Greasy Fork is available in English.

Youtube 双语字幕下载 v14 (中文+任选的一门双语,比如英语) (v13 开始自动字幕不再支持双语,自动字幕只能下载翻译后的中文)

字幕格式是 "中文 \n 英语"(\n 是换行符的意思)

// ==UserScript==
// @name           Youtube 双语字幕下载 v14 (中文+任选的一门双语,比如英语) (v13 开始自动字幕不再支持双语,自动字幕只能下载翻译后的中文)
// @include        https://*youtube.com/*
// @author         Cheng Zheng
// @version        14
// @copyright      Zheng Cheng
// @grant         GM_xmlhttpRequest
// @grant         unsafeWindow
// @description   字幕格式是 "中文 \n 英语"(\n 是换行符的意思)
// @license       MIT
// @namespace     https://greasyfork.org/users/5711
// @require        https://code.jquery.com/jquery-1.12.4.min.js
// @require      https://cdn.jsdelivr.net/npm/vue@2.7.14/dist/vue.js
// ==/UserScript==

/*
使用提示:
  如果本脚本不能使用,有一定概率是因为 jQuery 的 CDN 在你的网络环境下无法载入,
  就是文件顶部的这一句有问题:
  @require        https://code.jquery.com/jquery-1.12.4.min.js

  解决办法:去网上随便找一个 jQuery 的 CDN 地址,替换掉这个,比如
  https://cdn.bootcdn.net/ajax/libs/jquery/1.12.4/jquery.js
  https://cdn.staticfile.org/jquery/1.12.4/jquery.min.js

作者联系方式:
  QQ 1003211008
  邮件 guokrfans@gmail.com
  Github@1c7

本脚本使用场景:
  Tampermonkey (Chrome 上的一款插件)
  (意思是需要安装在 Tampermonkey 里来运行)

解决什么问题:
  下载中外双语的字幕,格式是 中文 \n 外语, \n 是换行符的意思。

术语说明:
  auto 自动字幕
  closed 完整字幕 (或者叫人工字幕也可以)

原理说明: 
  对于"完整字幕", Youtube 返回的时间轴完全一致,因此只需要结合在一起即可,相对比较简单。
  对于"自动字幕",中文是一个个句子,英文是一个个单词,格式不同,时间轴也不同
  因此,会基于中文的句子时间(时间轴),把英文放进去

特别感谢:
  ytian(tianyebj):解决英文字幕匹配错误的问题 (https://github.com/1c7/Youtube-Auto-Subtitle-Download/pull/11)

备忘:
  如果要把字符串保存下来, 使用: 
  downloadString(srt_string, "text/plain", file_name);

用于测试的视频: 
  https://www.youtube.com/watch?v=JfBZfnkg1uM


版本更新日志
  2022-12-23 把 v13 升级到 v14,因为 Youtube 又更新 UI 了。

  2021-12-21 把 v12 升级到 v13
  有多名用户分别在微博,微信,邮件,greasyfork, 四个渠道向我反馈了无法使用问题。
  此时版本是 v12
  测试视频1
    https://www.youtube.com/watch?v=OWaFPsVa3ig&t=16s (有1个字幕,英语(自动生成))
    绿色下拉菜单里选择 option "中文 + 英语 (自动生成)" 的确没有任何反应,无法下载。
    实测发现: 英文字幕长度 284行,中文字幕长度 162行,没法一一对应,所以出错了。
  测试视频2
    https://www.youtube.com/watch?v=3RkhZgRNC1k (有2个字幕,英语(美国),英语(自动生成))
    如果是中文+英语(美国) 那么两个都是885行,可以正常下载
    但是 中文+英语(自动生成),英语是937行,中文是471行。
  结论:
    如果是中文+完整字幕,那么长度是一一对应的。没问题
    如果是中文+自动生成字幕,那么长度不一样,就会有问题。
  解决办法:
    如果是自动字幕,只下载中文。
    完整字幕不需要做额外修改,保留现在这样就行,可以正常工作。
*/
;
(function () {
  'use strict'

  // Config
  var NO_SUBTITLE = '无字幕'
  var HAVE_SUBTITLE = '下载双语字幕 (中文+外语)'
  const NEW_LINE = '\n'
  const BUTTON_ID = 'youtube-dual-lang-downloader-by-1c7-last-update-2020-12-3'
  const anchor_element = "#above-the-fold #title";
  // Config

  var HASH_BUTTON_ID = `#${BUTTON_ID}`

  // initialize
  var first_load = true // indicate if first load this webpage or not
  var youtube_playerResponse_1c7 = null // for auto subtitle
  unsafeWindow.caption_array = [] // store all subtitle

