From dacfc42c7dbf996f81b6c143e715b4ccb532f68d Mon Sep 17 00:00:00 2001
From: Artur Arseniev <artur.catch@hotmail.it>
Date: Thu, 5 Aug 2021 12:54:03 +0200
Subject: [PATCH] Add DOMParser for HTML

---
 src/parser/config/config.js           | 16 +++++++-
 src/parser/model/BrowserParserHtml.js | 57 ++++++++++++++++-----------
 src/parser/model/ParserHtml.js        |  8 ++--
 3 files changed, 52 insertions(+), 29 deletions(-)

diff --git a/src/parser/config/config.js b/src/parser/config/config.js
index d33ad56f8..b6d08fec6 100644
--- a/src/parser/config/config.js
+++ b/src/parser/config/config.js
@@ -4,6 +4,20 @@ export default {
   // Custom CSS parser
   parserCss: null,
 
-  // Custom HTML parser
+  // Default DOMParser mime type
+  // @see https://developer.mozilla.org/en-US/docs/Web/API/DOMParser/parseFromString
+  htmlType: 'text/html',
+
+  // Custom function which tells which root Element to extract
+  // from the DOMParser.parseFromString result
+  // @see https://developer.mozilla.org/en-US/docs/Web/API/DOMParser/parseFromString
+  // @example
+  // // with htmlType as 'text/html' the function will be (default):
+  // htmlResult: (doc) => doc.body,
+  // // in case of `application/xml`
+  // htmlResult: (doc) => doc,
+  htmlResult: null,
+
+  // TODO: Custom HTML parser
   parserHtml: null
 };
diff --git a/src/parser/model/BrowserParserHtml.js b/src/parser/model/BrowserParserHtml.js
index 3a338908f..4c2c16871 100644
--- a/src/parser/model/BrowserParserHtml.js
+++ b/src/parser/model/BrowserParserHtml.js
@@ -1,4 +1,34 @@
-// PoC
+export default (str, config = {}) => {
+  const { htmlResult } = config;
+  const parser = new DOMParser();
+  const doc = parser.parseFromString(str, config.htmlType);
+
+  return htmlResult ? htmlResult(doc, str) : doc.body;
+};
+
+/**
+ * POC, custom html parser specs
+ * Parse an HTML string to an array of nodes
+ * example
+ * parse(`<div class="mycls" data-test>Hello</div><span>World <b>example</b></span>`)
+ * // result
+ * [
+ *  {
+ *      tagName: 'div',
+ *      attributes: { class: 'mycls', 'data-test': '' },
+ *      childNodes: ['Hello'],
+ *  },{
+ *      tagName: 'span',
+ *      childNodes: [
+ *          'World ',
+ *          {
+ *              tagName: 'b',
+ *              childNodes: ['example'],
+ *          }
+ *       ],
+ *  }
+ * ]
+ *
 
 export const parseNodes = nodes => {
   const result = [];
@@ -44,29 +74,7 @@ export const parseNode = el => {
   };
 };
 
-/**
- * Parse an HTML string to an array of nodes
- * @example
- * parse(`<div class="mycls" data-test>Hello</div><span>World <b>example</b></span>`)
- * // result
- * [
- *  {
- *      tagName: 'div',
- *      attributes: { class: 'mycls', 'data-test': '' },
- *      childNodes: ['Hello'],
- *  },{
- *      tagName: 'span',
- *      childNodes: [
- *          'World ',
- *          {
- *              tagName: 'b',
- *              childNodes: ['example'],
- *          }
- *       ],
- *  }
- * ]
- */
-export default (str, config) => {
+export default (str, config = {}) => {
   const result = [];
   const el = document.createElement('div');
   el.innerHTML = str;
@@ -79,3 +87,4 @@ export default (str, config) => {
 
   return result;
 };
+ */
diff --git a/src/parser/model/ParserHtml.js b/src/parser/model/ParserHtml.js
index 654031e73..036ca8512 100644
--- a/src/parser/model/ParserHtml.js
+++ b/src/parser/model/ParserHtml.js
@@ -1,4 +1,5 @@
 import { each, isString } from 'underscore';
+import BrowserParserHtml from './BrowserParserHtml';
 
 export default config => {
   var TEXT_NODE = 'span';
@@ -275,15 +276,14 @@ export default config => {
      */
     parse(str, parserCss) {
       const { em } = c;
-      const config = (em && em.get('Config')) || {};
+      const conf = (em && em.get('Config')) || {};
       const res = { html: '', css: '' };
-      const el = document.createElement('div');
-      el.innerHTML = str;
+      const el = BrowserParserHtml(str, config);
       const scripts = el.querySelectorAll('script');
       let i = scripts.length;
 
       // Remove all scripts
-      if (!config.allowScripts) {
+      if (!conf.allowScripts) {
         while (i--) scripts[i].parentNode.removeChild(scripts[i]);
       }