gocqhttp-twitter-bot/dist/webshot.js

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
const fs_1 = require("fs");
const util_1 = require("util");
const axios_1 = require("axios");
const CallableInstance = require("callable-instance");
const html_entities_1 = require("html-entities");
const pngjs_1 = require("pngjs");
const puppeteer = require("playwright");
const sharp = require("sharp");
const temp = require("temp");
const loggers_1 = require("./loggers");
const koishi_1 = require("./koishi");
const utils_1 = require("./utils");
const xmlEntities = new html_entities_1.XmlEntities();
const ZHType = (type) => new class extends String {
    constructor() {
        super(...arguments);
        this.type = super.toString();
        this.toString = () => `[${super.toString()}]`;
    }
}(type);
const typeInZH = {
    photo: ZHType('图片'),
    video: ZHType('视频'),
    animated_gif: ZHType('GIF'),
};
const logger = loggers_1.getLogger('webshot');
class Webshot extends CallableInstance {
    constructor(wsUrl, mode, onready) {
        super('webshot');
        this.connect = (onready) => axios_1.default.get(this.wsUrl)
            .then(res => {
            logger.info(`received websocket endpoint: ${JSON.stringify(res.data)}`);
            const browserType = Object.keys(res.data)[0];
            return puppeteer[browserType]
                .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));
        this.reconnect = (error, onready) => {
            logger.error(`connection error, reason: ${error}`);
            logger.warn('trying to reconnect in 2.5s...');
            return util_1.promisify(setTimeout)(2500)
                .then(() => this.connect(onready));
        };
        this.extendEntity = (media) => {
            logger.info('not working on a tweet');
        };
        this.truncateLongThread = (atId) => {
            logger.info('not working on a tweet');
        };
        this.renderWebshot = (url, height, webshotDelay, ...morePostProcessings) => {
            temp.track();
            const jpeg = (data) => data.pipe(sharp()).jpeg({ quality: 90, trellisQuantisation: true });
            const sharpToFile = (pic) => new Promise(resolve => {
                const webshotTempFilePath = temp.path({ suffix: '.jpg' });
                pic.toFile(webshotTempFilePath).then(() => resolve(`file://${webshotTempFilePath}`));
            });
            const promise = new Promise((resolve, reject) => {
                const width = 720;
                const zoomFactor = 2;
                logger.info(`shooting ${width}*${height} webshot for ${url}`);
                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(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.route('*:\/\/video.twimg.com\/**', route => { route.abort(); }))
                        .then(() => page.goto(url, { waitUntil: 'load', timeout: getTimeout() }))
                        .then(() => Promise.race([
                        page.waitForSelector('article', { state: 'attached', timeout: getTimeout() }),
                        page.click('#placeholder+#ScriptLoadFailure input[value="Try again"]', { timeout: getTimeout() }),
                    ]))
                        .then(() => page.addStyleTag({
                        content: 'header,#layers{display:none!important}article{background-color:transparent!important}' +
                            '[data-testid="caret"],[role="group"],[data-testid="tweet"] [class*=" "]+:last-child>*+[class*=" "]~div{display:none}',
                    }))
                        .then(() => page.addStyleTag({
                        content: '*{font-family:-apple-system,".Helvetica Neue DeskInterface",Hiragino Sans,Hiragino Sans GB,sans-serif!important}',
                    }))
                        .then(() => page.evaluate(() => {
                        const poll = setInterval(() => {
                            document.querySelectorAll('div[data-testid="placementTracking"]').forEach(container => {
                                if (container.querySelector('div[role="button"] svg')) {
                                    container.innerHTML = container.innerHTML;
                                    clearInterval(poll);
                                }
                            });
                        }, 250);
                    }))
                        .then(() => page.waitForSelector('xpath=//section/*/*/div[.//article[not(.//time[not(ancestor::div[@aria-labelledby])])]]', { state: 'attached', timeout: getTimeout() }))
                        .then(handle => handle.$$('xpath=..//a[contains(@href,"content_you_see")]/../../..//*[@role="button"]')
                        .then(sensitiveToggles => {
                        const count = sensitiveToggles.length;
                        if (count)
                            logger.info(`found ${count} sensitive ${count === 1 ? 'tweet' : 'tweets'} on page, uncollapsing...`);
                        return utils_1.chainPromises(sensitiveToggles.filter(toggle => toggle.isVisible()).map(toggle => () => toggle.click()));
                    })
                        .then(() => handle))
                        .then(handle => handle.$('[data-testid="tweet"]').then(owner => owner ? handle : null))
                        .catch((err) => {
                        if (err.name !== 'TimeoutError')
                            throw err;
                        logger.warn(`${err} (${getTimerTime()} ms)`);
                        return page.evaluate(() => document.documentElement.outerHTML).then(html => {
                            const path = temp.path({ suffix: '.html' });
                            fs_1.writeFileSync(path, html);
                            logger.warn(`saved debug html to ${path}`);
                        }).then(() => page.screenshot()).then(screenshot => {
                            sharpToFile(sharp(screenshot).jpeg({ quality: 90 })).then(fileUri => {