  // trigger when first load
  // $(document).ready(function () {
  //   console.log('开始执行??');
  //   start()
  // })

  // trigger when loading new page
  // (actually this would also trigger when first loading, that's not what we want, that's why we need to use firsr_load === false)
  // (new Material design version would trigger this "yt-navigate-finish" event. old version would not.)
  var body = document.getElementsByTagName('body')[0]
  body.addEventListener('yt-navigate-finish', function (event) {
    if (current_page_is_video_page() === false) {
      return
    }
    youtube_playerResponse_1c7 = event.detail.response.playerResponse // for auto subtitle
    unsafeWindow.caption_array = [] // clean up (important, otherwise would have more and more item and cause error)

    // if use click to another page, init again to get correct subtitle
    if (first_load === false) {
      remove_subtitle_download_button()
      init()
    }
  })

  // trigger when loading new page
  // (old version would trigger "spfdone" event. new Material design version not sure yet.)
  window.addEventListener('spfdone', function (e) {
    if (current_page_is_video_page()) {
      remove_subtitle_download_button()
      var checkExist = setInterval(function () {
        if ($('#watch7-headline').length) {
          init()
          clearInterval(checkExist)
        }
      }, 330)
    }
  })

  // return true / false
  // Detect [new version UI(material design)] OR [old version UI]
  // I tested this, accurated.
  function new_material_design_version() {
    var old_title_element = document.getElementById('watch7-headline')
    if (old_title_element) {
      return false
    } else {
      return true
    }
  }

  // return true / false
  function current_page_is_video_page() {
    return get_video_id() !== null
  }

  // return string like "RW1ChiWyiZQ",  from "https://www.youtube.com/watch?v=RW1ChiWyiZQ"
  // or null
  function get_video_id() {
    return getURLParameter('v')
  }

  //https://stackoverflow.com/questions/11582512/how-to-get-url-parameters-with-javascript/11582513#11582513
  function getURLParameter(name) {
    return (
      decodeURIComponent(
        (new RegExp('[?|&]' + name + '=' + '([^&;]+?)(&|#|;|$)').exec(
          location.search
        ) || [null, ''])[1].replace(/\+/g, '%20')
      ) || null
    )
  }

  function remove_subtitle_download_button() {
    $(HASH_BUTTON_ID).remove()
  }

  // 把 console.log 包装一层, 方便"开启"/"关闭"
  // 这样可以在代码里遗留很多 console.log,实际运行时"关闭"掉不输出, 调试时"开启"
  // function logging(...args) {
  //   if(typeof(console) !== 'undefined') {
  //     console.log(...args);
  //   }
  // }


  function inject_our_script() {
    var div = document.createElement('div'),
      select = document.createElement('select'),
      option = document.createElement('option');

    div.setAttribute(
      'style',
      `display: table; 
margin-top:4px;
border: 1px solid rgb(0, 183, 90); 
cursor: pointer; color: rgb(255, 255, 255); 
border-top-left-radius: 3px; 
border-top-right-radius: 3px; 
border-bottom-right-radius: 3px; 
border-bottom-left-radius: 3px; 
background-color: #00B75A;
`
    )

    div.id = BUTTON_ID
    div.title = 'Youtube Subtitle Downloader' // display when cursor hover

    select.id = 'captions_selector'
    select.disabled = true
    select.setAttribute(
      'style',
      `display:block; 
border: 1px solid rgb(0, 183, 90); 
cursor: pointer; 
color: rgb(255, 255, 255); 
background-color: #00B75A;
padding: 4px;
`
    )

    option.textContent = 'Loading...'
    option.selected = true
    select.appendChild(option)

    // 下拉菜单里,选择一项后触发下载
    select.addEventListener(
      'change',
      function () {
        download_subtitle(this)
      },
      false
    )

    div.appendChild(select) // put <select> into <div>

    // Put the div into page: new material design
    var title_element = document.querySelectorAll(
      '.title.style-scope.ytd-video-primary-info-renderer'
    )
    if (title_element) {
      $(title_element[0]).after(div)
    }

