The life's blood of any application is form input; this is the most effective way to gather significant information from the user. Whether it's a search form, a login screen or a multi-page registration wizard, forms are how the user really expresses themselves to the application.

Tapestry excels at creating forms and validating input. Input validation is declarative, meaning you simply tell Tapestry what validations to apply to a given field, and it takes care of it on the server and (once implemented) on the client as well.

Finally, Tapestry is able to not only present the errors back to the user, but to decorate the fields and the labels for the fields, marking them as containing errors (primarily, using CSS effects).

Form component

The core of Tapestry's form support is the Form component. The Form component encloses (wraps around) all the other field components such as TextField, TextArea, Checkbox, etc.

The Form component generates a number of component events that you may provide event handler methods for.

When rendering, the Form component emits two notifications: first, "prepareForRender", then "prepare". These allow the Form's container to setup any fields or properties that will be referenced in the form. For example, this is a good place to create a temporary entity object to be rendered, or to load an entity from a database to be edited.

When user submits the form on the client, a series of steps occur on the server.

First, the Form emits a "prepareForSubmit" notification, then a "prepare" notification. These allow the container to ensure that objects are set up and ready to receive information from the form submission.

Next, all the fields inside the form are activated to pull values out of the incoming request, validate them and (if valid) store the changes.

For Tapestry 4 Users: Tapestry 5 does not use the fragile "form rewind" approach from Tapestry 4. Instead, a hidden field generated during the render stores the information needed to process the form submission.

After the fields have done their processing, the Form emits a "validateForm" event. This is a chance to perform cross-form validation that can't be described declaratively.

Next, the Form determines if there have been any validation errors. If there have been, then the submission is considered a failure, and a "failure" event is emitted. If there have been no validation errors, then a "success" event is emitted.

Last, the Form emits a "submit" event (for logic that doesn't care about success or failure).

Tracking Validation Errors

Associated with the Form is an ValidationTracker that tracks all the provided user input and validation errors for every field in the form. The tracker can be provided to the Form via the Form's tracker parameter, but this is rarely necessary.

The Form includes methods isValid() and getHasErrors(), which are used to see if the Form's validation tracker contains any errors.

In your own logic, it is possible to record your own errors. Form includes two different versions of method recordError(), one of which specifies a Field (an interface implemented by all form element components), and one of which is for "global" errors, unassociated with any particular field.

Storing Data Between Requests

As with other action requests, the result of a form submission is to send a redirect to the client which re-renders the page. The ValidationTracker must be stored persistently between requests, or all the validation information will be lost (the default ValidationTracker provided by the Form is persistent).

Likewise, the individual fields updated by the components should also be persistent.

For example, a Login page, which collects a user name and a password, might look like:

public class Login
{
    @Persist
    private String userName;

    private String password;

    @Inject
    private UserAuthenticator authenticator;

    @Component(id = "password")
    private PasswordField passwordField;

    @Component
    private Form form;

    String onSuccess()
    {
        if (!authenticator.isValid(userName, password))
        {
            form.recordError(passwordField, "Invalid user name or password.");
            return null;
        }

        return "PostLogin";
    }

    public String getPassword()
    {
        return password;
    }

    public void setPassword(String password)
    {
        password = password;
    }

    public String getUserName()
    {
        return userName;
    }

    public void setUserName(String userName)
    {
        userName = userName;
    }
}

Because of the the fact that a form submission is two requests (the submission itself, then a re-render of the page), it is necessary to make the value stored in the userName field persist between the two requests. This would be necessary for the password field as well, except that the PasswordField component never renders a value.

Note that the onSuccess() method is not public; event handler methods can have any visibility, even private. Package private (that is, no modifier) is the typical use, as it allows the component to be tested, from a test case class in the same package.

The Form only emits a "success" event if the there are no prior validation errors. This means it is not necessary to write if (form.getHasErrors()) return; as the first line of the method.

