☰

OverDrive Transcriber

Transcribes books you read on OverDrive for offline reading

目前为 2018-04-25 提交的版本。查看 最新版本。

  1. // ==UserScript==
  2. // @name OverDrive Transcriber
  3. // @description Transcribes books you read on OverDrive for offline reading
  4. // @namespace Violentmonkey Scripts
  5. // @match *://*.overdrive.com/*
  6. // @match *://*.greasyfork.org/*
  7. // @grant GM_setValue
  8. // @grant GM_getValue
  9. // @grant GM_listValues
  10. // @grant GM_deleteValue
  11. // @grant GM_xmlhttpRequest
  12. // @run-at document-start
  13. // @version 0.1.3
  14. // @author qsniyg
  15. // ==/UserScript==
  16.  
  17. (function() {
  18.     var content_html_regex = /\.x?html?\?cmpt=/;
  19.     var content_html_match = /^(?:.*\/)?([^/.]*)\.x?html?/;
  20.     
  21.     var default_options = {
  22.         images: {
  23.             name: "Include images",
  24.             default: true
  25.         }
  26.     };
  27.     
  28.     var options = {};
  29.     
  30.     var img_requests = 0;
  31.     
  32.     for (var option in default_options) {
  33.         var value = GM_getValue("SETTINGS:" + option);
  34.         if (value !== undefined)
  35.             options[option] = JSON.parse(value);
  36.         else
  37.             options[option] = default_options[option].default;
  38.     }
  39.     
  40.     function onload(f) {
  41.         if (document.readyState === "interactive" || document.readyState === "complete") {
  42.             f();
  43.         } else {
  44.             document.addEventListener("DOMContentLoaded", f, false);
  45.         }
  46.     }
  47.     
  48.     function makebutton(el, bg) {
  49.         el.style.padding = ".5em 1em";
  50.         el.style.background = bg;
  51.         el.style.color = "white";
  52.         el.style.textDecoration = "none";
  53.         el.style.display = "inline-block";
  54.         el.style.margin = ".3em .5em";
  55.     }
  56.     
  57.     var fullurl = function(url, x) {
  58.         return urljoin(url, x);
  59.     };
  60.     
  61.     function urljoin(a, b) {
  62.         var protocol_split = a.split("://");
  63.         var protocol = protocol_split[0];
  64.         var splitted = protocol_split[1].split("/");
  65.         var domain = splitted[0];
  66.         var start = protocol + "://" + domain;
  67.  
  68.         if (b.length === 0)
  69.             return a;
  70.         if (b.match(/[a-z]*:\/\//))
  71.             return b;
  72.         if (b.length >= 2 && b.slice(0, 2) === "//")
  73.             return protocol + ":" + b;
  74.         if (b.length >= 1 && b.slice(0, 1) === "/")
  75.             return start + b;
  76.         
  77.         if (a.match(/\/$/))
  78.             return a + b.replace(/^\/*/, "");
  79.         else
  80.             return a.replace(/\/[^/]*$/, "/") + b.replace(/^\/*/, "");
  81.     }
  82.  
  83.     
  84.     // OverDrive section
  85.     function overdrive() {
  86.         var transcribe_btn;
  87.         var transcribed = false;
  88.         
  89.         function parse(url, content) {
  90.             var html = document.createElement("html");
  91.             html.innerHTML = content;
  92.  
  93.             var url_match = url.match(content_html_match);
  94.             if (!url_match)
  95.                 // shouldn't happen
  96.                 return;
  97.             //var book_id = url_match[2];
  98.             var book_id = bData["-odread-buid"];
  99.             var var_id = url_match[1];
  100.  
  101.             var titleel = html.getElementsByTagName("title")[0];
  102.             if (!titleel)
  103.                 return;
  104.  
  105.             var title = titleel.innerHTML;
  106.             /*console.log(book_id);
  107.             console.log(var_id);
  108.             console.log(title);*/
  109.  
