I'm working on a React app that is fed data from a Rails api. I'm currently working on a form that includes a nested association (i.e. in the model_a has many model_b's and you can create them in the same form).
The problem I'm having is that Rails expects nested association with a certain naming convention and the same field that controls how the parameter is named when its sent to rails also controls how React finds the right data when the Rails API responds.
This becomes problematic on the edit page because I want to show the models_a's (Retailers) already existing model_b's (SpendingThresholds in this case) and when I change the 'name' field to suit the rails side, React doesn't know where to look for that data anymore. When I try to pass the data directly it comes in as a different type of array and certain functions fail.
I think its easier to show than tell here so
initially I had this
<FieldArray
name="spending_thresholds"
component={renderSpendingThresholds}
/>
and data was coming through like
Object {_isFieldArray: true, forEach: function, get: function, getAll: function, insert: function…
to my React app from the Rails API, which worked, however that 'name' isn't to Rails liking (Rails wants it to be called 'spending_thresholds_attributes' for accepts_nested_attributes to work) so I changed it to
<FieldArray
name="spending_thresholds_attributes"
fields={this.props.retailer.spending_thresholds}
component={renderSpendingThresholds}
/>
and data start coming through to the renderSpendingThresholds component in this format
[Object]
0:Object
length:1
__proto__:Array(0)
which React doesn't like for some reason.
Anyone know how to fix this/why those two objects, which hold the same information from the Rails side anyway, are being treated differently?
EDITS
renderSpendingThresholds component
The fields attribute in the renderSpendingThresholds component is the object that's coming through differently depending on how I input it
const renderSpendingThresholds = ({ fields }) => (
<ul className="spending-thresholds">
<li>
<Button size="sm" color="secondary" onClick={(e) => {
fields.push({});
e.preventDefault();
}
}>
Add Spending Threshold
</Button>
</li>
{fields.map((spending_threshold, index) => (
<li key={index}>
<h4>Spending Threshold #{index + 1}</h4>
<Button
size="sm"
color="danger"
title="Remove Spending Threshold"
onClick={() => fields.remove(index)}
>
Remove
</Button>
<Field
name={`${spending_threshold}.spend_amount`}
type="number"
component={renderField}
label="Spend Amount"
placeholder="0"
/>
<Field
name={`${spending_threshold}.bonus_credits`}
type="number"
component={renderField}
label="Bonus Credits"
placeholder="0"
/>
</li>
))}
</ul>
);
It looks like you are passing fields through props and then destructuring the fields out of the props in the callback of the renderSpendingThresholds and discarding the rest. According to the docs, a specific redux-form object is passed through to the render callback. You're essentially overwriting this. Try changing {field} to something like member or spending_threshold. Then you can use the specific map function to iterate over the spending_threshold items. Your field prop should still be available under member.fields or something similar.
For the code that you currently show, who exactly handles the submission?
you use the original flow of form submit?
if so, so please handle that by yourself.
** this line of code, looks weird:
onClick={() => fields.remove(index)}
as you interact directly with the state values...
you need to update the state through
this.setState({fields: FIELDS_WITHOUT_ITEM})
and now when you need to handle your own submission, you don't really care of the input names. Because you are using the state as input.
ie:
class FormSpending extends Component {
handleSubmit() {
var fieldsData = this.state.fields.map(field => {
return {
whateverkey: field.dontcare,
otherKey: field.anotherDontCare
};
});
var formData = {
fields: fieldsData
};
ajaxLibrary.post(URL_HERE, formData).....
}
render() {
return (
...
<form onSubmit={()=>this.handleSubmit()}>
...
</form>
...
);
}
}
Related
(I'm new to Svelte so it is quite likely that I'm doing something wrong here)
UPDATE: I've added a second, slightly different REPL which may demonstrate the problem better. Try this one: https://svelte.dev/repl/ad7a65894f8440ad9081102946472544?version=3.20.1
I've encountered a problem attempting to bind a text input to a reactive value.
I'm struggling to describe the problem in words, so hopefully a reduced demo of the issue in the attached REPL will make more sense.
https://svelte.dev/repl/6c8068ed4cc048919f71d87f9d020696?version=3.20.1
The demo contains two custom <Selector> components on a page.
The first component is passed two string values ("one" and "two"):
<Selector valueOne="one" valueTwo="two"/>
Clicking the buttons next to the input field sets selectedValue to one of these values.
This, in turn, triggers the following reactive declaration to update:
$: value = selectedValue
The input field is bound to this reactive value:
<input type="text" bind:value>
So clicking the "One" button sets the input text to "one", and clicking the "Two" button sets the input field to "two".
Importantly though, you can still type anything into the input field.
The second component is passed two array values:
<Selector valueOne={[1, "one"]} valueTwo={[2, "two"]}/>
Again, clicking the buttons sets selectedValue to one of these.
However this time the reactive declaration depends on an array element:
$: value = selectedValue[1]
Everything works as before, except now you can no longer type into the input field at all.
So the question is - why does <input bind:value> behave differently for these two:
$: value = aString
vs
$: value = anArray[x]
It seems that this is only an issue when using two-way bindings.
By switching to a one-way and an on:input handler, the problem goes away:
i.e. instead of this:
<input type="text" bind:value={valX}/>
use this:
<input type="text" value={valX} on:input={e => valX = e.target.value}/>
I'm pretty sure your reactive declaration is overwriting your bound value as soon as it changes, which is with every key stroke on the input and every button press. Meaning it technically is working, you're just reverting it each time it changes. Check out this version of it that uses a watcher.
Also binding to a reactive declaration means you're never actually changing the variables with the input (which you can see in your JSON result on the first selector when you type in the input the value doesn't update only on button click).
Why not lose the reactive declaration and bind directly to the variable you want. Then use an {#if} block to switch between which version of the input you're showing based on the truthiness of index?
<script>
export let valueOne;
export let valueTwo;
export let index;
let selectedValue = index? [] : '';
let selectValue = (val) => selectedValue = val;
</script>
{#if index}
<input type="text" bind:value={selectedValue[index]} placeholder="Type anything...">
{:else}
<input type="text" bind:value={selectedValue} placeholder="Type anything...">
{/if}
<button on:click={() => selectValue(valueOne)}>One</button>
<button on:click={() => selectValue(valueTwo)}>Two</button>
<p>
<strong>Selected value:</strong> {JSON.stringify(selectedValue)}
</p>
By binding directly to the selectedValue or an index of it you have the added benefit of changing the value with the input. Here's a working example in the REPL
I'm creating a website with ASP.NET MVC5 and I'm using MaterializeCSS for the first time, which looks like a very exciting framework.
However, the checkboxes generated by CheckBoxFor helper become hidden !
When I write :
#Html.CheckBoxFor(m => m.IsAgreeTerms)
The generated HTML is :
<input name="IsAgreeTerms" type="hidden" value="false">
Why does Materialize change my type=checkbox into type=hidden ?
I tried to add type="checkbox" in the CheckboxFor helper, but it doesnt change anything. The only way is to modify in in my browser's console.
The only solution I found is this SO thread.
However, the accepted answer doesn't change anything for me.
The other answer works, but I think it's ugly to add some JS script to modify what Materialize modifies without my consent.
Is there any way to say "Hey, I ask for a type=checkbox, so just let my type=checkbox in the generated HTML" ?
Thank you
UPDATE :
My full ASP.NET MVC code is :
#Html.CheckBoxFor(m => m.IsAgreeTerms, new { #type = "checkbox" })
#Html.LabelFor(m => m.IsAgreeTerms, new { #class = "login-label" })
The full generated HTML is
<input data-val="true" data-val-required="Le champ IsAgreeTerms est requis." id="IsAgreeTerms" name="IsAgreeTerms" type="checkbox" value="true"
<input name="IsAgreeTerms" type="hidden" value="false">
<label class="login-label" for="IsAgreeTerms">IsAgreeTerms</label>
Here's a solution in the form of a html helper. It constructs a checkbox and label in the correct order:
public static IHtmlString CheckBoxWithLabelFor<TModel>(
this HtmlHelper<TModel> htmlHelper,
Expression<Func<TModel, bool>> expression,
string labelText,
object htmlAttributes = null
)
{
if (expression == null)
{
throw new ArgumentNullException(nameof(expression));
}
var checkBoxWithHidden = htmlHelper.CheckBoxFor(expression, htmlAttributes).ToHtmlString().Trim();
var pureCheckBox = checkBoxWithHidden.Substring(0, checkBoxWithHidden.IndexOf("<input", 1, StringComparison.Ordinal));
var labelHtml = htmlHelper.LabelFor(expression, labelText).ToHtmlString().Trim();
var result = pureCheckBox + Environment.NewLine + labelHtml + Environment.NewLine + $"<input type=\"hidden\" name=\"{ExpressionHelper.GetExpressionText(expression)}\" value=\"false\" />";
return new MvcHtmlString(result);
}
Is there other html generated by materialize.css? I think this happens because it is not possible apply a custom CSS to element input of type checkbox.
So, the checkbox becomes hidden and other html component represents visually the checkbox. Many components work like that.
UPDATE:
Why is Html checkbox generating an additional hidden input
OP here. Problem looks like more complex.
Actually, when using #Html.CheckBoxFor, MVC5 generates 3 fields, in that order :
Your input, with type="checkbox", binded to your model property
An hidden field (see the Claudio's link for an explaination)
Your label, generated by #Html.LabelFor
Problem is Materialize expects that in another order to work.
In your browser's console, just move the <label> element between the input and the hidden field, and everything works fine !
I found this very useful link, where, basically, it is said that the order of the generated fields by #Html.checkBoxFor will change ... In MVC6 !
As I'm working with MVC5, I use this very ugly solution in my _Layout:
$(":checkbox").each(function () {
$(this).nextAll("label").before($(this))
})
If anyone has a better idea, please feel free to post an elegant solution.
Here's something I thought might be a bit easier. Despite the specifics of the question, I'm interested in any method that will let me have a third form field auto-updated based on the content of two other fields with Polymer.dart.
Something like this, where the "[ ]" represent form fields.
Name: [given name] [family name]
Full name: [family_name, given_name]
So for example; if someone enters "John" and "Smith" in the first two fields. Then the "full name" line shows: [Smith, John], when either of the fields are updated.
I've based the following example on the classes and mark-up from the form Dart Polymer tutorial
Get Input from a Form tutorial
For a form like this ...
<polymer-element name="reference-form" extends="form" >
<template>
<style> ... </style>
<div id="slambookform" >
<div class="entry">
<label>Author:</label>
<input type="text" value="{{theData['authorGivenName']}}" >
<input type="text" value="{{theData['authorFamilyName']}}">
</div>
:
<div class="entry">
<label>Full name:</label>
<input disabled type="text" value="{{fullName}}" >
</div>
:
</div>
<template>
</polymer-element>
My initial attempt to make this happen was a function like:
#observable
String fullName(){
return theData['authorFamilyName'] +', '+ theData['authorGivenName'];
}
Which doesn't work. When I make 'fullName' to an #observable variable and update it with a button the form is updates as required. Hence my question, can I bind a third field to two (or more) others?
I think I will need some kind of event handler. For two fields, formatting on a change even is simple enough. I want to format several fields in the ultimate case, not just two fields.
While on this topic, is there a hook in dart-polymer or dart to supply a future or call-back? In my example, something like: 'after-change'. Just thinking out loud, something like that would be good.
Thanks in advance.
Along those lines (caution - code is not tested)
<polymer-element name="reference-form" extends="form" >
<template>
<style> ... </style>
<div id="slambookform" >
<div class="entry">
<label>Author:</label>
<input type="text" value="{{authorGivenName}}" >
<input type="text" value="{{authorFamilyName}}">
</div>
:
<div class="entry">
<label>Full name:</label>
<input disabled type="text" value="{{fullName}}" >
</div>
:
</div>
<template>
</polymer-element>
class reference_form.dart
String _authorGivenName;
#observable get authorGivenName => _authorGivenName;
set authorGivenName(String val) {
_authorGivenName = val;
notifyPropertyChange(#fullName, '${_authorGivenName} ${_authorFamilyName}',
'${val} ${_authorFamilyName}');
}
String _authorFamilyName;
#observable get authorFamilyName => _authorFamilyName;
set authorFamilyName(String val) {
_authorFamilyName = val;
notifyPropertyChange(#fullName, '${_authorGivenName} ${_authorFamilyName}',
'${_autorGivenName} ${val}');
}
#observable
String get fullName => '${_authorGivenName} ${_authorFamilyName}';
I have a workaround for this problem, standing on the shoulders of Günter Zöchbauer (comment above). My objective is to "bind" one field value to two in a read-only fashion. We are not quite there yet, however the pathway is educational in its own right.
Observer method
This solution is kind of a workaround for the objective I set myself. I've made some annotations on this code to explain what I saw, or why I think is happening.
The intention is for fullName to show both names in the form:
familyName, givenName; e.g.
Smith, John
reference-form.html:
<polymer-element name="reference-form" extends="form" >
<template>
<style> ... </style>
<div id="slambookform" >
<div class="entry">
<label>Author:</label>
<input type="text" value="{{theData['givenName']}}" >
<input type="text" value="{{familyName}}">
</div>
:
<div class="entry">
<label>Full name:</label>
<input disabled type="text" value="{{fullName}}" >
</div>
:
</div>
<template>
</polymer-element>
The code for the form properties, the things Polymer-dart binds to the HTML with the moustache syntax, "{{fullName}}". To keep things simple, I used just one 'notifier' field and this updates the fullName field from both familyName and givenName.
reference_form.dart:
//---- testing ----
String _familyName; // (1)
#observable // (2)
String get familyName => _familyName; // (3)
void set familyName( String nam ){ // (4)
_familyName = nam;
fullName = notifyPropertyChange( // (5)
#fullName,
"${fullName}",
"${nam}, ${theData['givenName']}" );
}
#observable
String fullName; // (6)
//---- end: testing ----
The private member, "_familyName", is a shadow for the public familyName property used in the template (snippet above).
Shadow (private) member, "_familyName", stores the data for the familyName pseudo property.
The next three lines declare an #observable property, familyName
Get familyName. Simply echo the value for the shadow variable.
Set familyName. Updates the shadow variable and the composite fullName property.
Note: the composit formatting could be done with two lines: _familyName = nam; fullName = nam; ... But we want to see all changes propagated see (#5).
The notifyPropertyChange() method updated all observers of the fullName property.
Note: I didn't hack around inside Polymer itself; inside the Observable class, fullName doesn't has no observers with the code shown.
Until I saw this, I assumed that the Polymer binding to the HTML template was via an observer (watcher), it would seem not. I may be mistaken. In any case, the call to notifyPropertyChange() for the '#fullName' symbol didn't change the results for this test case.
fullName property bound to the Polymer form.
Basically the {{fullName}} value will be updated every time there's a change to the familyName pseudo property.
Note on efficiency:
The familyName setter is called with every keystroke (observed while debugging). I understand that, and suggest it is not always really the best solution.
For me, I'd prefer to only call the setter when a user exits the field. However when I used onblur, the trigger was a blur of the form, not the field.
It seems that we might all benefit in terms of performance with a bit more insider information about these hooks, pathways and any options available to make things more efficient.
Comments and improvements welcome. This example is a workaround for me, so its definitely a work in progress. ;-)
Encapsulation method
I am evolving a solution closer to the original ambition and based on the 'observer method' above. This approach relies on the current, i.e. Dart v4, use of modules and libraries. I'll show the working code first and explain interesting stuff with notes.
reference_form.dart:
import 'package:exportable/exportable.dart'; // [1]
class _Data // [2]
extends Object with Exportable { // [3]
#export String publishDate; // [4]
#export String authorGivenName = '(given)';
#export String authorFamilyName = '(family)';
#export String authorUrl = '';
//--- attributes ---
String get fullName => "${authorFamilyName}, ${authorGivenName}"; // [5]
void set fullName( String nam ){ // [6]
//don't need this
}
//--- ctor ---
_Data(){
publishDate = new DateTime.now().toString(); // [7]
}
} //_Data
#CustomTag('reference-form')
class SlamBookComponent extends FormElement with Polymer, Observable {
SlamBookComponent.created() : super.created();
//---- testing ----
#observable
_Data data = new _Data(); // [8]
:
} //SlambookComponent
Notes:
Include Exportable mixin to convert to JSON. I'm not exporting 'fullName' because it is just formatting at the moment.
Add exportable to your pubspec.yaml and 'Run Pub get'.
The "_Data" class is private to the reference_form.dart module. I did a bit of testing of the scope rules because I do not want the internal data structure to leak, except for something catholic like JSON of course (small-c).
Bring-in the Exportable mixin.
I have tested Exportable, it implements exactly what I thought I'd have to write myself. Happy with this.
JSON is not a requirement of the original question; but I did want the (eventual) solution to be a first class artefact that can be serialized or saved is important in the majority of my use-cases.
This is a very good example of the facility to extend Dart quick and agile!
Use the #export modifier to identify fields specific to be interchanged as JSON.
Export the fullName attribute as a String (get).
There is no need for set operation. However Dart apparently insists that a Set method matches 'get'.
I am disappointed by this. I much prefer the idea that I can have READ-ONLY properties and attributes, e.g. like ruby.
As tested, Dart SDK v1.4.0; fails when a matching setter is not implemented/declared(??).
Use a constructor to set initial values for Date data attribute.
Declares an opaque public property called "data", as an (private) _Data instance.
The data formatting of key fields is encapsulated in the private _Data declaration.
The Exportable mixin interface is used to map the private class to a public JSON result.
Point #8 demonstrates a powerful aspect of dart, to enable an opaque implementation of objects and yet, you can 'deliver'/'share' details without specific internal details.
I have run this code and checked that the concepts work for hidden data (the _Data type) and opaque access and serialisation. Also you can't accidentally look at internal private type (accidentally, although explicit hacks may be possible). I don't apologise for accepting the C / C++ conscious responsibility paradigm -- I think this a the most powerful aspect of being a programmer; WE are responsible for effects/bugs stemming from the code we produce. I recommend testing 'bits of behaviour' in small mini-use-cases.
I put examples of the polymer markup; nothing surprising. For me this approach is less verbose and a bit more Object Oriented than the original (early) Dart tutorial
reference_form.html
<polymer-element name="reference-form" extends="form" >
<template>
<style> ... </style>
<div id="slambookform" >
<div class="entry">
<label>Author:</label>
<input type="text" value="{{data.authorGivenName}}" >
<input type="text" value="{{data.authorFamilyName}}">
</div>
<div class="entry">
<label>Published:</label>
<input type="date" value="{{data.publishDate}}">
</div>
</div>
</template>
<script type="application/dart" src="reference_form.dart"> </script>
</polymer-element>
In the Polymer mark-up can know (and has visibility over) internal field names. Why?
... Because the "reference_form.html" and "reference_form.dart" via Polymer-dart. It is quite nice really; although it seems that the ".dart" and ".html" components are closely coupled like ASP.NET and C#/VN.NET as (also) specified by convenience(??). I confess that's a completely different subject; there are things to resolve to keep things yar (yachting term).
Anyway for me, I feel the approach begun with the encapsulation shamble above is better suited to my needs for a small utility.
Polymer now supports this use case directly with #ObserveProperty
#observable String authorGivenName = '';
#observable String authorFamilyName = '';
#observable String get fullName => '${authorGivenName} ${authorFamilyName}';
#ObserveProperty('authorGivenName authorFamilyName')
void updateFullName(old) {
notifyPropertyChange(#fullName, old, fullName);
}
I have a model that contains a collection, such as this:
class MyModel
{
public List<MySubModel> SubModels { get; set; }
}
In the view, I want to dynamically add/remove from this list using Javascript before submitting. Right now I have this:
$("#new-submodel").click(function () {
var i = $("#submodels").children().size();
var html = '<div>\
<label for="SubModels[' + i + '].SomeProperty">SomeProperty</label>\
<input name="SubModels[' + i + '].SomeProperty" type="textbox" />\
</div>'
$("#submodels").append(html);
});
This works, but it's ugly. And, if I want to show those labels/textboxes for the existing items, there's no clean way to do that either (without duplicating).
I feel like I should be able to use Razor helpers or something to do this. Any ideas? Help me stay DRY.
You approach may lead to unexpected errors if you when you are removing or adding the divs. For example you have 4 items, you remove the first item, then $('#submodels').children().size() will return 3, but your last inserted div has the name attribute value set SubModels[3].SomeProperty which results in a conflict. And if your posted values contain SubModels[1] but not SubModels[0] the default model binder will fail to bind the list (it will bind it as null). I had to learn this the hard way...
To eliminate the aforementioned problem (and your's) I suggest you do something like this:
$("#addBtn").click(function() {
var html = '<div class="submodel">\
<label>SomeProperty</label>\
<input type="textbox" />\
</div>'; // you can convert this to a html helper!
$("#submodels").append(html);
refreshNames(); // trigger after html is inserted
});
$(refreshNames); // trigger on document ready, so the submodels generated by the server get inserted!
function refreshNames() {
$("#submodels").find(".submodel").each(function(i) {
$(this).find("label").attr('for', 'SubModels[' + i + '].SomeProperty');
$(this).find("label").attr('input', 'SubModels[' + i + '].SomeProperty');
});
}
Then your view (or even better an EditorTemplate for the SubModel type) can also generate code like:
<div class="submodel">
#Html.LabelFor(x => x.SomeProperty);
#Html.EditorFor(x => x.SomeProperty);
</div>
It would also be possible to convert the code generation to a html helper class, and use it in the EditorTemplate and in the JavaScript code
I would recommend you going through the following blog post.
I'm just now starting to learn ASP.NET MVC. How would I go about creating a reusable tri-state checbox? In WebForms this would be a control, but I don't know the MVC equivalent.
Add a TriStateCheckBox (or TriStateCheckBoxFor if you use the strongly typed overloads) extension method to HtmlHelper and add the namespace of that extension method class to the namespaces section of your web.config.
As for the implementation, I'd recommend having at look at the InputExtensions source on codeplex and using that to create your own.
Limitations:
View Rendering - When rendering HTML content, there is no attribute you can possibly place on an <input type="checkbox" /> that will give it the property indeterminate.
At some point, you'll have to use JavaScript to grab the element and set the indeterminate property:
// vanilla js
document.getElementById("myChk").indeterminate = true;
// jQuery
$("#myCheck).prop("indeterminate", true);
Form Data - model binding will always be limited to what values are actually sent in the request, either from the url or the data payload (on a POST).
In this simplified example, both unchecked and indeterminate checkboxes are treated identically:
And you can confirm that for yourself in this Stack Snippet:
label {
display: block;
margin-bottom: 3px;
}
<form action="#" method="post">
<label >
<input type="checkbox" name="chkEmpty">
Checkbox
</label>
<label >
<input type="checkbox" name="chkChecked" checked>
Checkbox with Checked
</label>
<label >
<input type="checkbox" name="chkIndeterminate" id="chkIndeterminate">
<script> document.getElementById("chkIndeterminate").indeterminate = true; </script>
Checkbox with Indeterminate
</label>
<label >
<input name="RegularBool" type="checkbox" value="true">
<input name="RegularBool" type="hidden" value="false">
RegularBool
</label>
<input type="submit" value="submit"/>
</form>
Model Binding - Further, model binding will only occur on properties that are actually sent. This actually poses a problem even for regular checkboxes, since they won't post a value when unchecked. Value types do always have a default value, however, if that's the only property in your model, MVC won't new up an entire class if it doesn't see any properties.
