Merge branch 'main' into sitemap

e2e/docs/.vuepress/config.ts

@@ -59,17 +59,17 @@ export default defineUserConfig({
 
     sidebar: {
       '/': ['/sidebar/'],
-      '/sidebar/': [
+      '/sidebar/heading/': 'heading',
+      '/sidebar/config/': [
         {
           text: 'Sidebar',
-          link: '/sidebar/',
+          link: '/sidebar/config/',
           children: [
-            { text: 'sidebar 1', link: '/sidebar/1.html' },
-            { text: 'sidebar 2', link: '/sidebar/2.html' },
+            { text: 'sidebar 1', link: '/sidebar/config/1.html' },
+            { text: 'sidebar 2', link: '/sidebar/config/2.html' },
           ],
         },
       ],
-      '/heading-sidebar/': 'heading',
     },
   }),
 }) as UserConfig

e2e/docs/sidebar/heading/1.md

@@ -0,0 +1,5 @@
+# Sidebar 1
+
+## Sidebar 1 Heading 1
+
+## Sidebar 1 Heading 2

e2e/docs/sidebar/heading/2.md

@@ -0,0 +1,5 @@
+# Sidebar 2
+
+## Sidebar 2 Heading 1
+
+## Sidebar 2 Heading 2

e2e/tests/theme-default/sidebar.cy.ts

@@ -1,15 +1,46 @@
-it('has heading sidebar', () => {
-  cy.visit('/sidebar/')
-  cy.get('.theme-default-content').then((el) => {
-    cy.wrap(el)
-      .get('a.sidebar-item')
-      .should('contain', 'Sidebar Heading 1')
-      .should('contain', 'Sidebar Heading 2')
+describe('has heading sidebar', () => {
+  it('frontmatter', () => {
+    cy.visit('/sidebar/auto.html')
+    cy.get('.theme-default-content').then((el) => {
+      cy.wrap(el)
+        .get('a.sidebar-item')
+        .should('contain', 'Sidebar Heading 1')
+        .should('contain', 'Sidebar Heading 2')
+    })
+  })
+
+  it('config', () => {
+    cy.visit('/sidebar/heading/1.html')
+
+    cy.get('.theme-default-content').then((el) => {
+      cy.wrap(el)
+        .get('h1')
+        .invoke('text')
+        .should('be.not.empty')
+        .then((text) => {
+          cy.get('.sidebar-heading')
+            .invoke('text')
+            .should('contain', text.replace('#', '').trim())
+        })
+
+      cy.wrap(el)
+        .get('h2')
+        .each((header) => {
+          cy.wrap(header)
+            .invoke('text')
+            .should('be.not.empty')
+            .then((headerText) => {
+              cy.get('a.sidebar-item').contains(
+                headerText.replace('#', '').trim(),
+              )
+            })
+        })
+    })
   })
 })
 
 it('has configured sidebar', () => {
-  cy.visit('/sidebar/1.html')
+  cy.visit('/sidebar/config/1.html')
   cy.get('.theme-default-content').then((el) => {
     cy.wrap(el)
       .get('a.sidebar-item')

tools/helper/src/node/page/index.ts

@@ -1 +1,2 @@
 export * from './excerpt.js'
+export * from './text.js'

tools/helper/src/node/page/text.ts

@@ -0,0 +1,159 @@
+// eslint-disable-next-line vue/prefer-import-from-vue
+import { isHTMLTag } from '@vue/shared'
+import type { AnyNode } from 'cheerio'
+import { load } from 'cheerio'
+import type { App, Page } from 'vuepress/core'
+import {} from 'vuepress/shared'
+import { isArray } from '../../shared/index.js'
+
+const MEDIA_WITH_ALT = ['img']
+
+const REMOVED_TAGS = [
+  // non content
+  'title',
+  'base',
+  'meta',
+  'template',
+  'script',
+  'style',
+  'canvas',
+  'slot',
+
+  // not main content
+  'nav',
+  'aside',
+  'footer',
+
+  // deleted
+  'del',
+  's',
+
+  // rich media
+  'audio',
+  'video',
+  'canvas',
+  'iframe',
+  'map',
+  'area',
+  'track',
+  'object',
+
+  // input
+  'input',
+  'textarea',
+  'select',
+  'option',
+  'optgroup',
+  'datalist',
+]
+
+export interface PageTextOptions {
+  /**
+   * Whether convert text to single line content
+   *
+   * 是否将文字转换成单行内容
+   *
+   * @default false
+   */
+  singleLine?: boolean
+
+  /**
+   * Length of text
+   *
+   * @description Text length will be the minimal possible length reaching this value
+   *
+   * 文字的长度
+   *
+   * @description 文字的长度会尽可能的接近这个值
+   *
+   * @default 300
+   */
+  length?: number
+
+  /**
+   * Tags to be removed
+   *
+   * @description Table and code blocks are removed by default.
+   *
+   * 需要移除的标签
+   *
+   * @description 默认情况下表格和代码块会被移除
+   *
+   * @default ['table', 'pre']
+   */
+  removedTags?: string[]
+}
+
+interface NodeOptions {
+  base: string
+  removedTags: string[]
+}
+
+const handleNode = (
+  node: AnyNode,
+  { base, removedTags }: NodeOptions,
+): string => {
+  if (node.type === 'tag') {
+    // toc should be dropped
+    if (
+      [node.attribs.class, node.attribs.id].some((item) =>
+        ['table-of-contents', 'toc'].includes(item),
+      )
+    )
+      return ''
+
+    // return alt text
+    if (MEDIA_WITH_ALT.includes(node.tagName)) {
+      return node.attribs.alt || ''
+    }
+
+    // html tags can be returned
+    if (
+      !REMOVED_TAGS.includes(node.tagName) &&
+      !removedTags.includes(node.tagName) &&
+      isHTMLTag(node.tagName)
+    ) {
+      return handleNodes(node.children, { base, removedTags })
+    }
+
+    return ''
+  }
+
+  if (node.type === 'text') return node.data
+
+  return ''
+}
+
+const handleNodes = (
+  nodes: AnyNode[] | null,
+  { base, removedTags }: NodeOptions,
+): string =>
+  isArray(nodes)
+    ? nodes.map((node) => handleNode(node, { base, removedTags })).join('')
+    : ''
+
+const $ = load('')
+
+export const getPageText = (
+  { options: { base } }: App,
+  { contentRendered }: Page,
+  {
+    length = 300,
+    singleLine,
+    removedTags = ['table', 'pre'],
+  }: PageTextOptions = {},
+): string => {
+  let result = ''
+  const rootNodes = $.parseHTML(contentRendered) || []
+
+  for (const node of rootNodes) {
+    const text = handleNode(node, { base, removedTags })
+
+    if (text) {
+      result += text
+      if (text.length >= length) break
+    }
+  }
+
+  return singleLine ? result.replace(/\n/g, ' ').replace(/\s+/g, ' ') : result
+}

tools/helper/tests/__fixtures__/src/markdown.md

@@ -102,7 +102,7 @@ Classic:
 |           center           |                    right | left                    |
 | :------------------------: | -----------------------: | :---------------------- |
 | For center align use `:-:` | For right align use `-:` | For left align use `:-` |
-|             b              |                aaaaaaaaa | aaaa                    |
+|             table text              |                aaaaaaaaa | aaaa                    |
 |             c              |                     aaaa | a                       |
 
 ## Codes