Fix plaintext markdown converter

5 files changed, 94 insertions, 23 deletions

diff --git a/server/controllers/feeds.ts b/server/controllers/feeds.ts
index 3c8680ca4..e6cdaf94b 100644
--- a/server/controllers/feeds.ts
+++ b/server/controllers/feeds.ts
@@ -1,6 +1,6 @@
 import express from 'express'
 import Feed from 'pfeed'
-import { mdToPlainText, toSafeHtml } from '@server/helpers/markdown'
+import { mdToOneLinePlainText, toSafeHtml } from '@server/helpers/markdown'
 import { getServerActor } from '@server/models/application/application'
 import { getCategoryLabel } from '@server/models/video/formatter/video-format-utils'
 import { VideoInclude } from '@shared/models'
@@ -236,7 +236,7 @@ function initFeed (parameters: {
 
   return new Feed({
     title: name,
-    description: mdToPlainText(description),
+    description: mdToOneLinePlainText(description),
     // updated: TODO: somehowGetLatestUpdate, // optional, default = today
     id: webserverUrl,
     link: webserverUrl,
@@ -299,7 +299,7 @@ function addVideosToFeed (feed, videos: VideoModel[]) {
       title: video.name,
       id: video.url,
       link: WEBSERVER.URL + video.getWatchStaticPath(),
-      description: mdToPlainText(video.getTruncatedDescription()),
+      description: mdToOneLinePlainText(video.getTruncatedDescription()),
       content: toSafeHtml(video.description),
       author: [
         {
diff --git a/server/helpers/markdown.ts b/server/helpers/markdown.ts
index 0b8c2fabc..25685ec6d 100644
--- a/server/helpers/markdown.ts
+++ b/server/helpers/markdown.ts
@@ -1,14 +1,14 @@
-import { getSanitizeOptions, TEXT_WITH_HTML_RULES } from '@shared/core-utils'
+import { getDefaultSanitizeOptions, getTextOnlySanitizeOptions, TEXT_WITH_HTML_RULES } from '@shared/core-utils'
 
-const sanitizeOptions = getSanitizeOptions()
+const defaultSanitizeOptions = getDefaultSanitizeOptions()
+const textOnlySanitizeOptions = getTextOnlySanitizeOptions()
 
 const sanitizeHtml = require('sanitize-html')
 const markdownItEmoji = require('markdown-it-emoji/light')
 const MarkdownItClass = require('markdown-it')
-const markdownIt = new MarkdownItClass('default', { linkify: true, breaks: true, html: true })
 
-markdownIt.enable(TEXT_WITH_HTML_RULES)
-markdownIt.use(markdownItEmoji)
+const markdownItWithHTML = new MarkdownItClass('default', { linkify: true, breaks: true, html: true })
+const markdownItWithoutHTML = new MarkdownItClass('default', { linkify: true, breaks: true, html: false })
 
 const toSafeHtml = (text: string) => {
   if (!text) return ''
@@ -17,29 +17,65 @@ const toSafeHtml = (text: string) => {
   const textWithLineFeed = text.replace(/<br.?\/?>/g, '\r\n')
 
   // Convert possible markdown (emojis, emphasis and lists) to html
-  const html = markdownIt.render(textWithLineFeed)
+  const html = markdownItWithHTML.enable(TEXT_WITH_HTML_RULES)
+                                 .use(markdownItEmoji)
+                                 .render(textWithLineFeed)
 
   // Convert to safe Html
-  return sanitizeHtml(html, sanitizeOptions)
+  return sanitizeHtml(html, defaultSanitizeOptions)
 }
 
-const mdToPlainText = (text: string) => {
+const mdToOneLinePlainText = (text: string) => {
   if (!text) return ''
 
-  // Convert possible markdown (emojis, emphasis and lists) to html
-  const html = markdownIt.render(text)
+  markdownItWithoutHTML.use(markdownItEmoji)
+                       .use(plainTextPlugin)
+                       .render(text)
 
   // Convert to safe Html
-  const safeHtml = sanitizeHtml(html, sanitizeOptions)
-
-  return safeHtml.replace(/<[^>]+>/g, '')
-                 .replace(/\n$/, '')
-                 .replace(/\n/g, ', ')
+  return sanitizeHtml(markdownItWithoutHTML.plainText, textOnlySanitizeOptions)
 }
 
