Before reading this document we highly recommend that you familiarise yourself with the basics of developing a custom Property Editor for Umbraco.

Setup your Property Editor as a Block Property Editor

In order for your editor to become a Block Editor you must setup your property editor through C#. The constructor of the class can be auto generated by your Integrated Development Environment (IDE).

using Umbraco.Cms.Core;
using Umbraco.Cms.Core.PropertyEditors;
using Umbraco.Cms.Core.WebAssets;
using Umbraco.Cms.Infrastructure.WebAssets;

namespace MyNamespace;

[DataEditor(
    "MyOwn.UnicornBlocksEditor",
    "Unicorn Blocks",
    "unicornblocks",
    ValueType = ValueTypes.Json,
    Group = Constants.PropertyEditors.Groups.Lists,
    Icon = "icon-thumbnail-list")]
[PropertyEditorAsset(AssetType.Javascript, "/App_Plugins/UnicornBlocks/UnicornBlocks.controller.js")]
public class UnicornBlocksPropertyEditor : BlockEditorPropertyEditor
{
    public UnicornBlocksPropertyEditor(IDataValueEditorFactory dataValueEditorFactory, PropertyEditorCollection propertyEditors)
        : base(dataValueEditorFactory, propertyEditors)
    {
    }
}

Notice how the PropertyEditorAsset attribute is used to load the UnicornBlocks.controller.js JavaScript file.

Your Property Editor will need a PropertyValueConverter. Read more about Property Value Converters.

Data structure of Block Editors

The Block Editor data structure consists of three main parts:

Layout: The Layout defines Blocks that each will reference (by UDI) a content item in the list of data. The Layout object pairs keys with Property Editor aliases and their value type varies based on setup.

ContentData: A list of content items based on ElementTypes (IPublishedElement).

SettingsData: A list of content items based on ElementTypes (IPublishedElement).

In the following example the layout object "MyOwn.UnicornBlocksEditor" is of type Array.

{
  "layout": {
    "MyOwn.UnicornBlocksEditor": [
      {
        "contentUdi": "umb://element/fffba547615b4e9ab4ab2a7674845bc9"
      },
      {
        "contentUdi": "umb://element/e7dba547615b4e9ab4ab2a7674845bc9",
        "settingsUdi": "umb://element/a47667bcba54845b15b4e9ab4ab2a7c8"
      }
    ]
  },
  "contentData": [
    {
      "udi": "umb://element/fffba547615b4e9ab4ab2a7674845bc9",
      "contentTypeKey": "09876595489865786897",
      "__yourPropertyAlias__": "Hello world"
    },
    {
      "udi": "umb://element/e7dba547615b4e9ab4ab2a7674845bc9",
      "contentTypeKey": "123456789097654323456",
      "__yourPropertyAlias1__": "Hello world",
      "__yourPropertyAlias2__": "Hello world",
      "__yourPropertyAlias3__": "Hello world"
    }
  ],
  "settingsData": [
    {
      "udi": "umb://element/a47667bcba54845b15b4e9ab4ab2a7c8",
      "contentTypeKey": "09876595489865786897",
      "__yourPropertyAlias__": "Hello world"
    }
  ]
}

Client side code

Basic knowledge for understanding how to work with Block Editor data

We created the BlockEditorModelObject to aid in managing the presented Data Structure in the Block Editor.

To get a better understanding of what the Model Object does for you, we need to look at some usages of the Model Object.

Maintain and work with the Layout of a Block Editor

The layout of a Block Editor can be any structure. Therefore the Model Object (BlockEditorModelObject) cannot maintain this data. Our usage of the Model Object becomes complex. We give it a reference to a layout entry and perform an action that may need to reflect changes back to the layout.

Since the origin of blocks is in the layout the Model Object only can serve as a helper to maintain and create data. Therefore the Property Editor code will be using the layout as origin, using the Model Object to help manage specific parts.

This is explained in more detail below.

