Advertisement
stuppid_bot

Untitled

Mar 28th, 2017
213
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
Python 19.59 KB | None | 0 0
  1. import re
  2. import requests
  3. import sys
  4.  
  5.  
  6. ARCHIVE_URL = 'http://salery.livejournal.com/calendar'
  7.  
  8. ARCHIVE_RE = (
  9.     '<tr><td class="caption">Years</td></tr><tr>\n'
  10.     '<td colspan="2" bgcolor="#ffffff">([\s\S]+?)</td>'
  11. )
  12.  
  13. YEAR_RE = 'href="([^"]+)'
  14.  
  15. DATE_RE = '<div align="center"><a href="([^"]+)'
  16.  
  17. POST_RE = '<a href="([^#]+)#\w+" class="ljcut-link-expand" title="Read more..." >Read more...</a>'
  18.  
  19.  
  20. def getArchiveLinks():
  21.     content = requests.get(ARCHIVE_URL).text
  22.     match = re.search(ARCHIVE_RE, content)
  23.     links = re.findall(YEAR_RE, match.group(1))
  24.     return links
  25.  
  26.  
  27. def getArchiveLinksByYear(link):
  28.     content = requests.get(link).text
  29.     links = re.findall(DATE_RE, content)
  30.     return links
  31.  
  32.  
  33. def getLinksByDate(link):
  34.     content = requests.get(link).text
  35.     links = re.findall(POST_RE, content)
  36.     return links
  37.  
  38.  
  39. def main(args):
  40.     links = []
  41.     for x in getArchiveLinks():
  42.         print(x)
  43.         for x1 in getArchiveLinksByYear(x):
  44.             print(x1)
  45.             links += getLinksByDate(x1)
  46.             print(links[-1])
  47.     with open('links.txt', 'w') as f:
  48.         f.write('\n'.join(links))
  49.     return 0
  50.  
  51.  
  52. if __name__ == '__main__':
  53.     sys.exit(main(sys.argv))
  54. """
  55.  
  56. http://salery.livejournal.com/339.html
  57. http://salery.livejournal.com/767.html
  58. http://salery.livejournal.com/947.html
  59. http://salery.livejournal.com/1097.html
  60. http://salery.livejournal.com/1729.html
  61. http://salery.livejournal.com/1868.html
  62. http://salery.livejournal.com/2303.html
  63. http://salery.livejournal.com/2474.html
  64. http://salery.livejournal.com/2668.html
  65. http://salery.livejournal.com/2977.html
  66. http://salery.livejournal.com/3274.html
  67. http://salery.livejournal.com/3338.html
  68. http://salery.livejournal.com/3788.html
  69. http://salery.livejournal.com/3889.html
  70. http://salery.livejournal.com/4259.html
  71. http://salery.livejournal.com/4427.html
  72. http://salery.livejournal.com/4807.html
  73. http://salery.livejournal.com/5109.html
  74. http://salery.livejournal.com/5364.html
  75. http://salery.livejournal.com/5535.html
  76. http://salery.livejournal.com/5803.html
  77. http://salery.livejournal.com/6323.html
  78. http://salery.livejournal.com/6640.html
  79. http://salery.livejournal.com/6669.html
  80. http://salery.livejournal.com/6976.html
  81. http://salery.livejournal.com/7468.html
  82. http://salery.livejournal.com/7839.html
  83. http://salery.livejournal.com/8190.html
  84. http://salery.livejournal.com/8409.html
  85. http://salery.livejournal.com/8636.html
  86. http://salery.livejournal.com/8875.html
  87. http://salery.livejournal.com/9147.html
  88. http://salery.livejournal.com/9346.html
  89. http://salery.livejournal.com/9631.html
  90. http://salery.livejournal.com/10691.html
  91. http://salery.livejournal.com/10773.html
  92. http://salery.livejournal.com/11173.html
  93. http://salery.livejournal.com/11480.html
  94. http://salery.livejournal.com/11633.html
  95. http://salery.livejournal.com/11810.html
  96. http://salery.livejournal.com/12188.html
  97. http://salery.livejournal.com/12511.html
  98. http://salery.livejournal.com/12722.html
  99. http://salery.livejournal.com/12990.html
  100. http://salery.livejournal.com/13059.html
  101. http://salery.livejournal.com/13685.html
  102. http://salery.livejournal.com/13888.html
  103. http://salery.livejournal.com/14309.html
  104. http://salery.livejournal.com/14574.html
  105. http://salery.livejournal.com/14746.html
  106. http://salery.livejournal.com/14922.html
  107. http://salery.livejournal.com/15281.html
  108. http://salery.livejournal.com/16156.html
  109. http://salery.livejournal.com/16456.html
  110. http://salery.livejournal.com/16697.html
  111. http://salery.livejournal.com/17111.html
  112. http://salery.livejournal.com/17343.html
