123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320 |
- import { writeFileSync } from 'fs';
- import { Readable } from 'stream';
- import { promisify } from 'util';
- import axios from 'axios';
- import * as CallableInstance from 'callable-instance';
- import { XmlEntities } from 'html-entities';
- import { PNG } from 'pngjs';
- import * as puppeteer from 'playwright';
- import * as sharp from 'sharp';
- import * as temp from 'temp';
- import { getLogger } from './loggers';
- import { Message } from './koishi';
- import { chainPromises } from './utils';
- import { linkBuilder, MediaItem, LazyMediaItem, WebshotHelpers } from './twitter';
- const xmlEntities = new XmlEntities();
- const ZHType = (type: string) => new class extends String {
- public type = super.toString();
- public toString = () => `[${super.toString()}]`;
- }(type);
- const typeInZH = {
- photo: ZHType('图片'),
- video: ZHType('视频'),
- };
- const logger = getLogger('webshot');
- export type Page = puppeteer.Page;
- export type Cookies = puppeteer.Cookie[];
- class Webshot extends CallableInstance<[LazyMediaItem[], (...args) => void, number], Promise<void>> {
- private browser: puppeteer.Browser;
- private mode: number;
- private wsUrl: string;
- private getCookies: () => Cookies;
- constructor(
- wsUrl: string, mode: number,
- getCookies: () => Cookies,
- onready?: (doOnNewPage?: typeof Webshot.prototype.performOnNewPage) => void
- ) {
- super('webshot');
- // tslint:disable-next-line: no-conditional-assignment
- // eslint-disable-next-line no-cond-assign
- if (this.mode = mode) {
- onready();
- } else {
- this.getCookies = getCookies;
- this.wsUrl = wsUrl;
- this.connect(() => onready(this.performOnNewPage));
- }
- }
- private connect = (onready?: (...args) => void): Promise<void> =>
- axios.get<{[key in 'chromium' | 'firefox' | 'webkit']?: string}>(this.wsUrl)
- .then(res => {
- logger.info(`received websocket endpoint: ${JSON.stringify(res.data)}`);
- const browserType = Object.keys(res.data)[0] as keyof typeof res.data;
- return (puppeteer[browserType] as puppeteer.BrowserType<puppeteer.Browser>)
- .connect({wsEndpoint: res.data[browserType]});
- })
- .then(browser => this.browser = browser)
- .then(() => {
- logger.info('launched puppeteer browser');
- if (onready) return onready();
- })
- .catch(error => this.reconnect(error, onready));
- private reconnect = (error, onready?: (...args) => void) => {
- logger.error(`connection error, reason: ${error}`);
- logger.warn('trying to reconnect in 2.5s...');
- return promisify(setTimeout)(2500)
- .then(() => this.connect(onready));
- };
- private performOnNewPage = <T>(action: (page: Page) => T | PromiseLike<T>, zoomFactor = 2, reconnectOnError = true) =>
- this.browser.newPage({
- bypassCSP: true,
- deviceScaleFactor: zoomFactor,
- locale: 'ja-JP',
- timezoneId: 'Asia/Tokyo',
- userAgent: 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/67.0.3396.99 Safari/537.36',
- }).then(action)
- .catch(error => {
- if (reconnectOnError) {
- return this.reconnect(error)
- .then((): Promise<T> => this.performOnNewPage(action, zoomFactor, reconnectOnError));
- }
- throw error;
- });
- private renderWebshot = (url: string, height: number, webshotDelay: number): Promise<string> => {
- temp.track();
- const jpeg = (data: Readable) => data.pipe(sharp()).jpeg({quality: 90, trellisQuantisation: true});
- const sharpToFile = (pic: sharp.Sharp) => new Promise<string>(resolve => {
- const webshotTempFilePath = temp.path({suffix: '.jpg'});
- pic.toFile(webshotTempFilePath).then(() => resolve(`file://${webshotTempFilePath}`));
- });
- const promise = new Promise<{ path: string, boundary: null | number }>((resolve, reject) => {
- const width = 720;
- const zoomFactor = 2;