  110.             GM_setValue("TITLE:" + book_id, title.toString());
  111.             GM_setValue("INFO:" + book_id, JSON.stringify(unsafeWindow.bData));
  112.  
  113.             var setcontents = function(contents) {
  114.                 GM_setValue("CONTENTS:" + book_id + ":" + var_id, contents);
  115.                 _contents = contents;
  116.             }
  117.  
  118.             var scripts = html.getElementsByTagName("script");
  119.             var regex = /^ *parent\.[^;(]*\(.*?['"](.*?)['"]/;
  120.             var set = false;
  121.             var _contents;
  122.             for (var i = 0; i < scripts.length; i++) {
  123.                 var matchobj = scripts[i].innerHTML.match(regex);
  124.                 if (matchobj) {
  125.                     var text = unsafeWindow.atob(matchobj[1]);
  126.                     setcontents(text);
  127.                     set = true;
  128.                 }
  129.             }
  130.  
  131.             if (!set) {
  132.                 if (html.querySelectorAll("body > p").length >= 0) {
  133.                     var body = html.getElementsByTagName("body")[0].cloneNode(true);
  134.                     body.removeAttribute("xmlns");
  135.                     body.removeAttribute("onload");
  136.                     setcontents(body.outerHTML);
  137.                 }
  138.             }
  139.             
  140.             if (options.images) {
  141.                 var doc = document.implementation.createHTMLDocument("preview");
  142.                 var newhtml = doc.createElement("html");
  143.                 newhtml.innerHTML = _contents;
  144.                 /*var parser = new DOMParser();
  145.                 var newhtml = parser.parseFromString(_contents, "text/xml");*/
  146.                 var images = newhtml.getElementsByTagName("img");
  147.                 for (var i = 0; i < images.length; i++) {
  148.                     console.log(images[i]);
  149.                     img_requests++;
  150.                     
  151.                     (function(src) {
  152.                         var full_url = fullurl(document.location.href, url);
  153.                         var newsrc = fullurl(full_url, src);
  154.                         new GM_xmlhttpRequest({
  155.                             method: 'GET',
  156.                             url: newsrc,
  157.                             overrideMimeType: 'text/plain; charset=x-user-defined',
  158.                             headers: {
  159.                                 "Origin": document.location.href.replace(/^([a-z]+:\/\/[^/]*).*?$/, "$1"),
  160.                                 "Referer": full_url
  161.                             },
  162.                             onload: function (resp) {
  163.                                 if (resp.status !== 200 && resp.status !== 304) {
  164.                                     console.dir(resp);
  165.                                     return;
  166.                                 }
  167.  
  168.                                 console.log(src);
  169.                                 img_requests--;
  170.                                 var retval = "";
  171.                                 for (var i = 0; i < resp.responseText.length; i++) {
  172.                                     retval += String.fromCharCode(resp.responseText.charCodeAt(i) & 0xff);
  173.                                 }
  174.                                 GM_setValue("IMAGE:" + book_id + ":" + src.replace(/.*?:\/\/[^/]*\/*/, ""), retval);
  175.  
  176.                                 if (img_requests === 0 && transcribed) {
  177.                                     transcribe_btn.innerHTML = "Done";
  178.                                 }
  179.                             }
  180.                         });
  181.                     })(images[i].getAttribute("src"));
  182.                 }
  183.             }
  184.         }
  185.  
  186.         var original_open = window.XMLHttpRequest.prototype.open;
  187.         window.XMLHttpRequest.prototype.open = function(method, url) {
  188.             if (!url)
  189.                 return;
  190.  
  191.             if (url.match(content_html_regex)) {
  192.                 this.addEventListener("readystatechange", function() {
  193.                     if (this.readyState === 4) {
  194.                         parse(url, this.responseText);
  195.                     }
  196.                 });
  197.             }
  198.             original_open.apply(this, arguments);
  199.         };
  200.  
  201.         function run_iframe(iframe) {
  202.             var ifdocument = iframe.contentDocument || iframe.contentWindow.document;
  203.             parse(iframe.src, ifdocument.documentElement.innerHTML)
  204.         }
  205.  
