,
fullOtherName = appendModelPrefix(other, prefix),
element = $(options.form).find(':input[name="' + fullOtherName + '"]')[0];
options.rules['requiredif'] = element;
if (options.message) {
options.messages['requiredif'] = options.message;
}
}
);
jQuery.validator.addMethod('requiredif', function (value, element, params) {
var otherValue = $(params).val();
if (otherValue != null && otherValue != '') {
return true;
}
return value != null && value != '';
}, '');
</script>
#using (Html.BeginForm())
{
<div>
#Html.LabelFor(x => x.Phone)
#Html.EditorFor(x => x.Phone)
#Html.ValidationMessageFor(x => x.Phone)
</div>
<div>
#Html.LabelFor(x => x.Mobile)
#Html.EditorFor(x => x.Mobile)
#Html.ValidationMessageFor(x => x.Mobile)
</div>
<button type="submit">OK</button>
}
Pretty sick stuff for something so extremely easy as validation rule that we encounter in our everyday lives. I don't know what the designers of ASP.NET MVC have been thinking when they decided to pick a declarative approach for validation instead of imperative.
Anyway, that's why I use FluentValidation.NET instead of data annotations to perform validations on my models. Implementing such simple validation scenarios is implemented in a way that it should be - simple.
I know this question is not so hot, because it was asked relatively long time ago, nevertheless I'm going to share with a slightly different idea of solving such an issue. I decided to implement mechanism which provides conditional attributes to calculate validation results based on other properties values and relations between them, which are defined in logical expressions.
Your problem can be defined and automatically solved by the usage of following annotations:
[RequiredIf("Mobile == null",
ErrorMessage = "At least email or phone should be provided.")]
public string Phone{ get; set; }
[RequiredIf("Phone == null",
ErrorMessage = "At least email or phone should be provided.")]
public string Mobile { get; set; }
If you feel it would be useful for your purposes, more information about ExpressiveAnnotations library can be found here. Client side validation is also supported out of the box.
Since nobody else suggested it, I'm going to tell you a different way to do this that we use.
If you create a notmapped field of a custom data type (in my example, a pair of gps points), you can put the validator on that and you don't even need to use reflection to get all the values.
[NotMapped]
[DCGps]
public GPS EntryPoint
{
get
{
return new GPS(EntryPointLat, EntryPointLon);
}
}
and the class, a standard getter/setter
public class GPS
{
public decimal? lat { get; set; }
public decimal? lon { get; set; }
public GPS(decimal? lat, decimal? lon)
{
this.lat = lat;
this.lon = lon;
}
}
and now the validator:
public class DCGps : DCValidationAttribute
{
protected override ValidationResult IsValid(object value, ValidationContext validationContext)
{
if (!(value is GPS)) {
return new ValidationResult("DCGps: This annotation only works with fields with the data type GPS.");
}
//value stored in the field.
//these come through as zero or emptry string. Normalize to ""
string lonValue = ((GPS)value).lonstring == "0" ? "" : ((GPS)value).lonstring;
string latValue = ((GPS)value).latstring == "0" ? "" : ((GPS)value).latstring;
//place validation code here. You have access to both values.
//If you have a ton of values to validate, you can do them all at once this way.
}
}
I'm currently working on a MVC.NET 3 application; I recently attended a course by "Uncle Bob" Martin which has inspired me (shamed me?) into taking a hard look at my current development practice, particularly my refactoring habits.
So: a number of my routes conform to:
{controller}/{action}/{type}
Where type typically determines the type of ActionResult to be returned, e.g:
public class ExportController
{
public ActionResult Generate(String type, String parameters)
{
switch (type)
{
case "csv":
//do something
case "html":
//do something else
case "json":
//do yet another thing
}
}
}
Has anyone successfully applied the "replace switch with polymorhism" refactoring to code like this? Is this even a good idea? Would be great to hear your experiences with this kind of refactoring.
Thanks in advance!
The way I am looking at it, this controller action is screaming for a custom action result:
public class MyActionResult : ActionResult
{
public object Model { get; private set; }
public MyActionResult(object model)
{
if (model == null)
{
throw new ArgumentNullException("Haven't you heard of view models???");
}
Model = model;
}
public override void ExecuteResult(ControllerContext context)
{
// TODO: You could also use the context.HttpContext.Request.ContentType
// instead of this type route parameter
var typeValue = context.Controller.ValueProvider.GetValue("type");
var type = typeValue != null ? typeValue.AttemptedValue : null;
if (type == null)
{
throw new ArgumentNullException("Please specify a type");
}
var response = context.HttpContext.Response;
if (string.Equals("json", type, StringComparison.OrdinalIgnoreCase))
{
var serializer = new JavaScriptSerializer();
response.ContentType = "text/json";
response.Write(serializer.Serialize(Model));
}
else if (string.Equals("xml", type, StringComparison.OrdinalIgnoreCase))
{
var serializer = new XmlSerializer(Model.GetType());
response.ContentType = "text/xml";
serializer.Serialize(response.Output, Model);
}
else if (string.Equals("csv", type, StringComparison.OrdinalIgnoreCase))
{
// TODO:
}
else
{
throw new NotImplementedException(
string.Format(
"Sorry but \"{0}\" is not a supported. Try again later",
type
)
);
}
}
}
and then:
public ActionResult Generate(string parameters)
{
MyViewModel model = _repository.GetMeTheModel(parameters);
return new MyActionResult(model);
}
A controller should not care about how to serialize the data. That's not his responsibility. A controller shouldn't be doing any plumbing like this. He should focus on fetching domain models, mapping them to view models and passing those view models to view results.
