fix: allow more meta

adobe · Oct 10, 2024 · a87b377 · a87b377
1 parent a2319d1
commit a87b377
Show file tree

Hide file tree

Showing 4 changed files with 103 additions and 6 deletions.
diff --git a/src/html2md.js b/src/html2md.js
@@ -32,6 +32,10 @@ import {
 } from './mdast-table-handler.js';
 import formatPlugin from './markdownFormatPlugin.js';
 
+const HELIX_META = {
+  viewport: true,
+};
+
 function m(type, children, props = {}) {
   return {
     type,
@@ -53,10 +57,6 @@ function image(url) {
   };
 }
 
-const HELIX_META = new Set(Array.from([
-  'viewport',
-]));
-
 function toGridTable(title, data) {
   return m(TYPE_GRID_TABLE, [
     m(
@@ -76,6 +76,11 @@ function toGridTable(title, data) {
   ]);
 }
 
+/**
+ * @param {string} str
+ * @returns {string}
+ * @throws {Error}
+ */
 function assertValidJSON(str) {
   try {
     return JSON.stringify(JSON.parse(str.trim()));
@@ -84,13 +89,53 @@ function assertValidJSON(str) {
   }
 }
 
+/**
+ * @param {string} str
+ * @param {number} [limit]
+ * @returns {string}
+ * @throws {Error}
+ */
 function assertMetaSizeLimit(str, limit = 128_000) {
   if (str && str.length > limit) {
     throw Error('metadata size limit exceeded');
   }
   return str;
 }
 
+/**
+ * Check if meta name is allowed:
+ *  - non-reserved
+ *  - not starting with 'twitter:'
+ *    - except 'twitter:label' and 'twitter:data'
+ * @param {string} name
+ * @returns {boolean}
+ */
+function isAllowedMetaName(name) {
+  if (typeof name !== 'string') {
+    return false;
+  }
+  return !HELIX_META[name] && (
+    !name.startsWith('twitter:')
+    || name.startsWith('twitter:label')
+    || name.startsWith('twitter:data')
+  );
+}
+
+/**
+ * Check if meta property is allowed:
+ *  - non-reserved
+ *  - og:type
+ *  - product:*
+ * @param {string|undefined} property
+ * @returns {boolean}
+ */
+function isAllowedMetaProperty(property) {
+  if (typeof property !== 'string') {
+    return false;
+  }
+  return !HELIX_META[property] && (property.startsWith('product:') || property === 'og:type');
+}
+
 function addMetadata(hast, mdast) {
   const meta = new Map();
 
@@ -99,13 +144,15 @@ function addMetadata(hast, mdast) {
     if (child.tagName === 'title') {
       meta.set(text('title'), text(assertMetaSizeLimit(toString(child))));
     } else if (child.tagName === 'meta') {
-      const { name, content } = child.properties;
-      if (name && !HELIX_META.has(name) && !name.startsWith('twitter:')) {
+      const { name, property, content } = child.properties;
+      if (isAllowedMetaName(name)) {
         if (name === 'image') {
           meta.set(text(name), image(assertMetaSizeLimit(content)));
         } else {
           meta.set(text(name), text(assertMetaSizeLimit(content)));
         }
+      } else if (isAllowedMetaProperty(property)) {
+        meta.set(text(property), text(assertMetaSizeLimit(content)));
       }
     } else if (child.tagName === 'script' && child.properties.type === 'application/ld+json') {
       const str = assertMetaSizeLimit(assertValidJSON(toString(child)));

diff --git a/test/fixtures/meta-tags.html b/test/fixtures/meta-tags.html
@@ -0,0 +1,25 @@
+<html>
+
+<head>
+  <meta property="og:type" content="product">
+  <meta property="product:availability" content="${inStock ? 'In stock' : 'Out of stock'}">
+  <meta property="product:price.amount" content="${prices.final.amount}">
+  <meta property="product:price.currency" content="${prices.final.currency}">
+  <meta property="invalid" content="foo">
+  <meta property="og:ignored" content="bar">
+  <meta name="twitter:label1" content="Price">
+  <meta name="twitter:data1" content="$3">
+  <meta name="twitter:label2" content="Availability">
+  <meta name="twitter:data2" content="In stock">
+  <meta name="twitter:image" content="this-is-removed">
+</head>
+
+<body>
+  <main>
+    <div>
+      <h1>Hello, World.</h1>
+    </div>
+  </main>
+</body>
+
+</html>
diff --git a/test/fixtures/meta-tags.md b/test/fixtures/meta-tags.md
@@ -0,0 +1,21 @@
+# Hello, World.
+
++-------------------------------------------------------------------+
+| Metadata                                                          |
++------------------------+------------------------------------------+
+| og:type                | product                                  |
++------------------------+------------------------------------------+
+| product:availability   | ${inStock ? 'In stock' : 'Out of stock'} |
++------------------------+------------------------------------------+
+| product:price.amount   | ${prices.final.amount}                   |
++------------------------+------------------------------------------+
+| product:price.currency | ${prices.final.currency}                 |
++------------------------+------------------------------------------+
+| twitter:label1         | Price                                    |
++------------------------+------------------------------------------+
+| twitter:data1          | $3                                       |
++------------------------+------------------------------------------+
+| twitter:label2         | Availability                             |
++------------------------+------------------------------------------+
+| twitter:data2          | In stock                                 |
++------------------------+------------------------------------------+
diff --git a/test/html2md.test.js b/test/html2md.test.js
@@ -107,6 +107,10 @@ describe('html2md Tests', () => {
   it('throws meaningful error when json-ld is too large', async () => {
     await assert.rejects(() => test('json-ld-too-large'), Error('metadata size limit exceeded'));
   });
+
+  it('convert a document with meta names and properties correctly', async () => {
+    await test('meta-tags');
+  });
 });
 
 describe('className to block type tests', () => {