index.js 6.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213
  1. const fs = require('fs-extra');
  2. const URL = require('url').URL;
  3. const iconv = require('iconv-lite');
  4. const chardet = require('chardet');
  5. const _ = require('lodash');
  6. const FileDetector = require('../FileDetector');
  7. const EasySAXParser = require('./easysaxmod');
  8. class BookConverter {
  9. constructor() {
  10. this.detector = new FileDetector();
  11. }
  12. async convertToFb2(inputFile, outputFile, url, callback) {
  13. const fileType = await this.detector.detectFile(inputFile);
  14. if (fileType && (fileType.ext == 'html' || fileType.ext == 'xml')) {
  15. const data = await fs.readFile(inputFile);
  16. if (data.toString().indexOf('<FictionBook') >= 0) {
  17. await fs.writeFile(outputFile, data);
  18. return;
  19. }
  20. const parsedUrl = new URL(url);
  21. if (parsedUrl.hostname == 'samlib.ru' ||
  22. parsedUrl.hostname == 'budclub.ru') {
  23. await fs.writeFile(outputFile, await this.convertSamlib(data));
  24. return;
  25. }
  26. //Заглушка
  27. await fs.writeFile(outputFile, data);
  28. callback(100);
  29. } else {
  30. if (fileType)
  31. throw new Error(`unknown file format: ${fileType.mime}`);
  32. else
  33. throw new Error(`unsupported file format: ${url}`);
  34. }
  35. }
  36. async convertSamlib(data) {
  37. let titleInfo = {};
  38. let desc = {_n: 'description', 'title-info': titleInfo};
  39. let pars = [];
  40. let body = {_n: 'body', section: {_a: [pars]}};
  41. let fb2 = [desc, body];
  42. let path = '';
  43. let tag = '';// eslint-disable-line no-unused-vars
  44. let inText = false;
  45. const newParagraph = () => {
  46. pars.push({_n: 'p', _t: ''});
  47. };
  48. const growParagraph = (text) => {
  49. const l = pars.length;
  50. if (l) {
  51. if (pars[l - 1]._t == '')
  52. text = text.trimLeft();
  53. pars[l - 1]._t += text;
  54. }
  55. };
  56. const parser = new EasySAXParser();
  57. parser.on('error', (msgError) => {// eslint-disable-line no-unused-vars
  58. });
  59. parser.on('startNode', (elemName, getAttr, isTagEnd, getStrNode) => {// eslint-disable-line no-unused-vars
  60. if (!inText) {
  61. path += '/' + elemName;
  62. tag = elemName;
  63. } else {
  64. if (elemName == 'p' || elemName == 'dd') {
  65. newParagraph();
  66. }
  67. }
  68. });
  69. parser.on('endNode', (elemName, isTagStart, getStrNode) => {// eslint-disable-line no-unused-vars
  70. if (!inText) {
  71. const oldPath = path;
  72. let t = '';
  73. do {
  74. let i = path.lastIndexOf('/');
  75. t = path.substr(i + 1);
  76. path = path.substr(0, i);
  77. } while (t != elemName && path);
  78. if (t != elemName) {
  79. path = oldPath;
  80. }
  81. let i = path.lastIndexOf('/');
  82. tag = path.substr(i + 1);
  83. }
  84. });
  85. parser.on('textNode', (text) => {// eslint-disable-line no-unused-vars
  86. switch (path) {
  87. case '/html/body/center/h2':
  88. titleInfo['book-title'] = text;
  89. return;
  90. case '/html/body/div/h3':
  91. if (!titleInfo.author)
  92. titleInfo.author = {};
  93. text = text.replace(':', '').trim().split(' ');
  94. if (text[0])
  95. titleInfo.author['last-name'] = text[0];
  96. if (text[1])
  97. titleInfo.author['first-name'] = text[1];
  98. if (text[2])
  99. titleInfo.author['middle-name'] = text[2];
  100. return;
  101. }
  102. if (inText)
  103. growParagraph(text);
  104. });
  105. parser.on('cdata', (data) => {// eslint-disable-line no-unused-vars
  106. });
  107. parser.on('comment', (text) => {// eslint-disable-line no-unused-vars
  108. if (text == '--------- Собственно произведение -------------')
  109. inText = true;
  110. if (text == '-----------------------------------------------')
  111. inText = false;
  112. });
  113. /*
  114. parser.on('progress', async(progress) => {
  115. callback(...........);
  116. });
  117. */
  118. const charsetAll = chardet.detectAll(data);
  119. let selected = 'ISO-8859-1';
  120. for (const charset of charsetAll) {
  121. if (charset.name.indexOf('ISO-8859') < 0) {
  122. selected = charset.name;
  123. break;
  124. }
  125. }
  126. await parser.parse(iconv.decode(data, selected));
  127. const title = (titleInfo['book-title'] ? titleInfo['book-title'] : '');
  128. let author = '';
  129. if (titleInfo.author) {
  130. author = _.compact([
  131. (titleInfo.author['last-name'] ? titleInfo.author['last-name'] : ''),
  132. (titleInfo.author['first-name'] ? titleInfo.author['first-name'] : ''),
  133. (titleInfo.author['middle-name'] ? titleInfo.author['middle-name'] : ''),
  134. ]).join(' ');
  135. }
  136. pars.unshift({_n: 'title', _a: [
  137. {_n: 'p', _t: author}, {_n: 'p', _t: ''},
  138. {_n: 'p', _t: title}, {_n: 'p', _t: ''},
  139. ]})
  140. return this.formatFb2(fb2);
  141. }
  142. formatFb2(fb2) {
  143. let out = '<?xml version="1.0" encoding="utf-8"?>';
  144. out += '<FictionBook xmlns="http://www.gribuser.ru/xml/fictionbook/2.0" xmlns:l="http://www.w3.org/1999/xlink">';
  145. out += this.formatFb2Node(fb2);
  146. out += '</FictionBook>';
  147. return out;
  148. }
  149. formatFb2Node(node, name) {
  150. let out = '';
  151. if (Array.isArray(node)) {
  152. for (const n of node) {
  153. out += this.formatFb2Node(n);
  154. }
  155. } else if (typeof node == 'string') {
  156. out += `<${name}>${node}</${name}>`;
  157. } else {
  158. if (node._n)
  159. name = node._n;
  160. if (!name)
  161. throw new Error(`malformed fb2 object`);
  162. out += `<${name}>`;
  163. if (node.hasOwnProperty('_t')) {
  164. out += node._t;
  165. } else {
  166. for (let nodeName in node) {
  167. if (nodeName == '_n')
  168. continue;
  169. const n = node[nodeName];
  170. out += this.formatFb2Node(n, nodeName);
  171. }
  172. }
  173. out += `</${name}>`;
  174. }
  175. return out;
  176. }
  177. }
  178. module.exports = BookConverter;