rikumi/scraper.js

## scraper.js
/*
  微机原理与接口技术 答案抓取器

  用法：
  1. 使用 Chrome 浏览器；
  2. 登录进入作业列表页面，要求所有作业都为已完成、且可查看答案的状态；
  3. 地址栏输入地址进入框架内部：223.3.65.249/student/Home_Management/Show_All_Homework_CN.aspx；
  4. 打开 F12 控制台，复制粘贴执行下列代码，待显示结果后，保存页面为 PDF。
*/

let initBody = function() {
  document.body.innerHTML = ''
  document.write('<style>h1,h2{display:block;margin:0;padding:15px 20px;font-size:24px;border-bottom:1px solid #f0f0f0}h2{font-weight:normal;font-size:18px}</style>')
}

if (/Show_All_Homework/i.test(location.href)) {
  let results = {}
  let links = [].slice.call(document.getElementsByTagName('a'))
  let schoolnum = /sno=(\d+)/.exec(document.cookie)[1]
  if (!schoolnum || schoolnum.length != 8) {
    alert('抓取学号失败，你需要重新登录')
  } else {
    initBody()
    document.write('<h1>抓取中…</h1>')
    Promise.all(links.map(link => {
      let homeworkTitle = link.innerText
      let homeworkLink = link.href
      let hno = /hno=(\d+)/.exec(homeworkLink)[1]
      results[homeworkTitle] = {}
      return fetch(homeworkLink, { credentials: "same-origin" }).then(r => r.text()).then(t => {
        let types = t.match(/<a class="courses_0 courses_1".*?\/a>/mg)
        return Promise.all(types.map(type => {
          let typeTitle = type.replace(/<.*?>/g, '').trim()
          let typeId = parseInt(/id="coursest(\d+)"/.exec(type)[1]) - 1
          let resultLink = 'http://223.3.65.249/student/Question_Bank_Management/Show_Mark_Question_CN.aspx'
          resultLink += '?type=' + typeId
          resultLink += '&hno=' + hno
          resultLink += '&sno=' + schoolnum
          return fetch(resultLink, { credentials: "same-origin" }).then(r => r.text()).then(t => {
            return '<h2>' + typeTitle + '</h2>' + /<body>([\s\S]*?)<\/body>/im.exec(t)[1]
          })
        })).then(array => '<h1>' + homeworkTitle + '</h1>' + array.join(''))
      })
    })).then(array => array.join('').replace(/<iframe.*?\/iframe>/img, '')).then(k => {
      initBody()
      k = k.replace(/<table id="DataList\d+_ctl\d+_results".*?\/table>/img, '')
      k = k.replace(/<img id="DataList\d+_ctl\d+_answerstatus".*?>/img, '')
      k = k.replace(/(<input.*?) checked(="checked")?(.*?>)/img, '$1$3')
      document.write(k)
    })
  }
}
	/*
	微机原理与接口技术答案抓取器

	用法：
	1. 使用 Chrome 浏览器；
	2. 登录进入作业列表页面，要求所有作业都为已完成、且可查看答案的状态；
	3. 地址栏输入地址进入框架内部：223.3.65.249/student/Home_Management/Show_All_Homework_CN.aspx；
	4. 打开 F12 控制台，复制粘贴执行下列代码，待显示结果后，保存页面为 PDF。
	*/

	let initBody = function() {
	document.body.innerHTML = ''
	document.write('<style>h1,h2{display:block;margin:0;padding:15px 20px;font-size:24px;border-bottom:1px solid #f0f0f0}h2{font-weight:normal;font-size:18px}</style>')
	}

	if (/Show_All_Homework/i.test(location.href)) {
	let results = {}
	let links = [].slice.call(document.getElementsByTagName('a'))
	let schoolnum = /sno=(\d+)/.exec(document.cookie)[1]
	if (!schoolnum \|\| schoolnum.length != 8) {
	alert('抓取学号失败，你需要重新登录')
	} else {
	initBody()
	document.write('<h1>抓取中…</h1>')
	Promise.all(links.map(link => {
	let homeworkTitle = link.innerText
	let homeworkLink = link.href
	let hno = /hno=(\d+)/.exec(homeworkLink)[1]
	results[homeworkTitle] = {}
	return fetch(homeworkLink, { credentials: "same-origin" }).then(r => r.text()).then(t => {
	let types = t.match(/<a class="courses_0 courses_1".*?\/a>/mg)
	return Promise.all(types.map(type => {
	let typeTitle = type.replace(/<.*?>/g, '').trim()
	let typeId = parseInt(/id="coursest(\d+)"/.exec(type)[1]) - 1
	let resultLink = 'http://223.3.65.249/student/Question_Bank_Management/Show_Mark_Question_CN.aspx'
	resultLink += '?type=' + typeId
	resultLink += '&hno=' + hno
	resultLink += '&sno=' + schoolnum
	return fetch(resultLink, { credentials: "same-origin" }).then(r => r.text()).then(t => {
	return '<h2>' + typeTitle + '</h2>' + /<body>([\s\S]*?)<\/body>/im.exec(t)[1]
	})
	})).then(array => '<h1>' + homeworkTitle + '</h1>' + array.join(''))
	})
	})).then(array => array.join('').replace(/<iframe.*?\/iframe>/img, '')).then(k => {
	initBody()
	k = k.replace(/<table id="DataList\d+_ctl\d+_results".*?\/table>/img, '')
	k = k.replace(/<img id="DataList\d+_ctl\d+_answerstatus".*?>/img, '')
	k = k.replace(/(<input.?) checked(="checked")?(.?>)/img, '$1$3')
	document.write(k)
	})
	}
	}