☰

Amazon Video - subtitle downloader

Allows you to download subtitles from Amazon Video

As of 2023-07-05. See the latest version.

  1. // ==UserScript==
  2. // @name        Amazon Video - subtitle downloader
  3. // @description Allows you to download subtitles from Amazon Video
  4. // @license     MIT
  5. // @version     1.9.8
  6. // @namespace   tithen-firion.github.io
  7. // @match       https://*.amazon.com/*
  8. // @match       https://*.amazon.de/*
  9. // @match       https://*.amazon.co.uk/*
  10. // @match       https://*.amazon.co.jp/*
  11. // @match       https://*.primevideo.com/*
  12. // @grant       unsafeWindow
  13. // @grant       GM.xmlHttpRequest
  14. // @grant       GM_xmlhttpRequest
  15. // @require     https://greasemonkey.github.io/gm4-polyfill/gm4-polyfill.js
  16. // @require     https://cdn.jsdelivr.net/gh/Tithen-Firion/UserScripts@7bd6406c0d264d60428cfea16248ecfb4753e5e3/libraries/xhrHijacker.js?version=1.0
  17. // @require     https://cdn.jsdelivr.net/gh/Stuk/jszip@579beb1d45c8d586d8be4411d5b2e48dea018c06/dist/jszip.min.js?version=3.1.5
  18. // @require     https://cdn.jsdelivr.net/gh/eligrey/FileSaver.js@283f438c31776b622670be002caf1986c40ce90c/dist/FileSaver.min.js?version=2018-12-29
  19. // ==/UserScript==
  20.  
  21. class ProgressBar {
  22.   constructor() {
  23.     let container = document.querySelector('#userscript_progress_bars');
  24.     if(container === null) {
  25.       container = document.createElement('div');
  26.       container.id = 'userscript_progress_bars'
  27.       document.body.appendChild(container)
  28.       container.style
  29.       container.style.position = 'fixed';
  30.       container.style.top = 0;
  31.       container.style.left = 0;
  32.       container.style.width = '100%';
  33.       container.style.background = 'red';
  34.       container.style.zIndex = '99999999';
  35.     }
  36.     self.container = container;
  37.   }
  38.  
  39.   init() {
  40.     this.current = 0;
  41.     this.max = 0;
  42.  
  43.     this.progressElement = document.createElement('div');
  44.     this.progressElement.style.width = 0;
  45.     this.progressElement.style.height = '10px';
  46.     this.progressElement.style.background = 'green';
  47.  
  48.     self.container.appendChild(this.progressElement);
  49.   }
  50.  
  51.   increment() {
  52.     this.current += 1;
  53.     if(this.current <= this.max)
  54.       this.progressElement.style.width = this.current / this.max * 100 + '%';
  55.   }
  56.  
  57.   incrementMax() {
  58.     this.max += 1;
  59.     if(this.current <= this.max)
  60.       this.progressElement.style.width = this.current / this.max * 100 + '%';
  61.   }
  62.  
  63.   destroy() {
  64.     this.progressElement.remove();
  65.   }
  66. }
  67.  
  68. var progressBar = new ProgressBar();
  69.  
  70. // add CSS style
  71. var s = document.createElement('style');
  72. s.innerHTML = 'p.download:hover { cursor:pointer }';
  73. document.head.appendChild(s);
  74.  
  75. // XML to SRT
  76. function parseTTMLLine(line, parentStyle, styles) {
  77.   const topStyle = line.getAttribute('style') || parentStyle;
  78.   let prefix = '';
  79.   let suffix = '';
  80.   let italic = line.getAttribute('tts:fontStyle') === 'italic';
  81.   let bold = line.getAttribute('tts:fontWeight') === 'bold';
  82.   let ruby = line.getAttribute('tts:ruby') === 'text';
  83.   if(topStyle !== null) {
  84.     italic = italic || styles[topStyle][0];
  85.     bold = bold || styles[topStyle][1];
  86.     ruby = ruby || styles[topStyle][2];
  87.   }
  88.  
  89.   if(italic) {
  90.     prefix = '<i>';
  91.     suffix = '</i>';
  92.   }
  93.   if(bold) {
  94.     prefix += '<b>';
  95.     suffix = '</b>' + suffix;
  96.   }
  97.   if(ruby) {
  98.     prefix += '(';
  99.     suffix = ')' + suffix;
  100.   }
  101.  
  102.   let result = '';
  103.  