    // 把按钮加到页面上。
    document.querySelector(anchor_element).appendChild(div)

    load_language_list(select)

    // <a> element is for download
    var a = document.createElement('a')
    a.style.cssText = 'display:none;'
    a.setAttribute('id', 'ForSubtitleDownload')
    var body = document.getElementsByTagName('body')[0]
    body.appendChild(a)
  }

  // Trigger when user select <option>
  async function download_subtitle(selector) {
    console.log('进入download_subtitle')
    // if user select first <option>
    // we just return, do nothing.
    if (selector.selectedIndex == 0) {
      return
    }

    // 核心概念
    // 对于完整字幕而言,英文和中文的时间轴是一致的,只需要一行行的 match 即可

    // 但是对于自动字幕就不是这样了,"自动字幕的英文"只能拿到一个个单词的开始时间和结束时间
    // "自动字幕的中文"只能拿到一个个句子
    // 现在的做法是,先拿到中文,处理成 SRT 格式,
    // 然后去拿英文,然后把英文的每个词,拿去和中文的每个句子的开始时间和结束时间进行对比
    // 如果"英文单词的开始时间"在"中文句子的开始-结束时间"区间内,那么认为这个英文单词属于这一句中文

    // 2021-8-11 更新
    // 自动字幕的改了,和完整字幕一样了。

    var caption = caption_array[selector.selectedIndex - 1] // because first <option> is for display, so index-1
    var lang_code = caption.lang_code
    var lang_name = caption.lang_name

    // 初始化2个变量
    var origin_url = null
    var translated_url = null

    // if user choose auto subtitle
    // 如果用户选的是自动字幕
    if (caption.lang_code == 'AUTO') {
      origin_url = get_auto_subtitle_xml_url()
      translated_url = origin_url + '&tlang=zh-Hans'
      var translated_xml = await get(translated_url)
      var translated_srt = parse_youtube_XML_to_object_list(translated_xml)
      var srt_string = object_array_to_SRT_string(translated_srt)
      var title = get_file_name(lang_name)
      downloadString(srt_string, 'text/plain', title)

      // after download, select first <option>
      selector.options[0].selected = true
      return // 别忘了 return
    }

    // 如果用户选的是完整字幕
    origin_url = await get_closed_subtitle_url(lang_code)
    translated_url = origin_url + '&tlang=zh-Hans'

    var original_xml = await get(origin_url)
    var translated_xml = await get(translated_url)

    // 根据时间轴融合这俩
    var original_srt = parse_youtube_XML_to_object_list(original_xml)
    var translated_srt = parse_youtube_XML_to_object_list(translated_xml)
    var dual_language_srt = merge_srt(original_srt, translated_srt)

    var srt_string = object_array_to_SRT_string(dual_language_srt)
    var title = get_file_name(lang_name)
    downloadString(srt_string, 'text/plain', title)

    // after download, select first <option>
    selector.options[0].selected = true
  }

  // 把两个语言的 srt 数组组合起来,
  // 比如把英文和中文的组合起来。
  // 这个函数假设两个数组的长度是一模一样的。如果不一样就会出错,比如 srt_A 是 284 个,srt_B 是 164 个元素。
  function merge_srt(srt_A, srt_B) {
    var dual_language_srt = []

    for (let index = 0; index < srt_A.length; index++) {
      const element_A = srt_A[index]
      const element_B = srt_B[index]

      var text = element_B.text + NEW_LINE + element_A.text // 中文 \n 英文
      var item = {
        startTime: element_A.startTime,
        endTime: element_A.endTime,
        text: text,
      }

      dual_language_srt.push(item)
    }
    return dual_language_srt
  }

  // Return something like: "(English)How Did Python Become A Data Science Powerhouse?.srt"
  function get_file_name(x) {
    return `(${x})${get_title()}.srt`
  }

  // Detect if "auto subtitle" and "closed subtitle" exist
  // And add <option> into <select>
  // 加载语言列表
  function load_language_list(select) {
    // auto
    var auto_subtitle_exist = false // 自动字幕是否存在(默认 false)

    // closed
    var closed_subtitle_exist = false

    // get auto subtitle
    var auto_subtitle_url = get_auto_subtitle_xml_url()
    if (auto_subtitle_url != false) {
      auto_subtitle_exist = true
    }

    // if there are "closed" subtitle?
    var captionTracks = get_captionTracks()
    if (
      captionTracks != undefined &&
      typeof captionTracks === 'object' &&
      captionTracks.length > 0
    ) {
      closed_subtitle_exist = true
    }

