webshot.ts 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335
  1. import { writeFileSync } from 'fs';
  2. import { Readable } from 'stream';
  3. import { promisify } from 'util';
  4. import axios from 'axios';
  5. import * as CallableInstance from 'callable-instance';
  6. import { XmlEntities } from 'html-entities';
  7. import { PNG } from 'pngjs';
  8. import * as puppeteer from 'playwright';
  9. import * as sharp from 'sharp';
  10. import * as temp from 'temp';
  11. import { getLogger } from './loggers';
  12. import { Message } from './koishi';
  13. import { chainPromises } from './utils';
  14. import { linkBuilder, MediaItem, LazyMediaItem, WebshotHelpers } from './twitter';
  15. const xmlEntities = new XmlEntities();
  16. const ZHType = (type: string) => new class extends String {
  17. public type = super.toString();
  18. public toString = () => `[${super.toString()}]`;
  19. }(type);
  20. const typeInZH = {
  21. photo: ZHType('图片'),
  22. video: ZHType('视频'),
  23. };
  24. const logger = getLogger('webshot');
  25. export type Page = puppeteer.Page;
  26. export type Cookies = puppeteer.Cookie[];
  27. class Webshot extends CallableInstance<[LazyMediaItem[], (...args) => void, number], Promise<void>> {
  28. private browser: puppeteer.Browser;
  29. private mode: number;
  30. private wsUrl: string;
  31. private getCookies: () => Cookies;
  32. constructor(
  33. wsUrl: string, mode: number,
  34. getCookies: () => Cookies,
  35. onready: (doOnNewPage?: typeof Webshot.prototype.performOnNewPage) => void
  36. ) {
  37. super('webshot');
  38. // tslint:disable-next-line: no-conditional-assignment
  39. // eslint-disable-next-line no-cond-assign
  40. if (this.mode = mode) {
  41. if (onready) onready();
  42. } else {
  43. this.getCookies = getCookies;
  44. this.wsUrl = wsUrl;
  45. this.connect(() => onready && onready(this.performOnNewPage));
  46. }
  47. }
  48. private connect = (onready?: (...args) => void): Promise<void> =>
  49. axios.get<{[key in 'chromium' | 'firefox' | 'webkit']?: string}>(this.wsUrl)
  50. .then(res => {
  51. logger.info(`received websocket endpoint: ${JSON.stringify(res.data)}`);
  52. const browserType = Object.keys(res.data)[0] as keyof typeof res.data;
  53. return (puppeteer[browserType] as puppeteer.BrowserType<puppeteer.Browser>)
  54. .connect({wsEndpoint: res.data[browserType]});
  55. })
  56. .then(browser => this.browser = browser)
  57. .then(() => {
  58. logger.info('launched puppeteer browser');
  59. if (onready) return onready();
  60. })
  61. .catch(error => this.reconnect(error, onready));
  62. private reconnect = (error, onready?: (...args) => void) => {
  63. logger.error(`connection error, reason: ${error}`);
  64. logger.warn('trying to reconnect in 2.5s...');
  65. return promisify(setTimeout)(2500)
  66. .then(() => this.connect(onready));
  67. };
  68. private performOnNewPage = <T>(action: (page: Page) => T | PromiseLike<T>, zoomFactor = 2, reconnectOnError = true) =>
  69. this.browser.newPage({
  70. bypassCSP: true,
  71. deviceScaleFactor: zoomFactor,
  72. locale: 'ja-JP',
  73. timezoneId: 'Asia/Tokyo',
  74. userAgent: 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/67.0.3396.99 Safari/537.36',
  75. }).then(action)
  76. .catch(error => {
  77. if (reconnectOnError) {
  78. return this.reconnect(error)
  79. .then((): Promise<T> => this.performOnNewPage(action, zoomFactor, reconnectOnError));
  80. }
  81. throw error;
  82. });
  83. private renderWebshot = (
  84. url: string, height: number, webshotDelay: number,
  85. ...morePostProcessings: ((page: Page) => Promise<any>)[]
  86. ): Promise<string> => {
  87. temp.track();
  88. const jpeg = (data: Readable) => data.pipe(sharp()).jpeg({quality: 90, trellisQuantisation: true});
  89. const sharpToFile = (pic: sharp.Sharp) => new Promise<string>(resolve => {
  90. const webshotTempFilePath = temp.path({suffix: '.jpg'});
  91. pic.toFile(webshotTempFilePath).then(() => resolve(`file://${webshotTempFilePath}`));
  92. });
  93. const promise = new Promise<{ path: string, boundary: null | number }>((resolve, reject) => {
  94. const width = 720;
  95. const zoomFactor = 2;
  96. logger.info(`shooting ${width}*${height} webshot for ${url}`);
  97. this.performOnNewPage(
  98. page => {
  99. const startTime = new Date().getTime();
  100. const getTimerTime = () => new Date().getTime() - startTime;
  101. const getTimeout = () => WebshotHelpers.isWaitingForLogin ? 0 : Math.max(500, webshotDelay - getTimerTime());
