draftjs-filters

Draft.js filters

Filter Draft.js content to preserve only the formatting you allow. Built for Draftail.

The main use case is to select what formatting to keep when copy-pasting rich text into an editor, for example from Word or Google Docs, addressing Draft.js limitations like #166 and #504. Check out the online demo!

If you want to learn more about how this is used in practice, have a look at Rethinking rich text pipelines with Draft.js.

Using the filters

First, grab the package from npm:

npm install --save draftjs-filters

Then, in your editor import filterEditorState and call it in the Draft.js onChange handler. This function takes two parameters: the filtering configuration, and the editorState.

import { filterEditorState } from "draftjs-filters"

function onChange(nextState) {
  const { editorState } = this.state
  let filteredState = nextState

  const shouldFilterPaste =
    nextState.getCurrentContent() !== editorState.getCurrentContent() &&
    nextState.getLastChangeType() === "insert-fragment"

  if (shouldFilterPaste) {
    filteredState = filterEditorState(
      {
        blocks: ["header-two", "header-three", "unordered-list-item"],
        styles: ["BOLD"],
        entities: [
          {
            type: "IMAGE",
            attributes: ["src"],
            whitelist: {
              src: "^http",
            },
          },
          {
            type: "LINK",
            attributes: ["url"],
          },
        ],
        maxNesting: 1,
        whitespacedCharacters: ["\n", "\t", "📷"],
      },
      filteredState,
    )
  }

  this.setState({ editorState: filteredState })
}

Here are all the available options:

// Whitelist of allowed block types. unstyled and atomic are always included.
blocks: $ReadOnlyArray<string>,
// Whitelist of allowed inline styles.
styles: $ReadOnlyArray<string>,
// Whitelist of allowed entities.
entities: $ReadOnlyArray<{
  // Entity type, eg. "LINK"
  type: string,
  // Allowed attributes. Other attributes will be removed. If this is omitted, all attributes are kept.
  attributes?: $ReadOnlyArray<string>,
  // Refine which entities are kept by whitelisting acceptable values with regular expression patterns.
  // It's also possible to use "true" to signify that a field is required to be present,
  // and "false" for fields required to be absent.
  // If this is omitted, all entities are kept.
  whitelist?: {
    [attribute: string]: string | boolean,
  },
}>,
// Maximum amount of depth for lists (0 = no nesting).
maxNesting: number,
// Characters to replace with whitespace.
whitespacedCharacters: Array<string>,
// Optional: Rules used to automatically convert blocks from one type to another
// based on the block’s text. Also supports setting the block depth.
// Defaults to the filters’ built-in block prefix rules.
blockTextRules?: $ReadOnlyArray<{
  // A regex as a string, to match against block text, e.g. "^(◦|o |o\t)".
  test: string,
  // The type to convert the block to if the test regex matches.
  type: string,
  // The depth to set (e.g. for list items with different prefixes per depth).
  depth: number,
}>,

Types

If your project uses Flow, type inference should just work. If you don't use Flow, it won't get in your way either.

Advanced usage

filterEditorState isn't very flexible. If you want more control over the filtering, simply compose your own filter function with the other single-purpose utilities. The Draft.js filters are published as ES6 modules using Rollup – module bundlers like Rollup and Webpack will tree shake (remove) the unused functions so you only bundle the code you use.

If using filters that remove blocks, be sure to use applyContentWithSelection to restore the selection where appropriate after filtering.

API

preserveAtomicBlocks

Creates atomic blocks where they would be required for a block-level entity to work correctly, when such an entity exists. Note: at the moment, this is only useful for IMAGE entities that Draft.js injects on arbitrary blocks on paste.

Parameters

• content ContentState

resetAtomicBlocks

Resets atomic blocks to have a single-space char and no styles. This is how they are stored by Draft.js by default.

Parameters

• content ContentState

removeInvalidAtomicBlocks

Removes atomic blocks for which the entity isn't whitelisted.

Parameters

• whitelist $ReadOnlyArray<{type: string}>
• content ContentState

removeInvalidDepthBlocks

Removes blocks that have a non-zero depth, and aren't list items. Happens with Apple Pages inserting unstyled items between list items.

Parameters

• content ContentState

preserveBlockByText

Changes block type and depth based on the block's text. – some word processors add a specific prefix within the text, eg. "· Bulleted list" in Word 2010. Also removes the matched text. This is meant first and foremost for list items where the list bullet or numeral ends up in the text. Other use cases may not be well covered.