    // if no subtitle at all, just say no and stop
    if (auto_subtitle_exist == false && closed_subtitle_exist == false) {
      select.options[0].textContent = NO_SUBTITLE
      disable_download_button()
      return false
    }

    // if at least one type of subtitle exist
    select.options[0].textContent = HAVE_SUBTITLE
    select.disabled = false

    var option = null // for <option>
    var caption_info = null // for our custom object

    // 自动字幕
    if (auto_subtitle_exist) {
      var auto_sub_name = get_auto_subtitle_name()
      var lang_name = `${auto_sub_name} 翻译的中文`
      caption_info = {
        lang_code: 'AUTO', // later we use this to know if it's auto subtitle
        lang_name: lang_name, // for display only
      }
      caption_array.push(caption_info)

      option = document.createElement('option')
      option.textContent = caption_info.lang_name
      select.appendChild(option)
    }

    // if closed_subtitle_exist
    if (closed_subtitle_exist) {
      for (var i = 0, il = captionTracks.length; i < il; i++) {
        var caption = captionTracks[i]
        if (caption.kind == 'asr') {
          continue
        }
        let lang_code = caption.languageCode
        let lang_translated = caption.name.simpleText
        var lang_name = `中文 + ${lang_code_to_local_name(
          lang_code,
          lang_translated
        )}`
        caption_info = {
          lang_code: lang_code, // for AJAX request
          lang_name: lang_name, // display to user
        }
        caption_array.push(caption_info)
        // 注意这里是加到 caption_array, 一个全局变量, 待会要靠它来下载
        option = document.createElement('option')
        option.textContent = caption_info.lang_name
        select.appendChild(option)
      }
    }
  }

  // 禁用下载按钮
  function disable_download_button() {
    $(HASH_BUTTON_ID)
      .css('border', '#95a5a6')
      .css('cursor', 'not-allowed')
      .css('background-color', '#95a5a6')
    $('#captions_selector')
      .css('border', '#95a5a6')
      .css('cursor', 'not-allowed')
      .css('background-color', '#95a5a6')

    if (new_material_design_version()) {
      $(HASH_BUTTON_ID).css('padding', '6px')
    } else {
      $(HASH_BUTTON_ID).css('padding', '5px')
    }
  }

  // 处理时间. 比如 start="671.33"  start="37.64"  start="12" start="23.029"
  // 处理成 srt 时间, 比如 00:00:00,090    00:00:08,460    00:10:29,350
  function process_time(s) {
    s = s.toFixed(3)
    // 超棒的函数, 不论是整数还是小数都给弄成3位小数形式
    // 举个柚子:
    // 671.33 -> 671.330
    // 671 -> 671.000
    // 注意函数会四舍五入. 具体读文档

    var array = s.split('.')
    // 把开始时间根据句号分割
    // 671.330 会分割成数组: [671, 330]

    var Hour = 0
    var Minute = 0
    var Second = array[0] // 671
    var MilliSecond = array[1] // 330
    // 先声明下变量, 待会把这几个拼好就行了

    // 我们来处理秒数.  把"分钟"和"小时"除出来
    if (Second >= 60) {
      Minute = Math.floor(Second / 60)
      Second = Second - Minute * 60
      // 把 秒 拆成 分钟和秒, 比如121秒, 拆成2分钟1秒

      Hour = Math.floor(Minute / 60)
      Minute = Minute - Hour * 60
      // 把 分钟 拆成 小时和分钟, 比如700分钟, 拆成11小时40分钟
    }
    // 分钟,如果位数不够两位就变成两位,下面两个if语句的作用也是一样。
    if (Minute < 10) {
      Minute = '0' + Minute
    }
    // 小时
    if (Hour < 10) {
      Hour = '0' + Hour
    }
    // 秒
    if (Second < 10) {
      Second = '0' + Second
    }
    return Hour + ':' + Minute + ':' + Second + ',' + MilliSecond
  }

  // Copy from: https://gist.github.com/danallison/3ec9d5314788b337b682
  // Thanks! https://github.com/danallison
  // Work in Chrome 66
  // Test passed: 2018-5-19
  function downloadString(text, fileType, fileName) {
    var blob = new Blob([text], {
      type: fileType,
    })
    var a = document.createElement('a')
    a.download = fileName
    a.href = URL.createObjectURL(blob)
    a.dataset.downloadurl = [fileType, a.download, a.href].join(':')
    a.style.display = 'none'
    document.body.appendChild(a)
    a.click()
    document.body.removeChild(a)
    setTimeout(function () {
      URL.revokeObjectURL(a.href)
    }, 1500)
  }