  104.   for(const node of line.childNodes) {
  105.     if(node.nodeType === Node.ELEMENT_NODE) {
  106.       const tagName = node.tagName.split(':').pop().toUpperCase();
  107.       if(tagName === 'BR') {
  108.         result += '\n';
  109.       }
  110.       else if(tagName === 'SPAN') {
  111.         result += parseTTMLLine(node, topStyle, styles);
  112.       }
  113.       else {
  114.         console.log('unknown node:', node);
  115.         throw 'unknown node';
  116.       }
  117.     }
  118.     else if(node.nodeType === Node.TEXT_NODE) {
  119.       result += prefix + node.textContent + suffix;
  120.     }
  121.   }
  122.  
  123.   return result;
  124. }
  125. function xmlToSrt(xmlString, lang) {
  126.   try {
  127.     let parser = new DOMParser();
  128.     var xmlDoc = parser.parseFromString(xmlString, 'text/xml');
  129.  
  130.     const styles = {};
  131.     for(const style of xmlDoc.querySelectorAll('head styling style')) {
  132.       const id = style.getAttribute('xml:id');
  133.       if(id === null) throw "style ID not found";
  134.       const italic = style.getAttribute('tts:fontStyle') === 'italic';
  135.       const bold = style.getAttribute('tts:fontWeight') === 'bold';
  136.       const ruby = style.getAttribute('tts:ruby') === 'text';
  137.       styles[id] = [italic, bold, ruby];
  138.     }
  139.  
  140.     const regionsTop = {};
  141.     for(const style of xmlDoc.querySelectorAll('head layout region')) {
  142.       const id = style.getAttribute('xml:id');
  143.       if(id === null) throw "style ID not found";
  144.       const origin = style.getAttribute('tts:origin') || "0% 80%";
  145.       const position = parseInt(origin.match(/\s(\d+)%/)[1]);
  146.       regionsTop[id] = position < 50;
  147.     }
  148.  
  149.     const topStyle = xmlDoc.querySelector('body').getAttribute('style');
  150.  
  151.     console.log(topStyle, styles, regionsTop);
  152.  
  153.     const lines = [];
  154.     const textarea = document.createElement('textarea');
  155.  
  156.     let i = 0;
  157.     for(const line of xmlDoc.querySelectorAll('body p')) {
  158.       let parsedLine = parseTTMLLine(line, topStyle, styles);
  159.       if(parsedLine != '') {
  160.         if(lang.indexOf('ar') == 0)
  161.           parsedLine = parsedLine.replace(/^(?!\u202B|\u200F)/gm, '\u202B');
  162.  
  163.         textarea.innerHTML = parsedLine;
  164.         parsedLine = textarea.value;
  165.  
  166.         const region = line.getAttribute('region');
  167.         if(regionsTop[region] === true) {
  168.           parsedLine = '{\\an8}' + parsedLine;
  169.         }
  170.  
  171.         lines.push(++i);
  172.         lines.push((line.getAttribute('begin') + ' --> ' + line.getAttribute('end')).replace(/\./g,','));
  173.         lines.push(parsedLine);
  174.         lines.push('');
  175.       }
  176.     }
  177.     return lines.join('\n');
  178.   }
  179.   catch(e) {
  180.     console.error(e);
  181.     alert('Failed to parse XML subtitle file, see browser console for more details');
  182.     return null;
  183.   }
  184. }
  185.  
  186. function sanitizeTitle(title) {
  187.   return title.replace(/[:*?"<>|\\\/]+/g, '_').replace(/ /g, '.');
  188. }
  189.  
  190. // download subs and save them
  191. function downloadSubs(url, title, downloadVars, lang) {
  192.   GM.xmlHttpRequest({
  193.     url: url,
  194.     method: 'get',
  195.     onload: function(resp) {
  196.  
  197.     progressBar.increment();
  198.     var srt = xmlToSrt(resp.responseText, lang);
  199.     if(srt === null) {
  200.       srt = resp.responseText;
  201.       title = title.replace(/\.[^\.]+$/, '.ttml2');
  202.     }
  203.     if(downloadVars) {
  204.       downloadVars.zip.file(title, srt);
  205.       --downloadVars.subCounter;
  206.       if((downloadVars.subCounter|downloadVars.infoCounter) === 0)
  207.         downloadVars.zip.generateAsync({type:"blob"})
  208.           .then(function(content) {
  209.             saveAs(content, sanitizeTitle(downloadVars.title) + '.zip');
  210.             progressBar.destroy();
  211.           });
  212.     }
  213.     else {
  214.       var blob = new Blob([srt], {type: 'text/plain;charset=utf-8'});
  215.       saveAs(blob, title, true);
  216.       progressBar.destroy();
  217.     }
  218.  