- logger.info(`shooting ${width}*${height} webshot for ${url}`);
- this.performOnNewPage(
- page => {
- const startTime = new Date().getTime();
- const getTimerTime = () => new Date().getTime() - startTime;
- const getTimeout = () => Math.max(500, webshotDelay - getTimerTime());
- page.setViewportSize({
- width: width / zoomFactor,
- height: height / zoomFactor,
- }).then(() => page.context().addCookies(this.getCookies()))
- .then(() => page.goto(url, {waitUntil: 'load', timeout: getTimeout()}))
- .then(() => WebshotHelpers.handleCookieConsent(page))
- .then(() =>
- (<T>(next: () => Promise<T>) => Promise.race([
- WebshotHelpers.handleLogin(page)
- .then(() => page.goto(url, {waitUntil: 'load', timeout: getTimeout()}))
- .then(next),
- next(),
- ]))(() => page.waitForSelector('article', {timeout: getTimeout()}))
- )
- .catch((err: Error): Promise<puppeteer.ElementHandle<Element> | null> => {
- if (err.name !== 'TimeoutError') throw err;
- logger.warn(`navigation timed out at ${getTimerTime()} ms`);
- return null;
- })
- // hide header, "more options" button, like and retweet count
- .then(() => page.addStyleTag({content:
- 'nav,footer,header+div,header+div+div>div>div+div,header div div div+div,' +
- 'article section,article section+div>ul>:not(div),article button,canvas{display:none!important} ' +
- 'section+div{overflow:hidden} section+*>*{position:relative!important} article{border-bottom:1px solid!important}',
- }))
- .then(() => page.addStyleTag({
- content: '*{font-family:-apple-system,".Helvetica Neue DeskInterface",Hiragino Sans,Hiragino Sans GB,sans-serif!important}',
- }))
- .then(() => page.evaluate(() => {
- let time: HTMLTimeElement;
- time = document.querySelector('div>div>time');
- if (time) time.parentElement.parentElement.style.display = 'none';
- time = document.querySelector('article section~div>a>time');
- if (time) {
- time.innerHTML = time.title + ' ' + new Date(time.dateTime).toLocaleTimeString().slice(0, -3);
- time.parentElement.parentElement.style.margin = '-24px 0 12px';
- }
- }))
- .then(() => page.screenshot())
- .then(screenshot => {
- new PNG({
- filterType: 4,
- deflateLevel: 0,
- }).on('parsed', function () {
- const idx = (x: number, y: number) => (this.width * y + x) << 2;
- let boundary: number = null;
- for (let y = this.height - 1; y > this.height - 1920; y -= zoomFactor) {
- if (
- this.data[idx(zoomFactor, y)] <= 38 &&
- this.data[idx(zoomFactor, y)] === this.data[idx(this.width - zoomFactor, y)] &&
- this.data[idx(zoomFactor, y + zoomFactor)] === this.data[idx(zoomFactor, y - 2 * zoomFactor)]
- ) {
- boundary = y - 1;
- break;
- }
- }
- if (boundary !== null) {
- logger.info(`found boundary at ${boundary}, cropping image`);
- this.data = this.data.slice(0, idx(this.width, boundary));
- this.height = boundary;
- sharpToFile(jpeg(this.pack())).then(path => {
- logger.info(`finished webshot for ${url}`);
- resolve({path, boundary});
- });
- } else if (height >= 8 * 1920) {
- logger.warn('too large, consider as a bug, returning');
- sharpToFile(jpeg(this.pack())).then(path => {
- resolve({path, boundary: 0});
- });
- } else {
- logger.info('unable to find boundary, try shooting a larger image');
- resolve({path: '', boundary});
- }
- }).parse(screenshot);
- })
- .catch(err => {
- if (err instanceof Error && err.name !== 'TimeoutError') throw err;
- logger.error(`error shooting webshot for ${url}, could not load web page of tweet`);
- resolve({path: '', boundary: 0});
- })
- .finally(() => { page.close(); });
- },
- zoomFactor, false)
- .catch(reject);
- });
- return promise.then(data => {
- if (data.boundary === null) return this.renderWebshot(url, height + 1920, webshotDelay);
- else return data.path;
- }).catch(error => this.reconnect(error)
- .then(() => this.renderWebshot(url, height, webshotDelay))
