BookParser.js 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382
  1. import EasySAXParser from './easysax';
  2. import {sleep} from '../../../share/utils';
  3. export default class BookParser {
  4. constructor() {
  5. // defaults
  6. this.p = 30;// px, отступ параграфа
  7. this.w = 300;// px, ширина страницы
  8. this.wordWrap = false;// перенос по слогам
  9. // заглушка
  10. this.measureText = (text, style) => {// eslint-disable-line no-unused-vars
  11. return text.length*10;
  12. };
  13. }
  14. async parse(data, callback) {
  15. if (!callback)
  16. callback = () => {};
  17. callback(0);
  18. this.data = data;
  19. if (data.indexOf('<FictionBook') < 0) {
  20. throw new Error('Неверный формат файла');
  21. }
  22. //defaults
  23. let fb2 = {
  24. firstName: '',
  25. middleName: '',
  26. lastName: '',
  27. bookTitle: '',
  28. };
  29. let path = '';
  30. let tag = '';
  31. let nextPerc = 0;
  32. let paraIndex = -1;
  33. let paraOffset = 0;
  34. let para = []; /*array of
  35. {
  36. index: Number,
  37. offset: Number, //сумма всех length до этого параграфа
  38. length: Number, //длина text без тегов
  39. text: String //текст параграфа (или title или epigraph и т.д) с вложенными тегами
  40. }
  41. */
  42. const newParagraph = (text, len) => {
  43. paraIndex++;
  44. let p = {
  45. index: paraIndex,
  46. offset: paraOffset,
  47. length: len,
  48. text: text
  49. };
  50. para[paraIndex] = p;
  51. paraOffset += p.length;
  52. };
  53. const growParagraph = (text, len) => {
  54. let p = para[paraIndex];
  55. if (p) {
  56. paraOffset -= p.length;
  57. if (p.text == ' ') {
  58. p.length = 0;
  59. p.text = '';
  60. }
  61. p.length += len;
  62. p.text += text;
  63. } else {
  64. p = {
  65. index: paraIndex,
  66. offset: paraOffset,
  67. length: len,
  68. text: text
  69. };
  70. }
  71. para[paraIndex] = p;
  72. paraOffset += p.length;
  73. };
  74. const parser = new EasySAXParser();
  75. parser.on('error', (msgError) => {// eslint-disable-line no-unused-vars
  76. });
  77. parser.on('startNode', (elemName, getAttr, isTagEnd, getStrNode) => {// eslint-disable-line no-unused-vars
  78. tag = elemName;
  79. path += '/' + elemName;
  80. if ((tag == 'p' || tag == 'empty-line') && path.indexOf('/FictionBook/body/section') == 0) {
  81. newParagraph(' ', 1);
  82. }
  83. });
  84. parser.on('endNode', (elemName, isTagStart, getStrNode) => {// eslint-disable-line no-unused-vars
  85. if (tag == elemName) {
  86. path = path.substr(0, path.length - tag.length - 1);
  87. let i = path.lastIndexOf('/');
  88. if (i >= 0) {
  89. tag = path.substr(i + 1);
  90. } else {
  91. tag = path;
  92. }
  93. }
  94. });
  95. parser.on('textNode', (text) => {
  96. if (text != ' ' && text.trim() == '')
  97. text = text.trim();
  98. if (text == '')
  99. return;
  100. switch (path) {
  101. case '/FictionBook/description/title-info/author/first-name':
  102. fb2.firstName = text;
  103. break;
  104. case '/FictionBook/description/title-info/author/middle-name':
  105. fb2.middleName = text;
  106. break;
  107. case '/FictionBook/description/title-info/author/last-name':
  108. fb2.lastName = text;
  109. break;
  110. case '/FictionBook/description/title-info/genre':
  111. fb2.genre = text;
  112. break;
  113. case '/FictionBook/description/title-info/date':
  114. fb2.date = text;
  115. break;
  116. case '/FictionBook/description/title-info/book-title':
  117. fb2.bookTitle = text;
  118. break;
  119. case '/FictionBook/description/title-info/id':
  120. fb2.id = text;
  121. break;
  122. }
  123. if (path.indexOf('/FictionBook/description/title-info/annotation') == 0) {
  124. if (!fb2.annotation)
  125. fb2.annotation = '';
  126. if (tag != 'annotation')
  127. fb2.annotation += `<${tag}>${text}</${tag}>`;
  128. else
  129. fb2.annotation += text;
  130. }
  131. if (path.indexOf('/FictionBook/body/title') == 0) {
  132. newParagraph(text, text.length);
  133. }
  134. if (path.indexOf('/FictionBook/body/section') == 0) {
  135. switch (tag) {
  136. case 'p':
  137. growParagraph(text, text.length);
  138. break;
  139. case 'section':
  140. case 'title':
  141. newParagraph(text, text.length);
  142. break;
  143. default:
  144. growParagraph(`<${tag}>${text}</${tag}>`, text.length);
  145. }
  146. }
  147. });
  148. parser.on('cdata', (data) => {// eslint-disable-line no-unused-vars
  149. });
  150. parser.on('comment', (text) => {// eslint-disable-line no-unused-vars
  151. });
  152. parser.on('progress', async(progress) => {