  219.     }
  220.   });
  221. }
  222.  
  223. // download episodes/movie info and start downloading subs
  224. function downloadInfo(url, downloadVars) {
  225.   var req = new XMLHttpRequest();
  226.   req.open('get', url);
  227.   req.withCredentials = true;
  228.   req.onload = function() {
  229.     var info = JSON.parse(req.response);
  230.     try {
  231.     var catalogMetadata = info.catalogMetadata;
  232.     if(typeof catalogMetadata === 'undefined')
  233.       catalogMetadata = {catalog:{type: 'MOVIE', title: info.returnedTitleRendition.asin}};
  234.     var epInfo = catalogMetadata.catalog;
  235.     var ep = epInfo.episodeNumber;
  236.     var title, season;
  237.     if(epInfo.type == 'MOVIE' || ep === 0) {
  238.       title = epInfo.title;
  239.       downloadVars.title = title;
  240.     }
  241.     else {
  242.       info.catalogMetadata.family.tvAncestors.forEach(function(tvAncestor) {
  243.         switch(tvAncestor.catalog.type) {
  244.           case 'SEASON':
  245.             season = tvAncestor.catalog.seasonNumber;
  246.             break;
  247.           case 'SHOW':
  248.             title = tvAncestor.catalog.title;
  249.             break;
  250.         }
  251.       });
  252.       title += '.S' + season.toString().padStart(2, '0');
  253.       if(downloadVars.type === 'all')
  254.         downloadVars.title = title;
  255.       title += 'E' + ep.toString().padStart(2, '0');
  256.       if(downloadVars.type === 'one')
  257.         downloadVars.title = title;
  258.       title += '.' + epInfo.title;
  259.     }
  260.     title = sanitizeTitle(title);
  261.     title += '.WEBRip.Amazon.';
  262.     var languages = new Set();
  263.  
  264.     var forced = info.forcedNarratives || [];
  265.     forced.forEach(function(forcedInfo) {
  266.       forcedInfo.languageCode += '-forced';
  267.     });
  268.  
  269.     var subs = (info.subtitleUrls || []).concat(forced);
  270.  
  271.     subs.forEach(function(subInfo) {
  272.       let lang = subInfo.languageCode;
  273.       if(subInfo.type === 'subtitle' || subInfo.type === 'subtitle') {}
  274.       else if(subInfo.type === 'shd')
  275.         lang += '[cc]';
  276.       else
  277.         lang += `[${subInfo.type}]`;
  278.       if(languages.has(lang)) {
  279.         let index = 0;
  280.         let newLang;
  281.         do {
  282.           newLang = `${lang}_${++index}`;
  283.         } while(languages.has(newLang));
  284.         lang = newLang;
  285.       }
  286.       languages.add(lang);
  287.       ++downloadVars.subCounter;
  288.       progressBar.incrementMax();
  289.       downloadSubs(subInfo.url, title + lang + '.srt', downloadVars, lang);
  290.     });
  291.     }
  292.     catch(e) {
  293.       console.log(info);
  294.       alert(e);
  295.     }
  296.     if(--downloadVars.infoCounter === 0 && downloadVars.subCounter === 0) {
  297.       alert("No subs found, make sure you're logged in and you have access to watch this video!");
  298.       progressBar.destroy();
  299.     }
  300.   };
  301.   req.send(null);
  302. }
  303.  
  304. function downloadThis(e) {
  305.   progressBar.init();
  306.   var id = e.target.getAttribute('data-id');
  307.   var downloadVars = {
  308.     type: 'one',
  309.     subCounter: 0,
  310.     infoCounter: 1,
  311.     zip: new JSZip()
  312.   };
  313.   downloadInfo(gUrl + id, downloadVars);
  314. }
  315. function downloadAll(e) {
  316.   progressBar.init();
  317.   var IDs = e.target.getAttribute('data-id').split(';');
  318.   var downloadVars = {
  319.     type: 'all',
  320.     subCounter: 0,
  321.     infoCounter: IDs.length,
  322.     zip: new JSZip()
  323.   };
  324.   IDs.forEach(function(id) {
  325.     downloadInfo(gUrl + id, downloadVars);
  326.   });
  327. }
  328.  
