☰

HTML2FB2Lib

This is a library for converting HTML to FB2.

当前为 2023-06-23 提交的版本，查看 最新版本。

此脚本不应直接安装。它是供其他脚本使用的外部库，要使用该库请加入元指令 // @require https://update.cn-greasyfork.org/scripts/468831/1209382/HTML2FB2Lib.js 

  1. // ==UserScript==
  2. // @name           HTML2FB2Lib
  3. // @namespace      90h.yy.zz
  4. // @version        0.4.0
  5. // @author         Ox90
  6. // @match          http://*
  7. // @match          https://*
  8. // @description    This is a library for converting HTML to FB2.
  9. // @run-at         document-start
  10. // @license        MIT
  11. // ==/UserScript==
  12.  
  13. class FB2Parser {
  14.   constructor() {
  15.     this._stop = null;
  16.   }
  17.  
  18.   async parse(htmlNode, fromNode) {
  19.     const that = this;
  20.     async function _parse(node, from, fb2el, depth) {
  21.       let n = from || node.firstChild;
  22.       while (n) {
  23.         const nn = that.startNode(n, depth);
  24.         if (nn) {
  25.           const f = that.processElement(FB2Element.fromHTML(nn, false), depth);
  26.           if (f) {
  27.             if (fb2el) fb2el.children.push(f);
  28.             await _parse(nn, null, f, depth + 1);
  29.           }
  30.           that.endNode(nn, depth);
  31.         }
  32.         if (that._stop) break;
  33.         n = n.nextSibling;
  34.       }
  35.     }
  36.     await _parse(htmlNode, fromNode, null, 0);
  37.     return this._stop;
  38.   }
  39.  
  40.   startNode(node, depth) {
  41.     return node;
  42.   }
  43.  
  44.   processElement(fb2el, depth) {
  45.     return fb2el;
  46.   }
  47.  
  48.   endNode(node, depth) {
  49.   }
  50. }
  51.  
  52. class FB2Document {
  53.   constructor() {
  54.     this.binaries = [];
  55.     this.bookAuthors = [];
  56.     this.annotation = null;
  57.     this.genres = [];
  58.     this.chapters = [];
  59.     this.xmldoc = null;
  60.   }
  61.  
  62.   toString() {
  63.     this._ensureXMLDocument();
  64.     const root = this.xmldoc.documentElement;
  65.     this._markBinaries();
  66.     root.appendChild(this._makeDescriptionElement());
  67.     root.appendChild(this._makeBodyElement());
  68.     this._makeBinaryElements().forEach(el => root.appendChild(el));
  69.     const res = (new XMLSerializer()).serializeToString(this.xmldoc);
  70.     this.xmldoc = null;
  71.     return res;
  72.   }
  73.  
  74.   createElement(name) {
  75.     this._ensureXMLDocument();
  76.     return this.xmldoc.createElementNS(this.xmldoc.documentElement.namespaceURI, name);
  77.   }
  78.  
  79.   createTextNode(value) {
  80.     this._ensureXMLDocument();
  81.     return this.xmldoc.createTextNode(value);
  82.   }
  83.  
  84.   createDocumentFragment() {
  85.     this._ensureXMLDocument();
  86.     return this.xmldoc.createDocumentFragment();
  87.   }
  88.  
  89.   _ensureXMLDocument() {
  90.     if (!this.xmldoc) {
  91.       this.xmldoc = new DOMParser().parseFromString(
  92.         '<?xml version="1.0" encoding="UTF-8"?><FictionBook xmlns="http://www.gribuser.ru/xml/fictionbook/2.0"/>',
  93.         "application/xml"
  94.       );
  95.       this.xmldoc.documentElement.setAttribute("xmlns:l", "http://www.w3.org/1999/xlink");
  96.     }
  97.   }
  98.  