  153. if (progress > nextPerc) {
  154. await sleep(1);
  155. callback(progress);
  156. nextPerc += 10;
  157. }
  158. });
  159. await parser.parse(data);
  160. this.fb2 = fb2;
  161. this.para = para;
  162. callback(100);
  163. await sleep(10);
  164. return {fb2};
  165. }
  166. findParaIndex(bookPos) {
  167. let result = undefined;
  168. //дихотомия
  169. let first = 0;
  170. let last = this.para.length - 1;
  171. while (first < last) {
  172. let mid = first + Math.floor((last - first)/2);
  173. if (bookPos <= this.para[mid].offset + this.para[mid].length - 1)
  174. last = mid;
  175. else
  176. first = mid + 1;
  177. }
  178. if (last >= 0) {
  179. const ofs = this.para[last].offset;
  180. if (bookPos >= ofs && bookPos < ofs + this.para[last].length)
  181. result = last;
  182. }
  183. return result;
  184. }
  185. removeTags(s) {
  186. let result = '';
  187. const parser = new EasySAXParser();
  188. parser.on('textNode', (text) => {
  189. result += text;
  190. });
  191. parser.parse(`<p>${s}</p>`);
  192. return result;
  193. }
  194. parsePara(paraIndex) {
  195. const para = this.para[paraIndex];
  196. if (para.parsed &&
  197. para.parsed.w === this.w &&
  198. para.parsed.p === this.p &&
  199. para.parsed.wordWrap === this.wordWrap &&
  200. para.parsed.font === this.font
  201. )
  202. return para.parsed;
  203. const parsed = {
  204. w: this.w,
  205. p: this.p,
  206. wordWrap: this.wordWrap,
  207. font: this.font,
  208. };
  209. const lines = [];
  210. /* array of
  211. {
  212. begin: Number,
  213. end: Number,
  214. first: Boolean,
  215. last: Boolean,
  216. parts: array of {
  217. style: 'bold'|'italic',
  218. text: String,
  219. }
  220. }*/
  221. const text = this.removeTags(para.text);
  222. const words = text.split(' ');
  223. let line = {begin: para.offset, parts: []};
  224. let prevPart = '';
  225. let part = '';
  226. let prevW = 0;
  227. let j = 0;
  228. // тут начинается самый замес, перенос и стилизация
  229. for (let i = 0; i < words.length; i++) {
  230. const word = words[i];
  231. part += word;
  232. let w = this.measureText(part) + (j == 0 ? parsed.p : 0);
  233. if (w > parsed.w) {
  234. line.parts.push({style: '', text: prevPart});
  235. line.end = line.begin + prevPart.length;//нет -1 !!!
  236. line.width = prevW;
  237. line.first = (j == 0);
  238. line.last = false;
  239. lines.push(line);
  240. line = {begin: line.end + 1, parts: []};
  241. part = word;
  242. j++;
  243. }
  244. prevW = w;
  245. prevPart = part;
  246. part += ' ';
  247. }
  248. line.parts.push({style: '', text: prevPart});
  249. line.end = line.begin + prevPart.length - 1;
  250. line.width = prevW;
  251. line.first = (j == 0);
  252. line.last = true;
  253. lines.push(line);
  254. parsed.lines = lines;
  255. para.parsed = parsed;
  256. return parsed;
  257. }
  258. findLineIndex(bookPos, lines) {
  259. let result = undefined;
  260. //дихотомия
  261. let first = 0;
  262. let last = lines.length - 1;
  263. while (first < last) {
  264. let mid = first + Math.floor((last - first)/2);
  265. if (bookPos <= lines[mid].end)
  266. last = mid;
  267. else
  268. first = mid + 1;
  269. }
  270. if (last >= 0) {
  271. if (bookPos >= lines[last].begin && bookPos <= lines[last].end)
  272. result = last;
  273. }
  274. return result;
  275. }
  276. getLines(bookPos, n) {
  277. const result = [];
  278. let paraIndex = this.findParaIndex(bookPos);
  279. if (paraIndex === undefined)
  280. return result;
  281. if (n > 0) {
  282. let parsed = this.parsePara(paraIndex);
  283. let i = this.findLineIndex(bookPos, parsed.lines);
  284. if (i === undefined)
  285. return result;
  286. while (n > 0) {
  287. result.push(parsed.lines[i]);
  288. i++;
  289. if (i >= parsed.lines.length) {
  290. paraIndex++;
  291. if (paraIndex < this.para.length)
  292. parsed = this.parsePara(paraIndex);
  293. else
  294. return result;
  295. i = 0;
  296. }
  297. n--;
  298. }
  299. } else if (n < 0) {
  300. n = -n;
  301. let parsed = this.parsePara(paraIndex);
  302. let i = this.findLineIndex(bookPos, parsed.lines);
  303. if (i === undefined)
  304. return result;
  305. while (n > 0) {
  306. result.push(parsed.lines[i]);
  307. i--;
  308. if (i < 0) {
  309. paraIndex--;
  310. if (paraIndex >= 0)
  311. parsed = this.parsePara(paraIndex);
  312. else
  313. return result;
  314. i = parsed.lines.length - 1;
  315. }
  316. n--;
  317. }
  318. }
  319. return result;
  320. }
  321. }