  329. // remove unnecessary parameters from URL
  330. function parseURL(url) {
  331.   var filter = ['consumptionType', 'deviceID', 'deviceTypeID', 'firmware', 'gascEnabled', 'marketplaceID', 'userWatchSessionId', 'videoMaterialType', 'clientId', 'operatingSystemName', 'operatingSystemVersion', 'customerID', 'token'];
  332.   var urlParts = url.split('?');
  333.   var params = ['desiredResources=CatalogMetadata%2CSubtitleUrls%2CForcedNarratives'];
  334.   urlParts[1].split('&').forEach(function(param) {
  335.     var p = param.split('=');
  336.     if(filter.indexOf(p[0]) > -1)
  337.       params.push(param);
  338.   });
  339.   params.push('resourceUsage=CacheResources');
  340.   params.push('titleDecorationScheme=primary-content');
  341.   params.push('subtitleFormat=TTMLv2');
  342.   params.push('asin=');
  343.   urlParts[1] = params.join('&');
  344.   return urlParts.join('?');
  345. }
  346.  
  347. function createDownloadButton(id, type) {
  348.   var p = document.createElement('p');
  349.   p.classList.add('download');
  350.   p.setAttribute('data-id', id);
  351.   p.innerHTML = 'Download subs for this ' + type;
  352.   p.addEventListener('click', (type == 'season' ? downloadAll : downloadThis));
  353.   return p;
  354. }
  355.  
  356. function findMovieID() {
  357.   for(const templateElement of document.querySelectorAll('script[type="text/template"]')) {
  358.     let data;
  359.     try {
  360.       data = JSON.parse(templateElement.innerHTML);
  361.     }
  362.     catch(ignore) {
  363.       continue;
  364.     }
  365.     const args = data.initArgs || data.args;
  366.     if(typeof args !== 'undefined' && typeof args.titleID !== 'undefined')
  367.       return args.titleID;
  368.   }
  369.   throw Error("Couldn't find movie ID");
  370. }
  371.  
  372. function allLoaded(resolve, epCount) {
  373.   if(epCount !== document.querySelectorAll('.js-node-episode-container, li[id^=av-ep-episodes-]').length)
  374.     resolve();
  375.   else
  376.     window.setTimeout(allLoaded, 200, resolve, epCount);
  377. }
  378.  
  379. function showAll() {
  380.   return new Promise(resolve => {
  381.     let btn = document.querySelector('[data-automation-id="ep-expander"]');
  382.     if(btn === null)
  383.       resolve();
  384.  
  385.     let epCount = document.querySelectorAll('.js-node-episode-container, li[id^=av-ep-episodes-]').length;
  386.     btn.click();
  387.     allLoaded(resolve, epCount);
  388.   });
  389. }
  390.  
  391. // add download buttons
  392. async function init(url) {
  393.   initialied = true;
  394.   gUrl = parseURL(url);
  395.   console.log(gUrl);
  396.  
  397.   await showAll();
  398.  
  399.   let button;
  400.   let epElems = document.querySelectorAll('.dv-episode-container, .avu-context-card, .js-node-episode-container, li[id^=av-ep-episodes-]');
  401.   if(epElems.length > 0) {
  402.     let IDs = [];
  403.     for(let i=epElems.length; i--; ) {
  404.       let selector, id, el;
  405.       if((el = epElems[i].querySelector('input[name="highlight-list-selector"]')) !== null) {
  406.         id = el.id.replace('selector-', '');
  407.         selector = '.js-episode-offers';
  408.       }
  409.       else if((el = epElems[i].querySelector('input[name="ep-list-selector"]')) !== null) {
  410.         id = el.value;
  411.         selector = '.av-episode-meta-info';
  412.       }
  413.       else if(id = epElems[i].getAttribute('data-aliases'))
  414.         selector = '.dv-el-title';
  415.       else
  416.         continue;
  417.       id = id.split(',')[0];
  418.       epElems[i].querySelector(selector).parentNode.appendChild(createDownloadButton(id, 'episode'));
  419.       IDs.push(id);
  420.     }
  421.     button = createDownloadButton(IDs.join(';'), 'season');
  422.   }
  423.   else {
  424.     let id = findMovieID();
  425.     id = id.split(',')[0];
  426.     button = createDownloadButton(id, 'movie');
  427.   }
  428.   document.querySelector('.dv-node-dp-badges, .av-badges').appendChild(button);
  429. }
  430.  
  431. var initialied = false, gUrl;
  432. // hijack xhr, we need to find out tokens and other parameters needed for subtitle info
  433. xhrHijacker(function(xhr, id, origin, args) {
  434.   if(!initialied && origin === 'open')
  435.     if(args[1].indexOf('/GetPlaybackResources') > -1) {
  436.       init(args[1])
  437.         .catch(error => {
  438.           console.log(error);
  439.           alert(`subtitle downloader error: ${error.message}`);
  440.         });
  441.     }
  442. });