feat: allow to retain all empty paragraphs

thematters · Apr 16, 2024 · 119a7ef · 119a7ef
1 parent 1ab6675
commit 119a7ef
Show file tree

Hide file tree

Showing 3 changed files with 88 additions and 7 deletions.
diff --git a/src/transformers/lib/rehypeSqueezeParagraphs.ts b/src/transformers/lib/rehypeSqueezeParagraphs.ts
@@ -8,7 +8,8 @@ import { type Root, type RootContent } from 'hast'
  * =>
  * <p><br></p><p><br></p>
  *
- * @param {number} maxCount
+ * @param {number} maxCount: maximum number of empty paragraphs, -1 to retain all
+ *
  */
 export const rehypeSqueezeParagraphs =
   ({ maxCount }: { maxCount: number }) =>
@@ -18,6 +19,7 @@ export const rehypeSqueezeParagraphs =
     }
 
     const children: RootContent[] = []
+    const isRetainAll = maxCount < 0
     let count = 0
     let touched = false
 
@@ -47,9 +49,9 @@ export const rehypeSqueezeParagraphs =
         return
       }
 
-      // cap empty paragraphs
+      // cap empty paragraphs or retain all by adding <br>
       count++
-      if (count <= maxCount) {
+      if (count <= maxCount || isRetainAll) {
         children.push({
           type: 'element',
           tagName: 'p',
@@ -68,7 +70,7 @@ export const rehypeSqueezeParagraphs =
       }
     })
 
-    if (touched) {
+    if (touched || isRetainAll) {
       tree.children = children
     }
   }
diff --git a/src/transformers/normalize-sanitize.test.ts b/src/transformers/normalize-sanitize.test.ts
@@ -40,7 +40,7 @@ const expectProcessCommentHTML = (
 }
 
 describe('Sanitize and normalize article', () => {
-  test('squeeze empty paragraphys', () => {
+  test('squeeze empty paragraphs', () => {
     expectProcessArticleHTML(
       stripIndent`
         <p>abc</p>
@@ -75,10 +75,50 @@ describe('Sanitize and normalize article', () => {
       { maxEmptyParagraphs: 2 },
     )
   })
+
+  test('squeeze and retain all empty paragraphs', () => {
+    expectProcessArticleHTML(
+      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/><br/></p>
+        <p></p>
+      `,
+      stripIndent`
+        <p>abc</p>
+        <p><br class="smart"></p>
+        <p><br class="smart"></p>
+        <p>abc</p>
+        <p><br class="smart"></p>
+        <p>abc</p>
+        <p><br class="smart"></p>
+        <p><br class="smart"></p>
+        <p><br class="smart"></p>
+        <p>abc</p>
+        <p><br class="smart"></p>
+        <p><br class="smart"></p>
+        <p><br class="smart"></p>
+        <p><br class="smart"></p>
+        <p><br class="smart"></p>
+      `,
+      { maxEmptyParagraphs: -1 },
+    )
+  })
 })
 
 describe('Sanitize and normalize comment', () => {
-  test('skip squeezing empty paragraphys', () => {
+  test('skip squeezing empty paragraphs', () => {
     expectProcessCommentHTML(
       stripIndent`
         <p>abc</p>

diff --git a/src/transformers/sanitize.test.ts b/src/transformers/sanitize.test.ts
@@ -24,7 +24,7 @@ describe('Sanitization: custom', () => {
     )
   })
 
-  test('allow max two empty paragraphys', () => {
+  test('allow max two empty paragraphs', () => {
     expectSanitizeHTML(
       stripIndent`
         <p>abc</p>
@@ -59,6 +59,45 @@ describe('Sanitization: custom', () => {
     )
   })
 
+  test('squeeze and retain all empty paragraphs', () => {
+    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/><br/></p>
+        <p></p>
+      `,
+      stripIndent`
+        <p>abc</p>
+        <p><br></p>
+        <p><br></p>abc
+        <p><br></p>
+        <p>abc</p>
+        <p><br></p>
+        <p><br></p>
+        <p><br></p>
+        <p>abc</p>
+        <p><br></p>
+        <p><br></p>
+        <p><br></p>
+        <p><br></p>
+        <p><br></p>
+      `,
+      { maxEmptyParagraphs: -1 },
+    )
+  })
+
   test('skip squeezing empty paragraphs', () => {
     expectSanitizeHTML(
       stripIndent`