wip: get roominfo

chenfan0 · Oct 29, 2024 · 7cf68e2 · 7cf68e2
1 parent 36d8bf7
commit 7cf68e2
Show file tree

Hide file tree

Showing 17 changed files with 282 additions and 78 deletions.
diff --git a/src/const.ts b/src/const.ts
@@ -1,6 +1,7 @@
 export const NAV_BY_DEFAULT_BROWSER = 'NAV_BY_DEFAULT_BROWSER'
 export const SELECT_DIR = 'SELECT_DIR'
 export const GET_LIVE_URLS = 'GET_LIVE_URLS'
+export const GET_ROOM_INFO = 'GET_ROOM_INFO'
 export const START_STREAM_RECORD = 'START_STREAM_RECORD'
 export const STOP_STREAM_RECORD = 'STOP_STREAM_RECORD'
 export const STREAM_RECORD_END = 'STREAM_RECORD_END'

diff --git a/src/main/crawler/capture-error.js b/src/main/crawler/capture-error.js
@@ -14,21 +14,15 @@ export function captureError(fn) {
     const realArgs = args.slice(0, args.length - 1)
     try {
       const res = await fn.apply(this, realArgs)
-      writeLog(`Fetch Live Res: ${JSON.stringify(res, null, 2)}`)
+      writeLog(`Fetch Res: ${JSON.stringify(res, null, 2)}`)
       return res
     } catch (e) {
       const message = e.message
 
-      writeLog(`Fetch Live Error: ${message}`)
+      writeLog(`Fetch Error: ${message}`)
 
       log('error:', message)
 
-      // if (message.includes('timeout')) {
-      //   return {
-      //     code: CRAWLER_ERROR_CODE.TIMEOUT
-      //   }
-      // }
-
       if (message === ERROR_MESSAGE.INVALID_PROXY) {
         return {
           code: CRAWLER_ERROR_CODE.INVALID_PROXY

diff --git a/src/main/crawler/index.js b/src/main/crawler/index.js
@@ -1,91 +1,201 @@
 import debug from 'debug'
 
