series.ts 3.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105
  1. 'use strict';
  2. import cheerio = require('cheerio');
  3. import episode from './episode';
  4. import fs = require('fs');
  5. import request = require('./request');
  6. import path = require('path');
  7. import url = require('url');
  8. import log = require('./log');
  9. var persistent = '.crpersistent';
  10. /**
  11. * Streams the series to disk.
  12. */
  13. export default function(config: IConfig, address: string, done: (err: Error) => void) {
  14. var persistentPath = path.join(config.output || process.cwd(), persistent);
  15. fs.readFile(persistentPath, 'utf8', (err, contents) => {
  16. var cache = config.cache ? {} : JSON.parse(contents || '{}');
  17. page(config, address, (err, page) => {
  18. if (err) return done(err);
  19. var i = 0;
  20. (function next() {
  21. if (i >= page.episodes.length) return done(null);
  22. download(cache, config, address, page.episodes[i], (err, ignored) => {
  23. if (err) return done(err);
  24. if ((ignored == false) || (ignored == undefined))
  25. {
  26. var newCache = JSON.stringify(cache, null, ' ');
  27. fs.writeFile(persistentPath, newCache, err => {
  28. if (err) return done(err);
  29. i += 1;
  30. next();
  31. });
  32. }
  33. else
  34. {
  35. i += 1;
  36. next();
  37. }
  38. });
  39. })();
  40. });
  41. });
  42. }
  43. /**
  44. * Downloads the episode.
  45. */
  46. function download(cache: {[address: string]: number},
  47. config: IConfig,
  48. baseAddress: string,
  49. item: ISeriesEpisode,
  50. done: (err: Error, ign: boolean) => void) {
  51. if (!filter(config, item)) return done(null, false);
  52. var address = url.resolve(baseAddress, item.address);
  53. if (cache[address]) return done(null, false);
  54. episode(config, address, (err, ignored) => {
  55. if (err) return done(err, false);
  56. cache[address] = Date.now();
  57. done(null, ignored);
  58. });
  59. }
  60. /**
  61. * Filters the item based on the configuration.
  62. */
  63. function filter(config: IConfig, item: ISeriesEpisode) {
  64. // Filter on chapter.
  65. var episodeFilter = config.episode;
  66. if (episodeFilter > 0 && parseInt(item.episode, 10) <= episodeFilter) return false;
  67. if (episodeFilter < 0 && parseInt(item.episode, 10) >= -episodeFilter) return false;
  68. // Filter on volume.
  69. var volumeFilter = config.volume;
  70. if (volumeFilter > 0 && item.volume <= volumeFilter) return false;
  71. if (volumeFilter < 0 && item.volume >= -volumeFilter) return false;
  72. return true;
  73. }
  74. /**
  75. * Requests the page and scrapes the episodes and series.
  76. */
  77. function page(config: IConfig, address: string, done: (err: Error, result?: ISeries) => void) {
  78. request.get(config, address, (err, result) => {
  79. if (err) return done(err);
  80. var $ = cheerio.load(result);
  81. var title = $('span[itemprop=name]').text();
  82. if (!title) return done(new Error('Invalid page.(' + address + ')'));
  83. log.info("Checking availability for " + title);
  84. var episodes: ISeriesEpisode[] = [];
  85. $('.episode').each((i, el) => {
  86. if ($(el).children('img[src*=coming_soon]').length) return;
  87. var volume = /([0-9]+)\s*$/.exec($(el).closest('ul').prev('a').text());
  88. var regexp = /Episode\s+((PV )?[S0-9][P0-9.]*[a-fA-F]?)\s*$/i;
  89. var episode = regexp.exec($(el).children('.series-title').text());
  90. var address = $(el).attr('href');
  91. if (!address || !episode) return;
  92. episodes.push({
  93. address: address,
  94. episode: episode[1],
  95. volume: volume ? parseInt(volume[0], 10) : 1
  96. });
  97. });
  98. done(null, {episodes: episodes.reverse(), series: title});
  99. });
  100. }