                                logger.warn(`saved debug screenshot to ${fileUri.substring(7)}`);
                            });
                        }).then(() => null);
                    })
                        .then((handle) => {
                        if (handle === null)
                            throw new puppeteer.errors.TimeoutError();
                        return handle.evaluate(div => {
                            try {
                                const selector = '[data-testid="tweet"] :nth-child(2)>:first-child a';
                                const getProfileUrl = () => (div.querySelector(selector) || { href: '' }).href;
                                const ownerProfileUrl = getProfileUrl();
                                const bottom = div;
                                while (div = div.previousElementSibling) {
                                    if (getProfileUrl() !== ownerProfileUrl || div === bottom.previousElementSibling)
                                        continue;
                                    const top = document.documentElement.scrollTop = window.scrollY + div.getBoundingClientRect().top;
                                    if (top > 10)
                                        return div.querySelector('article a[aria-label]').href.replace(/.*\/status\//, '');
                                }
                            }
                            catch (_a) { }
                            document.documentElement.scrollTop = 0;
                        }).then(this.truncateLongThread).then(() => handle);
                    })
                        .then(handle => handle.evaluate(div => {
                        const cardImg = div.querySelector('div[data-testid^="card.layout"][data-testid$=".media"] img');
                        if (typeof (cardImg === null || cardImg === void 0 ? void 0 : cardImg.getAttribute('src')) === 'string') {
                            const match = /^(.*\/card_img\/(\d+)\/.+\?format=.*)&name=/.exec(cardImg === null || cardImg === void 0 ? void 0 : cardImg.getAttribute('src'));
                            if (match) {
                                const [media_url_https, id_str] = match.slice(1);
                                return {
                                    media_url: media_url_https.replace(/^https/, 'http'),
                                    media_url_https,
                                    url: '',
                                    display_url: '',
                                    expanded_url: '',
                                    type: 'photo',
                                    id: Number(id_str),
                                    id_str,
                                    sizes: undefined,
                                };
                            }
                        }
                    }))
                        .then(cardImg => {
                        if (cardImg)
                            this.extendEntity(cardImg);
                    })
                        .then(() => utils_1.chainPromises(morePostProcessings.map(func => () => func(page))))
                        .then(() => util_1.promisify(setTimeout)(getTimeout()))
                        .then(() => page.evaluate(() => document.activeElement.blur()))
                        .then(() => page.screenshot())
                        .then(screenshot => {
                        new pngjs_1.PNG({
                            filterType: 4,
                            deflateLevel: 0,
                        }).on('parsed', function () {
                            const idx = (x, y) => (this.width * y + x) << 2;
                            let boundary = null;
                            const x = zoomFactor * 2;
                            for (let y = x; y < this.height; y += zoomFactor) {
                                if (this.data[idx(x, y)] !== this.data[idx(x, y - zoomFactor)] &&
                                    this.data[idx(x, y)] === this.data[idx(x + zoomFactor * 10, y)]) {
                                    boundary = y;
                                    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(); });
                })
                    .catch(reject);
            });
            return promise.then(data => {
                if (data.boundary === null) {
                    return this.renderWebshot(url, height + 1920, webshotDelay, ...morePostProcessings);
                }
                else
                    return data.path;
            }).catch(error => this.reconnect(error)
                .then(() => this.renderWebshot(url, height, webshotDelay, ...morePostProcessings)));
        };
        this.fetchMedia = (url) => new Promise((resolve, reject) => {
            logger.info(`fetching ${url}`);
            axios_1.default({
                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}: ${err instanceof Error ? err.message : err}`);
                reject();
            });
        }).then(data => {
            var _a;
            return (ext => {
                const mediaTempFilePath = temp.path({ suffix: `.${ext}` });
                fs_1.writeFileSync(mediaTempFilePath, Buffer.from(data));
                const path = `file://${mediaTempFilePath}`;
                switch (ext) {
                    case 'jpg':
                    case 'png':
                        return koishi_1.Message.Image(path);
                    case 'mp4':
                        return koishi_1.Message.Video(path);
                }
                logger.warn('unable to find MIME type of fetched media, failing this fetch');
                throw Error();
            })(((_a = (/\?format=([a-z]+)&/.exec(url))) !== null && _a !== void 0 ? _a : (/.*\/.*\.([^?]+)/.exec(url)))[1]);
        });
        if (this.mode = mode) {
            onready();
        }
        else {
            this.wsUrl = wsUrl;
            this.connect(onready);
        }
    }
    webshot(tweets, callback, webshotDelay) {
        let promise = new Promise(resolve => {
            resolve();
        });
        tweets.forEach(twi => {
            promise = promise.then(() => {
                logger.info(`working on ${twi.user.screen_name}/${twi.id_str}`);
            });
            const originTwi = twi.retweeted_status || twi;
            let messageChain = '';
            let truncatedAt;
            let author = `${twi.user.name} (@${twi.user.screen_name}):\n`;
            if (twi.retweeted_status)
                author += `RT @${twi.retweeted_status.user.screen_name}: `;
            let text = originTwi.full_text;
            promise = promise.then(() => {
                if (originTwi.entities && originTwi.entities.urls && originTwi.entities.urls.length) {
                    originTwi.entities.urls.forEach(url => {
                        text = text.replace(new RegExp(url.url, 'gm'), url.expanded_url);
                    });
                }
                if (originTwi.extended_entities) {
                    originTwi.extended_entities.media.forEach(media => {
                        text = text.replace(new RegExp(media.url, 'gm'), this.mode === 1 ? typeInZH[media.type] : '');
                    });
                }
                if (this.mode > 0)
                    messageChain += (author + xmlEntities.decode(text));
            });
            if (this.mode === 0) {
                const url = `https://mobile.twitter.com/${twi.user.screen_name}/status/${twi.id_str}`;
                this.extendEntity = (cardImg) => {
                    var _a, _b;
                    originTwi.extended_entities = Object.assign(Object.assign({}, originTwi.extended_entities), { media: [
                            ...(_b = (_a = originTwi.extended_entities) === null || _a === void 0 ? void 0 : _a.media) !== null && _b !== void 0 ? _b : [],
                            cardImg,
                        ] });
                };
                this.truncateLongThread = (atId) => {
                    if (!atId)
                        return;
                    logger.info(`thread too long, truncating at tweet ${atId}...`);
                    truncatedAt = atId;
                };
                promise = promise.then(() => this.renderWebshot(url, 1920, webshotDelay))
                    .then(fileurl => {
                    if (fileurl)
                        return koishi_1.Message.Image(fileurl);
                    return '[截图不可用] ' + author + text;
                })
                    .then(msg => {
                    if (msg)
                        messageChain += msg;
                });
            }
            if (1 - this.mode % 2)
                promise = promise.then(() => {
                    if (originTwi.extended_entities) {
                        return utils_1.chainPromises(originTwi.extended_entities.media.map(media => () => {
                            let url;
                            if (media.type === 'photo') {
                                url = media.media_url_https.replace(/\.([a-z]+)$/, '?format=$1') + '&name=orig';
                            }
                            else {
                                url = media.video_info.variants
                                    .filter(variant => variant.bitrate !== undefined)
                                    .sort((var1, var2) => var2.bitrate - var1.bitrate)
                                    .map(variant => variant.url)[0];
                            }
                            const altMessage = `\n[失败的${typeInZH[media.type].type}：${url}]`;
                            return this.fetchMedia(url)
                                .catch(error => {
                                logger.warn('unable to fetch media, sending plain text instead...');
                                return altMessage;
                            })
                                .then(msg => { messageChain += msg; });
                        }));
                    }
                });
            if (this.mode === 0) {
                if (originTwi.entities && originTwi.entities.urls && originTwi.entities.urls.length) {
                    promise = promise.then(() => {
                        const urls = originTwi.entities.urls
                            .filter(urlObj => urlObj.indices[0] < originTwi.display_text_range[1])
                            .map(urlObj => `\n\ud83d\udd17 ${urlObj.expanded_url}`);
                        if (urls.length) {
                            messageChain += urls.join('');
                        }
                    });
                }
            }
            promise = promise.then(() => {
                if (truncatedAt) {
                    messageChain += `\n回复此命令查看对话串中更早的推文：\n/twitter_view ${truncatedAt}`;
                }
            });
            if (originTwi.is_quote_status) {
                promise = promise.then(() => {
                    var _a, _b;
                    const match = /\/status\/(\d+)/.exec((_a = originTwi.quoted_status_permalink) === null || _a === void 0 ? void 0 : _a.expanded);
                    const blockQuoteIdStr = match ? match[1] : (_b = originTwi.quoted_status) === null || _b === void 0 ? void 0 : _b.id_str;
                    if (blockQuoteIdStr)
                        messageChain += `\n回复此命令查看引用的推文：\n/twitter_view ${blockQuoteIdStr}`;
                });
            }
            promise.then(() => {
                logger.info(`done working on ${twi.user.screen_name}/${twi.id_str}, message chain:`);
                logger.info(JSON.stringify(koishi_1.Message.ellipseBase64(messageChain)));
                const twiId = twi.retweeted_status ? twi.retweeted_status.id_str : twi.id_str;
                callback(twiId, messageChain, xmlEntities.decode(text), author);
            });
        });
        return promise;
    }
}
exports.default = Webshot;