Parameters

• rules $ReadOnlyArray<{test: string, type: string, depth: number}>
• content ContentState

limitBlockDepth

Resets the depth of all the content to at most max.

Parameters

• max number
• content ContentState

filterBlockTypes

Converts all block types not present in the whitelist to unstyled. Also sets depth to 0 (for potentially nested list items).

Parameters

• whitelist $ReadOnlyArray<string>
• content ContentState

filterInlineStyles

Removes all styles not present in the whitelist.

Parameters

• whitelist $ReadOnlyArray<string>
• content ContentState

cloneEntities

Clones entities in the entityMap, so each range points to its own entity instance. This only clones entities as necessary – if an entity is only referenced in a single range, it won't be changed.

Parameters

• content ContentState

filterEntityRanges

Filters entity ranges (where entities are applied on text) based on the result of the callback function. Returning true keeps the entity range, false removes it. Draft.js automatically removes entities if they are not applied on any text.

Parameters

• filterFn function (content: ContentState, entityKey: string, block: BlockNode): boolean
• content ContentState

shouldKeepEntityType

Keeps all entity types (images, links, documents, embeds) that are enabled.

Parameters

• whitelist $ReadOnlyArray<{type: string}>
• type string

shouldRemoveImageEntity

Removes invalid images – they should only be in atomic blocks. This only removes the image entity, not the camera emoji (📷) that Draft.js inserts.

Parameters

• entityType string
• blockType string

shouldKeepEntityByAttribute

Filters entities based on the data they contain.

Parameters

• entityTypes $ReadOnlyArray<{type: string, whitelist: {}?}>
• entityType string
• data {}

filterEntityData

Filters data on an entity to only retain what is whitelisted. This is crucial for IMAGE and LINK, where Draft.js adds a lot of unneeded attributes (width, height, etc).

Parameters

• entityTypes $ReadOnlyArray<{type: string, attributes: $ReadOnlyArray<string>?}>
• content ContentState

replaceTextBySpaces

Replaces the given characters by their equivalent length of spaces, in all blocks.

Parameters

• characters $ReadOnlyArray<string>
• content ContentState

applyContentWithSelection

Applies the new content to the editor state, optionally moving the selection to be on a valid block, inserting one if needed. See https://github.com/thibaudcolas/draftjs-filters/issues/27.

Parameters

• editorState EditorState
• content ContentState
• nextContent ContentState

filterEditorState

Applies whitelist and blacklist operations to the editor content, to enforce it's shaped according to the options. Will not alter the editor state if there are no changes to make.

Parameters

• options FilterOptions
• editorState EditorStateType

Browser support and polyfills

The Draft.js filters follow the browser support targets of Draft.js. Be sure to have a look at the required Draft.js polyfills.

Word processor support

Have a look at our test data in pasting/.

Editor - Browser	Chrome Windows	Chrome macOS	Firefox Windows	Firefox macOS	Edge Windows	IE11 Windows	Safari macOS	Safari iOS	Chrome Android
Word 2016								N/A	N/A
Word 2010		N/A		N/A			N/A	N/A	N/A
Apple Pages	N/A		N/A		N/A	N/A			N/A
Google Docs
Word Online						Unsupported		?	?
Dropbox Paper						Unsupported		?	?
Draft.js

Use the Draft.js Cut/Copy/Paste testing plan. We target specific external sources, and have ready-made test documents available to test them:

External sources

Here are external sources we want to pay special attention to, and for which we have ready-made test documents with diverse rich content.

• Microsoft Word 2016
• Microsoft Word 2010
• Google Docs
• Dropbox Paper
• Apple Pages
• Microsoft Word Online

IE11

There are known Draft.js issues with pasting in IE11. For now, we advise users to turn on stripPastedStyles in IE11 only so that Draft.js removes all formatting but preserves whitespace:

const IS_IE11 = !window.ActiveXObject && "ActiveXObject" in window

const editor = <Editor stripPastedStyles={IS_IE11} />

Contributing

See anything you like in here? Anything missing? We welcome all support, whether on bug reports, feature requests, code, design, reviews, tests, documentation, and more. Please have a look at our contribution guidelines.

Credits

View the full list of contributors. MIT licensed. Website content available as CC0.

Microsoft Word toolbars screenshot from PCWorld – Microsoft Word Turns 25 article.

Changelog

2.3.0 (2020-01-03)

Features

• api: add blockTextRules parameter to options. Fix #65 (#127) (2850d72)