  102. page.setViewportSize({
  103. width: width / zoomFactor,
  104. height: height / zoomFactor,
  105. }).then(() => page.context().addCookies(this.getCookies()))
  106. .then(() => page.goto(url, {waitUntil: 'load', timeout: getTimeout()}))
  107. .then(() => WebshotHelpers.handleCookieConsent(page))
  108. .then(() =>
  109. (<T>(next: () => Promise<T>) => Promise.race([
  110. WebshotHelpers.handleLogin(page)
  111. .then(() => page.goto(url, {waitUntil: 'load', timeout: getTimeout()}))
  112. .then(next),
  113. next(),
  114. ]))(() => promisify(setTimeout)(2000).then(() => page.waitForSelector('article', {timeout: getTimeout()})))
  115. )
  116. .catch((err: Error): Promise<puppeteer.ElementHandle<Element> | null> => {
  117. if (err.name !== 'TimeoutError') throw err;
  118. logger.warn(`navigation timed out at ${getTimerTime()} ms`);
  119. return null;
  120. })
  121. // hide header, footer, "more options" button, like and share count, avatar stories/live indicator
  122. // person tags, carousel navigator, 2nd avatar, and comments; adjust layout for center alignment
  123. .then(() => page.addStyleTag({content:
  124. 'nav,footer,header+div,header~div>div>div+div,article button,canvas,article section,article section+div>ul>:not(div),' +
  125. 'article section+div>ul>div [role="button"],header~div [tabindex="0"]>*>[tabindex="-1"]~div{display:none!important} ' +
  126. 'section+div{overflow:hidden} section+*>*{position:relative!important} article{border-bottom:1px solid!important} ' +
  127. 'article section+div>ul>div>li{padding:6px 2px 12px!important}',
  128. }))
  129. .then(() => page.addStyleTag({
  130. content: '*{font-family:-apple-system,".Helvetica Neue DeskInterface",Hiragino Sans,Hiragino Sans GB,sans-serif!important}',
  131. }))
  132. // display absolute date and time
  133. .then(() => page.evaluate(() => {
  134. let time: HTMLTimeElement;
  135. time = document.querySelector('div>div>time');
  136. if (time) time.parentElement.parentElement.style.display = 'none';
  137. time = document.querySelector('article section~div>a>time');
  138. if (time) {
  139. time.innerHTML = time.title + ' ' + new Date(time.dateTime).toLocaleTimeString().slice(0, -3);
  140. time.parentElement.parentElement.style.margin = '-24px 2px 12px';
  141. }
  142. }))
  143. .then(() => chainPromises(morePostProcessings.map(func => () => func(page))))
  144. .then(() => promisify(setTimeout)(getTimeout()))
  145. .then(() => page.screenshot())
  146. .then(screenshot => {
  147. new PNG({
  148. filterType: 4,
  149. deflateLevel: 0,
  150. }).on('parsed', function () {
  151. const idx = (x: number, y: number) => (this.width * y + x) << 2;
  152. let boundary: number = null;
  153. for (let y = this.height - 1; y > this.height - 3840; y -= zoomFactor) {
  154. if (
  155. this.data[idx(zoomFactor, y)] <= 38 &&
  156. this.data[idx(zoomFactor, y)] === this.data[idx(this.width - zoomFactor, y)] &&
  157. this.data[idx(zoomFactor, y + zoomFactor)] === this.data[idx(zoomFactor, y - 2 * zoomFactor)]
  158. ) {
  159. boundary = y - 1;
  160. break;
  161. }
  162. }
  163. if (boundary !== null) {
  164. logger.info(`found boundary at ${boundary}, cropping image`);
  165. this.data = this.data.slice(0, idx(this.width, boundary));
  166. this.height = boundary;
  167. sharpToFile(jpeg(this.pack())).then(path => {
  168. logger.info(`finished webshot for ${url}`);
  169. resolve({path, boundary});
  170. });
  171. } else if (height >= 8 * 1920) {
  172. logger.warn('too large, consider as a bug, returning');
  173. sharpToFile(jpeg(this.pack())).then(path => {
  174. resolve({path, boundary: 0});
  175. });
  176. } else {
  177. logger.info('unable to find boundary, try shooting a larger image');
  178. resolve({path: '', boundary});
  179. }
  180. }).parse(screenshot);
  181. })
  182. .catch(err => {
  183. if (err instanceof Error && err.name !== 'TimeoutError') throw err;
  184. logger.error(`error shooting webshot for ${url}, could not load web page of tweet`);
  185. resolve({path: '', boundary: 0});
  186. })
  187. .finally(() => { page.close(); });
  188. },
  189. zoomFactor, false)
  190. .catch(reject);
  191. });
  192. return promise.then(data => {
  193. if (data.boundary === null) {
  194. return this.renderWebshot(url, height + 3840, webshotDelay, ...morePostProcessings);
  195. } else return data.path;
  196. }).catch(error => this.reconnect(error)
  197. .then(() => this.renderWebshot(url, height, webshotDelay, ...morePostProcessings))
  198. );
  199. };