  // https://css-tricks.com/snippets/javascript/unescape-html-in-js/
  // turn HTML entity back to text, example: &quot; should be "
  function htmlDecode(input) {
    var e = document.createElement('div')
    e.class =
      'dummy-element-for-tampermonkey-Youtube-cn-other-subtitle-script-to-decode-html-entity-2021-8-11'
    e.innerHTML = input
    return e.childNodes.length === 0 ? '' : e.childNodes[0].nodeValue
  }

  // 获得自动字幕的地址
  // return URL or null;
  // later we can send a AJAX and get XML subtitle
  // 例子输出: https://www.youtube.com/api/timedtext?v=JfBZfnkg1uM&asr_langs=de,en,es,fr,it,ja,ko,nl,pt,ru&caps=asr&exp=xftt,xctw&xorp=true&xoaf=5&hl=zh-CN&ip=0.0.0.0&ipbits=0&expire=1628691971&sparams=ip,ipbits,expire,v,asr_langs,caps,exp,xorp,xoaf&signature=55984444BD75E34DB9FE809058CCF7DE5B1AB3B5.193DC32A1E0183D8D627D229C9C111E174FF56FF&key=yt8&kind=asr&lang=en
  /*
    如果直接访问这个地址,里面的格式是 XML,比如
    <transcript>
      <text start="0.589" dur="6.121">hello in this video I would like to</text>
      <text start="3.6" dur="5.88">share what I&#39;ve learned about setting up</text>
      <text start="6.71" dur="5.08">shadows and shadow casting and shadow</text>
      <text start="9.48" dur="5.6">occlusion and stuff like that in a</text>
    </transcript>
  */
  function get_auto_subtitle_xml_url() {
    try {
      var captionTracks = get_captionTracks()
      for (var index in captionTracks) {
        var caption = captionTracks[index]
        if (typeof caption.kind === 'string' && caption.kind == 'asr') {
          return captionTracks[index].baseUrl
        }
        // ASR – A caption track generated using automatic speech recognition.
        // https://developers.google.com/youtube/v3/docs/captions
      }
    } catch (error) {
      return false
    }
  }

  // Input: lang_code like 'en'
  // Output: URL (String)
  async function get_closed_subtitle_url(lang_code) {
    try {
      var captionTracks = get_captionTracks()
      for (var index in captionTracks) {
        var caption = captionTracks[index]
        if (caption.languageCode === lang_code && caption.kind != 'asr') {
          var url = captionTracks[index].baseUrl
          return url
        }
      }
    } catch (error) {
      console.log(error)
      return false
    }
  }

  // 把 ajax 请求简单 wrap 一下方便使用
  // Usage: var result = await get(url)
  function get(url) {
    return $.ajax({
      url: url,
      type: 'get',
      success: function (r) {
        return r
      },
      fail: function (error) {
        return error
      },
    })
  }

  // Input: XML (provide by Youtube)
  // Output: Array of object
  // each object look like:
  /*
    {
      startTime: "",
      endTime: "",
      text: ""
    }
  */
  // it's intermediate representation for SRT
  function parse_youtube_XML_to_object_list(youtube_xml_string) {
    if (
      youtube_xml_string === '' ||
      youtube_xml_string === undefined ||
      youtube_xml_string === null
    ) {
      return false
    }
    var result_array = []
    var text_nodes = youtube_xml_string.getElementsByTagName('text')
    var len = text_nodes.length
    for (var i = 0; i < len; i++) {
      var text = text_nodes[i].textContent.toString()
      text = text.replace(/(<([^>]+)>)/gi, '') // remove all html tag.
      text = htmlDecode(text)

      var attr_start = text_nodes[i].getAttribute('start') // 开始时间
      var attr_dur = text_nodes[i].getAttribute('dur')

      var start = parseFloat(attr_start)
      var dur = parseFloat(attr_dur)
      var end = start + dur // 结束时间