-import { getBilibiliLiveUrlsPlugin } from './plugins/bilibili'
-import { getCCLiveUrlsPlugin } from './plugins/cc'
-import { getDouYinLiveUrlsPlugin } from './plugins/douyin'
-import { getDouyuLiveUrlsPlugin } from './plugins/douyu'
-import { getHuyaLiveUrlsPlugin } from './plugins/huya'
-import { getKuaishouLiveUrlsPlugin } from './plugins/kuaishou'
-import { getYoutubeLiveUrlsPlugin } from './plugins/youtube'
-import { getTwitchLiveUrlsPlugin } from './plugins/twitch'
-import { getTiktokLiveUrlsPlugin } from './plugins/tiktok'
-import { getWeiboLiveUrlsPlugin } from './plugins/weibo'
-import { getHuaJiaoLiveUrlsPlugin } from './plugins/huajiao'
-import { getTaobaoLiveUrlsPlugin } from './plugins/taobao'
-import { getBigoLiveUrlsPlugin } from './plugins/bigo'
-import { getYYLiveUrlsPlugin } from './plugins/yy'
-import { getJDLiveUrlsPlugin } from './plugins/jd'
-import { getMomoLiveUrlsPlugin } from './plugins/momo'
-import { getShiGuangLiveUrlsPlugin } from './plugins/shiguang'
-import { getVvLiveUrlsPlugin } from './plugins/vv'
-import { getChangLiaoLiveUrlsPlugin } from './plugins/changliao'
-import { get17LiveUrlsPlugin } from './plugins/17live'
-import { getXhsUrlsPlugin } from './plugins/xhs'
-import { getKilakilaLiveUrlsPlugin } from './plugins/kilakila'
-import { getAcFunLiveUrlsPlugin } from './plugins/acfun'
+import { getBilibiliLiveUrlsPlugin, getBilibiliRoomInfoPlugin } from './plugins/bilibili'
+import { getCCLiveUrlsPlugin, getCCRoomInfoPlugin } from './plugins/cc'
+import { getDouYinLiveUrlsPlugin, getDouYinRoomInfoPlugin } from './plugins/douyin'
+import { getDouyuLiveUrlsPlugin, getDouyuRoomInfoPlugin } from './plugins/douyu'
+import { getHuyaLiveUrlsPlugin, getHuyaRoomInfoPlugin } from './plugins/huya'
+import { getKuaishouLiveUrlsPlugin, getKuaishouRoomInfoPlugin } from './plugins/kuaishou'
+import { getYoutubeLiveUrlsPlugin, getYoutubeRoomInfoPlugin } from './plugins/youtube'
+import { getTwitchLiveUrlsPlugin, getTwitchRoomInfoPlugin } from './plugins/twitch'
+import { getTiktokLiveUrlsPlugin, getTiktokRoomInfoPlugin } from './plugins/tiktok'
+import { getWeiboLiveUrlsPlugin, getWeiboRoomInfoPlugin } from './plugins/weibo'
+import { getHuaJiaoLiveUrlsPlugin, getHuaJiaoRoomInfoPlugin } from './plugins/huajiao'
+import { getTaobaoLiveUrlsPlugin, getTaobaoRoomInfoPlugin } from './plugins/taobao'
+import { getBigoLiveUrlsPlugin, getBigoRoomInfoPlugin } from './plugins/bigo'
+import { getYYLiveUrlsPlugin, getYYRoomInfoPlugin } from './plugins/yy'
+import { getJDLiveUrlsPlugin, getJDRoomInfoPlugin } from './plugins/jd'
+import { getMomoLiveUrlsPlugin, getMomoRoomInfoPlugin } from './plugins/momo'
+import { getShiGuangLiveUrlsPlugin, getShiGuangRoomInfoPlugin } from './plugins/shiguang'
+import { getVvLiveUrlsPlugin, getVvRoomInfoPlugin } from './plugins/vv'
+import { getChangLiaoLiveUrlsPlugin, getChangLiaoRoomInfoPlugin } from './plugins/changliao'
+import { get17LiveUrlsPlugin, get17LiveRoomInfoPlugin } from './plugins/17live'
+import { getXhsUrlsPlugin, getXhsRoomInfoPlugin } from './plugins/xhs'
+import { getKilakilaLiveUrlsPlugin, getKilakilaRoomInfoPlugin } from './plugins/kilakila'
+import { getAcFunLiveUrlsPlugin, getAcFunRoomInfoPlugin } from './plugins/acfun'
 
 import { CRAWLER_ERROR_CODE } from '../../code'
 
 const log = debug('fideo-crawler')
 
 const hostnameToPlatformCrawlerFnMap = {
-  'www.youtube.com': getYoutubeLiveUrlsPlugin,
-  'youtube.com': getYoutubeLiveUrlsPlugin,
+  'www.youtube.com': {
+    getLiveUrlsFn: getYoutubeLiveUrlsPlugin,
+    getRoomInfoFn: getYoutubeRoomInfoPlugin
+  },
+  'youtube.com': {
+    getLiveUrlsFn: getYoutubeLiveUrlsPlugin,
+    getRoomInfoFn: getYoutubeRoomInfoPlugin
+  },
 
-  'www.twitch.tv': getTwitchLiveUrlsPlugin,
-  'twitch.tv': getTwitchLiveUrlsPlugin,
+  'www.twitch.tv': {
+    getLiveUrlsFn: getTwitchLiveUrlsPlugin,
+    getRoomInfoFn: getTwitchRoomInfoPlugin
+  },
+  'twitch.tv': {
+    getLiveUrlsFn: getTwitchLiveUrlsPlugin,
+    getRoomInfoFn: getTwitchRoomInfoPlugin
+  },
 
-  'www.tiktok.com': getTiktokLiveUrlsPlugin,
-  'tiktok.com': getTiktokLiveUrlsPlugin,
+  'www.tiktok.com': {
+    getLiveUrlsFn: getTiktokLiveUrlsPlugin,
+    getRoomInfoFn: getTiktokRoomInfoPlugin
+  },
+  'tiktok.com': {
+    getLiveUrlsFn: getTiktokLiveUrlsPlugin,
+    getRoomInfoFn: getTiktokRoomInfoPlugin
+  },
 
