feat: header extraction improvement (close: #238) (#271)

* feat: sidebar header extraction improvements (close: #238)

* chore: add more markdown tokens to be removed

* fix: typo

Author: ulivz

lib/util/index.js

@@ -1,7 +1,4 @@
-const parseEmojis = str => {
-  const emojiData = require('markdown-it-emoji/lib/data/full.json')
-  return str.replace(/:(.+?):/g, (placeholder, key) => emojiData[key] || placeholder)
-}
+const parseHeaders = require('./parseHeaders')
 
 exports.normalizeHeadTag = tag => {
   if (typeof tag === 'string') {
@@ -35,11 +32,11 @@ exports.inferTitle = function (frontmatter) {
     return 'Home'
   }
   if (frontmatter.data.title) {
-    return parseEmojis(frontmatter.data.title)
+    return parseHeaders(frontmatter.data.title)
   }
   const match = frontmatter.content.trim().match(/^#+\s+(.*)/)
   if (match) {
-    return parseEmojis(match[1])
+    return parseHeaders(match[1])
   }
 }
 
@@ -71,7 +68,7 @@ exports.extractHeaders = (content, include = [], md) => {
   const res = []
   tokens.forEach((t, i) => {
     if (t.type === 'heading_open' && include.includes(t.tag)) {
-      const title = parseEmojis(tokens[i + 1].content)
+      const title = parseHeaders(tokens[i + 1].content)
       const slug = t.attrs.find(([name]) => name === 'id')[1]
       res.push({
         level: parseInt(t.tag.slice(1), 10),

lib/util/parseHeaders.js

@@ -0,0 +1,29 @@
+const parseEmojis = str => {
+  const emojiData = require('markdown-it-emoji/lib/data/full.json')
+  return str.replace(/:(.+?):/g, (placeholder, key) => emojiData[key] || placeholder)
+}
+
+const unescapeHtml = html => html
+  .replace(/"/g, '"')
+  .replace(/'/g, '\'')
+  .replace(/:/g, ':')
+  .replace(/&lt;/g, '<')
+  .replace(/&gt;/g, '>')
+
+const removeMarkdownToken = str => str
+  .replace(/`(.*)`/, '$1')  // ``
+  .replace(/\[(.*)\]\(.*\)/, '$1')  // []()
+  .replace(/\*\*(.*)\*\*/, '$1')  // **
+  .replace(/\*(.*)\*/, '$1')  // *
+  .replace(/_(.*)_/, '$1')  // _
+
+// put here to avoid circular references
+const compose = (...processors) => {
+  if (processors.length === 0) return input => input
+  if (processors.length === 1) return processors[0]
+  return processors.reduce((prev, next) => {
+    return (...args) => next(prev(...args))
+  })
+}
+
+module.exports = compose(unescapeHtml, parseEmojis, removeMarkdownToken)