If you wanted to "replace switch with polymorphism" in this case, you could create three overloaded Generate() ActionResult methods. Using custom model binding, make the Type parameter a strongly-typed enum called DataFormat (or whatever.) Then you'd have:
public ActionResult Generate(DataFormat.CSV, String parameters)
{
}
public ActionResult Generate(DataFormat.HTML, String parameters)
{
}
public ActionResult Generate(DataFormat.JSON, String parameters)
{
}
Once you get to this point, you can refactor further to get the repetition out of your Controller.
It seems that when MVC validates a Model that it runs through the DataAnnotation attributes (like required, or range) first and if any of those fail it skips running the Validate method on my IValidatableObject model.
Is there a way to have MVC go ahead and run that method even if the other validation fails?
You can manually call Validate() by passing in a new instance of ValidationContext, like so:
[HttpPost]
public ActionResult Create(Model model) {
if (!ModelState.IsValid) {
var errors = model.Validate(new ValidationContext(model, null, null));
foreach (var error in errors)
foreach (var memberName in error.MemberNames)
ModelState.AddModelError(memberName, error.ErrorMessage);
return View(post);
}
}
A caveat of this approach is that in instances where there are no property-level (DataAnnotation) errors, the validation will be run twice. To avoid that, you could add a property to your model, say a boolean Validated, which you set to true in your Validate() method once it runs and then check before manually calling the method in your controller.
So in your controller:
if (!ModelState.IsValid) {
if (!model.Validated) {
var validationResults = model.Validate(new ValidationContext(model, null, null));
foreach (var error in validationResults)
foreach (var memberName in error.MemberNames)
ModelState.AddModelError(memberName, error.ErrorMessage);
}
return View(post);
}
And in your model:
public bool Validated { get; set; }
public IEnumerable<ValidationResult> Validate(ValidationContext validationContext) {
// perform validation
Validated = true;
}
There's a way to do it without requiring boilerplate code at the top of each controller action.
You'll need to replace the default model binder with one of your own:
protected void Application_Start()
{
// ...
ModelBinderProviders.BinderProviders.Clear();
ModelBinderProviders.BinderProviders.Add(new CustomModelBinderProvider());
// ...
}
Your model binder provider looks like this:
public class CustomModelBinderProvider : IModelBinderProvider
{
public IModelBinder GetBinder(Type modelType)
{
return new CustomModelBinder();
}
}
Now create a custom model binder that actually forces the validation. This is where the heavy lifting's done:
public class CustomModelBinder : DefaultModelBinder
{
protected override void OnModelUpdated(ControllerContext controllerContext, ModelBindingContext bindingContext)
{
base.OnModelUpdated(controllerContext, bindingContext);
ForceModelValidation(bindingContext);
}
private static void ForceModelValidation(ModelBindingContext bindingContext)
{
var model = bindingContext.Model as IValidatableObject;
if (model == null) return;
var modelState = bindingContext.ModelState;
var errors = model.Validate(new ValidationContext(model, null, null));
foreach (var error in errors)
{
foreach (var memberName in error.MemberNames)
{
// Only add errors that haven't already been added.
// (This can happen if the model's Validate(...) method is called more than once, which will happen when
// there are no property-level validation failures.)
var memberNameClone = memberName;
var idx = modelState.Keys.IndexOf(k => k == memberNameClone);
if (idx < 0) continue;
if (modelState.Values.ToArray()[idx].Errors.Any()) continue;
modelState.AddModelError(memberName, error.ErrorMessage);
}
}
}
}
You'll need an IndexOf extension method, too. This is a cheap implementation but it'll work:
public static int IndexOf<T>(this IEnumerable<T> source, Func<T, bool> predicate)
{
if (source == null) throw new ArgumentNullException("source");
if (predicate == null) throw new ArgumentNullException("predicate");
var i = 0;
foreach (var item in source)
{
if (predicate(item)) return i;
i++;
}
return -1;
}