  99.   _makeDescriptionElement() {
  100.     const desc = this.createElement("description");
  101.     // title-info
  102.     const t_info = this.createElement("title-info");
  103.     desc.appendChild(t_info);
  104.     this.genres.forEach(g => t_info.appendChild(g.xml(this)));
  105.     (this.bookAuthors.length ? this.bookAuthors : [ new FB2Author("Неизвестный автор") ]).forEach(a => {
  106.       t_info.appendChild(a.xml(this));
  107.     });
  108.     t_info.appendChild((new FB2Element("book-title", this.bookTitle)).xml(this));
  109.     if (this.annotation) t_info.appendChild(this.annotation.xml(this));
  110.     if (this.keywords) t_info.appendChild(this.keywords.xml(this));
  111.     if (this.bookDate) {
  112.       const el = this.createElement("date");
  113.       el.setAttribute("value", FB2Utils.dateToAtom(this.bookDate));
  114.       el.textContent = this.bookDate.getFullYear();
  115.       t_info.appendChild(el);
  116.     }
  117.     if (this.coverpage) {
  118.       const el = this.createElement("coverpage");
  119.       (Array.isArray(this.coverpage) ? this.coverpage : [ this.coverpage ]).forEach(img => {
  120.         el.appendChild(img.xml(this));
  121.       });
  122.       t_info.appendChild(el);
  123.     }
  124.     const lang = this.createElement("lang");
  125.     lang.textContent = "ru";
  126.     t_info.appendChild(lang);
  127.     if (this.sequence) {
  128.       const el = this.createElement("sequence");
  129.       el.setAttribute("name", this.sequence.name);
  130.       if (this.sequence.number) el.setAttribute("number", this.sequence.number);
  131.       t_info.appendChild(el);
  132.     }
  133.     // document-info
  134.     const d_info = this.createElement("document-info");
  135.     desc.appendChild(d_info);
  136.     d_info.appendChild((new FB2Author("Ox90")).xml(this));
  137.     if (this.programName) d_info.appendChild((new FB2Element("program-used", this.programName)).xml(this));
  138.     d_info.appendChild((() => {
  139.       const f_time = new Date();
  140.       const el = this.createElement("date");
  141.       el.setAttribute("value", FB2Utils.dateToAtom(f_time));
  142.       el.textContent = f_time.toUTCString();
  143.       return el;
  144.     })());
  145.     if (this.sourceURL) {
  146.       d_info.appendChild((new FB2Element("src-url", this.sourceURL)).xml(this));
  147.     }
  148.     d_info.appendChild((new FB2Element("id", this._genBookId())).xml(this));
  149.     d_info.appendChild((new FB2Element("version", "1.0")).xml(this));
  150.     return desc;
  151.   }
  152.  
  153.   _makeBodyElement() {
  154.     const body = this.createElement("body");
  155.     if (this.bookTitle || this.bookAuthors.length) {
  156.       const title = this.createElement("title");
  157.       body.appendChild(title);
  158.       if (this.bookAuthors.length) title.appendChild((new FB2Paragraph(this.bookAuthors.join(", "))).xml(this));
  159.       if (this.bookTitle) title.appendChild((new FB2Paragraph(this.bookTitle)).xml(this));
  160.     }
  161.     this.chapters.forEach(ch => body.appendChild(ch.xml(this)));
  162.     return body;
  163.   }
  164.  
  165.   _markBinaries() {
  166.     let idx = 0;
  167.     this.binaries.forEach(img => {
  168.       if (!img.id) img.id = "image" + (++idx) + img.suffix();
  169.     });
  170.   }
  171.  
  172.   _makeBinaryElements() {
  173.     return this.binaries.reduce((list, img) => {
  174.       if (img.value) list.push(img.xmlBinary(this));
  175.       return list;
  176.     }, []);
  177.   }
  178.  
  179.   _genBookId() {
  180.     let str = this.sourceURL || this.bookTitle || "";
  181.     let hash = 0;
  182.     const slen = str.length;
  183.     for (let i = 0; i < slen; ++i) {
  184.       const ch = str.charCodeAt(i);
  185.       hash = ((hash << 5) - hash) + ch;
  186.       hash = hash & hash; // Convert to 32bit integer
  187.     }
  188.     return this.idPrefix || "h2f2l_" + Math.abs(hash).toString() + (hash > 0 ? "1" : "");
  189.   }
  190. }
  191.  
  192. class FB2Element {
  193.   constructor(name, value) {
  194.     this.name = name;
  195.     this.value = value !== undefined ? value : null;
  196.     this.children = [];
  197.   }
  198.  