  206.         function find_iframes() {
  207.             return;
  208.             var iframes = document.getElementsByTagName("iframe");
  209.             for (var i = 0; i < iframes.length; i++) {
  210.                 if (iframes[i].src.match(content_html_regex)) {
  211.                     (function(iframes, i) {
  212.                         iframes[i].onload = function() {
  213.                             run_iframe(iframes[i]);
  214.                         }
  215.  
  216.                         run_iframe(iframes[i]);
  217.                     })(iframes, i);
  218.                 }
  219.             }
  220.         }
  221.         
  222.         function transcribe(el) {
  223.             el.innerHTML = "Transcribing...";
  224.             
  225.             var info = unsafeWindow.bData;
  226.             
  227.             var i = 0;
  228.             function do_request() {
  229.                 if (i >= info.spine.length) {
  230.                     console.log("Done text");
  231.                     transcribed = true;
  232.                     
  233.                     if (img_requests === 0) {
  234.                         el.innerHTML = "Done";
  235.                     }
  236.                     return;
  237.                 }
  238.                 
  239.                 var path = info.spine[i].path;
  240.                 i++;
  241.                 console.log(path);
  242.                 if (!path.match(content_html_regex)) {
  243.                     console.log("Skipping: " + path);
  244.                     do_request();
  245.                     return;
  246.                 }
  247.  
  248.                 var oReq = new XMLHttpRequest();
  249.                 oReq.addEventListener("load", do_request);
  250.                 oReq.open("GET", path);
  251.                 oReq.send();
  252.             }
  253.             
  254.             do_request();
  255.         }
  256.  
  257.         function start() {
  258.             new MutationObserver(find_iframes).observe(document.documentElement, {
  259.                 attributes: true,
  260.                 childList: true
  261.             });
  262.             find_iframes();
  263.             
  264.             if (unsafeWindow.bData) {
  265.                 var outer_div = document.createElement("div");
  266.                 outer_div.style.width = "100%";
  267.  
  268.                 transcribe_btn = document.createElement("a");
  269.                 transcribe_btn.innerHTML = "Transcribe";
  270.                 transcribe_btn.onclick = function() {
  271.                     transcribe(transcribe_btn);
  272.                 };
  273.                 transcribe_btn.href = "javascript:void(0)";
  274.                 transcribe_btn.style.zIndex = 999999999;
  275.                 transcribe_btn.style.position = "absolute";
  276.                 makebutton(transcribe_btn, "#1070a0");
  277.  
  278.                 outer_div.appendChild(transcribe_btn);
  279.                 document.body.appendChild(outer_div);
  280.             }
  281.         }
  282.  
  283.         onload(start);
  284.     }
  285.     
  286.     // GreasyFork section
  287.     function greasyfork() {
  288.         function start() {
  289.             var insert = document.getElementById("overdrive-insert");
  290.             
  291.             if (!insert) {
  292.                 var addto = document.querySelector(".script-author-description");
  293.                 addto.innerHTML = "<div id='overdrive-insert'></div>" + addto.innerHTML;
  294.             
  295.                 insert = document.getElementById("overdrive-insert");
  296.             }
  297.             
  298.             insert.innerHTML = "";
  299.             
  300.             var table = document.createElement("table");
  301.             table.style.border = "1px solid black";
  302.             table.style.background = "white";
  303.             table.style.width = "100%";
  304.             
  305.             var preview_iframe = document.createElement("iframe");
  306.             preview_iframe.style.width = "100%";
  307.             preview_iframe.style.display = "block";
  308.             preview_iframe.style.border = "0";
  309.             var preview_tr = document.createElement("tr");
  310.             preview_tr.style.padding = 0;
  311.             preview_tr.style.margin = 0;
  312.             var preview_td = document.createElement("td");
  313.             preview_td.style.padding = 0;
  314.             preview_td.style.margin = 0;
  315.             preview_td.style.borderBottom = "1px solid black";
  316.             preview_td.style.display = "none";
  317.             preview_td.setAttribute("colspan", 10);
  318.             preview_td.appendChild(preview_iframe);
  319.             preview_tr.appendChild(preview_td);
  320.             table.appendChild(preview_tr);
  321.             
