With custom extensions you can add new content types and new functionalities, on top of what already exists or from scratch. Let’s start with a few common examples of how you can extend existing nodes, marks and extensions.
You’ll learn how you start from scratch at the end, but you’ll need the same knowledge for extending existing and creating new extensions.
Let’s say, you’d like to change the keyboard shortcut for the bullet list. You should start with looking at the source code of the extension, in that case [the `BulletList` node](https://github.com/ueberdosis/tiptap/blob/main/packages/extension-bullet-list/src/bullet-list.ts). For the bespoken example to overwrite the keyboard shortcut, your code could look like that:
The same applies to every aspect of an existing extension, except to the name. Let’s look at all the things that you can change through the extend method. We focus on one aspect in every example, but you can combine all those examples and change multiple aspects in one `extend()` call too.
The extension name is used in a whole lot of places and changing it isn’t too easy. If you want to change the name of an existing extension, you can copy the whole extension and change the name in all occurrences.
The extension name is also part of the JSON. If you [store your content as JSON](/guide/output#option-1-json), you need to change the name there too.
### Priority
The priority defines the order in which extensions are registered. The default priority is `100`, that’s what most extension have. Extensions with a higher priority will be loaded earlier.
```js
import Link from '@tiptap/extension-link'
const CustomLink = Link.extend({
priority: 1000,
})
```
The order in which extensions are loaded influences two things:
The [`Link`](/api/marks/link) mark for example has a higher priority, which means it will be rendered as `<a href="…"><strong>Example</strong></a>` instead of `<strong><a href="…">Example</a></strong>`.
All settings can be configured through the extension anyway, but if you want to change the default settings, for example to provide a library on top of Tiptap for other developers, you can do it like that:
tiptap works with a strict schema, which configures how the content can be structured, nested, how it behaves and many more things. You [can change all aspects of the schema](/api/schema) for existing extensions. Let’s walk through a few common use cases.
The default `Blockquote` extension can wrap other nodes, like headings. If you want to allow nothing but paragraphs in your blockquotes, set the `content` attribute accordingly:
You can use attributes to store additional information in the content. Let’s say you want to extend the default `Paragraph` node to have different colors:
That is already enough to tell Tiptap about the new attribute, and set `'pink'` as the default value. All attributes will be rendered as a HTML attribute by default, and parsed from the content when initiated.
Let’s stick with the color example and assume you want to add an inline style to actually color the text. With the `renderHTML` function you can return HTML attributes which will be rendered in the output.
You can also control how the attribute is parsed from the HTML. Maybe you want to store the color in an attribute called `data-color` (and not just `color`), here’s how you would do that:
Attributes can be applied to multiple extensions at once. That’s useful for text alignment, line height, color, font family, and other styling related attributes.
Take a closer look at [the full source code](https://github.com/ueberdosis/tiptap/tree/main/packages/extension-text-align) of the [`TextAlign`](/api/extensions/text-align) extension to see a more complex example. But here is how it works in a nutshell:
With the `renderHTML` function you can control how an extension is rendered to HTML. We pass an attributes object to it, with all local attributes, global attributes, and configured CSS classes. Here is an example from the `Bold` extension:
```js
renderHTML({ HTMLAttributes }) {
return ['strong', HTMLAttributes, 0]
},
```
The first value in the array should be the name of HTML tag. If the second element is an object, it’s interpreted as a set of attributes. Any elements after that are rendered as children.
The number zero (representing a hole) is used to indicate where the content should be inserted. Let’s look at the rendering of the `CodeBlock` extension with two nested tags:
```js
renderHTML({ HTMLAttributes }) {
return ['pre', ['code', HTMLAttributes, 0]]
},
```
If you want to add some specific attributes there, import the `mergeAttributes` helper from `@tiptap/core`:
The `parseHTML()` function tries to load the editor document from HTML. The function gets the HTML DOM element passed as a parameter, and is expected to return an object with attributes and their values. Here is a simplified example from the [`Bold`](/api/marks/bold) mark:
```js
parseHTML() {
return [
{
tag: 'strong',
},
]
},
```
This defines a rule to convert all `<strong>` tags to `Bold` marks. But you can get more advanced with this, here is the full example from the extension:
// <spanstyle="font-weight: bold"> and <spanstyle="font-weight: 700">
{
style: 'font-weight',
getAttrs: value => /^(bold(er)?|[5-9]\d{2,})$/.test(value as string) && null,
},
]
},
```
This looks for `<strong>` and `<b>` tags, and any HTML tag with an inline style setting the `font-weight` to bold.
