gocqhttp-twitter-bot/src/webshot.ts

import { writeFileSync } from 'fs';

import axios from 'axios';
import * as CallableInstance from 'callable-instance';
import { XmlEntities } from 'html-entities';
import * as temp from 'temp';

import { getLogger } from './loggers';
import { Message } from './koishi';
import { MediaItem } 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');

class Webshot extends CallableInstance<[MediaItem[], (...args) => void, number], Promise<void>> {
  constructor(_wsUrl: string, _mode: number, onready?: (...args) => void) {
    super('webshot');
    onready();
  }

  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();
    })(/\/.+\.(?:.*?(?<=[?&])stp=dst-(jpg)|(.+?)\?)/.exec(url).filter(g => g)[1])
  );

  public fetchBestCandidate = ({image_versions2, video_versions}: MediaItem) => {
    const candidates: (
      Partial<typeof video_versions[0]> & typeof image_versions2.candidates[0]
    )[] = video_versions || image_versions2.candidates;
    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${typeInZH[video_versions ? 'video' : 'photo'].type}：${url}]`;
    return this.fetchMedia(url)
      .catch(error => {
        logger.warn('unable to fetch media, sending plain text instead...');
        return altMessage;
      });
  };

  public webshot(
    mediaItems: MediaItem[],
    callback: (msgs: string, text: string, author: string) => void,
    webshotDelay: number
  ): Promise<void> {
    const promises = mediaItems.map(item => {
      let promise = Promise.resolve();
      logger.info(`working on ${item.user.username}/${item.code}`);
      let messageChain = '';

      // text processing
      const author = `${item.user.full_name} (@${item.user.username}):\n`;
      const date = `${new Date(item.taken_at * 1000)}\n`;
      messageChain += author + date;

      // fetch extra entities
      promise = promise.then(() => this.fetchBestCandidate(item))
        .then(msg => { messageChain += msg; });
      return 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(item.caption), author);
      });
    });
    return Promise.all(promises).then();
  }
}

export default Webshot;