  200. private fetchMedia = (url: string): Promise<string> => new Promise<ArrayBuffer>((resolve, reject) => {
  201. logger.info(`fetching ${url}`);
  202. axios({
  203. method: 'get',
  204. url,
  205. responseType: 'arraybuffer',
  206. timeout: 150000,
  207. }).then(res => {
  208. if (res.status === 200) {
  209. logger.info(`successfully fetched ${url}`);
  210. resolve(res.data);
  211. } else {
  212. logger.error(`failed to fetch ${url}: ${res.status}`);
  213. reject();
  214. }
  215. }).catch (err => {
  216. logger.error(`failed to fetch ${url}: ${err instanceof Error ? err.message : err}`);
  217. reject();
  218. });
  219. }).then(data =>
  220. (ext => {
  221. const mediaTempFilePath = temp.path({suffix: `.${ext}`});
  222. writeFileSync(mediaTempFilePath, Buffer.from(data));
  223. const path = `file://${mediaTempFilePath}`;
  224. switch (ext) {
  225. case 'jpg':
  226. case 'png':
  227. return Message.Image(path);
  228. case 'mp4':
  229. return Message.Video(path);
  230. }
  231. logger.warn('unable to find MIME type of fetched media, failing this fetch');
  232. throw Error();
  233. })(/\/.*\.(.+?)\?/.exec(url)[1])
  234. );
  235. public webshot(
  236. lazyMediaItems: LazyMediaItem[],
  237. callback: (msgs: string, text: string, author: string) => void,
  238. webshotDelay: number
  239. ): Promise<void> {
  240. let grandPromise = Promise.resolve();
  241. // eslint-disable-next-line @typescript-eslint/no-misused-promises
  242. lazyMediaItems.forEach(lazyItem => grandPromise = grandPromise.then(lazyItem.item).then(item => {
  243. let promise = Promise.resolve();
  244. promise = promise.then(() => {
  245. logger.info(`working on ${item.user.username}/${item.code}`);
  246. });
  247. let messageChain = '';
  248. // text processing
  249. const author = `${item.user.full_name} (@${item.user.username}):\n`;
  250. const text = item.caption?.text || '';
  251. if (this.mode > 0) messageChain += (author + xmlEntities.decode(text));
  252. // invoke webshot
  253. if (this.mode === 0) {
  254. const url = linkBuilder({postUrlSegment: item.code});
  255. promise = promise.then(() => this.renderWebshot(url, 3840, webshotDelay, page =>
  256. // display full name
  257. page.addStyleTag({content:
  258. 'header>div>div+div{font-size:12px; line-height:15px; padding-top:0!important}' +
  259. `header>div>div+div::before{content:"${item.user.full_name}"; color:#8e8e8e; font-weight:bold}`,
  260. })
  261. ))
  262. .then(fileurl => {
  263. if (fileurl) return Message.Image(fileurl);
  264. return author + text;
  265. })
  266. .then(msg => {
  267. if (msg) messageChain += msg;
  268. });
  269. }
  270. // fetch extra entities
  271. const type = (mediaItem): keyof typeof typeInZH =>
  272. (mediaItem as MediaItem).video_versions ? 'video' : 'photo';
  273. const fetchBestCandidate =(
  274. candidates: (Partial<typeof item.video_versions[0]> & typeof item.image_versions2.candidates[0])[],
  275. mediaType: keyof typeof typeInZH
  276. ) => {
  277. const url = candidates
  278. .sort((var1, var2) => var2.width + (var2?.type || 0) - var1.width - (var1?.type || 0))
  279. .map(variant => variant.url)[0]; // largest media
  280. const altMessage = `\n[失败的${typeInZH[mediaType].type}:${url}]`;
  281. return this.fetchMedia(url)
  282. .catch(error => {
  283. logger.warn('unable to fetch media, sending plain text instead...');
  284. return altMessage;
  285. })
  286. .then(msg => { messageChain += msg; });
  287. };
  288. // tslint:disable-next-line: curly
  289. // eslint-disable-next-line curly
  290. if (1 - this.mode % 2) promise = promise.then(() => {
  291. if (item.carousel_media) {
  292. return chainPromises(item.carousel_media.map(carouselItem =>
  293. () => fetchBestCandidate(
  294. (carouselItem as unknown as MediaItem).video_versions ||
  295. carouselItem.image_versions2.candidates,
  296. type(carouselItem)
  297. )
  298. ));
  299. } else if (item.video_versions) {
  300. return fetchBestCandidate(item.video_versions, type(item));
  301. } else if (item.image_versions2) {
  302. return fetchBestCandidate(item.image_versions2.candidates, type(item));
  303. }
  304. });
  305. promise.then(() => {
  306. logger.info(`done working on ${item.user.username}/${item.code}, message chain:`);
  307. logger.info(JSON.stringify(Message.ellipseBase64(messageChain)));
  308. callback(messageChain, xmlEntities.decode(text), author);
  309. });
  310. return promise;
  311. }));
  312. return grandPromise;
  313. }
  314. }
  315. export default Webshot;