メタ情報を取得する

index.js

'use strict'

import { getMetaInfo }from './ogp'

const marked = require('marked')
const async = require('async')

// https://github.com/markedjs/marked/issues/1279
const linkSupport = /^!?\[((?:\[[^\[\]]*\]|\\[\[\]]?|`[^`]*`|[^\[\]\\])*?)\]\(\s*(<(?:\\[<>]?|[^\s<>\\])*>|(?:\\[()]?|\([^\s\x00-\x1f()\\]*\)|[^\s\x00-\x1f()\\])*?(?:\s+=(?:[\w%]+)?x(?:[\w%]+)?)?)(?:\s+("(?:\\"?|[^"\\])*"|'(?:\\'?|[^'\\])*'|\((?:\\\)?|[^)\\])*\)))?\s*\)/;

module.exports = async function markdownLinkViewer(markdown, opt = {}, callback) {
    let links = []

    const renderer = new marked.Renderer()

    marked.InlineLexer.rules.normal.link = linkSupport
    marked.InlineLexer.rules.gfm.link = linkSupport
    marked.InlineLexer.rules.breaks.link = linkSupport

    renderer.link = function (href, title, text) {
        links.push(href)
    }

    renderer.image = function (href, title, text) {
        href = href.replace(/ =\d*%?x\d*%?$/, "")
        links.push(href)
    }

    marked(markdown, {
        renderer: renderer
    })

    let ogpValues = []

    for (let link of links) {
        const { defaults } = await getMetaInfo(link)
        ogpValues.push(defaults)

        const props = ['title', 'description']
        for (let prop of props) {
            console.log(`og ${prop}: `)
            console.log(defaults[prop])
            console.log('-----')
        }
    }

    async.map(ogpValues, function(ogpValue, callback) {
        callback(null, ogpValue)
    }, callback)
}

ogp.js

/**
 * メタ情報を取得する
 * @param {*} url URL
 */
export const getMetaInfo = async (url) => {
    const jsdom = require('jsdom')
    const { JSDOM } = jsdom
    const virtualConsole = new jsdom.VirtualConsole()

    let defaults = {}

    try {
        const dom = await JSDOM.fromURL(url, {
            virtualConsole
        })
        const doc = dom.window.document
        // const metaEl = doc.getElementsByTagName('meta')
        const titleEl = doc.getElementsByTagName('title')
        const description = doc.querySelector('meta[name=\'description\']')

        if (titleEl.length) {
            defaults.title = titleEl[0].textContent || ''
        }

        if (description) {
            defaults.description = description.content || ''
        }

        return { defaults }
    } catch(err) {
        throw err
    }
}