From 3d462180da7a2e705b4f9976d6efbdbc06dfc94e Mon Sep 17 00:00:00 2001
From: Peter Bengtsson <peterbe@github.com>
Date: Tue, 11 Oct 2022 19:52:03 +0200
Subject: [PATCH] optimize render of textOnly snippets (#31581)

Co-authored-by: Robert Sese <734194+rsese@users.noreply.github.com>
---
 lib/render-content/renderContent.js | 27 +++++++++++++++++++++++----
 1 file changed, 23 insertions(+), 4 deletions(-)

diff --git a/lib/render-content/renderContent.js b/lib/render-content/renderContent.js
index a5db3a2ddb..c22ace3d2d 100644
--- a/lib/render-content/renderContent.js
+++ b/lib/render-content/renderContent.js
@@ -1,6 +1,6 @@
 import liquid from './liquid.js'
 import cheerio from 'cheerio'
-import { encode } from 'html-entities'
+import { encode, decode } from 'html-entities'
 import stripHtmlComments from 'strip-html-comments'
 import createProcessor from './create-processor.js'
 
@@ -52,17 +52,21 @@ async function renderContent(template = '', context = {}, options = {}) {
     let html = vFile.toString()
 
     // Remove unwanted newlines (which appear as spaces) from inline tags inside tables
-    if (html.includes('<table>')) html = removeNewlinesFromInlineTags(html)
+    if (html.includes('<table>')) {
+      html = removeNewlinesFromInlineTags(html)
+    }
 
     if (options.textOnly) {
-      html = cheerio.load(html).text().trim()
+      html = fastTextOnly(html)
     }
 
     if (options.cheerioObject) {
       return cheerio.load(html, { xmlMode: true })
     }
 
-    if (options.encodeEntities) html = encode(html)
+    if (options.encodeEntities) {
+      html = encode(html)
+    }
 
     return html.trim()
   } catch (error) {
@@ -85,6 +89,21 @@ function removeNewlinesFromInlineTags(html) {
   return $('body').html()
 }
 
+// Given a piece of HTML return it without HTML. E.g.
+// `<p>Foo &amp; bar</p>` becomes `Foo & bar`
+// and `A <a href="">link</a> and <code>code</code>` becomes `A link and code`.
+// Take advantage of the subtle fact that a lot of the times, the html value
+// we get here is a single line that starts with `<p>` and ends with `</p>`
+// and contains no longer HTML tags.
+function fastTextOnly(html) {
+  if (!html) return ''
+  if (html.startsWith('<p>') && html.endsWith('</p>')) {
+    const middle = html.slice(3, -4)
+    if (!middle.includes('<')) return decode(middle.trim())
+  }
+  return cheerio.load(html, { xmlMode: true }).text().trim()
+}
+
 renderContent.liquid = liquid
 
 export default renderContent