Decode HTML entities for previews and search index

fixes #15

we use the "he" library for this as the stripping has to happen at build
time in `extendPageData`, i.e. in a NodeJS process that does not have
access to the browser facilities for converting HTML entities.

We don't just strip the entities as depending on the language, they may
be important searchable characters (e.g. special non-ASCII letters of
the language).

index.js

@@ -1,4 +1,5 @@
 const { path } = require('@vuepress/shared-utils')
+const he = require('he');
 
 module.exports = (options) => ({
     extendPageData ($page) {
@@ -7,9 +8,11 @@ module.exports = (options) => ({
         }
         // _strippedContent does not contain the YAML frontmatter
         const { html } = $page._context.markdown.render($page._strippedContent)
-        const text = html
-            .replace(/(<[^>]+>)+/g, " ") // remove HTML tags
-            .replace(/^\s*#\s/gm, "") // remove header anchors inserted by vuepress
+        const text = he.decode( // decode HTML entities like &quot;
+			html
+				.replace(/(<[^>]+>)+/g, " ") // remove HTML tags
+				.replace(/^\s*#\s/gm, "") // remove header anchors inserted by vuepress
+		)
         $page.content = text
     },
     alias: {

package.json

@@ -29,7 +29,8 @@
   "license": "MIT",
   "homepage": "https://github.com/z3by/vuepress-plugin-flexsearch#readme",
   "devDependencies": {
-    "vuepress": "^1.4.1"
+    "vuepress": "^1.4.1",
+    "he": "^1.2.0"
   },
   "dependencies": {
     "flexsearch": "nextapps-de/flexsearch"