webshot.ts 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333
  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. onready();
  42. } else {
  43. this.getCookies = getCookies;
  44. this.wsUrl = wsUrl;
  45. this.connect(() => 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 = () => 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. ]))(() => 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, carousel navigator, and comments
  122. .then(() => page.addStyleTag({content:
  123. 'nav,footer,header+div,header+div+div>div>div+div,header div div div+div,' +
  124. 'article section,article section+div>ul>:not(div),article button,canvas{display:none!important} ' +
  125. 'section+div{overflow:hidden} section+*>*{position:relative!important} article{border-bottom:1px solid!important}',
  126. }))
  127. .then(() => page.addStyleTag({
  128. content: '*{font-family:-apple-system,".Helvetica Neue DeskInterface",Hiragino Sans,Hiragino Sans GB,sans-serif!important}',
  129. }))
  130. // display absolute date and time
  131. .then(() => page.evaluate(() => {
  132. let time: HTMLTimeElement;
  133. time = document.querySelector('div>div>time');
  134. if (time) time.parentElement.parentElement.style.display = 'none';
  135. time = document.querySelector('article section~div>a>time');
  136. if (time) {
  137. time.innerHTML = time.title + ' ' + new Date(time.dateTime).toLocaleTimeString().slice(0, -3);
  138. time.parentElement.parentElement.style.margin = '-24px 0 12px';
  139. }
  140. }))
  141. .then(() => chainPromises(morePostProcessings.map(func => () => func(page))))
  142. .then(() => page.screenshot())
  143. .then(screenshot => {
  144. new PNG({
  145. filterType: 4,
  146. deflateLevel: 0,
  147. }).on('parsed', function () {
  148. const idx = (x: number, y: number) => (this.width * y + x) << 2;
  149. let boundary: number = null;
  150. for (let y = this.height - 1; y > this.height - 1920; y -= zoomFactor) {
  151. if (
  152. this.data[idx(zoomFactor, y)] <= 38 &&
  153. this.data[idx(zoomFactor, y)] === this.data[idx(this.width - zoomFactor, y)] &&
  154. this.data[idx(zoomFactor, y + zoomFactor)] === this.data[idx(zoomFactor, y - 2 * zoomFactor)]
  155. ) {
  156. boundary = y - 1;
  157. break;
  158. }
  159. }
  160. if (boundary !== null) {
  161. logger.info(`found boundary at ${boundary}, cropping image`);
  162. this.data = this.data.slice(0, idx(this.width, boundary));
  163. this.height = boundary;
  164. sharpToFile(jpeg(this.pack())).then(path => {
  165. logger.info(`finished webshot for ${url}`);
  166. resolve({path, boundary});
  167. });
  168. } else if (height >= 8 * 1920) {
  169. logger.warn('too large, consider as a bug, returning');
  170. sharpToFile(jpeg(this.pack())).then(path => {
  171. resolve({path, boundary: 0});
  172. });
  173. } else {
  174. logger.info('unable to find boundary, try shooting a larger image');
  175. resolve({path: '', boundary});
  176. }
  177. }).parse(screenshot);
  178. })
  179. .catch(err => {
  180. if (err instanceof Error && err.name !== 'TimeoutError') throw err;
  181. logger.error(`error shooting webshot for ${url}, could not load web page of tweet`);
  182. resolve({path: '', boundary: 0});
  183. })
  184. .finally(() => { page.close(); });
  185. },
  186. zoomFactor, false)
  187. .catch(reject);
  188. });
  189. return promise.then(data => {
  190. if (data.boundary === null) {
  191. return this.renderWebshot(url, height + 1920, webshotDelay, ...morePostProcessings);
  192. } else return data.path;
  193. }).catch(error => this.reconnect(error)
  194. .then(() => this.renderWebshot(url, height, webshotDelay, ...morePostProcessings))
  195. );
  196. };
  197. private fetchMedia = (url: string): Promise<string> => new Promise<ArrayBuffer>((resolve, reject) => {
  198. logger.info(`fetching ${url}`);
  199. axios({
  200. method: 'get',
  201. url,
  202. responseType: 'arraybuffer',
  203. timeout: 150000,
  204. }).then(res => {
  205. if (res.status === 200) {
  206. logger.info(`successfully fetched ${url}`);
  207. resolve(res.data);
  208. } else {