Finally, notice how business logic fits into validation. The UserAuthenticator service is responsible for ensuring that the userName and (plaintext) password are valid. When it returns false, we ask the Form component to record an error. We provide the PasswordField instance as the first parameter; this ensures that the password field, and its label, are decorated when the Form is re-rendered, to present the errors to the user.

Configuring Fields and Labels

The template for a page contains a minimal amount of Tapestry instrumentation:

<html xmlns:t="http://tapestry.apache.org/schema/tapestry_5_1_0.xsd">
    <head>
        <title>Login</title>
    </head>
    <body>
        <h1>Please Login</h1>

        <form t:type="form" t:id="form">

            <t:errors/>

            <t:label for="userName"/>:
            <input t:type="TextField" t:id="userName" t:validate="required,minlength=3" size="30"/>
            <br/>
            <t:label for="password"/>:
            <input t:type="PasswordField" t:id="password" t:validate="required,minlength=3" size="30"/>
            <br/>
            <input type="submit" value="Login"/>
        </form>
    </body>
</html>

The Tapestry Form component is responsible for creating the necessary URL for the form submission (this is Tapestry's responsibility, not yours).

The Errors component must be placed inside a Form, it outputs all of the errors for all the fields within the Form as a single list. It uses some simple styling to make the result more presentable.

Each field component, such as the TextField, is paired with a Label component. The Label will render out a <label> element connected to the field. This is very important for usability, especially for users with visual disabilities. It also means you can click on the label text to move the cursor to the corresponding field.

The for parameter of the Label is the id of a component.

For the TextField, we provide a component id, userName. We could specify the value parameter, but the default is to match the TextField's id against a property of the container, the Login page, if such a property exists.

As a rule of thumb, you should always give your fields a specific id (this id will be used to generate the name and id attributes of the rendered tag). Being allowed to omit the value parameter helps to keep the template from getting too cluttered.

The validate parameter identifies what validations should occur for the field. This is a list of validator names. Validators are configured within Tapestry, and the list of available validators is extensible. "required" is a name of one of the built-in validators, that ensures that the submitted value is not the empty string. Likewise, "minlen" ensures that the value has the specified minimum length.

The validate parameter was placed within the Tapestry namespace using the t: prefix. This is not strictly necessary, as the template is well formed either way. However, putting the Tapestry specific values into the Tapestry namespace ensures that the template will itself be valid.

Errors and Decorations

Note: This section has not been updated to reflect the introduction of client-side input validation.

When you first activate the Login page, the fields and forms will render normally, awaiting input:

Notice how the Label components are displaying the textual names for the fields. Given that we have not done any explicit configuration, what's happened is that the component's ids ("userName" and "password") have been converted to "User Name" and "Password".

If you just submit the form as is, the fields will violate the "required" constraint and the page will be redisplayed to present those errors to the user:

There's a couple of subtle things going on here. First, Tapestry tracks all the errors for all the fields. The Errors component has displayed them at the top of the form. Further, the default validation decorator has added decorations to the labels and the fields, adding "t-error" to the CSS class for the fields and labels. Tapestry provides a default CSS stylesheet that combines with the "t-error" class to make things turn red.

Next, we'll fill in the user name but not provide enough characters for password.

The user name field is OK, but there's an error on just the password field. The PasswordField component always displays a blank value by default, otherwise we'd see the partial password displayed inside.

If you type in enough characters and submit, we see how the logic inside the Login page can attach errors to fields:

This is nice and seamless; the same look and feel and behavior for both the built-in validators, and for errors generated based on application logic.

Centralizing Validation with @Validate

The @Validate annotation can take the place of the validate parameter of TextField, PasswordField, TextArea and other components. When the validate parameter is not bound, the component will check for the @Validate annotation and use its value as the validation definition.

The annotation may be placed on the getter or setter method, or on the field itself.

Customizing Validation Messages

Each validator (such as "required" or "minlength") has a default message used (on the client side and the server side) when the constraint is violated; that is, when the user input is not valid.

The message can be customized by adding an entry to the page's message catalog (or the containing component's message catalog). As with any localized property, this can also go into the application's message catalog.

