Merge pull request #456 from thematters/feat/comment-style

feat: squeeze empty paragraphs to max 2

src/transformers/sanitize.test.ts

@@ -1,3 +1,4 @@
+import { stripIndent } from 'common-tags'
 import { describe, expect, test } from 'vitest'
 
 import { sanitizeHTML } from './sanitize'
@@ -18,6 +19,41 @@ describe('Sanitization: custom', () => {
       '<a class="">pp</a>',
     )
   })
+
+  test('allow max two empty paragraphys', () => {
+    expectSanitizeHTML(
+      stripIndent`
+        <p>abc</p>
+        <p></p>
+        <p></p>
+        abc
+        <p></p>
+        <p>abc</p>
+        <p></p>
+        <p></p>
+        <p></p>
+        <p>abc</p>
+        <p></p>
+        <p><br></p>
+        <p><br/></p>
+        <p><br></br></p>
+        <p><br/><br/><br/></p>
+        <p></p>
+      `,
+      stripIndent`
+        <p>abc</p>
+        <p></p>
+        <p></p>abc
+        <p></p>
+        <p>abc</p>
+        <p></p>
+        <p></p>
+        <p>abc</p>
+        <p></p>
+        <p></p>
+      `,
+    )
+  })
 })
 
 // via https://github.com/leizongmin/js-xss/blob/master/test/test_xss.js

src/transformers/sanitize.ts

@@ -1,3 +1,4 @@
+import { type Root, type RootContent } from 'hast'
 import rehypeFormat from 'rehype-format'
 import rehypeParse from 'rehype-parse'
 import rehypeRaw from 'rehype-raw'
@@ -11,14 +12,81 @@ import {
   rehypeStringifyOptions,
 } from './options'
 
+/**
+ * Squeeze empty paragraphs to a maximum of N
+ *
+ * e.g.
+ * <p></p><p></p><p></p><p></p><p></p><p></p>
+ * =>
+ * <p></p><p></p>
+ *
+ * @param {number} maxCount
+ */
+const rehypeSqueezeParagraphs =
+  ({ maxCount }: { maxCount: number }) =>
+  (tree: Root) => {
+    if (tree.type !== 'root') {
+      return
+    }
+
+    const children: RootContent[] = []
+    let count = 0
+    let touched = false
+
+    tree.children.forEach((node) => {
+      // skip empty text nodes
+      if (node.type === 'text' && node.value.replace(/\s/g, '') === '') {
+        children.push(node)
+        return
+      }
+
+      // skip non-paragraph nodes
+      if (node.type !== 'element' || node.tagName !== 'p') {
+        count = 0
+        children.push(node)
+        return
+      }
+
+      // skip non-empty paragraphs:
+      // - <p></p>
+      // - <p><br/></p>
+      const isEmptyParagraph =
+        node.children.length === 0 ||
+        node.children.every((n) => n.type === 'element' && n.tagName === 'br')
+      if (!isEmptyParagraph) {
+        count = 0
+        children.push(node)
+        return
+      }
+
+      // cap empty paragraphs
+      count++
+      if (count <= maxCount) {
+        children.push({
+          type: 'element',
+          tagName: 'p',
+          properties: {},
+          children: [],
+        })
+      } else {
+        touched = true
+      }
+    })
+
+    if (touched) {
+      tree.children = children
+    }
+  }
+
 const formatter = unified()
   .use(rehypeParse, rehypeParseOptions)
   .use(rehypeRaw)
   .use(rehypeSanitize, rehypeSanitizeOptions)
+  .use(rehypeSqueezeParagraphs, { maxCount: 2 })
   .use(rehypeFormat)
   .use(rehypeStringify, rehypeStringifyOptions)
 
-export const sanitizeHTML = (md: string): string => {
-  const result = formatter.processSync(md)
+export const sanitizeHTML = (html: string): string => {
+  const result = formatter.processSync(html)
   return String(result)
 }