auto-close unclosed list items

deno.json

@@ -1,6 +1,6 @@
 {
   "name": "@dbushell/hyperless",
-  "version": "0.11.0",
+  "version": "0.12.0",
   "exports": {
     ".": "./mod.ts"
   },

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@dbushell/hyperless",
-  "version": "0.11.0",
+  "version": "0.12.0",
   "repository": {
     "type": "git",
     "url": "git+https://github.com/dbushell/hyperless.git"

src/html-parser.ts

@@ -5,6 +5,9 @@ import {anyTag, comment} from './regexp.ts';
 /** Regular expression to match HTML comment or tag */
 const commentTag = new RegExp(`^${comment.source}|^${anyTag.source}`);
 
+/** List of valid <li> parents */
+const listTags = new Set(['ol', 'ul', 'menu']);
+
 /** HTML parser state */
 type ParseState = 'DATA' | 'RAWTEXT';
 
@@ -107,7 +110,15 @@ export const parseHTML = (html: string, options = parseOptions): Node => {
     }
     // Append opening tag and ascend
     else {
-      const node = new Node(parent, 'ELEMENT', tagText, tagRaw);
+      const node = new Node(null, 'ELEMENT', tagText, tagRaw);
+      // Close unclosed `<li>` if new `<li>` item is found
+      if (
+        node.tag === 'li' &&
+        parent.tag === 'li' &&
+        listTags.has(parent.parent?.tag!)
+      ) {
+        parent = parent.parent!;
+      }
       parent.append(node);
       parent = node;
     }

test/html_test.ts

@@ -64,6 +64,35 @@ Deno.test('leftover text', () => {
   assertEquals(root.at(1)!.raw, 'Leftover');
 });
 
+Deno.test('unclosed <li>', () => {
+  const html = `
+<ul>
+  <li>1<b>b</b>
+  <li>2</li>
+  <li>3
+    <ol>
+      <li>4<i>i</i>
+      <li>5
+    </ol>
+</ul>`.replace(/\s+/g, '');
+  const root = parseHTML(html);
+  assertEquals(root.at(0)!.size, 3);
+  assertEquals(
+    root.toString(),
+    `
+<ul>
+  <li>1<b>b</b></li>
+  <li>2</li>
+  <li>3
+    <ol>
+      <li>4<i>i</i></li>
+      <li>5</li>
+    </ol>
+  </li>
+</ul>`.replace(/\s+/g, '')
+  );
+});
+
 Deno.test('comments', () => {
   const html = `
 <!--