2.1. Dialects

If you’ve read the Using Thymeleaf tutorial before getting here —which you should have done— you should know that what you’ve been learning all this time was not exactly Thymeleaf, but rather its Standard Dialect (or the SpringStandard Dialect, if you’ve also read the Thymeleaf + Spring tutorial).

What does that mean? It means that all those th:x attributes you learned to use are only a standard, out-of-the-box set of features, but you can define your own set of attributes (or tags) with the names you wish and use them in Thymeleaf to process your templates. You can define your own dialects.

Dialects are objects implementing the org.thymeleaf.dialect.IDialect interface, which cannot be any simpler:

public interface IDialect {

    public String getName();

}

The only core requirement of a dialect is to have a name that can be used for its identification. But this alone is of little use, so dialects will normally implement one or several subinterfaces of IDialect, depending on what they provide to the Thymeleaf engine:

• IProcessorDialect for dialects that provide processors.
• IPreProcessorDialect for dialects that provide pre-processors.
• IPostProcessorDialect for dialects that provide post-processors.
• IExpressionObjectDialect for dialects that provide expression objects.
• IExecutionAttributeDialect for dialects that provide execution attributes.

public interface IProcessorDialect extends IDialect {

    public String getPrefix();
    public int getDialectProcessorPrecedence();
    public Set<IProcessor> getProcessors(final String dialectPrefix);

}

public interface IPreProcessorDialect extends IDialect {

    public int getDialectPreProcessorPrecedence();
    public Set<IPreProcessor> getPreProcessors();

}

public interface IPostProcessorDialect extends IDialect {

    public int getDialectPostProcessorPrecedence();
    public Set<IPostProcessor> getPostProcessors();

}

public interface IExpressionObjectDialect extends IDialect {

    public IExpressionObjectFactory getExpressionObjectFactory();

}

public interface IExpressionObjectFactory {

    public Map<String,ExpressionObjectDefinition> getObjectDefinitions();

    public Object buildObject(final IProcessingContext processingContext, final String expressionObjectName);

}

public interface IExecutionAttributeDialect extends IDialect {

    public Map<String,Object> getExecutionAttributes();

}

2.2. Processors

Processors are objects implementing the org.thymeleaf.processor.IProcessor interface, and they contain the real logic to be applied on the different parts of a template (which we will represent as events, given Thymeleaf is an event-based engine). This interface looks like this:

public interface IProcessor {

    public TemplateMode getTemplateMode();
    public int getPrecedence();

}

As with dialects, this is a very simple interface that only specified the template mode in which the processor can be applied and its precedence.

But there are several types of processor, one for each possible type of event:

• Template start/end
• Element Tags
• Texts
• Comments
• CDATA Sections
• DOCTYPE Clauses
• XML Declarations
• Processing Instructions

And also for models: sequences of events representing an entire element, i.e. an element with its entire body, including any nested elements or any other kind of artifacts that might appear inside. If the modelled element is a standalone element, the model will only contain its corresponding event; but if the modelled element has a body, the model will contain every event from its open tag to its close tag, both included.

All these types of processors are created by implementing a specific interface, or by extending one of the available abstract implementations. All these artifacts conforming the Thymeleaf 3.0 Processor API live at the org.thymeleaf.processor package.

public interface IElementProcessor extends IProcessor {

    public MatchingElementName getMatchingElementName();
    public MatchingAttributeName getMatchingAttributeName();

}

public interface IElementTagProcessor extends IElementProcessor {

    public void process(
            final ITemplateContext context, 
            final IProcessableElementTag tag,
            final IElementTagStructureHandler structureHandler);

}

// Obtain the attribute value
String attributeValue = tag.getAttributeValue(attributeName);

// Unescape the attribute value
attributeValue = 
    EscapedAttributeUtils.unescapeAttribute(context.getTemplateMode(), attributeValue);

// Instruct the structureHandler to remove the attribute from the tag
structureHandler.removeAttribute(attributeName);