  199.   static fromHTML(node, recursive) {
  200.     let fb2el = null;
  201.     const names = new Map([
  202.       [ "U", "emphasis" ], [ "EM", "emphasis" ], [ "EMPHASIS", "emphasis" ], [ "I", "emphasis" ],
  203.       [ "S", "strikethrough" ], [ "DEL", "strikethrough" ], [ "STRIKE", "strikethrough" ],
  204.       [ "STRONG", "strong" ], [ "BLOCKQUOTE", "cite" ],
  205.       [ "#comment", null ]
  206.     ]);
  207.     const node_name = node.nodeName;
  208.     if (names.has(node_name)) {
  209.       const name = names.get(node_name);
  210.       if (!name) return null;
  211.       fb2el = new FB2Element(names.get(node_name));
  212.     } else {
  213.       switch (node_name) {
  214.         case "#text":
  215.           return new FB2Text(node.textContent);
  216.         case "SPAN":
  217.           fb2el = new FB2Text();
  218.           break;
  219.         case "P":
  220.         case "LI":
  221.           fb2el = new FB2Paragraph();
  222.           break;
  223.         case "SUBTITLE":
  224.           fb2el = new FB2Subtitle();
  225.           break;
  226.         case "A":
  227.           fb2el = new FB2Link(node.href || node.getAttribute("l:href"));
  228.           break;
  229.         case "OL":
  230.           fb2el = new FB2OrderedList();
  231.           break;
  232.         case "UL":
  233.           fb2el = new FB2UnorderedList();
  234.           break;
  235.         case "BR":
  236.           return new FB2EmptyLine();
  237.         case "HR":
  238.           return new FB2Paragraph("---");
  239.         case "IMG":
  240.           return new FB2Image(node.src);
  241.         default:
  242.           throw new FB2UnknownNodeError("Неизвестный HTML блок: " + node.nodeName);
  243.       }
  244.     }
  245.     if (recursive) fb2el.appendContentFromHTML(node);
  246.     return fb2el;
  247.   }
  248.  
  249.   hasValue() {
  250.     return ((this.value !== undefined && this.value !== null) || !!this.children.length);
  251.   }
  252.  
  253.   setContentFromHTML(data, fb2doc, log) {
  254.     this.children = [];
  255.     this.appendContentFromHTML(data, fb2doc, log);
  256.   }
  257.  
  258.   appendContentFromHTML(data, fb2doc, log) {
  259.     for (const node of data.childNodes) {
  260.       let fe = FB2Element.fromHTML(node, true);
  261.       if (fe) this.children.push(fe);
  262.     }
  263.   }
  264.  
  265.   normalize() {
  266.     const _normalize = function(list) {
  267.       let done = true;
  268.       let res_list = list.reduce((accum, cur_el) => {
  269.         accum.push(cur_el);
  270.         const tmp_ch = cur_el.children;
  271.         cur_el.children = [];
  272.         tmp_ch.forEach(el => {
  273.           if (el instanceof FB2EmptyLine || el instanceof FB2Subtitle) {
  274.             accum.push(el);
  275.             const nm = cur_el.name;
  276.             cur_el = new cur_el.constructor();
  277.             if (!cur_el.name) cur_el.name = nm;
  278.             accum.push(cur_el);
  279.             done = false;
  280.           } else {
  281.             let cnt = 0;
  282.             el.normalize().forEach(e => {
  283.               if (!e.value && e.children.length === 1 && e.name === e.children[0].name) {
  284.                 e = e.children[0];
  285.               }
  286.               if (e !== el) done = false;
  287.               if (e.hasValue()) cur_el.children.push(e);
  288.             });
  289.           }
  290.         });
  291.         return accum;
  292.       }, []);
  293.       return { list: res_list, done: done };
  294.     }
  295.     //--
  296.     let result = _normalize([ this ]);
  297.     while (!result.done) {
  298.       result = _normalize(result.list);
  299.     }
  300.     return result.list;
  301.   }
  302.  
  303.   xml(doc) {
  304.     const el = doc.createElement(this.name);
  305.     if (this.value !== null) el.textContent = this.value;
  306.     this.children.forEach(ch => el.appendChild(ch.xml(doc)));
  307.     return el;
  308.   }
  309. }
  310.  