The basic setup for a Block Editor

Instantiate a Model Object and load dependencies. Provide the basic structure for the layout property when receiving the reference to it:

// We must get a scope that exists in all the lifetime of this data. Across variants and split-view.
var scopeOfExistence = $scope;
// Setup your component to require umbVariantContentEditors and vm.umbElementEditorContent. If one of them is avaiable use the method getScope to retrieve a shared scope for multiple editors of this content.
if(vm.umbVariantContentEditors && vm.umbVariantContentEditors.getScope) {
    scopeOfExistence = vm.umbVariantContentEditors.getScope();
} else if(vm.umbElementEditorContent && vm.umbElementEditorContent.getScope) {
    scopeOfExistence = vm.umbElementEditorContent.getScope();
}
// Define variables for layout and modelObject as you will be using these throughout your property editor.
vm.layout;
var modelObject;

// When we are ready we can instantiate the Model Object and load any dependencies.
vm.$onInit = function() {
  modelObject = blockEditorService.createModelObject(vm.model.value, vm.model.editor, vm.model.config.blocks, scopeOfExistence);
  modelObject.load().then(onLoaded);
}
function onLoaded() {
  // Define the default layout, this is used when there is no data stored for this property.
  var defaultLayout = [];
  // We store a reference to layout as we have to maintain this.
  // The getLayout method gives us a reference to the layout-object based on the property-editor alias. The defaultLayout will be initialized if it does not exist.
  vm.layout = modelObject.getLayout(defaultLayout);
}

Create a Block

Use the Model Object to create a Block and append the returned layout-entry to the layout.

In the following example we will create a new block and append it at the appropriate location in the 'layout' object:

// continuing from previous example.

// Creates a block and returns a layout entry. The layout entry is not part of layout yet as its not managed by the Model Object.
var layoutEntry = modelObject.create(contentTypeKey);
if (layoutEntry === null) {
  // The creation was not successful, therefore exit and without appending anything to our 'layout' object.
  return false;
}
// If we reach this line, we are good to add the layoutEntry to layout model.
// In this example our layout is an array and we would like to append the new block as the last entry.
vm.layout.push(layoutEntry);

Working with Blocks

The layout-entries alone do not provide much value when displaying or editing Blocks.

Our Model Object allows obtaining a Block Object by parsing a block's layout-entry for a specific Block.

The Block Object provides data of interest. The most important of these properties are: Block configuration, a label and the Block content in the Element Type Data Model format. This Content-model is useful for building the UI for editing the Content of a Block.

This example uses the Model Object to retrive a Block Object for outputting its label in the console.

// continuing from the basic setup example.

if (vm.layout.length > 0) {
  // Get first entry of from the layout, which is an array in this example.
  var firstLayoutEntry = vm.layout[0];

  // Create a Block Object for that entry.
  var block = modelObject.getBlockObject(firstLayoutEntry);

  // Check if the Block Object creation went well. (If a block is supported by the configuration of the Property Editor)
  if (block !== null) {
    console.log(block.label);
  }
}

This similar example uses the Block Object for setting a value on the first property in the Blocks Content.

// continuing from the basic setup example.

if (vm.layout.length > 0) {
  // Get first entry of from the layout, which is an array in this example.
  var firstLayoutEntry = vm.layout[0];

  // Create a Block Object for that entry.
  var block = modelObject.getBlockObject(firstLayoutEntry);

  // Check if the Block Object creation went well. (If a block is supported by the configuration of the Property Editor)
  if (block !== null) {

    // This line edits the value of 'myProp' directly, available by 'block.data':
    block.data.myProp = "Hello world";

    // Alternatively you can use this line, which edits the data throught the Element Type data model avaiable in 'block.content':
    block.content.variants[0].tabs[0].properties[0].value = "Hello world"; // This value will automatically be synced to the Property Editors Data Model, ´block.data.myProp´
  }
}

See blockEditorModelObject for the getBlockObject method for more information on the properties avaiable on a Block Object.

Remove a Block

Removing a Block and destroying its data is done by calling the removeDataAndDestroyModel method of the Model Object, which allows us to maintain the 'layout' object.

Your code will be based on working with Block Objects and therefore removal of a Block is be done by referring to a Block Object.

This example shows how to remove the first Block of our imaginary Block Editor and remove the block from our layout.