... // do something with that attributeValue

public interface IElementTagStructureHandler {

    public void reset();

    public void setLocalVariable(final String name, final Object value);
    public void removeLocalVariable(final String name);

    public void setAttribute(final String attributeName, final String attributeValue);
    public void setAttribute(final String attributeName, final String attributeValue, 
                             final AttributeValueQuotes attributeValueQuotes);

    public void replaceAttribute(final AttributeName oldAttributeName, 
                                 final String attributeName, final String attributeValue);
    public void replaceAttribute(final AttributeName oldAttributeName, 
                                 final String attributeName, final String attributeValue, 
                                 final AttributeValueQuotes attributeValueQuotes);

    public void removeAttribute(final String attributeName);
    public void removeAttribute(final String prefix, final String name);
    public void removeAttribute(final AttributeName attributeName);

    public void setSelectionTarget(final Object selectionTarget);

    public void setInliner(final IInliner inliner);

    public void setTemplateData(final TemplateData templateData);

    public void setBody(final String text, final boolean processable);
    public void setBody(final IModel model, final boolean processable);

    public void insertBefore(final IModel model); // cannot be processable
    public void insertImmediatelyAfter(final IModel model, final boolean processable);

    public void replaceWith(final String text, final boolean processable);
    public void replaceWith(final IModel model, final boolean processable);


    public void removeElement();
    public void removeTags();
    public void removeBody();
    public void removeAllButFirstChild();

    public void iterateElement(final String iterVariableName, 
                               final String iterStatusVariableName, 
                               final Object iteratedObject);

}

public interface IElementModelProcessor extends IElementProcessor {

    public void process(
            final ITemplateContext context, 
            final IModel model,
            final IElementModelStructureHandler structureHandler);

}

final IModelFactory modelFactory =  context.getModelFactory();

int n = model.size();
while (n-- != 0) {
    final ITemplateEvent event = model.get(n);
    if (event instanceof IText) {
        final IComment comment =
                modelFactory.createComment(((IText)event).getText());
        model.insert(n, comment);
        model.remove(n + 1);
    }
}

public interface IElementModelStructureHandler {

    public void reset();

    public void setLocalVariable(final String name, final Object value);
    public void removeLocalVariable(final String name);

    public void setSelectionTarget(final Object selectionTarget);

    public void setInliner(final IInliner inliner);

    public void setTemplateData(final TemplateData templateData);

}

public interface ITemplateBoundariesProcessor extends IProcessor {

    public void processTemplateStart(
            final ITemplateContext context,
            final ITemplateStart templateStart,
            final ITemplateBoundariesStructureHandler structureHandler);

    public void processTemplateEnd(
            final ITemplateContext context,
            final ITemplateEnd templateEnd,
            final ITemplateBoundariesStructureHandler structureHandler);

}

public interface ITemplateBoundariesStructureHandler {

    public void reset();

    public void setLocalVariable(final String name, final Object value);
    public void removeLocalVariable(final String name);

    public void setSelectionTarget(final Object selectionTarget);

    public void setInliner(final IInliner inliner);

    public void insert(final String text, final boolean processable);
    public void insert(final IModel model, final boolean processable);

}

public interface ITextProcessor extends IProcessor {

    public void process(
            final ITemplateContext context, 
            final IText text,
            final ITextStructureHandler structureHandler);

}

public interface ITextStructureHandler {

    public void reset();

    public void setText(final CharSequence text);

    public void replaceWith(final IModel model, final boolean processable);

    public void removeText();

}

3.1. extraThyme: a website for Thymeland’s football league

Football is a popular sport in Thymeland¹. There is a 10-team league going on there each season, and its organizers have just asked us to create a website for it called “extraThyme”.

This website will be very simple: just a table with:

• The team names.
• How many matches they won, drew or lost, as well as the total points earned.
• A remark explaining whether their position in the table qualifies them for higher-level competitions next year or else mean their relegation to regional leagues.

Above the league table, a banner will be displaying headlines with the results of recent matches, and also there will be a clearly visible banner that warns users every Sunday that Sundays are match days, and therefore they should be going to the stadium instead of browsing the internet.

We will use HTML5, Spring MVC and the SpringStandard dialect for our application, and we will be extending Thymeleaf by creating a score dialect that includes:

• A score:remarkforposition attribute that outputs internationalized text for the Remarks column in the table. This text should explain whether the team’s position in the table qualifies it for the World Champions League, the Continental Play-Offs, or relegates it to the Regional League.
• A score:classforposition attribute that establishes a CSS class for the table rows depending on the team’s remarks: blue background for the World Champions League, green for the Continental Play-Offs, and red for relegation.
• A score:headlines tag for drawing the yellow box at the top with the results of recent matches. This tag should support an order attribute with values random (for showing a randomly selected match) and latest (default, for showing only the last match).
• A score:match-day-today attribute that can be added to the league table header in order to (conditionally, if it is Sunday) add a banner warning the user that today is a match day.