  113. http://salery.livejournal.com/17513.html
  114. http://salery.livejournal.com/17684.html
  115. http://salery.livejournal.com/18140.html
  116. http://salery.livejournal.com/18383.html
  117. http://salery.livejournal.com/18586.html
  118. http://salery.livejournal.com/19135.html
  119. http://salery.livejournal.com/19211.html
  120. http://salery.livejournal.com/19590.html
  121. http://salery.livejournal.com/19781.html
  122. http://salery.livejournal.com/20098.html
  123. http://salery.livejournal.com/20291.html
  124. http://salery.livejournal.com/20573.html
  125. http://salery.livejournal.com/20865.html
  126. http://salery.livejournal.com/21189.html
  127. http://salery.livejournal.com/21429.html
  128. http://salery.livejournal.com/22006.html
  129. http://salery.livejournal.com/22156.html
  130. http://salery.livejournal.com/22518.html
  131. http://salery.livejournal.com/22766.html
  132. http://salery.livejournal.com/23949.html
  133. http://salery.livejournal.com/24287.html
  134. http://salery.livejournal.com/24534.html
  135. http://salery.livejournal.com/24766.html
  136. http://salery.livejournal.com/24922.html
  137. http://salery.livejournal.com/25241.html
  138. http://salery.livejournal.com/25541.html
  139. http://salery.livejournal.com/25767.html
  140. http://salery.livejournal.com/26012.html
  141. http://salery.livejournal.com/26593.html
  142. http://salery.livejournal.com/26638.html
  143. http://salery.livejournal.com/27058.html
  144. http://salery.livejournal.com/27201.html
  145. http://salery.livejournal.com/27615.html
  146. http://salery.livejournal.com/28316.html
  147. http://salery.livejournal.com/28747.html
  148. http://salery.livejournal.com/29089.html
  149. http://salery.livejournal.com/29412.html
  150. http://salery.livejournal.com/29608.html
  151. http://salery.livejournal.com/29951.html
  152. http://salery.livejournal.com/30650.html
  153. http://salery.livejournal.com/30938.html
  154. http://salery.livejournal.com/31151.html
  155. http://salery.livejournal.com/31623.html
  156. http://salery.livejournal.com/31898.html
  157. http://salery.livejournal.com/32146.html
  158. http://salery.livejournal.com/32452.html
  159. http://salery.livejournal.com/32918.html
  160. http://salery.livejournal.com/33113.html
  161. http://salery.livejournal.com/33295.html
  162. http://salery.livejournal.com/33542.html
  163. http://salery.livejournal.com/34420.html
  164. http://salery.livejournal.com/34564.html
  165. http://salery.livejournal.com/35237.html
  166. http://salery.livejournal.com/35394.html
  167. http://salery.livejournal.com/35789.html
  168. http://salery.livejournal.com/36002.html
  169. http://salery.livejournal.com/36223.html
  170. http://salery.livejournal.com/36494.html
  171. http://salery.livejournal.com/36654.html
  172. http://salery.livejournal.com/36957.html
  173. http://salery.livejournal.com/37228.html
  174. http://salery.livejournal.com/37556.html
  175. http://salery.livejournal.com/37822.html
  176. http://salery.livejournal.com/38227.html
  177. http://salery.livejournal.com/38792.html
  178. http://salery.livejournal.com/39160.html
  179. http://salery.livejournal.com/40132.html
  180. http://salery.livejournal.com/40637.html
  181. http://salery.livejournal.com/40855.html
  182. http://salery.livejournal.com/41039.html
  183. http://salery.livejournal.com/41230.html
  184. http://salery.livejournal.com/41894.html
  185. http://salery.livejournal.com/42122.html
  186. http://salery.livejournal.com/42694.html
  187. http://salery.livejournal.com/42785.html
  188. http://salery.livejournal.com/43134.html
  189. http://salery.livejournal.com/44022.html
  190. http://salery.livejournal.com/44112.html
  191. http://salery.livejournal.com/44297.html
  192. http://salery.livejournal.com/44580.html
  193. http://salery.livejournal.com/44819.html
  194. http://salery.livejournal.com/45188.html
  195. http://salery.livejournal.com/45523.html
  196. http://salery.livejournal.com/45675.html
  197. http://salery.livejournal.com/45835.html
  198. http://salery.livejournal.com/46310.html
  199. http://salery.livejournal.com/46614.html
  200. http://salery.livejournal.com/47417.html
  201. http://salery.livejournal.com/47752.html