  311. class FB2BlockElement extends FB2Element {
  312.   normalize() {
  313.     // Предварительная нормализация
  314.     this.children = this.children.reduce((list, ch) => {
  315.       ch.normalize().forEach(cc => list.push(cc));
  316.       return list;
  317.     }, []);
  318.     // Удалить пустоты справа
  319.     while (this.children.length) {
  320.       const el = this.children[this.children.length - 1];
  321.       if (el instanceof FB2Text) el.trimRight();
  322.       if (!el.hasValue()) {
  323.         this.children.pop();
  324.         continue;
  325.       }
  326.       break;
  327.     }
  328.     // Удалить пустоты слева
  329.     while (this.children.length) {
  330.       const el = this.children[0];
  331.       if (el instanceof FB2Text) el.trimLeft();
  332.       if (!el.hasValue()) {
  333.         this.children.shift();
  334.         continue;
  335.       }
  336.       break;
  337.     }
  338.     // Окончательная нормализация
  339.     return super.normalize();
  340.   }
  341. }
  342.  
  343. /**
  344.  * FB2 элемент верхнего уровня section
  345.  */
  346. class FB2Chapter extends FB2Element {
  347.   constructor(title) {
  348.     super("section");
  349.     this.title = title;
  350.   }
  351.  
  352.   normalize() {
  353.     // Обернуть текстовые ноды в параграфы и удалить пустые элементы
  354.     this.children = this.children.reduce((list, el) => {
  355.       if (el instanceof FB2Text) {
  356.         const pe = new FB2Paragraph();
  357.         pe.children.push(el);
  358.         el = pe;
  359.       }
  360.       el.normalize().forEach(el => {
  361.         if (el.hasValue()) list.push(el);
  362.       });
  363.       return list;
  364.     }, []);
  365.     return [ this ];
  366.   }
  367.  
  368.   xml(doc) {
  369.     const el = super.xml(doc);
  370.     if (this.title) {
  371.       const t_el = doc.createElement("title");
  372.       const p_el = doc.createElement("p");
  373.       p_el.textContent = this.title;
  374.       t_el.appendChild(p_el);
  375.       el.prepend(t_el);
  376.     }
  377.     return el;
  378.   }
  379. }
  380.  
  381. /**
  382.  * FB2 элемент верхнего уровня annotation
  383.  */
  384. class FB2Annotation extends FB2Element {
  385.   constructor() {
  386.     super("annotation");
  387.   }
  388.  
  389.   normalize() {
  390.     // Обернуть неформатированный текст, разделенный <br> в параграфы
  391.     let lp = null;
  392.     const newParagraph = list => {
  393.       lp = new FB2Paragraph();
  394.       list.push(lp);
  395.     };
  396.     this.children = this.children.reduce((list, el) => {
  397.       if (el.name === "empty-line") {
  398.         newParagraph(list);
  399.       } else if (el instanceof FB2BlockElement) {
  400.         list.push(el);
  401.         lp = null;
  402.       } else {
  403.         if (!lp) newParagraph(list);
  404.         lp.children.push(el);
  405.       }
  406.       return list;
  407.     }, []);
  408.     // Запустить собственную нормализацию дочерних элементов
  409.     // чтобы предотвратить их дальнейшее всплытие
  410.     this.children = this.children.reduce((list, el) => {
  411.       el.normalize().forEach(el => {
  412.         if (el.hasValue()) list.push(el);
  413.       });
  414.       return list;
  415.     }, []);
  416.   }
  417. }
  418.  
  419. class FB2Subtitle extends FB2BlockElement {
  420.   constructor(value) {
  421.     super("subtitle", value);
  422.   }
  423. }
  424.  
  425. class FB2Paragraph extends FB2BlockElement {
  426.   constructor(value) {
  427.     super("p", value);
  428.   }
  429. }
  430.  
  431. class FB2EmptyLine extends FB2Element {
  432.   constructor() {
  433.     super("empty-line");
  434.   }
  435.  
  436.   hasValue() {
  437.     return true;
  438.   }
  439. }
  440.  
  441. class FB2Text extends FB2Element {
  442.   constructor(value) {
  443.     super("text", value);
  444.   }
  445.  