Our markup will therefore look like this, making use of both th and score attributes:

<!DOCTYPE html>

<!--/* Note the xmlns:* here are completely optional and only meant to     */-->
<!--/* avoid IDEs from complaining about tags/attributes they may not know */-->
<html xmlns:th="http://www.thymeleaf.org" xmlns:score="http://thymeleafexamples">

  <head>
    <title>extraThyme: Thymeland's football website</title>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
    <link rel="stylesheet" type="text/css" media="all"
          href="../../css/extrathyme.css" th:href="@{/css/extrathyme.css}"/>
  </head>

  <body>

    <div>
      <img src="../../images/extrathymelogo.png" 
           alt="extraThyme logo" title="extraThyme logo"
           th:src="@{/images/extrathymelogo.png}" th:alt-title="#{title.application}"/>
    </div>

    <score:headlines order="random" />

    <div class="leaguetable">

      <h2 score:match-day-today th:text="#{title.leaguetable(${execInfo.now.time})}">
        League table for 07 July 2011
      </h2>
      
      <table>
        <thead>
          <tr>
            <th th:text="#{team.name}">Team</th>
            <th th:text="#{team.won}" class="matches">Won</th>
            <th th:text="#{team.drawn}" class="matches">Drawn</th>
            <th th:text="#{team.lost}" class="matches">Lost</th>
            <th th:text="#{team.points}" class="points">Points</th>
            <th th:text="#{team.remarks}">Remarks</th>
          </tr>
        </thead>
        <tbody>
          <tr th:each="t : ${teams}" score:classforposition="${tStat.count}">
            <td th:text="|${t.name} (${t.code})|">The Winners (TWN)</td>
            <td th:text="${t.won}" class="matches">1</td>
            <td th:text="${t.drawn}" class="matches">0</td>
            <td th:text="${t.lost}" class="matches">0</td>
            <td th:text="${t.points}" class="points">3</td>
            <td score:remarkforposition="${tStat.count}">Great winner!</td>
          </tr>
          <!--/*-->
          <tr>
            <td>The First Losers (TFL)</td>
            <td class="matches">0</td>
            <td class="matches">1</td>
            <td class="matches">0</td>
            <td class="points">1</td>
            <td>Little loser!</td>
          </tr>
          <tr>
            <td>The Last Losers (TLL)</td>
            <td class="matches">0</td>
            <td class="matches">0</td>
            <td class="matches">1</td>
            <td class="points">0</td>
            <td>Big loooooser</td>
          </tr>
          <!--*/-->
        </tbody>
      </table>

    </div>

  </body>

</html>

(Note that we’ve added a second and third rows to our table, surrounded by parser-level comments <!--/* ... */--> so that our template shows nicely as a prototype when directly opened in a browser.)

3.2. Changing the CSS class by team position

The first attribute processor we will develop will be ClassForPositionAttributeTagProcessor, which we will implement as a subclass of a convenience abstract class provided by Thymeleaf called AbstractAttributeTagProcessor.

This abstract class is the base for all tag processors (i.e. processors that act on tag events and not models) which match (i.e. are selected for execution) based on the existence of a specific attribute in such tag. In this case, score:classforposition.

The idea is that we will use this processor for setting a new value to the class attribute of the tag score:classforposition is living in.

Let’s have a look at our code:

public class ClassForPositionAttributeTagProcessor extends AbstractAttributeTagProcessor {

    private static final String ATTR_NAME = "classforposition";
    private static final int PRECEDENCE = 10000;


    public ClassForPositionAttributeTagProcessor(final String dialectPrefix) {
        super(
            TemplateMode.HTML, // This processor will apply only to HTML mode
            dialectPrefix,     // Prefix to be applied to name for matching
            null,              // No tag name: match any tag name
            false,             // No prefix to be applied to tag name
            ATTR_NAME,         // Name of the attribute that will be matched
            true,              // Apply dialect prefix to attribute name
            PRECEDENCE,        // Precedence (inside dialect's own precedence)
            true);             // Remove the matched attribute afterwards
    }