  322.             var keys = GM_listValues();
  323.             if (keys.length === 0) {
  324.                 var tr = document.createElement("tr");
  325.                 var td = document.createElement("td");
  326.                 td.innerHTML = "<i>No books yet</>";
  327.                 tr.appendChild(td);
  328.                 table.appendChild(tr);
  329.             }
  330.             
  331.             for (var i = 0; i < keys.length; i++) {
  332.                 if (!keys[i].match(/^TITLE:/)) {
  333.                     continue;
  334.                 }
  335.                 
  336.                 (function(key) {
  337.                     var title = GM_getValue(key);
  338.                     var id = key.replace(/.*:/, "");
  339.                     var info = JSON.parse(GM_getValue("INFO:" + id));
  340.  
  341.                     var tr = document.createElement("tr");
  342.                     tr.style.border = "1px solid black";
  343.  
  344.                     var name_td = document.createElement("td");
  345.                     name_td.innerHTML = "<b>" + title + "</b>";
  346.                     
  347.                     var html = '<html><head><title>' + title + '</title><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /></head><body>';
  348.                     var size = 0;
  349.                     var items = {};
  350.                     var images = {};
  351.                     for (var i = 0; i < keys.length; i++) {
  352.                         if (keys[i].indexOf("CONTENTS:" + id + ":") !== 0 &&
  353.                             keys[i].indexOf("IMAGE:" + id + ":") !== 0) {
  354.                             continue;
  355.                         }
  356.  
  357.                         var contents = GM_getValue(keys[i]);
  358.                         size += contents.length;
  359.                         
  360.                         if (keys[i].indexOf("IMAGE:") === 0) {
  361.                             images[keys[i].replace(/.*:/, "")] = contents;
  362.                             continue;
  363.                         }
  364.                         
  365.                         var element = document.createElement("html");
  366.                         element.innerHTML = '<head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /></head>';
  367.                         element.innerHTML += contents;
  368.                         
  369.                         var text = element.getElementsByTagName("body")[0].innerHTML;
  370.  
  371.                         //html += text;
  372.                         items[keys[i].replace(/.*:/, "")] = text;
  373.                     }
  374.                     
  375.                     var have = 0;
  376.                     var total = 0;
  377.                     for (var i = 0; i < info.spine.length; i++) {
  378.                         if (!info.spine[i].path.match(content_html_regex))
  379.                             continue;
  380.                         total++;
  381.                         var spinematch = info.spine[i].path.match(content_html_match);
  382.                         if (!spinematch)
  383.                             continue;
  384.                         var spineid = spinematch[1];
  385.                         if (spineid in items) {
  386.                             html += items[spineid];
  387.                             have++;
  388.                         } else {
  389.                             console.log("Missing " + spineid);
  390.                         }
  391.                     }
  392.                     html += "</body></html>";
  393.                     
  394.                     if (Object.keys(images).length > 0) {
  395.                         var doc = document.implementation.createHTMLDocument("preview");
  396.                         var htmlel = doc.createElement("html");
  397.                         htmlel.innerHTML = html;
  398.                         var img_els = htmlel.getElementsByTagName("img");
  399.                         for (var i = 0; i < img_els.length; i++) {
  400.                             var src = img_els[i].getAttribute("src");
  401.                             if (src in images) {
  402.                                 var ext = src.replace(/.*\.([a-zA-Z]*).*?$/, "$1").toLowerCase();
  403.                                 var mime = "image/" + ext;
  404.                                 if (ext === "jpg")
  405.                                     mime = "image/jpeg";
  406.                                 img_els[i].setAttribute("src", "data:" + mime + ";base64," + btoa(images[src]));
  407.                             }
  408.                         }
  409.                         html = htmlel.innerHTML;
  410.                     }
  411.  