- );
- };
- private fetchMedia = (url: string): Promise<string> => new Promise<ArrayBuffer>((resolve, reject) => {
- logger.info(`fetching ${url}`);
- axios({
- method: 'get',
- url,
- responseType: 'arraybuffer',
- timeout: 150000,
- }).then(res => {
- if (res.status === 200) {
- logger.info(`successfully fetched ${url}`);
- resolve(res.data);
- } else {
- logger.error(`failed to fetch ${url}: ${res.status}`);
- reject();
- }
- }).catch (err => {
- logger.error(`failed to fetch ${url}: ${err instanceof Error ? err.message : err}`);
- reject();
- });
- }).then(data =>
- (ext => {
- const mediaTempFilePath = temp.path({suffix: `.${ext}`});
- writeFileSync(mediaTempFilePath, Buffer.from(data));
- const path = `file://${mediaTempFilePath}`;
- switch (ext) {
- case 'jpg':
- case 'png':
- return Message.Image(path);
- case 'mp4':
- return Message.Video(path);
- }
- logger.warn('unable to find MIME type of fetched media, failing this fetch');
- throw Error();
- })(/\/.*\.(.+?)\?/.exec(url)[1])
- );
- public webshot(
- lazyMediaItems: LazyMediaItem[],
- callback: (msgs: string, text: string, author: string) => void,
- webshotDelay: number
- ): Promise<void> {
- let grandPromise = Promise.resolve();
- // eslint-disable-next-line @typescript-eslint/no-misused-promises
- lazyMediaItems.forEach(lazyItem => grandPromise = grandPromise.then(lazyItem.item).then(item => {
- let promise = Promise.resolve();
- promise = promise.then(() => {
- logger.info(`working on ${item.user.username}/${item.code}`);
- });
- let messageChain = '';
- // text processing
- const author = `${item.user.full_name} (@${item.user.username}):\n`;
- const text = item.caption?.text || '';
- if (this.mode > 0) messageChain += (author + xmlEntities.decode(text));
- // invoke webshot
- if (this.mode === 0) {
- const url = linkBuilder({postUrlSegment: item.code});
- promise = promise.then(() => this.renderWebshot(url, 1920, webshotDelay))
- .then(fileurl => {
- if (fileurl) return Message.Image(fileurl);
- return author + text;
- })
- .then(msg => {
- if (msg) messageChain += msg;
- });
- }
- // fetch extra entities
- const type = (mediaItem): keyof typeof typeInZH =>
- (mediaItem as MediaItem).video_versions ? 'video' : 'photo';
- const fetchBestCandidate =(
- candidates: (Partial<typeof item.video_versions[0]> & typeof item.image_versions2.candidates[0])[],
- mediaType: keyof typeof typeInZH
- ) => {
- const url = candidates
- .sort((var1, var2) => var2.width + (var2?.type || 0) - var1.width - (var1?.type || 0))
- .map(variant => variant.url)[0]; // largest media
- const altMessage = `\n[失败的${typeInZH[mediaType].type}:${url}]`;
- return this.fetchMedia(url)
- .catch(error => {
- logger.warn('unable to fetch media, sending plain text instead...');
- return altMessage;
- })
- .then(msg => { messageChain += msg; });
- };
- // tslint:disable-next-line: curly
- // eslint-disable-next-line curly
- if (1 - this.mode % 2) promise = promise.then(() => {
- if (item.carousel_media) {
- return chainPromises(item.carousel_media.map(carouselItem =>
- fetchBestCandidate(
- (carouselItem as unknown as MediaItem).video_versions ||
- carouselItem.image_versions2.candidates,
- type(carouselItem)
- )
- ));
- } else if (item.video_versions) {
- return fetchBestCandidate(item.video_versions, type(item));
- } else if (item.image_versions2) {
- return fetchBestCandidate(item.image_versions2.candidates, type(item));
- }
- });
- promise.then(() => {
- logger.info(`done working on ${item.user.username}/${item.code}, message chain:`);
- logger.info(JSON.stringify(Message.ellipseBase64(messageChain)));
- callback(messageChain, xmlEntities.decode(text), author);
- });
- return promise;
- }));
- return grandPromise;
- }
- }
- export default Webshot;
|