    @Override
    protected void doProcess(
            final ITemplateContext context, final IProcessableElementTag tag,
            final AttributeName attributeName, final String attributeValue,
            final IElementTagStructureHandler structureHandler) {

        final IEngineConfiguration configuration = context.getConfiguration();

        /*
         * Obtain the Thymeleaf Standard Expression parser
         */
        final IStandardExpressionParser parser =
                StandardExpressions.getExpressionParser(configuration);

        /*
         * Parse the attribute value as a Thymeleaf Standard Expression
         */
        final IStandardExpression expression = parser.parseExpression(context, attributeValue);

        /*
         * Execute the expression just parsed
         */
        final Integer position = (Integer) expression.execute(context);

        /*
         * Obtain the remark corresponding to this position in the league table.
         */
        final Remark remark = RemarkUtil.getRemarkForPosition(position);

        /*
         * Select the adequate CSS class for the element.
         */
        final String newValue;
        if (remark == Remark.WORLD_CHAMPIONS_LEAGUE) {
            newValue = "wcl";
        } else if (remark == Remark.CONTINENTAL_PLAYOFFS) {
            newValue = "cpo";
        } else if (remark == Remark.RELEGATION) {
            newValue = "rel";
        } else {
            newValue = null;
        }

        /*
         * Set the new value into the 'class' attribute (maybe appending to an existing value)
         */
        if (newValue != null) {
            String currentClass = tag.getAttribute("class").getValue();
            if (currentClass != null) {
                structureHandler.setAttribute("class", currentClass + " " + newValue);
            } else {
                structureHandler.setAttribute("class", newValue);
            }
        }

    }

}

The basic logic flow is easy to see and understand: get the value of the attribute, use it for computing what we need, and finally use the structureHandler to instruct the engine about the modifications needed as a result.

It is important to note that we are creating this attribute with the ability of executing expressions written in the Standard Syntax (used by both the Standard and the SpringStandard dialects). This is, the ability to be set values like ${var}, #{messageKey}, conditionals, etc. See how we use this in our template:

<tr th:each="t : ${teams}" score:classforposition="${tStat.count}">

In order to evaluate these expressions (also called Thymeleaf Standard Expressions) we need to first obtain the Standard Expression Parser, then parse the attribute value, and finally execute the parsed expresion:

final IStandardExpressionParser parser =
        StandardExpressions.getExpressionParser(configuration);

final IStandardExpression expression = parser.parseExpression(context, attributeValue);

final Integer position = (Integer) expression.execute(context);

It’s also interesting the way we use the structureHandler to add a new attribute to the host tag (remember the tag object is immutable):

if (newValue != null) {
    String currentClass = tag.getAttribute("class").getValue();
    if (currentClass != null) {
        structureHandler.setAttribute("class", currentClass + " " + newValue);
    } else {
        structureHandler.setAttribute("class", newValue);
    }
}

Last, note that HTML-escaping texts and attribute is our responsibility, but in this case we know all the possible values of the newValue variable and they require no escaping, so for the sake of simplicity we are skipping that operation.

3.3. Displaying an internationalized remark

The next thing to do is creating an attribute processor able to display the remark text. This will be very similar to the ClassForPositionAttrProcessor, but with a couple of important differences:

• We will not be setting a value for an attribute in the host tag, but rather the text body (content) of the tag, in the same way a th:text attribute does.
• We need to access the message externalization (internationalization) system from our code so that we can display the text corresponding to the selected locale.

We will be using the same AbstractAttributeTagProcessor base class. And this will be our code:

public class RemarkForPositionAttributeTagProcessor extends AbstractAttributeTagProcessor {

    private static final String ATTR_NAME = "remarkforposition";
    private static final int PRECEDENCE = 12000;


    public RemarkForPositionAttributeTagProcessor(final String dialectPrefix) {
        super(
            TemplateMode.HTML, // This processor will apply only to HTML mode
            dialectPrefix,     // Prefix to be applied to name for matching
            null,              // No tag name: match any tag name
            false,             // No prefix to be applied to tag name
            ATTR_NAME,         // Name of the attribute that will be matched
            true,              // Apply dialect prefix to attribute name
            PRECEDENCE,        // Precedence (inside dialect's precedence)
            true);             // Remove the matched attribute afterwards
    }


