docs/lib/get-mini-toc-items.js

114 строки
4.0 KiB
JavaScript

import cheerio from 'cheerio'
import { range } from 'lodash-es'
export default function getMiniTocItems(html, maxHeadingLevel = 2, headingScope = '') {
const $ = cheerio.load(html, { xmlMode: true })
// eg `h2, h3` or `h2, h3, h4` depending on maxHeadingLevel
const selector = range(2, maxHeadingLevel + 1)
.map((num) => `${headingScope} h${num}`)
.join(', ')
const headings = $(selector)
// return an array of objects containing each heading's contents, level, and optional platform.
// Article layout uses these as follows:
// - `title` and `link` to render the mini TOC headings
// - `headingLevel` the `2` in `h2`; used for determining required indentation
// - `platform` to show or hide platform-specific headings via client JS
// H1 = highest importance, H6 = lowest importance
let mostImportantHeadingLevel
const flatToc = headings
.get()
.filter((item) => {
if (!item.parent || !item.parent.attribs) return true
// Hide any items that belong to a hidden div
const { attribs } = item.parent
return !('hidden' in attribs)
})
.map((item) => {
// remove any <span> tags including their content
$('span', item).remove()
// Capture the anchor tag nested within the header, get its href and remove it
const anchor = $('a.doctocat-link', item)
const href = anchor.attr('href')
anchor.remove()
// remove any <strong> tags but leave content
$('strong', item).map((i, el) => $(el).replaceWith($(el).contents()))
const contents = `<a class="d-block width-auto" href="${href}">${$(item).html()}</a>`
const headingLevel = parseInt($(item)[0].name.match(/\d+/)[0], 10) || 0 // the `2` from `h2`
const platform = $(item).parent('.extended-markdown').attr('class') || ''
// track the most important heading level while we're looping through the items
if (headingLevel < mostImportantHeadingLevel || mostImportantHeadingLevel === undefined) {
mostImportantHeadingLevel = headingLevel
}
return { contents, headingLevel, platform }
})
.map((item) => {
// set the indentation level for each item based on the most important
// heading level in the current article
return {
...item,
indentationLevel: item.headingLevel - mostImportantHeadingLevel,
}
})
// convert the flatToc to a nested structure to simplify semantic rendering on the client
const nestedToc = buildNestedToc(flatToc)
return nestedToc
}
// Recursively build a tree from the list of allItems
function buildNestedToc(allItems, startIndex = 0) {
const startItem = allItems[startIndex]
if (!startItem) {
return []
}
let curLevelIndentation = startItem.indentationLevel
const currentLevel = []
for (let cursor = startIndex; cursor < allItems.length; cursor++) {
const cursorItem = allItems[cursor]
const nextItem = allItems[cursor + 1]
const nextItemIsNested = nextItem && nextItem.indentationLevel > cursorItem.indentationLevel
// if it's the current indentation level, push it on and keep going
if (curLevelIndentation === cursorItem.indentationLevel) {
currentLevel.push({
...cursorItem,
items: nextItemIsNested ? buildNestedToc(allItems, cursor + 1) : [],
})
continue
}
// these items were already handled via recursion
if (curLevelIndentation < cursorItem.indentationLevel) {
continue
}
// current root indentation is _greater_ than our current cursor item,
if (curLevelIndentation > cursorItem.indentationLevel) {
// special scenario where the initial list started with "less important" headers
// so we need to reset our expectations of what level to judge the indentation on
if (startIndex === 0) {
curLevelIndentation = cursorItem.indentationLevel
currentLevel.push({
...cursorItem,
items: nextItemIsNested ? buildNestedToc(allItems, cursor + 1) : [],
})
continue
}
break
}
}
return currentLevel
}