如何利用node把别人的html变成你想要的json

const express = require('express')
const cheerio = require('cheerio')
const app = express()
const request = require('request')
var rp = require('request-promise');
const Iconv = require('iconv-lite')
const pc = require('phantomjs-cheerio')
var Rx = require('rxjs')


async function getImages(val) {
    return await pc.request({url: val.link}).then(($) => {
        let images = []
        $('.view-box .view-main p').each(function () {
            let image = 'http://www.xiaoliaoba.cn'.concat($(this).find('img').attr('data-src'))
            images.push(image)
        })
        val.images = images
        return val
    }).catch(function (err) {
        return val
    })
}

async function list(req, res) {
    let page = parseInt(req.params.page)
    let url = `http://www.xiaoliaoba.cn/tags/%E5%8A%A8%E6%80%81%E5%9B%BE?page=${page}`
    let option = {
        url: url,
        encoding: null,
        transform: function (body) {
            body = Iconv.decode(body, 'utf-8')
            return cheerio.load(body);
        }
    }

    let results = await rp(option).then(function ($) {
        links = []
        $('.cont-item').each(function () {
            let header = $(this).find('.cont-editor-head a img').attr('src')
            let name = $(this).find('.view-editor-name a').text()
            let time = $(this).find('.view-editor-name span').text()
            let id = $(this).children().find('.cont-list-in a').attr('href').match(/\\d+/)[0]
            let cover = $(this)
                .children('.cont-list-main')
                .find('img')
                .attr('data-src')
            let count = $(this)
                .children('.cont-list-main')
                .find('span')
                .text()
            if (cover != undefined) {
                let tmp = {
                    name: name,
                    header: header,
                    time: time,
                    title: $(this)
                        .children('.cont-list-title')
                        .text(),
                    cover: 'http://www.xiaoliaoba.cn'.concat(cover),
                    count:count,
                    id: id
                }
                links.push(tmp)
            }
        })
        return links
    }).catch(function (err) {
        console.log(err)
        return []
    })
    res.send({
        code: 200,
        data: results,
        msg: ''
    })
    //使用套一个异步
    // for (result of results) {
    //     console.log(result)
    //     result.images = await getImages(result.link)
    // }
    //使用rxjs的concatmap来做，一样的，你大爷，二次异步请求，网页加载，太耗时！！！
    // const source = Rx.Observable.of(results)
    // const example = source.pipe(Rx.Operator.concatMap(val=>getImages(val)),Rx.Operator.toArray())
    // const subscribe = example.subscribe(val =>
    //     res.send({
    //         code: 200,
    //         data: val,
    //         msg: ''
    //     })
    // );

}

app.get('/:page', function (req, res) {
    if (isNaN(req.params.page)) {
        res.send({
            msg: '请正确填写page参数 int类型',
            code: 2
        })
        return false
    }
    list(req, res)
})
module.exports = app