 // ---------------------------------------------------------------------------
 
 export {
   toSafeHtml,
-  mdToPlainText
+  mdToOneLinePlainText
+}
+
+// ---------------------------------------------------------------------------
+
+// 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, ' ')
+  }
+
+  function scan (tokens: any[]) {
+    let text = ''
+
+    for (const token of tokens) {
+      if (token.children !== null) {
+        text += scan(token.children)
+        continue
+      }
+
+      if (token.type === 'list_item_close') {
+        lastSeparator = ', '
+      } else if (/[a-zA-Z]+_close/.test(token.type)) {
+        lastSeparator = ' '
+      } else if (token.content) {
+        text += lastSeparator
+        text += token.content
+      }
+    }
+
+    return text
+  }
+
+  markdownIt.core.ruler.push('plainText', plainTextRule)
 }
diff --git a/server/lib/client-html.ts b/server/lib/client-html.ts
index 74788af52..19354ab70 100644
--- a/server/lib/client-html.ts
+++ b/server/lib/client-html.ts
@@ -12,7 +12,7 @@ import { HttpStatusCode } from '../../shared/models/http/http-error-codes'
 import { VideoPlaylistPrivacy, VideoPrivacy } from '../../shared/models/videos'
 import { isTestInstance } from '../helpers/core-utils'
 import { logger } from '../helpers/logger'
-import { mdToPlainText } from '../helpers/markdown'
+import { mdToOneLinePlainText } from '../helpers/markdown'
 import { CONFIG } from '../initializers/config'
 import {
   ACCEPT_HEADERS,
@@ -103,7 +103,7 @@ class ClientHtml {
       res.status(HttpStatusCode.NOT_FOUND_404)
       return html
     }
-    const description = mdToPlainText(video.description)
+    const description = mdToOneLinePlainText(video.description)
 
     let customHtml = ClientHtml.addTitleTag(html, video.name)
     customHtml = ClientHtml.addDescriptionTag(customHtml, description)
@@ -164,7 +164,7 @@ class ClientHtml {
       return html
     }
 
-    const description = mdToPlainText(videoPlaylist.description)
+    const description = mdToOneLinePlainText(videoPlaylist.description)
 
     let customHtml = ClientHtml.addTitleTag(html, videoPlaylist.name)
     customHtml = ClientHtml.addDescriptionTag(customHtml, description)
@@ -263,7 +263,7 @@ class ClientHtml {
       return ClientHtml.getIndexHTML(req, res)
     }
 
-    const description = mdToPlainText(entity.description)
+    const description = mdToOneLinePlainText(entity.description)
 
     let customHtml = ClientHtml.addTitleTag(html, entity.getDisplayName())
     customHtml = ClientHtml.addDescriptionTag(customHtml, description)
diff --git a/server/tests/helpers/index.ts b/server/tests/helpers/index.ts
index 66db93c99..91d11e25d 100644
--- a/server/tests/helpers/index.ts
+++ b/server/tests/helpers/index.ts
@@ -1,4 +1,5 @@
 import './image'
 import './core-utils'
 import './comment-model'
+import './markdown'
 import './request'
diff --git a/server/tests/helpers/markdown.ts b/server/tests/helpers/markdown.ts
new file mode 100644
index 000000000..0488a1a05
--- /dev/null
+++ b/server/tests/helpers/markdown.ts
@@ -0,0 +1,34 @@
+/* eslint-disable @typescript-eslint/no-unused-expressions,@typescript-eslint/require-await */
+
+import 'mocha'
+import { mdToOneLinePlainText } from '@server/helpers/markdown'
+import { expect } from 'chai'
+
+describe('Markdown helpers', function () {
+
+  describe('Plain text', function () {
+
+    it('Should convert a list to plain text', function () {
+      const result = mdToOneLinePlainText(`* list 1
+* list 2
+* list 3`)
+
+      expect(result).to.equal('list 1, list 2, list 3')
+    })
+
+    it('Should convert a list with indentation to plain text', function () {
+      const result = mdToOneLinePlainText(`Hello:
+  * list 1
+  * list 2
+  * list 3`)
+
+      expect(result).to.equal('Hello: list 1, list 2, list 3')
+    })
+
+    it('Should convert HTML to plain text', function () {
+      const result = mdToOneLinePlainText(`**Hello** <strong>coucou</strong>`)
+
+      expect(result).to.equal('Hello coucou')
+    })
+  })
+})