    @Override
    protected void doProcess(
            final ITemplateContext context, final IProcessableElementTag tag,
            final AttributeName attributeName, final String attributeValue,
            final IElementTagStructureHandler structureHandler) {

        final IEngineConfiguration configuration = context.getConfiguration();

        /*
         * Obtain the Thymeleaf Standard Expression parser
         */
        final IStandardExpressionParser parser =
                StandardExpressions.getExpressionParser(configuration);

        /*
         * Parse the attribute value as a Thymeleaf Standard Expression
         */
        final IStandardExpression expression =
                parser.parseExpression(context, attributeValue);

        /*
         * Execute the expression just parsed
         */
        final Integer position = (Integer) expression.execute(context);

        /*
         * Obtain the remark corresponding to this position in the league table
         */
        final Remark remark = RemarkUtil.getRemarkForPosition(position);
        
        /*
         * If no remark is to be applied, just set an empty body to this tag
         */
        if (remark == null) {
            structureHandler.setBody("", false); // false == 'non-processable'
            return;
        }
        
        /*
         * Message should be internationalized, so we ask the engine to resolve
         * the message 'remarks.{REMARK}' (e.g. 'remarks.RELEGATION'). No
         * parameters are needed for this message.
         *
         * Also, we will specify to "use absent representation" so that, if this
         * message entry didn't exist in our resource bundles, an absent-message
         * label will be shown.
         */
        final String i18nMessage =
                context.getMessage(
                        RemarkForPositionAttributeTagProcessor.class, 
                        "remarks." + remark.toString(), 
                        new Object[0], 
                        true);

        /*
         * Set the computed message as the body of the tag, HTML-escaped and
         * non-processable (hence the 'false' argument)
         */
        structureHandler.setBody(HtmlEscape.escapeHtml5(i18nMessage), false);
        
    }

}

final String i18nMessage =
        context.getMessage(
                RemarkForPositionAttributeTagProcessor.class, 
                "remarks." + remark.toString(), 
                new Object[0], 
                true);

structureHandler.setBody(HtmlEscape.escapeHtml5(i18nMessage), false);

3.4. An element processor for our headlines

The third processor we will write is an element (tag) processor. Note we call this an element tag processor in contrast with the two previous processors, which were attribute tag processors. The reason is, in this case we want our processor to match (i.e. to be selected for execution) based on the name of the tag, not on the name of one of its attributes.

This kind of tag processor has one advantage and also one disadvantage with respect to attribute tag processors:

• Advantage: elements can contain multiple attributes, and so your element processors can receive a richer and more complex set of configuration parameters.
• Disadvantage: custom elements/tags are unknown to browsers, and so if you are developing a web application using custom tags you might have to sacrifice one of the most interesting features of Thymeleaf: the ability to statically display templates as prototypes (natural templating).

This processor will extend AbstractElementTagProcessor, the base class to be used for tag processors that do not match on a specific attribute:

public class HeadlinesElementTagProcessor extends AbstractElementTagProcessor {

    private static final String TAG_NAME = "headlines";
    private static final int PRECEDENCE = 1000;


    private final Random rand = new Random(System.currentTimeMillis());


    public HeadlinesElementTagProcessor(final String dialectPrefix) {
        super(
            TemplateMode.HTML, // This processor will apply only to HTML mode
            dialectPrefix,     // Prefix to be applied to name for matching
            TAG_NAME,          // Tag name: match specifically this tag
            true,              // Apply dialect prefix to tag name
            null,              // No attribute name: will match by tag name
            false,             // No prefix to be applied to attribute name
            PRECEDENCE);       // Precedence (inside dialect's own precedence)
    }


    @Override
    protected void doProcess(
            final ITemplateContext context, final IProcessableElementTag tag,
            final IElementTagStructureHandler structureHandler) {

        /*
         * Obtain the Spring application context.
         */
        final ApplicationContext appCtx = SpringContextUtils.getApplicationContext(context);

        /*
         * Obtain the HeadlineRepository bean from the application context, and ask
         * it for the current list of headlines.
         */
        final HeadlineRepository headlineRepository = appCtx.getBean(HeadlineRepository.class);
        final List<Headline> headlines = headlineRepository.findAllHeadlines();

        /*
         * Read the 'order' attribute from the tag. This optional attribute in our tag 
         * will allow us to determine whether we want to show a random headline or
         * only the latest one ('latest' is default).
         */
        final String order = tag.getAttributeValue("order");