  412.                     
  413.                     var link = "data:text/html," + encodeURIComponent(html);
  414.                     name_td.innerHTML += "  (" + have + "/" + total + ")";
  415.                     
  416.  
  417.                     var preview = document.createElement("a");
  418.                     preview.onclick = function() {
  419.                         preview_iframe.src = link;
  420.                         preview_td.style.display = "table-cell";
  421.                     };
  422.                     preview.href = "javascript:void(0)";
  423.                     preview.innerHTML = "Preview";
  424.                     makebutton(preview, "#105210");
  425.                     
  426.                     var download = document.createElement("a");
  427.                     download.href = link;
  428.                     download.setAttribute("download", title + ".html");
  429.                     download.innerHTML = "Download (" + (html.length / 1024).toFixed(1) + "KB)";
  430.                     makebutton(download, "#1070a0");
  431.                     
  432.                     var del = document.createElement("a");
  433.                     del.href = "javascript:void(0)";
  434.                     del.onclick = function() {
  435.                         if (!confirm("Delete '" + title + "'?"))
  436.                             return;
  437.                         
  438.                         for (var i = 0; i < keys.length; i++) {
  439.                             if (keys[i] === ("TITLE:" + id) ||
  440.                                 keys[i] === ("INFO:" + id) ||
  441.                                 keys[i].indexOf("CONTENTS:" + id + ":") >= 0 ||
  442.                                 keys[i].indexOf("IMAGE:" + id + ":") >= 0) {
  443.                                 console.log(keys[i]);
  444.                                 GM_deleteValue(keys[i]);
  445.                             }
  446.                         }
  447.                         
  448.                         start();
  449.                     }
  450.                     del.innerHTML = "Delete (" + (size / 1024).toFixed(1) + "KB)";
  451.                     del.style.float = "right";
  452.                     makebutton(del, "#a02010");
  453.                     
  454.                     var actions_td = document.createElement("td");
  455.                     actions_td.appendChild(preview);
  456.                     actions_td.appendChild(download);
  457.                     actions_td.appendChild(del);
  458.                     
  459.                     tr.appendChild(name_td);
  460.                     tr.appendChild(actions_td);
  461.                     
  462.                     table.appendChild(tr);
  463.                 })(keys[i]);
  464.             }
  465.             
  466.             var hr_tr = document.createElement("tr");
  467.             var hr_td = document.createElement("td");
  468.             hr_td.setAttribute("colspan", 10);
  469.             hr_td.style.borderBottom = "1px solid black";
  470.             hr_tr.appendChild(hr_td);
  471.             table.appendChild(hr_tr);
  472.             
  473.             for (var option in default_options) {
  474.                 (function(option) {
  475.                     var tr = document.createElement("tr");
  476.                     var name_td = document.createElement("td");
  477.                     name_td.innerHTML = default_options[option].name;
  478.                     tr.appendChild(name_td);
  479.                     
  480.                     var value_td = document.createElement("td");
  481.                     var value_input = document.createElement("input");
  482.                     value_input.type = "checkbox";
  483.                     
  484.                     if (options[option])
  485.                         value_input.setAttribute("checked", "");
  486.                     
  487.                     value_input.onclick = function() {
  488.                         var val = false;
  489.                         if (value_input.checked)
  490.                             val = true;
  491.                         GM_setValue("SETTINGS:" + option, JSON.stringify(val));
  492.                     };
  493.                     
  494.                     value_td.appendChild(value_input);
  495.                     tr.appendChild(value_td);
  496.                     table.appendChild(tr);
  497.                 })(option);
  498.             }
  499.             
  500.             insert.appendChild(table);
  501.         }
  502.         
  503.         onload(start);
  504.     }
  505.     
  506.     if (document.location.href.match(/:\/\/[^/]*\.overdrive\.com\//))
  507.         overdrive();
  508.     else if (document.location.href.match(/\/41000-overdrive-transcriber(?:\/?[?#].*)?$/))
  509.         greasyfork();
  510. })();