// continuing from the basic setup example.

if (vm.layout.length > 0) {
  // Get first entry of from the layout, which is an array in this sample.
  var firstLayoutEntry = vm.layout[0];

  // Create a Block Object for that entry.
  var block = modelObject.getBlockObject(firstLayoutEntry);

  // Check if the Block Object creation went well. (If a block isnt supported by the configuration of the Property Editor)
  if(block !== null) {

    modelObject.removeDataAndDestroyModel(block);// Removing the data of our block and destroying the Block Object for performance reasons.

    // We need to maintain the 'layout' object, so therefor its up to our code to remove the block from the 'layout' object.
    const index = array.indexOf(5);
    if (index > -1) {
      vm.layout.splice(index, 1);
    }
  }
}

Manage Block Objects for general use through out your Property Editor

Block Objects are used extensively and should be available throughout your Block Editor's runtime, rather than created for each action.

You probably want to use BlockObjects for your property-editor view. We append them to layout entries, so you can access them as properties from the layout object.

// continuing from the basic setup example.

var invalidLayoutItems = [];

// Append the blockObjects to our layout.
vm.layout.forEach(entry => {

// As we might have the same property-editor displayed multiple times on screen (splitview) we only need to initialize a BlockObject if its not already in place.

    // $block must have the 'data' property to be a valid BlockObject, if not it is considered a destroyed blockObject.
    if (entry.$block === undefined || entry.$block === null || entry.$block.data === undefined) {
        var block = modelObject.getBlockObject(entry);

        // If this entry was not supported by our property-editor it would return 'null'.
        if (block !== null) {
            entry.$block = block;
        } else {
          // We didnt succeed initializing this Block, therefore we need to filter this out of 'layout'. This only happens if the content data of the block couldn't be found.
            invalidLayoutItems.push(entry);
        }
    }
});

// remove the ones that are invalid
invalidLayoutItems.forEach(entry => {
    var index = vm.layout.findIndex(x => x === entry);
    if (index >= 0) {
        vm.layout.splice(index, 1);
    }
});

The following example loops through a layout array to display the contentUdi of each block:

<div ng-repeat="layoutEntry in vm.layout track by layoutEntry.$block.key">

    <h3 ng-bind="layout.$block.data.udi"></h3>

</div>

A Property Value Converter converts a property editor's database-stored value to another type. The converted value can be accessed from MVC Razor or any other Published Content API.

For example the standard Umbraco Core "Content Picker" stores a nodeId as String type. However if you implement a converter it could return an IPublishedContent object.

Published property values have four "Values":

• Source - The raw data stored in the database, this is generally a String

• Intermediate - An object of a type that is appropriate to the property, for example a nodeId should be an Int or a collection of nodeIds would be an integer array, Int[]

• Object - The object to be used when accessing the property using a Published Content API, for example UmbracoHelper's GetPropertyValue<T> method

• XPath - The object to be used when the property is accessed by XPath; This should generally be a String or an XPathNodeIterator

Registering PropertyValueConverters

PropertyValueConverters are automatically registered when implementing the interface. Any given PropertyEditor can only utilize a single PropertyValueConverter.

If you are implementing a PropertyValueConverter for a PropertyEditor that doesn't already have one, creating the PropertyValueConverter will automatically enable it. No further actions are needed.

If you aim to override an existing PropertyValueConverter, possibly from Umbraco or a package, additional steps are necessary. Deregister the existing one to prevent conflicts in this scenario.

The built-in PropertyValueConverters included with Umbraco, are currently marked as internal. This means you will not be able to remove them by type since the type isn't accessible outside of the namespace. In order to remove such PropertyValueConverters, you will need to look up the instance by name and then deregister it by the instance. This could be the case for other PropertyValueConverters included by packages as well, depending on the implementation details.

Implementing the Interface

Implement IPropertyValueConverter from the Umbraco.Cms.Core.PropertyEditors namespace on your class

Methods - Information

IsConverter(IPublishedPropertyType propertyType)

This method is called for each PublishedPropertyType (Document Type Property) at application startup. By returning True your value converter will be registered for that property type and your conversion methods will be executed whenever that value is requested.

Example: Checking if the IPublishedPropertyType EditorAlias property is equal to the alias of the core content editor. This check is a string comparison but we recommend creating a constant for it to avoid spelling errors:

IsValue(object value, PropertyValueLevel level)

This method is called to determine if the passed-in value is a value, and is of the level specified. There's a basic implementation of this in PropertyValueConverterBase.