  209. logger.error(`failed to fetch ${url}: ${res.status}`);
  210. reject();
  211. }
  212. }).catch (err => {
  213. logger.error(`failed to fetch ${url}: ${err instanceof Error ? err.message : err}`);
  214. reject();
  215. });
  216. }).then(data =>
  217. (ext => {
  218. const mediaTempFilePath = temp.path({suffix: `.${ext}`});
  219. writeFileSync(mediaTempFilePath, Buffer.from(data));
  220. const path = `file://${mediaTempFilePath}`;
  221. switch (ext) {
  222. case 'jpg':
  223. case 'png':
  224. return Message.Image(path);
  225. case 'mp4':
  226. return Message.Video(path);
  227. }
  228. logger.warn('unable to find MIME type of fetched media, failing this fetch');
  229. throw Error();
  230. })(/\/.*\.(.+?)\?/.exec(url)[1])
  231. );
  232. public webshot(
  233. lazyMediaItems: LazyMediaItem[],
  234. callback: (msgs: string, text: string, author: string) => void,
  235. webshotDelay: number
  236. ): Promise<void> {
  237. let grandPromise = Promise.resolve();
  238. // eslint-disable-next-line @typescript-eslint/no-misused-promises
  239. lazyMediaItems.forEach(lazyItem => grandPromise = grandPromise.then(lazyItem.item).then(item => {
  240. let promise = Promise.resolve();
  241. promise = promise.then(() => {
  242. logger.info(`working on ${item.user.username}/${item.code}`);
  243. });
  244. let messageChain = '';
  245. // text processing
  246. const author = `${item.user.full_name} (@${item.user.username}):\n`;
  247. const text = item.caption?.text || '';
  248. if (this.mode > 0) messageChain += (author + xmlEntities.decode(text));
  249. // invoke webshot
  250. if (this.mode === 0) {
  251. const url = linkBuilder({postUrlSegment: item.code});
  252. promise = promise.then(() => this.renderWebshot(url, 1920, webshotDelay, page =>
  253. // display full name
  254. page.addStyleTag({content:
  255. 'header div div div *{text-align:left!important} header div div div a::after{' +
  256. `content:"${item.user.full_name}";` +
  257. 'display:block; font-size:smaller; color: #8e8e8e; line-height:1.25}',
  258. })
  259. ))
  260. .then(fileurl => {
  261. if (fileurl) return Message.Image(fileurl);
  262. return author + text;
  263. })
  264. .then(msg => {
  265. if (msg) messageChain += msg;
  266. });
  267. }
  268. // fetch extra entities
  269. const type = (mediaItem): keyof typeof typeInZH =>
  270. (mediaItem as MediaItem).video_versions ? 'video' : 'photo';
  271. const fetchBestCandidate =(
  272. candidates: (Partial<typeof item.video_versions[0]> & typeof item.image_versions2.candidates[0])[],
  273. mediaType: keyof typeof typeInZH
  274. ) => {
  275. const url = candidates
  276. .sort((var1, var2) => var2.width + (var2?.type || 0) - var1.width - (var1?.type || 0))
  277. .map(variant => variant.url)[0]; // largest media
  278. const altMessage = `\n[失败的${typeInZH[mediaType].type}:${url}]`;
  279. return this.fetchMedia(url)
  280. .catch(error => {
  281. logger.warn('unable to fetch media, sending plain text instead...');
  282. return altMessage;
  283. })
  284. .then(msg => { messageChain += msg; });
  285. };
  286. // tslint:disable-next-line: curly
  287. // eslint-disable-next-line curly
  288. if (1 - this.mode % 2) promise = promise.then(() => {
  289. if (item.carousel_media) {
  290. return chainPromises(item.carousel_media.map(carouselItem =>
  291. () => fetchBestCandidate(
  292. (carouselItem as unknown as MediaItem).video_versions ||
  293. carouselItem.image_versions2.candidates,
  294. type(carouselItem)
  295. )
  296. ));
  297. } else if (item.video_versions) {
  298. return fetchBestCandidate(item.video_versions, type(item));
  299. } else if (item.image_versions2) {
  300. return fetchBestCandidate(item.image_versions2.candidates, type(item));
  301. }
  302. });
  303. promise.then(() => {
  304. logger.info(`done working on ${item.user.username}/${item.code}, message chain:`);
  305. logger.info(JSON.stringify(Message.ellipseBase64(messageChain)));
  306. callback(messageChain, xmlEntities.decode(text), author);
  307. });
  308. return promise;
  309. }));
  310. return grandPromise;
  311. }
  312. }
  313. export default Webshot;