      // 如果下一行的"开始时间", 在当前行的"结束时间"之后,那么代表这一行和下一行没有串行。屏幕上不会同时显示两行字幕。
      // 如果下一行的"开始时间",早于当前行的"结束时间",那么我们把当前行的结束时间,改为下一行的开始时间
      var next_line_index = i + 1
      if (next_line_index < len) {
        var next_line = text_nodes[next_line_index]
        var next_line_attr_start = next_line.getAttribute('start')
        var next_line_start = parseFloat(next_line_attr_start)
        if (end > next_line_start) {
          end = next_line_start
        }
      }

      var start_time = process_time(parseFloat(start))
      var end_time = process_time(parseFloat(end))

      var item = {
        startTime: start_time,
        endTime: end_time,
        text: text,
      }
      result_array.push(item)
    }

    return result_array
  }

  /*
    Input: [ {startTime: "", endTime: "", text: ""}, {...}, {...} ]
    Output: SRT
  */
  function object_array_to_SRT_string(object_array) {
    var result = ''
    var BOM = '\uFEFF'
    result = BOM + result // store final SRT result

    for (var i = 0; i < object_array.length; i++) {
      var item = object_array[i]
      var index = i + 1
      var start_time = item.startTime
      var end_time = item.endTime
      var text = item.text

      var new_line = NEW_LINE
      result = result + index + new_line

      result = result + start_time
      result = result + ' --> '
      result = result + end_time + new_line

      result = result + text + new_line + new_line
    }

    return result
  }

  // return "English (auto-generated)" or a default name;
  function get_auto_subtitle_name() {
    try {
      var captionTracks = get_captionTracks()
      for (var index in captionTracks) {
        var caption = captionTracks[index]
        if (typeof caption.kind === 'string' && caption.kind == 'asr') {
          return captionTracks[index].name.simpleText
        }
      }
      return 'Auto Subtitle'
    } catch (error) {
      return 'Auto Subtitle'
    }
  }

  // return player_response
  // or return null
  function get_youtube_data() {
    return document.getElementsByTagName('ytd-app')[0].data.playerResponse
  }

  function get_captionTracks() {
    let data = get_youtube_data()
    var captionTracks =
      data?.captions?.playerCaptionsTracklistRenderer?.captionTracks
    return captionTracks
  }

  // Input a language code, output that language name in current locale
  // 如果当前语言是中文简体, Input: "de" Output: 德语
  // if current locale is English(US), Input: "de" Output: "Germany"
  function lang_code_to_local_name(languageCode, fallback_name) {
    try {
      var captionTracks = get_captionTracks()
      for (var i in captionTracks) {
        var caption = captionTracks[i]
        if (caption.languageCode === languageCode) {
          let simpleText = captionTracks[i].name.simpleText
          if (simpleText) {
            return simpleText
          } else {
            return fallback_name
          }
        }
      }
    } catch (error) {
      return fallback_name
    }
  }

  // 获取当前视频的标题 (String), 比如 "How Does the Earth Move? Crash Course Geography #5"
  // https://www.youtube.com/watch?v=ljjLV-5Sa98
  // 获取视频标题
  function get_title() {
    // 方法1:先尝试拿到标题
    var title_element = document.querySelector(
      'h1.title.style-scope.ytd-video-primary-info-renderer'
    )
    if (title_element != null) {
      var title = title_element.innerText
      // 能拿到就返回
      if (title != undefined && title != null && title != '') {
        return title
      }
    }
    // 方法2:如果方法1失效用这个
    return ytplayer.config.args.title // 这个会 delay, 如果页面跳转了,这个获得的标题还是旧的
  }

  const wait = (ms) => new Promise((resolve) => setTimeout(resolve, ms))

  // 等待一个元素存在
  // https://stackoverflow.com/questions/5525071/how-to-wait-until-an-element-exists
  function waitForElm(selector) {
    return new Promise((resolve) => {
      if (document.querySelector(selector)) {
        return resolve(document.querySelector(selector))
      }

      const observer = new MutationObserver((mutations) => {
        if (document.querySelector(selector)) {
          resolve(document.querySelector(selector))
          observer.disconnect()
        }
      })

      observer.observe(document.body, {
        childList: true,
        subtree: true,
      })
    })
  }

  function init() {
    // console.log('进入 init');
    inject_our_script()
    first_load = false
  }

  async function main() {
    // console.log("[开始工作] Youtube 双语字幕下载");
    await waitForElm(anchor_element)
    init()
  }

  setTimeout(main, 2000);
})()