GetPropertyValueType(IPublishedPropertyType propertyType)

This is where you can specify the type returned by this Converter. This type will be used by ModelsBuilder to return data from properties using this Converter in the proper type.

Example: Content Picker data is being converted to IPublishedContent.

PropertyCacheLevel GetPropertyCacheLevel(IPublishedPropertyType propertyType)

Here you specify which level the property value is cached at.

A property value can be cached at the following levels:

PropertyCacheLevel.Unknown

Do not use this cache level unless you know exactly what you're doing. We recommend using the PropertyCacheLevel.Element level.

PropertyCacheLevel.Element

The property value will be cached until its element is modified. The element is what holds (or owns) the property. For example:

• For properties used at the page level, the element is the entire page.

• For properties contained within Block List items, the element is the individual Block List item.

This is the most commonly used cache level and should be your default, unless you have specific reasons to do otherwise.

PropertyCacheLevel.Elements

The property value will be cached until any element (see above) is changed. This means that any change to any page will clear the property value cache.

This is particularly useful for property values that contain references to other content or elements. For example, this cache level is utilized by the Content Picker to clear its property values from the cache upon content updates.

PropertyCacheLevel.Snapshot

The property value will only be cached for the duration of the current snapshot.

A snapshot represents a point in time. For example, a snapshot is created for every content request from the frontend. When accessing a property in a snapshot using this cache level, it gets converted, cached throughout the snapshot, and later cleared.

For all intents and purposes, think of this cache level as "per request". If your property value should only be cached per request, this is the cache level you should use. Use it with caution, as the added property conversions incur a performance penalty.

PropertyCacheLevel.None

The property value will never be cached. Every time a property value is accessed (even within the same snapshot) property conversion is performed explicitly.

Use this cache level with extreme caution, as it incurs a massive performance penalty.

Methods - Conversion

There are a few different levels of conversion which can occur.

ConvertSourceToIntermediate(IPublishedElement owner, IPublishedPropertyType propertyType, object source, bool preview)

This method should convert the raw data value into an appropriate type. For example, a node identifier stored as a String should be converted to an Int or Udi.

Include a using Umbraco.Extensions; to be able to use the TryConvertTo extension method.

ConvertIntermediateToObject(IPublishedElement owner, IPublishedPropertyType propertyType, PropertyCacheLevel referenceCacheLevel, object inter, bool preview)

This method converts the Intermediate to an Object. The returned value is used by the GetPropertyValue<T> method of IPublishedContent.

The below example converts the nodeId (converted to Int or Udi by ConvertSourceToIntermediate) into an 'IPublishedContent' object.

ConvertIntermediateToXPath(IPublishedElement owner, IPublishedPropertyType propertyType, PropertyCacheLevel referenceCacheLevel, object inter, bool preview)

This method converts the Intermediate to XPath. The return value should generally be of type String or XPathNodeIterator.

In the example below, we convert the nodeId (converted by ConvertSourceToIntermediate) back into a String.

Sample

Generally Umbraco supports two different ways to declare a property editor. Most commonly one would create a package.manifest file, and then use it for declaring one or more property editors. But as an alternative, property editors can also be declared using C#.

A property editor consists of a number of mandatory properties, and some optional ones as well. As such, the outer JSON object for the property editor has the following properties:

The editor object then has the following properties:

Using a Package Manifest

A package manifest is a file specific to your package or custom code. This file is always stored in a folder in /App_Plugins/{YourPackageName}, and with the name package.manifest :

This example manifest specifies a Sir Trevor property editor via the propertyEditors collection, and also adds a single JavaScript file via the javascript property.

The actual Sir Trevor property editor has some additional configuration. It's a block based editor, so for instance it has a prevalue for setting the maximum amount of blocks allowed. In full, the package.manifest file for the Sir Trevor package looks like:

Using Csharp

The same property editor can be declared using C# instead using the DataEditor class and decorating the class with the DataEditor attribute:

Also notice how the PropertyEditorAsset attribute is used to load the SirTrevor.controller.js JavaScript file.

DataEditor attribute

PropertyEditorAsset attribute

The constructor of the attribute takes the type of the assets as the first parameter. Possible values are either AssetType.Javascript or AssetType.Css. The second parameter is the URL of the asset.

DataEditor class

In the example above, the SirTrevorEditor class doesn't really do much. For more basic property editors, the C# approach may require a bit more work compared to that of package.manifest files. But as property editors grow in complexity, using C# becomes a bit more useful - and also lets you do things not possible with package.manifest files.

Virtual methods are methods declared in a parent class. These methods have a default implementation that can be overridden in classes that inherit from the parent class. For instance in the example below, we can override the method and provide our own SirTrevorConfigurationEditor instead of what Umbraco returns by default.

In this case, the SirTrevorConfigurationEditor class doesn't do much either - but notice that it inherits from ConfigurationEditor<SirTrevorConfiguration>, meaning the configuration will be of type SirTrevorConfiguration:

The referenced SirTrevorConfiguration class is then what declares the configuration fields of when editing a Data Type using the Sir Trevor property editor:

A benefit of this approach (opposed to package.manifest files) is that we can now refer to the configuration using a strongly typed model - eg. as in this example Razor view:

Both instances of IDataType and PublishedDataType have a Configuration property. When looking across all data types and property editors, there is no common type for the configuration, so the return value is object. To get the strongly typed model, you can either cast the configuration value on your own, or use the generic ConfigurationAs extension method as shown above.

Like mentioned before, the SirTrevorConfigurationEditor class doesn't really do much in this example with the Sir Trevor property editor. But the Multi Node Tree Picker and others of Umbraco's build in property editors also override the ToValueEditor method.

This section describes how to work with and create Property Editors. A property editor is the editor used to insert content into Umbraco.

Tutorials - Creating a property editor

Reference for the package.manifest JSON file format to register one or more property editors for Umbraco.

Convert the stored property data value to a useful object returned by the Published Content APIs.

Use Property Actions to add additional functionaility to your custom property editors.

Learn how to build your own Block Editors.

Learn how to extend Property editors to track entity references inside the property editor.

More information

Property Actions are a built-in feature that provide a generic place for secondary functionality for property editors.

Property Actions appear as a small button next to the label of the property, which expands to show the available actions. They are defined and implemented in the Property Editor, making it open as to what a Property Action is.

Data Structure of Property Actions

Property Actions are an array of objects defining each action. An action is defined by the following properties:

We use labelKey and labelTokens to retrieve a localized string that is displayed as the Actions label.

isDisabled is used to disable an Action, which change the visual appearance and prevents interaction. Use this option when an action wouldn't provide any change. In the example above, the action remove all entries would not have any impact if there is no entries.

Implementation

The implementation of Property Actions varies depending on whether your Property Editor is implemented with a Controller or as a Component.

Controller Implementation

When your Property Editor is implemented with a Controller, use the following approach for the Property Action:

Component Implementation

Follow this guide if your Property Editor is implemented as a Component. The Component must be configured to retrieve an optional reference to umbProperty. The requirement must be optional because property-editors are implemented in scenarios where it's not presented.

See the following example:

See the following example for implementation of Property Actions in a Component, notice the difference is that we are parsing actions to this.umbProperty.setPropertyActions(...).

Property editors can be extended further to track entity references that may be selected or referenced inside the property editor. For example in the core of the CMS we have added this to numerous property editors.

A good example of this is the Media Picker. The CMS stores a reference to the selected media item, enabling the identification of content nodes that use that particular media item. This avoids it being accidentally deleted if it is being used.

When a content node is saved it will save the entity references as relations.

Viewing References

For Media Items

1. Go to the Media section.

2. Select a media item and click the Info tab.

For Content Nodes

1. Go to the Settings section.

2. Under the Relation Types folder, select Related Document relations and click Relations.

For Data Types

1. Go to the Settings section.

2. Expand the Data Types folder.

3. Select the Data Type you wish to view the references

4. Navigate to the Info tab.

Example

The following example shows how to implement tracking for the inbuilt CMS property editor Content Picker. It will always add a specific media reference, regardless of what value is picked in the content picker. In your own implementations, you will need to parse the value stored from the property editor you are implementing. You will also need to find any references to picked items in order to track their references.