As you can see, you can optionally pass a `getAttrs` callback, to add more complex checks, for example for specific HTML attributes. The callback gets passed the HTML DOM node, except when checking for the `style` attribute, then it’s the value.
The `getAttrs` function you’ve probably noticed in the example has two purposes:
1. Check the HTML attributes to decide whether a rule matches (and a mark or node is created from that HTML). When the function returns `false`, it’s not matching.
2. Get the DOM Element and use the HTML attributes to set your mark or node attributes accordingly:
```js
parseHTML() {
return [
{
tag: 'span',
getAttrs: element => {
// Check if the element has an attribute
element.hasAttribute('style')
// Get an inline style
element.style.color
// Get a specific attribute
element.getAttribute('data-color')
},
},
]
},
```
You can return an object with the attribute as the key and the parsed value to set your mark or node attribute. We would recommend to use the `parseHTML` inside `addAttributes()`, though. That will keep your code cleaner.
```js
addAttributes() {
return {
color: {
// Set the color attribute according to the value of the `data-color` attribute
parseHTML: element => element.getAttribute('data-color'),
:::warning Use the commands parameter inside of addCommands
To access other commands inside `addCommands` use the `commands` parameter that’s passed to it.
:::
### Keyboard shortcuts
Most core extensions come with sensible keyboard shortcut defaults. Depending on what you want to build, you’ll likely want to change them though. With the `addKeyboardShortcuts()` method you can overwrite the predefined shortcut map:
```js
// Change the bullet list keyboard shortcut
import BulletList from '@tiptap/extension-bullet-list'
By default text between two tildes on both sides is transformed to ~~striked text~~. If you want to think one tilde on each side is enough, you can overwrite the input rule like this:
Paste rules work like input rules (see above) do. But instead of listening to what the user types, they are applied to pasted content.
There is one tiny difference in the regular expression. Input rules typically end with a `$` dollar sign (which means “asserts position at the end of a line”), paste rules typically look through all the content and don’t have said `$` dollar sign.
Taking the example from above and applying it to the paste rule would look like the following example.
```js
// Check pasted content for the ~single tilde~ markdown syntax
After all, Tiptap is built on ProseMirror and ProseMirror has a pretty powerful plugin API, too. To access that directly, use `addProseMirrorPlugins()`.
To hook into events, for example a click, double click or when content is pasted, you can pass [event handlers](https://prosemirror.net/docs/ref/#view.EditorProps) to `editorProps` on the [editor](/api/editor#editor-props).
For advanced use cases, where you need to execute JavaScript inside your nodes, for example to render a sophisticated interface around an image, you need to learn about node views.
They are really powerful, but also complex. In a nutshell, you need to return a parent DOM element, and a DOM element where the content should be rendered in. Look at the following, simplified example:
There is a whole lot to learn about node views, so head over to the [dedicated section in our guide about node views](/guide/node-views) for more information. If you are looking for a real-world example, look at the source code of the [`TaskItem`](/api/nodes/task-item) node. This is using a node view to render the checkboxes.
If you think of the document as a tree, then [nodes](/api/nodes) are just a type of content in that tree. Good examples to learn from are [`Paragraph`](/api/nodes/paragraph), [`Heading`](/api/nodes/heading), or [`CodeBlock`](/api/nodes/code-block).
One or multiple marks can be applied to [nodes](/api/nodes), for example to add inline formatting. Good examples to learn from are [`Bold`](/api/marks/bold), [`Italic`](/api/marks/italic) and [`Highlight`](/api/marks/highlight).
Extensions add new capabilities to Tiptap and you’ll read the word extension here very often, even for nodes and marks. But there are literal extensions. Those can’t add to the schema (like marks and nodes do), but can add functionality or change the behaviour of the editor.