ASP.NET solves this problem by emitting two inputs per checkbox:
Note: The hidden input guarantees that a 'false' value will be sent even when the checkbox is not checked. When the checkbox is checked, HTTP is allowed to submit multiple values with the same name, but ASP.NET MVC will only take the first instance, so it will return true like we'd expect.
Render Only Solution
We can render a checkbox for a nullable boolean, however this really only works to guarantee a bool by converting null → false when rendering. It is still difficult to share the indeterminate state across server and client. If you don't need to ever post back indeterminate, this is probably the cleanest / easiest implementation.
Roundtrip Solution
As there are serious limitations to using a HTML checkbox to capture and post all 3 visible states, let's separate out the view of the control (checkbox) with the tri-state values that we want to persist, and then keep them synchronized via JavsScript. Since we already need JS anyway, this isn't really increasing our dependency chain.
Start with an Enum that will hold our value:
/// <summary> Specifies the state of a control, such as a check box, that can be checked, unchecked, or set to an indeterminate state.</summary>
/// <remarks> Adapted from System.Windows.Forms.CheckState, but duplicated to remove dependency on Forms.dll</remarks>
public enum CheckState
{
Checked,
Indeterminate,
Unchecked
}
Then add the following property to your Model instead of a boolean:
public CheckState OpenTasks { get; set; }
Then create an EditorTemplate for the property that will render the actual property we want to persist inside of a hidden input PLUS a checkbox control that we'll use to update that property
Views/Shared/EditorTemplates/CheckState.cshtml:
#model CheckState
#Html.HiddenFor(model => model, new { #class = "tri-state-hidden" })
#Html.CheckBox(name: "",
isChecked: (Model == CheckState.Checked),
htmlAttributes: new { #class = "tri-state-box" })
Note: We're using the same hack as ASP.NET MVC to submit two fields with the same name, and placing the HiddenFor value that we want to persist first so it wins. This just makes it easy to traverse the DOM and find the corresponding value, but you could use different names to prevent any possible overlap.
Then, in your view, you can render both the property + checkbox using the editor template the same way you would have used a checkbox, since it renders both. So just add this to your view:
#Html.EditorFor(model => model.OpenTasks)
The finally piece is to keep them synchronized via JavaScript on load and whenever the checkbox changes like this:
// on load, set indeterminate
$(".tri-state-hidden").each(function() {
var isIndeterminate = this.value === "#CheckState.Indeterminate";
if (isIndeterminate) {
var $box = $(".tri-state-box[name='" + this.name + "'][type='checkbox']");
$box.prop("indeterminate", true);
}
});
// on change, keep synchronized
$(".tri-state-box").change(function () {
var newValue = this.indeterminate ? "#CheckState.Indeterminate"
: this.checked ? "#CheckState.Checked"
: "#CheckState.Unchecked";
var $hidden = $(".tri-state-hidden[name='" + this.name + "'][type='hidden']");
$hidden.val(newValue);
});
Then you can use however you'd like in your business model. For example, if you wanted to map to a nullable boolean, you could use the CheckState property as a backing value and expose/modify via getters/setters in a bool? like this:
public bool? OpenTasksBool
{
get
{
if (OpenTasks == CheckState.Indeterminate) return null;
return OpenTasks == CheckState.Checked;
}
set
{
switch (value)
{
case null: OpenTasks = CheckState.Indeterminate; break;
case true: OpenTasks = CheckState.Checked; break;
case false: OpenTasks = CheckState.Unchecked; break;
}
}
}
Alternative Solution
Also, depending on your domain model, you could just use Yes, No, ⁿ/ₐ radio buttons
ASP.NET MVC certainly doesn't provide such component, actually it simply relies on the standard elements available in HTML but you may want to check out this solution.