Make DOMPurify configurable

Expose Trix.config.dompurify to allow users to configure DOMPurify.
Additionally avoid stripping data-trix-* attributes.

README.md

@@ -548,6 +548,16 @@ localStorage["editorState"] = JSON.stringify(element.editor)
 element.editor.loadJSON(JSON.parse(localStorage["editorState"]))
 ```
 
+## HTML Sanitization
+
+Trix uses [DOMPurify](https://github.com/cure53/DOMPurify/) to sanitize the editor content. You can set the DOMPurify config via `Trix.config.dompurify`.
+
+For example if you want to keep a custom tag, you can access do that via.
+
+```js
+Trix.config.dompurify.ADD_TAGS = [ "my-custom-tag" ]
+```
+
 ## Observing Editor Changes
 
 The `<trix-editor>` element emits several events which you can use to observe and respond to changes in editor state.

src/test/unit.js

@@ -5,6 +5,7 @@ import "test/unit/composition_test"
 import "test/unit/document_test"
 import "test/unit/document_view_test"
 import "test/unit/html_parser_test"
+import "test/unit/html_sanitizer_test"
 import "test/unit/location_mapper_test"
 import "test/unit/mutation_observer_test"
 import "test/unit/serialization_test"

src/test/unit/html_sanitizer_test.js

@@ -0,0 +1,54 @@
+import {
+  assert,
+  test,
+  testGroup,
+} from "test/test_helper"
+
+import { HTMLSanitizer } from "../../trix/models"
+import * as config from "../../trix/config"
+
+testGroup("HTMLSanitizer", () => {
+  test("strips custom tags", () => {
+    const html = "<custom-tag></custom-tag>"
+    const expectedHTML = ""
+    const document = HTMLSanitizer.sanitize(html).body.innerHTML
+    assert.equal(document, expectedHTML)
+  })
+
+  test("keeps custom tags configured for DOMPurify", () => {
+    const config = {
+      ADD_TAGS: [ "custom-tag" ],
+      RETURN_DOM: true,
+    }
+    withDOMPurifyConfig(config, () => {
+      const html = "<custom-tag></custom-tag>"
+      const expectedHTML = "<custom-tag></custom-tag>"
+      const document = HTMLSanitizer.sanitize(html).body.innerHTML
+      assert.equal(document, expectedHTML)
+    })
+  })
+})
+
+const withDOMPurifyConfig = (attrConfig = {}, fn) => {
+  withConfig("dompurify", attrConfig, fn)
+}
+
+const withConfig = (section, newConfig = {}, fn) => {
+  const originalConfig = Object.assign({}, config[section])
+  const copy = (section, properties) => {
+    for (const [ key, value ] of Object.entries(properties)) {
+      if (value) {
+        config[section][key] = value
+      } else {
+        delete config[section][key]
+      }
+    }
+  }
+
+  try {
+    copy(section, newConfig)
+    fn()
+  } finally {
+    copy(section, originalConfig)
+  }
+}

src/trix/config/dompurify.js

@@ -0,0 +1,4 @@
+export default {
+  ADD_ATTR: [ "language" ],
+  RETURN_DOM: true
+}

src/trix/config/index.js

@@ -2,6 +2,7 @@ export { default as attachments } from "./attachments"
 export { default as blockAttributes } from "./block_attributes"
 export { default as browser } from "./browser"
 export { default as css } from "./css"
+export { default as dompurify } from "./dompurify"
 export { default as fileSize } from "./file_size_formatting"
 export { default as input } from "./input"
 export { default as keyNames } from "./key_names"

src/trix/models/html_sanitizer.js

@@ -2,6 +2,14 @@ import BasicObject from "trix/core/basic_object"
 
 import { nodeIsAttachmentElement, removeNode, tagName, walkTree } from "trix/core/helpers"
 import DOMPurify from "dompurify"
+import * as config from "trix/config"
+
+DOMPurify.addHook("uponSanitizeAttribute", function (node, data) {
+  const allowedAttributePattern = /^data-trix-/
+  if (allowedAttributePattern.test(data.attrName)) {
+    data.forceKeepAttr = true
+  }
+})
 
 const DEFAULT_ALLOWED_ATTRIBUTES = "style href src width height language class".split(" ")
 const DEFAULT_FORBIDDEN_PROTOCOLS = "javascript:".split(" ")
@@ -31,7 +39,10 @@ export default class HTMLSanitizer extends BasicObject {
   sanitize() {
     this.sanitizeElements()
     this.normalizeListElementNesting()
-    return DOMPurify.sanitize(this.body, { ADD_ATTR: [ "language" ], RETURN_DOM: true } )
+    DOMPurify.setConfig(config.dompurify)
+    this.body = DOMPurify.sanitize(this.body)
+
+    return this.body
   }
 
   getHTML() {