  202. http://salery.livejournal.com/47937.html
  203. http://salery.livejournal.com/48220.html
  204. http://salery.livejournal.com/48443.html
  205. http://salery.livejournal.com/49350.html
  206. http://salery.livejournal.com/49641.html
  207. http://salery.livejournal.com/49727.html
  208. http://salery.livejournal.com/50067.html
  209. http://salery.livejournal.com/51013.html
  210. http://salery.livejournal.com/53135.html
  211. http://salery.livejournal.com/53331.html
  212. http://salery.livejournal.com/54418.html
  213. http://salery.livejournal.com/54553.html
  214. http://salery.livejournal.com/55079.html
  215. http://salery.livejournal.com/55361.html
  216. http://salery.livejournal.com/56657.html
  217. http://salery.livejournal.com/57068.html
  218. http://salery.livejournal.com/58992.html
  219. http://salery.livejournal.com/60081.html
  220. http://salery.livejournal.com/60824.html
  221. http://salery.livejournal.com/61029.html
  222. http://salery.livejournal.com/61551.html
  223. http://salery.livejournal.com/61937.html
  224. http://salery.livejournal.com/62387.html
  225. http://salery.livejournal.com/62486.html
  226. http://salery.livejournal.com/62954.html
  227. http://salery.livejournal.com/63203.html
  228. http://salery.livejournal.com/63523.html
  229. http://salery.livejournal.com/64415.html
  230. http://salery.livejournal.com/64761.html
  231. http://genrogge.ru/hwj/" target="_blank" title="Полное издание, №№ 1–84. Свыше 1 500 фотографий и иллюстраций. 1699 страниц, включая оглавления и отдельные наградные списки в распознанном тексте" rel="nofollow">&laquo;Летопись войны с&nbsp;Японией&raquo;</a> и <a href="http://genrogge.ru/wwi-1914-1917/" target="_blank" title="Полное издание, №№ 1–132. Свыше 2 000 фотографий и иллюстраций. 2300 страниц, включая сообщения Официального отдела." rel="nofollow">&laquo;Летопись войны 1914&ndash;15&ndash;16&ndash;17&nbsp;гг.&raquo;</a>&nbsp;&mdash; с&nbsp;тысячами фотографий).<br /></p><p>Собственно, чтобы представить страну, достаточно посмотреть на лица. Помню, в 80-х на Петровских Линиях кто-то устроил фотовыставку (приуроченную к Первой мировой). Всего-то две комнатки, но она пользовалась огромной популярностью. Это был культурный шок: сразу несколько сот &laquo;старых&raquo; лиц под звуки маршей и вальсов того времени. Как жаль, что никто до сих пор не додумался переиздать (хотя бы в более скромном виде) что-нибудь из альбомов, издававшихся в России в начале ХХ&nbsp;в. Я вынужден был ограничиться изучаемым контингентом, но вообще-то там много всего - от польских помещиков до аксакалов среднеазиатского местного управления (очень интересны, кстати, многочисленные портреты предпринимателей средней руки). <br /></p><p>Довольно любопытны фото одних и тех же лиц в разном возрасте. Кто-то из писателей (кажется, Акунин) даже призвал на сайте присылать ему такие на конкурс. Делать из этого какой-то &laquo;конкурс&raquo;, конечно, пошлость, но сама идея очень неплоха. Старых (дорев.) фото такого рода, правда, мало, но встречаются. Мне тоже попадались. Вот, например (тут даже люди специально сфотографировались в тех же самых позах), между фото&nbsp;&mdash; 12 лет и примерно столько же тыс.&nbsp;км.</p><b data-widget='ljcut' data-widget-options='{"sticky":false,"cutid":1,"journalid":"11178575","placeholders":0,"ditemid":64905}' class="ljcut-link lj-widget"><span class='ljcut-brace'>(&nbsp;</span><span class="ljcut-decor"><a href="http://salery.livejournal.com/64905.html
  232. http://salery.livejournal.com/65421.html
  233. http://salery.livejournal.com/65584.html
  234. http://salery.livejournal.com/65970.html
  235. http://salery.livejournal.com/66238.html
  236. http://salery.livejournal.com/66585.html
  237. http://salery.livejournal.com/67227.html
  238. http://salery.livejournal.com/67498.html
  239. http://salery.livejournal.com/67777.html
  240. http://salery.livejournal.com/68479.html
  241. http://salery.livejournal.com/68782.html
  242. http://salery.livejournal.com/69668.html
  243. http://salery.livejournal.com/69963.html
  244. http://salery.livejournal.com/70719.html
  245. http://salery.livejournal.com/71271.html
  246. http://salery.livejournal.com/71556.html
  247. http://salery.livejournal.com/71746.html
  248. http://salery.livejournal.com/72482.html
