Advanced conversion concepts — attributes
This guide extends the Introduction to CKEditor 5 editing engine architecture guide, which we highly recommend reading first. Also, the Implementing a block widget and Implementing an inline widget tutorials explain the basics of conversion with examples, hence reading them is recommended as well.
In this guide we will dive deeper into some of the conversion concepts.
# Inline and block content
Generally speaking, there are two main types of content in the editor view and data output: inline and block.
The inline content means elements like
<div>, the inline elements do not structure the data. Instead, they format some text in a specific (visual and semantical) way. These elements are a characteristic of a text. For instance, you could say that some part of the text is bold, or is linked, etc. This concept has its reflection in the model of the rich-text editor where
<strong> are not represented as elements. Instead, they are the attributes of the text.
For example — in the model, you might have a
<paragraph> element with the “Foo bar” text, where “bar” has the
bold attribute set to
true. A pseudo–code of this model data structure could look as follows:
<paragraph> "Foo " // no attributes "bar" // bold=true </paragraph>
Throughout the rest of this guide the following, shorter convention will be used to represent model text attributes for the sake of clarity:
<paragraph>Foo <$text bold="true">bar</$text></paragraph>
Note that there is no
<strong> or any other additional element there, it is just some text with an attribute.
So, when does this text become wrapped with a
<strong> element? This happens during the conversion to the view. It is also important to know what type of a view element needs to be used. In the case of the elements that represent inline formatting, this should be an
# Conversion of multiple text attributes
A model text node may have multiple attributes (e.g. be bolded and linked) and all of them are converted into their respective view elements by independent converters.
Keep in mind that in the model, attributes do not have any specific order. This is contrary to the editor view or HTML output, where inline elements are nested in one another. Fortunately, the nesting happens automatically during the conversion from the model to the view. This makes working in the model simpler, as features do not need to take care of breaking or rearranging elements in the model.
For instance, consider the following model structure:
<paragraph> <$text bold="true" linkHref="url">Foo </$text> <$text linkHref="url">bar</$text> <$text bold="true"> baz</$text> </paragraph>
During the conversion, it will be converted to the following view structure:
<p> <a href="url"><strong>Foo </strong>bar</a><strong> baz</strong> </p>
Note that the
<a> element is converted in such way that it always becomes the “topmost” element. This is intentional so that no element ever breaks a link, which would otherwise look as follows:
<p> <strong><a href="url">Foo </a></strong><a href="url">bar</a><strong> baz</strong> </p>
There are two links with the same
href attribute next to each other in the generated view (editor output), which is semantically wrong. To make sure that it never happens, the view element that represents a link must have a priority defined. Most elements, like for instance
<strong>, do not care about it and stick to the default priority (
10). The link feature ensures that all
<a> view elements have the priority set to
5 therefore they are kept outside other elements.
# Merging attribute elements during conversion
Most of the simple view inline elements like
<em> do not have any attributes. Some of them have just one, for instance
<a> has its
But it is easy to come up with features that style a part of a text in a more complex way. An example would be the font family feature. When used, it adds the
fontFamily attribute to the text in the model, which is later converted to a
<span> element with a corresponding
So what would happen if several attributes were set on the same part of the text? Take this model example where
fontSize is used next to
<paragraph> <$text fontFamily="Tahoma" fontSize="big">foo</$text> </paragraph>
CKEditor 5 features are implemented in a granular way, which means that e.g. the font size converter is completely independent from the font family converter. This means that the above example is converted as follows:
<span style="font-family: value;">,
And, in theory, you could expect the following HTML as a result:
<p> <span style="font-family: Tahoma;"> <span class="text-big">foo</span> </span> </p>
But this is not the most optimal output you can get from the rich-text editor. Why not have just one
<span> element instead?
<p> <span style="font-family: Tahoma;" class="text-big">foo</span> </p>
Obviously a single
<span> makes more sense. And thanks to the merging mechanism built into the conversion process, this would be the actual output of the conversion.
Why is it so? In the above scenario, two model attributes are converted to
<span> elements. When the first attribute (say,
fontFamily) is converted, there is no
<span> in the view yet. So the first
<span> is added with the
style attribute. But then, when
fontSize is converted, the
<span> is already in the view. The downcast writer recognizes it and checks whether these elements can be merged, following these 3 rules:
- Both elements must have the same name.
- Both elements must have the same priority.
- Neither can have an ID.
Once you understand more about the conversion of model attributes, you can check some examples of:
- Extending the editor output — How to extend the output of existing CKEditor 5 features.
- Preserving custom content — How to make CKEditor 5 accept more content.
- Custom element conversion — How to deal with complex view structures during the model-to-view conversion.