From 2f19481147b12e2ae503ed3d1f28621c94447ac3 Mon Sep 17 00:00:00 2001
From: Chocobozzz <me@florianbigard.com>
Date: Mon, 7 Mar 2022 11:48:53 +0100
Subject: Optimize markdown renderer

---
 server/helpers/markdown.ts       | 54 ++++++++++++++++++++++++----------------
 server/tests/helpers/markdown.ts |  6 +++++
 2 files changed, 38 insertions(+), 22 deletions(-)

diff --git a/server/helpers/markdown.ts b/server/helpers/markdown.ts
index 25685ec6d..f9291bdad 100644
--- a/server/helpers/markdown.ts
+++ b/server/helpers/markdown.ts
@@ -7,8 +7,13 @@ const sanitizeHtml = require('sanitize-html')
 const markdownItEmoji = require('markdown-it-emoji/light')
 const MarkdownItClass = require('markdown-it')
 
-const markdownItWithHTML = new MarkdownItClass('default', { linkify: true, breaks: true, html: true })
-const markdownItWithoutHTML = new MarkdownItClass('default', { linkify: true, breaks: true, html: false })
+const markdownItForSafeHtml = new MarkdownItClass('default', { linkify: true, breaks: true, html: true })
+  .enable(TEXT_WITH_HTML_RULES)
+  .use(markdownItEmoji)
+
+const markdownItForPlainText = new MarkdownItClass('default', { linkify: false, breaks: true, html: false })
+  .use(markdownItEmoji)
+  .use(plainTextPlugin)
 
 const toSafeHtml = (text: string) => {
   if (!text) return ''
@@ -17,9 +22,7 @@ const toSafeHtml = (text: string) => {
   const textWithLineFeed = text.replace(/<br.?\/?>/g, '\r\n')
 
   // Convert possible markdown (emojis, emphasis and lists) to html
-  const html = markdownItWithHTML.enable(TEXT_WITH_HTML_RULES)
-                                 .use(markdownItEmoji)
-                                 .render(textWithLineFeed)
+  const html = markdownItForSafeHtml.render(textWithLineFeed)
 
   // Convert to safe Html
   return sanitizeHtml(html, defaultSanitizeOptions)
@@ -28,12 +31,10 @@ const toSafeHtml = (text: string) => {
 const mdToOneLinePlainText = (text: string) => {
   if (!text) return ''
 
-  markdownItWithoutHTML.use(markdownItEmoji)
-                       .use(plainTextPlugin)
-                       .render(text)
+  markdownItForPlainText.render(text)
 
   // Convert to safe Html
-  return sanitizeHtml(markdownItWithoutHTML.plainText, textOnlySanitizeOptions)
+  return sanitizeHtml(markdownItForPlainText.plainText, textOnlySanitizeOptions)
 }
 
 // ---------------------------------------------------------------------------
@@ -47,30 +48,39 @@ export {
 
 // Thanks: https://github.com/wavesheep/markdown-it-plain-text
 function plainTextPlugin (markdownIt: any) {
-  let lastSeparator = ''
-
   function plainTextRule (state: any) {
     const text = scan(state.tokens)
 
-    markdownIt.plainText = text.replace(/\s+/g, ' ')
+    // markdownIt.plainText = text.replace(/\s+/g, ' ')
+    markdownIt.plainText = text
   }
 
   function scan (tokens: any[]) {
+    let lastSeparator = ''
     let text = ''
 
-    for (const token of tokens) {
-      if (token.children !== null) {
-        text += scan(token.children)
-        continue
-      }
-
+    function buildSeparator (token: any) {
       if (token.type === 'list_item_close') {
         lastSeparator = ', '
-      } else if (/[a-zA-Z]+_close/.test(token.type)) {
+      }
+
+      if (token.tag === 'br' || token.type === 'paragraph_close') {
         lastSeparator = ' '
-      } else if (token.content) {
-        text += lastSeparator
-        text += token.content
+      }
+    }
+
+    for (const token of tokens) {
+      buildSeparator(token)
+
+      if (token.type !== 'inline') continue
+
+      for (const child of token.children) {
+        buildSeparator(child)
+
+        if (!child.content) continue
+
+        text += lastSeparator + child.content
+        lastSeparator = ''
       }
     }
 
diff --git a/server/tests/helpers/markdown.ts b/server/tests/helpers/markdown.ts
index 0488a1a05..8177477f6 100644
--- a/server/tests/helpers/markdown.ts
+++ b/server/tests/helpers/markdown.ts
@@ -30,5 +30,11 @@ describe('Markdown helpers', function () {
 
       expect(result).to.equal('Hello coucou')
     })
+
+    it('Should convert tags to plain text', function () {
+      const result = mdToOneLinePlainText(`#déconversion\n#newage\n#histoire`)
+
+      expect(result).to.equal('#déconversion #newage #histoire')
+    })
   })
 })
-- 
cgit v1.2.3