  249. http://salery.livejournal.com/72839.html
  250. http://salery.livejournal.com/73036.html
  251. http://salery.livejournal.com/73460.html
  252. http://salery.livejournal.com/73627.html
  253. http://salery.livejournal.com/74487.html
  254. http://salery.livejournal.com/74580.html
  255. http://salery.livejournal.com/74922.html
  256. http://salery.livejournal.com/75231.html
  257. http://salery.livejournal.com/75771.html
  258. http://salery.livejournal.com/75929.html
  259. http://salery.livejournal.com/76440.html
  260. http://salery.livejournal.com/76662.html
  261. http://salery.livejournal.com/77235.html
  262. http://salery.livejournal.com/77500.html
  263. http://salery.livejournal.com/77600.html
  264. http://salery.livejournal.com/78299.html
  265. http://salery.livejournal.com/78586.html
  266. http://salery.livejournal.com/78760.html
  267. http://salery.livejournal.com/79051.html
  268. http://salery.livejournal.com/79866.html
  269. http://salery.livejournal.com/80702.html
  270. http://salery.livejournal.com/80946.html
  271. http://salery.livejournal.com/81205.html
  272. http://salery.livejournal.com/81456.html
  273. http://salery.livejournal.com/81892.html
  274. http://salery.livejournal.com/81986.html
  275. http://salery.livejournal.com/82176.html
  276. http://salery.livejournal.com/82570.html
  277. http://salery.livejournal.com/83748.html
  278. http://salery.livejournal.com/84191.html
  279. http://salery.livejournal.com/84252.html
  280. http://salery.livejournal.com/85175.html
  281. http://salery.livejournal.com/85391.html
  282. http://salery.livejournal.com/86335.html
  283. http://salery.livejournal.com/86840.html
  284. http://salery.livejournal.com/87078.html
  285. http://salery.livejournal.com/87375.html
  286. http://salery.livejournal.com/87663.html
  287. http://salery.livejournal.com/88140.html
  288. http://salery.livejournal.com/88334.html
  289. http://salery.livejournal.com/88788.html
  290. http://salery.livejournal.com/89048.html
  291. http://salery.livejournal.com/89497.html
  292. http://salery.livejournal.com/89707.html
  293. http://salery.livejournal.com/90074.html
  294. http://salery.livejournal.com/90243.html
  295. http://salery.livejournal.com/90532.html
  296. http://salery.livejournal.com/90810.html
  297. http://salery.livejournal.com/90921.html
  298. http://salery.livejournal.com/91218.html
  299. http://salery.livejournal.com/91641.html
  300. http://salery.livejournal.com/91678.html
  301. http://salery.livejournal.com/91927.html
  302. http://salery.livejournal.com/92274.html
  303. http://salery.livejournal.com/92433.html
  304. http://salery.livejournal.com/92810.html
  305. http://salery.livejournal.com/93105.html
  306. http://salery.livejournal.com/93223.html
  307. http://salery.livejournal.com/93444.html
  308. http://salery.livejournal.com/93865.html
  309. http://salery.livejournal.com/94036.html
  310. http://salery.livejournal.com/94222.html
  311. http://salery.livejournal.com/94557.html
  312. http://salery.livejournal.com/94778.html
  313. http://salery.livejournal.com/95105.html
  314. http://salery.livejournal.com/95351.html
  315. http://salery.livejournal.com/95567.html
  316. http://salery.livejournal.com/95797.html
  317. http://salery.livejournal.com/96009.html
  318. http://salery.livejournal.com/96424.html
  319. http://salery.livejournal.com/96727.html
  320. http://salery.livejournal.com/96802.html
  321. http://salery.livejournal.com/97198.html
  322. http://salery.livejournal.com/97711.html
  323. http://salery.livejournal.com/98284.html
  324. http://salery.livejournal.com/98381.html
  325. http://salery.livejournal.com/98751.html
  326. http://salery.livejournal.com/99020.html
  327. http://salery.livejournal.com/99179.html
  328. http://salery.livejournal.com/99480.html
  329. http://salery.livejournal.com/99663.html
  330. http://salery.livejournal.com/99852.html
  331. http://salery.livejournal.com/100515.html
  332. http://salery.livejournal.com/100709.html
  333. http://salery.livejournal.com/101451.html
  334. http://salery.livejournal.com/101662.html
  335. http://salery.livejournal.com/101938.html
  336. http://salery.livejournal.com/102303.html
  337. http://salery.livejournal.com/102512.html
  338. http://salery.livejournal.com/102682.html