  446.   trimLeft() {
  447.     if (typeof(this.value) === "string") this.value = this.value.trimLeft() || null;
  448.     if (!this.value) {
  449.       while (this.children.length) {
  450.         const first_child = this.children[0];
  451.         if (first_child instanceof FB2Text) first_child.trimLeft();
  452.         if (first_child.hasValue()) break;
  453.         this.children.shift();
  454.       }
  455.     }
  456.   }
  457.  
  458.   trimRight() {
  459.     while (this.children.length) {
  460.       const last_child = this.children[this.children.length - 1];
  461.       if (last_child instanceof FB2Text) last_child.trimRight();
  462.       if (last_child.hasValue()) break;
  463.       this.children.pop();
  464.     }
  465.     if (!this.children.length && typeof(this.value) === "string") {
  466.       this.value = this.value.trimRight() || null;
  467.     }
  468.   }
  469.  
  470.   xml(doc) {
  471.     if (!this.value && this.children.length) {
  472.       let fr = doc.createDocumentFragment();
  473.       for (const ch of this.children) {
  474.         fr.appendChild(ch.xml(doc));
  475.       }
  476.       return fr;
  477.     }
  478.     return doc.createTextNode(this.value);
  479.   }
  480. }
  481.  
  482. class FB2Link extends FB2Element {
  483.   constructor(href) {
  484.     super("a");
  485.     this.href = href;
  486.   }
  487.  
  488.   xml(doc) {
  489.     const el = super.xml(doc);
  490.     el.setAttribute("l:href", this.href);
  491.     return el;
  492.   }
  493. }
  494.  
  495. class FB2OrderedList extends FB2Element {
  496.   constructor() {
  497.     super("list");
  498.   }
  499.  
  500.   xml(doc) {
  501.     const fr = doc.createDocumentFragment();
  502.     let pos = 0;
  503.     for (const ch of this.children) {
  504.       const ch_el = ch.xml(doc);
  505.       if (ch.hasValue()) {
  506.         ++pos;
  507.         ch_el.prepend(`${pos}. `);
  508.       }
  509.       fr.appendChild(ch_el);
  510.     }
  511.     return fr;
  512.   }
  513. }
  514.  
  515. class FB2UnorderedList extends FB2Element {
  516.   constructor() {
  517.     super("list");
  518.   }
  519.  
  520.   xml(doc) {
  521.     const fr = doc.createDocumentFragment();
  522.     for (const ch of this.children) {
  523.       if (ch.hasValue()) fr.appendChild(ch.xml(doc));
  524.     }
  525.     return fr;
  526.   }
  527. }
  528.  
  529. class FB2Author extends FB2Element {
  530.   constructor(s) {
  531.     super("author");
  532.     const a = s.split(" ");
  533.     switch (a.length) {
  534.       case 1:
  535.         this.nickName = s;
  536.         break;
  537.       case 2:
  538.         this.firstName = a[0];
  539.         this.lastName = a[1];
  540.         break;
  541.       default:
  542.         this.firstName = a[0];
  543.         this.middleName = a.slice(1, -1).join(" ");
  544.         this.lastName = a[a.length - 1];
  545.         break;
  546.     }
  547.     this.homePage = null;
  548.   }
  549.  
  550.   hasValue() {
  551.     return (!!this.firstName || !!this.lastName || !!this.middleName);
  552.   }
  553.  
  554.   toString() {
  555.     if (!this.firstName) return this.nickName;
  556.     return [ this.firstName, this.middleName, this.lastName ].reduce((list, name) => {
  557.       if (name) list.push(name);
  558.       return list;
  559.     }, []).join(" ");
  560.   }
  561.  
  562.   xml(doc) {
  563.     let a_el = super.xml(doc);
  564.     [
  565.       [ "first-name", this.firstName ], [ "middle-name", this.middleName ],
  566.       [ "last-name", this.lastName ], [ "home-page", this.homePage ],
  567.       [ "nickname", this.nickName ]
  568.     ].forEach(it => {
  569.       if (it[1]) {
  570.         const e = doc.createElement(it[0]);
  571.         e.textContent = it[1];
  572.         a_el.appendChild(e);
  573.       }
  574.     });
  575.     return a_el;
  576.   }
  577. }
  578.  
  579. class FB2Image extends FB2Element {
  580.   constructor(value) {
  581.     super("image");
  582.     if (typeof(value) === "string") {
  583.       this.url = value;
  584.     } else {
  585.       this.value = value;
  586.     }
  587.   }
  588.  
  589.   async load(onprogress) {
  590.     if (this.url) {
  591.       const bin = await this._load(this.url, { responseType: "binary", onprogress: onprogress });
  592.       this.type = bin.type;
  593.       this.size = bin.size;
  594.       return new Promise((resolve, reject) => {
  595.         const reader = new FileReader();
  596.         reader.addEventListener("loadend", (event) => resolve(event.target.result));
  597.         reader.readAsDataURL(bin);
  598.       }).then(base64str => {
  599.         this.value = base64str.substr(base64str.indexOf(",") + 1);
  600.       }).catch(err => {
  601.         throw new Error("Ошибка загрузки изображения");
  602.       });
  603.     }
  604.   }
  605.  
  606.   hasValue() {
  607.     return true;
  608.   }
  609.  
  610.   xml(doc) {
  611.     if (this.value) {
  612.       const el = doc.createElement(this.name);
  613.       el.setAttribute("l:href", "#" + this.id);
  614.       return el
  615.     }
  616.     const id = this.id || "изображение";
  617.     return doc.createTextNode(`[ ${id} ]`);
  618.   }
  619.  
  620.   xmlBinary(doc) {
  621.     const el = doc.createElement("binary");
  622.     el.setAttribute("id", this.id);
  623.     el.setAttribute("content-type", this.type);
  624.     el.textContent = this.value
  625.     return el;
  626.   }
  627.  
  628.   suffix() {
  629.     switch (this.type) {
  630.       case "image/png":
  631.         return ".png";
  632.       case "image/jpeg":
  633.         return ".jpg";
  634.       case "image/gif":
  635.         return ".gif";
  636.       case "image/webp":
  637.         return ".webp";
  638.     }
  639.     return "";
  640.   }
  641.  
  642.   async _load(...args) {
  643.     return FB2Loader.addJob(...args);
  644.   }
  645. }
  646.  
  647. class FB2Loader {
  648.   static async addJob(url, params) {
  649.     params ||= {};
  650.     const fp = {};
  651.     fp.method = params.method || "GET";
  652.     fp.credentials = "same-origin";
  653.     fp.signal = this._getSignal();
  654.     const resp = await fetch(url, fp);
  655.     if (!resp.ok) throw new Error(`Сервер вернул ошибку (${resp.status})`);
  656.     const reader = resp.body.getReader();
  657.     const type = resp.headers.get("Content-Type");
  658.     const total = +resp.headers.get("Content-Length");
  659.     let loaded = 0;
  660.     const chunks = [];
  661.     const onprogress = (total && typeof(params.onprogress) === "function") ? params.onprogress : null;
  662.     while (true) {
  663.       const { done, value } = await reader.read();
  664.       if (done) break;
  665.       chunks.push(value);
  666.       loaded += value.length;
  667.       if (onprogress) onprogress(loaded, total);
  668.     }
  669.     switch (params.responseType) {
  670.       case "binary":
  671.         return new Blob(chunks, { type: type });
  672.       default:
  673.         {
  674.           let pos = 0;
  675.           const data = new Uint8Array(loaded);
  676.           for (let ch of chunks) {
  677.             data.set(ch, pos);
  678.             pos += ch.length;
  679.           }
  680.           return (new TextDecoder("utf-8")).decode(data);
  681.         }
  682.     }
  683.   }
  684.  
  685.   static abortAll() {
  686.     if (this._controller) {
  687.       this._controller.abort();
  688.       this._controller = null;
  689.     }
  690.   }
  691.  
  692.   static _getSignal() {
  693.     let controller = this._controller;
  694.     if (!controller) this._controller = controller = new AbortController();
  695.     return controller.signal;
  696.   }
  697. }
  698.  
  699. class FB2Utils {
  700.   static dateToAtom(date) {
  701.     const m = date.getMonth() + 1;
  702.     const d = date.getDate();
  703.     return "" + date.getFullYear() + '-' + (m < 10 ? "0" : "") + m + "-" + (d < 10 ? "0" : "") + d;
  704.   }
  705. }
  706.  
  707. class FB2UnknownNodeError extends Error {
  708.   constructor(message) {
  709.     super(message);
  710.     this.name = "UnknownNodeError";
  711.   }
  712. }
  713.