        String headlineText = null;
        if (order != null && order.trim().toLowerCase().equals("random")) {
            // Order is random 

            final int r = this.rand.nextInt(headlines.size());
            headlineText = headlines.get(r).getText();
            
        } else {
            // Order is "latest", only the latest headline will be shown
            
            Collections.sort(headlines);
            headlineText = headlines.get(headlines.size() - 1).getText();
            
        }

        /*
         * Create the DOM structure that will be substituting our custom tag.
         * The headline will be shown inside a '<div>' tag, and so this must
         * be created first and then a Text node must be added to it.
         */
        final IModelFactory modelFactory = context.getModelFactory();

        final IModel model = modelFactory.createModel();

        model.add(modelFactory.createOpenElementTag("div", "class", "headlines"));
        model.add(modelFactory.createText(HtmlEscape.escapeHtml5(headlineText)));
        model.add(modelFactory.createCloseElementTag("div"));

        /*
         * Instruct the engine to replace this entire element with the specified model.
         */
        structureHandler.replaceWith(model, false);
        
    }

}

The first interesting part of the code above is showing how to access Spring’s ApplicationContext in order to obtain one of our beans from it (the HeadlineRepository):

final ApplicationContext appCtx = SpringContextUtils.getApplicationContext(context);

Also, this processor is different to the previous ones in that we will need to create markup as a result of its execution: we are going to replace the original <score:headlines .../> tag with a <div>...</div> fragment, so we will need to make use of the model factory.

final IModelFactory modelFactory = context.getModelFactory();

final IModel model = modelFactory.createModel();

model.add(modelFactory.createOpenElementTag("div", "class", "headlines"));
model.add(modelFactory.createText(HtmlEscape.escapeHtml5(headlineText)));
model.add(modelFactory.createCloseElementTag("div"));

final IOpenElemenTag newTag = modelFactory.setAttribute(tag, "class", "newvalue");

final IModel model = 
        modelFactory.parse(
                context.getTemplateData(), 
                "<div class='headlines'>Some headlines</div>");

3.5. A model processor for our “Match Day Today” banner

The last processor we will include in our dialect is of a different nature than the ones we’ve seen so far: it is a model processor, not a tag processor.

As already mentioned in a previous section, model processors do not execute on a specific tag event, but on the complete sequence of events (i.e. the model) that contains the entire element they are matching.

So if we have a model processor that matches <p> tags with attribute score:matcher, and a fragment of template such as:

<p score:matcher="whatever">
    This is some body
</p>

That model processor will receive as an argument of its doProcess() method an IModel containing 3 events: <p score:matcher="whatever"> (open tag), \n This is some body\n (text) and </p> (close tag).

So back to our requirements: we wanted a model processor matching a scrore:match-day-today, that we can apply to the league table header and make it display, below this header, a banner warning the user that sundays are match days:

<h2 score:match-day-today th:text="#{title.leaguetable(${execInfo.now.time})}">
    League table for 07 July 2011
</h2>

Note that we don’t need a value for this score:match-day-today attribute, so we can just ignore it. Our code will like like this:

public class MatchDayTodayModelProcessor extends AbstractAttributeModelProcessor {

    private static final String ATTR_NAME = "match-day-today";
    private static final int PRECEDENCE = 100;


    public MatchDayTodayModelProcessor(final String dialectPrefix) {
        super(
            TemplateMode.HTML, // This processor will apply only to HTML mode
            dialectPrefix,     // Prefix to be applied to name for matching
            null,              // No tag name: match any tag name
            false,             // No prefix to be applied to tag name
            ATTR_NAME,         // Name of the attribute that will be matched
            true,              // Apply dialect prefix to attribute name
            PRECEDENCE,        // Precedence (inside dialect's own precedence)
            true);             // Remove the matched attribute afterwards
    }


    protected void doProcess(
            final ITemplateContext context, final IModel model,
            final AttributeName attributeName, final String attributeValue,
            final IElementModelStructureHandler structureHandler) {


        if (!checkPositionInMarkup(context)) {
            throw new TemplateProcessingException(
                    "The " + ATTR_NAME + " attribute can only be used inside a " +
                    "markup element with class \"leaguetable\"");
        }

        final Calendar now = Calendar.getInstance(context.getLocale());
        final int dayOfWeek = now.get(Calendar.DAY_OF_WEEK);