  339. http://salery.livejournal.com/102996.html
  340. http://salery.livejournal.com/103418.html
  341. http://salery.livejournal.com/103530.html
  342. http://salery.livejournal.com/103761.html
  343. http://salery.livejournal.com/103967.html
  344. http://salery.livejournal.com/104255.html
  345. http://salery.livejournal.com/104465.html
  346. http://salery.livejournal.com/104750.html
  347. http://salery.livejournal.com/105086.html
  348. http://salery.livejournal.com/105718.html
  349. http://salery.livejournal.com/105855.html
  350. http://salery.livejournal.com/106206.html
  351. http://salery.livejournal.com/106460.html
  352. http://salery.livejournal.com/106635.html
  353. http://salery.livejournal.com/106753.html
  354. http://salery.livejournal.com/107114.html
  355. http://salery.livejournal.com/107401.html
  356. http://salery.livejournal.com/107655.html
  357. http://salery.livejournal.com/108371.html
  358. http://salery.livejournal.com/108754.html
  359. http://salery.livejournal.com/109000.html
  360. http://salery.livejournal.com/109179.html
  361. http://salery.livejournal.com/109414.html
  362. http://salery.livejournal.com/109658.html
  363. http://salery.livejournal.com/109970.html
  364. http://salery.livejournal.com/110321.html
  365. http://salery.livejournal.com/110817.html
  366. http://salery.livejournal.com/110980.html
  367. http://salery.livejournal.com/111449.html
  368. http://salery.livejournal.com/112181.html
  369. http://salery.livejournal.com/112453.html
  370. http://salery.livejournal.com/112706.html
  371. http://salery.livejournal.com/113100.html
  372. http://salery.livejournal.com/113265.html
  373. http://salery.livejournal.com/113509.html
  374. http://salery.livejournal.com/113714.html
  375. http://salery.livejournal.com/113952.html
  376. http://salery.livejournal.com/114295.html
  377. http://salery.livejournal.com/114484.html
  378. http://salery.livejournal.com/114702.html
  379. http://salery.livejournal.com/115102.html
  380. http://salery.livejournal.com/115445.html
  381. http://salery.livejournal.com/115479.html
  382. http://salery.livejournal.com/116186.html
  383. http://salery.livejournal.com/116260.html
  384. http://salery.livejournal.com/116483.html
  385. http://salery.livejournal.com/116852.html
  386. http://salery.livejournal.com/117190.html
  387. http://salery.livejournal.com/117370.html
  388. http://salery.livejournal.com/117519.html
  389. http://salery.livejournal.com/117984.html
  390. http://salery.livejournal.com/118149.html
  391. http://salery.livejournal.com/118757.html
  392. http://salery.livejournal.com/118981.html
  393. http://salery.livejournal.com/119040.html
  394. http://salery.livejournal.com/119320.html
  395. http://salery.livejournal.com/119573.html
  396. http://salery.livejournal.com/119923.html
  397. http://salery.livejournal.com/120242.html
  398. http://salery.livejournal.com/120480.html
  399. http://salery.livejournal.com/120622.html
  400. http://salery.livejournal.com/121006.html
  401. http://salery.livejournal.com/121709.html
  402. http://salery.livejournal.com/122345.html
  403. http://salery.livejournal.com/122434.html
  404. http://salery.livejournal.com/123062.html
  405. http://salery.livejournal.com/123443.html
  406. http://salery.livejournal.com/123826.html
  407. http://salery.livejournal.com/124098.html
  408. http://salery.livejournal.com/124427.html
  409. http://salery.livejournal.com/124738.html
  410. http://salery.livejournal.com/125148.html
  411. http://salery.livejournal.com/125335.html
  412. http://salery.livejournal.com/125653.html
  413. http://salery.livejournal.com/125784.html
  414. http://salery.livejournal.com/126114.html
  415. http://salery.livejournal.com/126808.html
  416. http://salery.livejournal.com/127106.html
  417. http://salery.livejournal.com/127570.html
  418. http://salery.livejournal.com/127825.html
  419. http://salery.livejournal.com/128175.html
  420. http://salery.livejournal.com/128642.html
  421. http://salery.livejournal.com/128917.html
  422. http://salery.livejournal.com/129223.html
  423. http://salery.livejournal.com/129372.html
  424. http://salery.livejournal.com/129539.html
  425.  
  426. """
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement