fix: avoid memory leak and concurrency issues (#157)

Author: tripodsan

lib/HtmlDiff.js

@@ -15,8 +15,20 @@ import { defaults } from './defaults.js';
  * @param {Boolean}  [options.ignoreSelfClosingSlash=false]
  */
 export class HtmlDiff extends Diff {
-  constructor(options) {
+  constructor(options, tokens) {
     super(options);
     this.options = defaults(options);
+    this._tokenOverride = tokens;
+  }
+
+  /**
+     * Tokenizes the given HTML
+     *
+     * This overrides the default tokenizer from the `diff`
+     * @param {String} html
+     * @returns {Array}
+     */
+  tokenize(html) {
+    return this._tokenOverride[html];
   }
 }

lib/index.js

@@ -3,23 +3,11 @@ import modifyHtmlAccordingToOptions from './utils/modify.js';
 import { defaults } from './defaults.js';
 import handleMasks from './utils/mask.js';
 
-let htmlDifferOptions = {};
-const modifiedTokens = {};
-
-async function saveModifiedTokens(html) {
-  const tokens = await modifyHtmlAccordingToOptions(html, htmlDifferOptions);
-  modifiedTokens[html] = tokens.split(/({{.+?}}(?!})|[{}\(\)\[\]#\*`=:;,.<>"'\/]|\s+)/).filter(i => i);
+async function getModifiedTokens(html, options) {
+  const tokens = await modifyHtmlAccordingToOptions(html, options);
+  return tokens.split(/({{.+?}}(?!})|[{}\(\)\[\]#\*`=:;,.<>"'\/]|\s+)/).filter(i => i);
 }
 
-/**
- * Tokenizes the given HTML
- * @param {String} html
- * @returns {Array}
- */
-HtmlDiff.prototype.tokenize = function(html) {
-  return modifiedTokens[html];
-};
-
 /**
  * @class HtmlDiffer
  * @constructor
@@ -33,9 +21,7 @@ HtmlDiff.prototype.tokenize = function(html) {
  */
 export class HtmlDiffer {
   constructor(options) {
-    options = defaults(options);
-
-    htmlDifferOptions = options;
+    this.options = defaults(options);
   }
 
   /**
@@ -47,11 +33,18 @@ export class HtmlDiffer {
    * @returns {Diff}
    */
   async diffHtml(html1, html2) {
-    await saveModifiedTokens(html1);
-    await saveModifiedTokens(html2);
+    // precompute the tokenized html here, since it can't do it in `Diff.tokenize()` because
+    // the sax parser is async
+    const tokens = {
+      html1: await getModifiedTokens(html1, this.options),
+      html2: await getModifiedTokens(html2, this.options)
+    };
+
+    const htmlDiffer = new HtmlDiff(this.options, tokens);
 
-    const htmlDiffer = new HtmlDiff(htmlDifferOptions);
-    const diff = htmlDiffer.diff(html1, html2);
+    // just pass in keys for the left and right html, the original strings are no longer used
+    // after tokenize()
+    const diff = htmlDiffer.diff('html1', 'html2');
 
     return handleMasks(diff);
   };