        // Sundays are Match Days!!
        if (dayOfWeek == Calendar.SUNDAY) {

            // The Model Factory will allow us to create new events
            final IModelFactory modelFactory = context.getModelFactory();

            // We will be adding the "Today is Match Day" banner just after
            // the element we are processing for:
            //
            // <h4 class="matchday">Today is MATCH DAY!</h4>
            //
            model.add(modelFactory.createOpenElementTag("h4", "class", "matchday")); //
            model.add(modelFactory.createText("Today is MATCH DAY!"));
            model.add(modelFactory.createCloseElementTag("h4"));

        }

    }


    private static boolean checkPositionInMarkup(final ITemplateContext context) {

        /*
         * We want to make sure this processor is being applied inside a container tag which has
         * class="leaguetable". So we need to check the second-to-last entry in the element stack
         * (the last entry is the tag being processed itself).
         */

        final List<IProcessableElementTag> elementStack = context.getElementStack();
        if (elementStack.size() < 2) {
            return false;
        }

        final IProcessableElementTag container = elementStack.get(elementStack.size() - 2);
        if (!(container instanceof IOpenElementTag)) {
            return false;
        }

        final String classValue = container.getAttributeValue("class");
        return classValue != null && classValue.equals("leaguetable");

    }


}

The first thing to note is that we are performing a check on the position the attribute is being used at: we will only allow it inside a container with class="leaguetable". So our checkPositionInMarkup(...) method makes use of the element stack in order to know the list of tags that had to be processed in order to process the current one.

Also, regarding the way the new banner element is created (an <h4>) notice how what we are doing is modifying the model attribute passed as an argument to doProcess(...). No new model object is being created:

final IModelFactory modelFactory = context.getModelFactory();

model.add(modelFactory.createOpenElementTag("h4", "class", "matchday")); //
model.add(modelFactory.createText("Today is MATCH DAY!"));
model.add(modelFactory.createCloseElementTag("h4"));

3.6. Declaring it all: the Dialect

The last step we need to take in order to complete our dialect is, of course, the dialect class itself.

As seen in a previous section, dialects might implement different interfaces depending on what they provide to the template engine. In this case, our dialect is only providing processors so it will be implementing IProcessorDialect.

In fact, we will extend an abstract convenience implementation that will ease the implementation of the interface: AbstractProcessorDialect:

public class ScoreDialect extends AbstractProcessorDialect {

    private static final String DIALECT_NAME = "Score Dialect";


    public ScoreDialect() {
        // We will set this dialect the same "dialect processor" precedence as
        // the Standard Dialect, so that processor executions can interleave.
        super(DIALECT_NAME, "score", StandardDialect.PROCESSOR_PRECEDENCE);
    }

    /*
     * Two attribute processors are declared: 'classforposition' and
     * 'remarkforposition'. Also one element processor: the 'headlines'
     * tag.
     */
    public Set<IProcessor> getProcessors(final String dialectPrefix) {
        final Set<IProcessor> processors = new HashSet<IProcessor>();
        processors.add(new ClassForPositionAttributeTagProcessor(dialectPrefix));
        processors.add(new RemarkForPositionAttributeTagProcessor(dialectPrefix));
        processors.add(new HeadlinesElementTagProcessor(dialectPrefix));
        processors.add(new MatchDayTodayModelProcessor(dialectPrefix));
        // This will remove the xmlns:score attributes we might add for IDE validation
        processors.add(new StandardXmlNsTagProcessor(TemplateMode.HTML, dialectPrefix));
        return processors;
    }


}

Once our dialect is created, we will need to add it to our Template Engine object in order to use it. This being a Spring-enabled application, we will modify the declared template engine bean:

@Bean
public SpringTemplateEngine templateEngine(){
    SpringTemplateEngine templateEngine = new SpringTemplateEngine();
    templateEngine.setTemplateResolver(templateResolver());
    templateEngine.addDialect(new ScoreDialect());
    return templateEngine;
}

Note that the addDialect(...) call there will add the Score Dialect to the one already configured by default in a SpringTemplateEngine: the SpringStandard dialect.

And that’s it! Our dialect is ready to run now, and our league table will display in exactly the way we wanted.