series.ts 5.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224
  1. 'use strict';
  2. import cheerio = require('cheerio');
  3. import episode from './episode';
  4. // import fs = require('fs');
  5. import fs = require('fs-extra');
  6. import my_request = require('./my_request');
  7. import path = require('path');
  8. import url = require('url');
  9. import log = require('./log');
  10. const persistent = '.crpersistent';
  11. /**
  12. * Check if a file exist..
  13. */
  14. function fileExist(path: string)
  15. {
  16. try
  17. {
  18. fs.statSync(path);
  19. return true;
  20. } catch (e)
  21. {
  22. return false;
  23. }
  24. }
  25. /**
  26. * Streams the series to disk.
  27. */
  28. export default function(config: IConfig, task: IConfigTask, done: (err: Error) => void)
  29. {
  30. const persistentPath = path.join(config.output || process.cwd(), persistent);
  31. /* Make a backup of the persistent file in case of */
  32. if (fileExist(persistentPath))
  33. {
  34. fs.copySync(persistentPath, persistentPath + '.backup');
  35. }
  36. fs.readFile(persistentPath, 'utf8', (err: Error, contents: string) =>
  37. {
  38. const cache = config.cache ? {} : JSON.parse(contents || '{}');
  39. pageScrape(config, task, (errP, page) =>
  40. {
  41. if (errP)
  42. {
  43. return done(errP);
  44. }
  45. let i = 0;
  46. (function next()
  47. {
  48. if (config.debug)
  49. {
  50. log.dumpToDebug('Episode ' + i, JSON.stringify(page.episodes[i]));
  51. }
  52. if (i >= page.episodes.length) return done(null);
  53. download(cache, config, task, page.episodes[i], (errD, ignored) =>
  54. {
  55. if (errD)
  56. {
  57. if (page.episodes[i].retry <= 0)
  58. {
  59. log.error(errD);
  60. log.error('Cannot fetch episode "s' + page.episodes[i].volume + 'e' + page.episodes[i].episode +
  61. '", please rerun later');
  62. /* Go to the next on the list */
  63. i += 1;
  64. }
  65. else
  66. {
  67. if ((config.verbose) || (config.debug))
  68. {
  69. if (config.debug)
  70. {
  71. log.dumpToDebug('series address', task.address);
  72. log.dumpToDebug('series error', errD.stack || errD);
  73. log.dumpToDebug('series data', JSON.stringify(page));
  74. }
  75. log.error(errD);
  76. }
  77. log.warn('Retrying to fetch episode "s' + page.episodes[i].volume + 'e' + page.episodes[i].episode +
  78. '" - Retry ' + page.episodes[i].retry + ' / ' + config.retry);
  79. page.episodes[i].retry -= 1;
  80. }
  81. next();
  82. }
  83. else
  84. {
  85. if ((ignored === false) || (ignored === undefined))
  86. {
  87. const newCache = JSON.stringify(cache, null, ' ');
  88. fs.writeFile(persistentPath, newCache, (errW: Error) =>
  89. {
  90. if (errW)
  91. {
  92. return done(errW);
  93. }
  94. i += 1;
  95. next();
  96. });
  97. }
  98. else
  99. {
  100. i += 1;
  101. next();
  102. }
  103. }
  104. });
  105. })();
  106. });
  107. });
  108. }
  109. /**
  110. * Downloads the episode.
  111. */
  112. function download(cache: {[address: string]: number}, config: IConfig,
  113. task: IConfigTask, item: ISeriesEpisode,
  114. done: (err: Error, ign: boolean) => void)
  115. {
  116. const episodeNumber = parseInt(item.episode, 10);
  117. if ( (episodeNumber < task.episode_min) ||
  118. (episodeNumber > task.episode_max) )
  119. {
  120. return done(null, false);
  121. }
  122. const address = url.resolve(task.address, item.address);
  123. if (cache[address])
  124. {
  125. return done(null, false);
  126. }
  127. episode(config, address, (err, ignored) =>
  128. {
  129. if (err)
  130. {
  131. return done(err, false);
  132. }
  133. cache[address] = Date.now();
  134. done(null, ignored);
  135. });
  136. }
  137. /**
  138. * Requests the page and scrapes the episodes and series.
  139. */
  140. function pageScrape(config: IConfig, task: IConfigTask, done: (err: Error, result?: ISeries) => void)
  141. {
  142. if (task.address[0] === '@')
  143. {
  144. log.info('Trying to fetch from ' + task.address.substr(1));
  145. const episodes: ISeriesEpisode[] = [];
  146. episodes.push({
  147. address: task.address.substr(1),
  148. episode: '',
  149. volume: 0,
  150. retry: config.retry,
  151. });
  152. done(null, {episodes: episodes.reverse(), series: ''});
  153. }
  154. else
  155. {
  156. let episodeCount = 0;
  157. my_request.get(config, task.address, (err, result) => {
  158. if (err) {
  159. return done(err);
  160. }
  161. const $ = cheerio.load(result);
  162. const title = $('span[itemprop=name]').text();
  163. if (config.debug)
  164. {
  165. log.dumpToDebug('serie page', $.html());
  166. }
  167. if (!title) {
  168. if (config.debug)
  169. {
  170. log.dumpToDebug('missing title', task.address);
  171. }
  172. return done(new Error('Invalid page.(' + task.address + ')'));
  173. }
  174. log.info('Checking availability for ' + title);
  175. const episodes: ISeriesEpisode[] = [];
  176. $('.episode').each((i, el) => {
  177. if ($(el).children('img[src*=coming_soon]').length) {
  178. return;
  179. }
  180. const volume = /([0-9]+)\s*$/.exec($(el).closest('ul').prev('a').text());
  181. const regexp = /Episode\s+((PV )?[S0-9][\-P0-9.]*[a-fA-F]?)\s*$/i;
  182. const episode = regexp.exec($(el).children('.series-title').text());
  183. const url = $(el).attr('href');
  184. if ((!url) || (!episode)) {
  185. return;
  186. }
  187. episodeCount += 1;
  188. episodes.push({
  189. address: url,
  190. episode: episode[1],
  191. volume: volume ? parseInt(volume[0], 10) : 1,
  192. retry: config.retry,
  193. });
  194. });
  195. if (episodeCount === 0)
  196. {
  197. log.warn('No episodes found for ' + title + '. Could it be a movie?');
  198. }
  199. done(null, {episodes: episodes.reverse(), series: title});
  200. });
  201. }
  202. }