The first key checked is formId-fieldId-validatorName-message.

• formId: the local component id of the Form component
• fieldId: the local component id of the field (TextField, etc.)
• validatorName: the name of the validator, i.e., "required" or "minlength"
  If there is not message for that key, a second check is made, for fieldId-validatorName-message.

If that does not match a message, then the built-in default validation message is used.

Customizing Validation Messages / BeanEditForm

The BeanEditForm component also supports validation message customizing. The search is the similar; the formId is the component id of the BeanEditForm component (not the Form component it contains). The fieldId is the property name.

Configuring Validator Contraints in the Message Catalog

It is possible to omit the validation constraint from the validator parameter (or @Validator annotation), in which case it is expected to be stored in the message catalog.

This is useful when the validation constraint is awkward to enter inline, such as a regular expression for use with the regexp validator.

The key here is similar to customizing the validation message: formId-fieldId-validatorName or just fieldId-validatorName.

For example, your template may have the following:

  <t:textfield t:id="ssn" validate="required,regexp"/>

And your message catalog can contain:

ssn-regexp=\d{3}-\d{2}-\d{4}
ssn-regexp-message=Social security numbers are in the format 12-34-5678.

This technique also works with the BeanEditForm; as with validation messages, the formId is the BeanEditForm component's id, and the fieldId is the name of the property being editted.

Available Validators

Currently Tapestry provides the following built-in validators:

Validation Macros

Since Tapestry 5.2, we can create validation macros, wich will contain a list of validators. This mechanism is very useful for combining your validators. We just have to contribute to the ValidatorMacro Service in your AppModule, by adding a new entry to the configuration object. The first parameter is the name of your macro, the second is a comma-separated list of validators.

@Contribute(ValidatorMacro.class)
public static void combinePasswordValidators(MappedConfiguration<String, String> configuration) {
      configuration.add("password","required,minlength=5,maxlength=15,");
}

Then, we can use this new macro in our Template or Java Class.

<input t:type="textField" t:id="password" t:validate="password" />

@Validate("password")
private String password;

Overriding the Translator with Events

The TextField, PasswordField and TextArea components all have a translate parameter, a FieldTranslator object that is used to convert values on the server side to strings on the client side.

In most cases, the translate parameter is not set explicitly; Tapestry derives an appropriate value based on the type of property being editted by the field.

In certain cases, you may want to override the translator. This can be accomplished using two events triggered on the component, "toclient" and "parseclient".

The "toclient" event is passed the current object value and returns a string, which will be the default value for the field. When there is no event handler, or when the event handler returns null, the default Translator is used to convert the server side value to a string.

For example, you may have a quantity field that you wish to display as blank, rather than zero, initially:

  <t:textfield t:id="quantity" size="10"/>

  . . .

  private int quantity;

  String onToClientFromQuantity()
  {
    if (quantity == 0) return "";

    return null;
  }

This is good so far, but if the field is optional and the user submits the form, you'll get a validation error, because the empty string is not valid as an integer.

That's where the "parseclient" event comes in:

  Object onParseClientFromQuantity(String input)
  {
    if ("".equals(input)) return 0;

    return null;
  }

The event handler method has precedence over the translator. Here it checks for the empty string (and note that the input may be null!) and evaluates that as zero.

Again, returning null lets the normal translator do its work.

The event handler may also throw a ValidationException to indicate a value that can't be parsed.

Now, what if you want to perform your own custom validation? That's another event: "validate":

  void onValidateFromCount(Integer value) throws ValidationException
  {
    if (value.equals(13)) throw new ValidationException("Thirteen is an unlucky number.");
  }

This event gets fired after the normal validators. It is passed the parsed value (not the string from the client, but the object value from the translator, or from the "parseclient" event handler).

The method may not return a value, but may throw a ValidationException to indicate a problem with the value.

Caution: These events are exclusively on the server side. This means that, in certain circumstances, an input value will be rejected on the client side even though it is valid on the server side. You may need to disable client-side validation in order to use this feature.