-  'live.douyin.com': getDouYinLiveUrlsPlugin,
-  'v.douyin.com': getDouYinLiveUrlsPlugin,
+  'live.douyin.com': {
+    getLiveUrlsFn: getDouYinLiveUrlsPlugin,
+    getRoomInfoFn: getDouYinRoomInfoPlugin
+  },
+  'v.douyin.com': {
+    getLiveUrlsFn: getDouYinLiveUrlsPlugin,
+    getRoomInfoFn: getDouYinRoomInfoPlugin
+  },
 
-  'live.kuaishou.com': getKuaishouLiveUrlsPlugin,
-  'live.bilibili.com': getBilibiliLiveUrlsPlugin,
+  'live.kuaishou.com': {
+    getLiveUrlsFn: getKuaishouLiveUrlsPlugin,
+    getRoomInfoFn: getKuaishouRoomInfoPlugin
+  },
 
-  'cc.163.com': getCCLiveUrlsPlugin,
+  'live.bilibili.com': {
+    getLiveUrlsFn: getBilibiliLiveUrlsPlugin,
+    getRoomInfoFn: getBilibiliRoomInfoPlugin
+  },
 
-  'www.huajiao.com': getHuaJiaoLiveUrlsPlugin,
-  'huajiao.com': getHuaJiaoLiveUrlsPlugin,
+  'cc.163.com': {
+    getLiveUrlsFn: getCCLiveUrlsPlugin,
+    getRoomInfoFn: getCCRoomInfoPlugin
+  },
 
-  'weibo.com': getWeiboLiveUrlsPlugin,
-  'www.weibo.com': getWeiboLiveUrlsPlugin,
+  'www.huajiao.com': {
+    getLiveUrlsFn: getHuaJiaoLiveUrlsPlugin,
+    getRoomInfoFn: getHuaJiaoRoomInfoPlugin
+  },
 
-  'www.douyu.com': getDouyuLiveUrlsPlugin,
-  'douyu.com': getDouyuLiveUrlsPlugin,
+  'huajiao.com': {
+    getLiveUrlsFn: getHuaJiaoLiveUrlsPlugin,
+    getRoomInfoFn: getHuaJiaoRoomInfoPlugin
+  },
 
-  'tbzb.taobao.com': getTaobaoLiveUrlsPlugin,
+  'weibo.com': {
+    getLiveUrlsFn: getWeiboLiveUrlsPlugin,
+    getRoomInfoFn: getWeiboRoomInfoPlugin
+  },
+  'www.weibo.com': {
+    getLiveUrlsFn: getWeiboLiveUrlsPlugin,
+    getRoomInfoFn: getWeiboRoomInfoPlugin
+  },
 
-  'www.bigo.tv': getBigoLiveUrlsPlugin,
-  'bigo.tv': getBigoLiveUrlsPlugin,
+  'www.douyu.com': {
+    getLiveUrlsFn: getDouyuLiveUrlsPlugin,
+    getRoomInfoFn: getDouyuRoomInfoPlugin
+  },
+  'douyu.com': {
+    getLiveUrlsFn: getDouyuLiveUrlsPlugin,
+    getRoomInfoFn: getDouyuRoomInfoPlugin
+  },
 
-  'www.yy.com': getYYLiveUrlsPlugin,
-  'yy.com': getYYLiveUrlsPlugin,
+  'tbzb.taobao.com': {
+    getLiveUrlsFn: getTaobaoLiveUrlsPlugin,
+    getRoomInfoFn: getTaobaoRoomInfoPlugin
+  },
 
-  'www.huya.com': getHuyaLiveUrlsPlugin,
-  'huya.com': getHuyaLiveUrlsPlugin,
+  'www.bigo.tv': {
+    getLiveUrlsFn: getBigoLiveUrlsPlugin,
+    getRoomInfoFn: getBigoRoomInfoPlugin
+  },
+  'bigo.tv': {
+    getLiveUrlsFn: getBigoLiveUrlsPlugin,
+    getRoomInfoFn: getBigoRoomInfoPlugin
+  },
 
-  'lives.jd.com': getJDLiveUrlsPlugin,
+  'www.yy.com': {
+    getLiveUrlsFn: getYYLiveUrlsPlugin,
+    getRoomInfoFn: getYYRoomInfoPlugin
+  },
+  'yy.com': {
+    getLiveUrlsFn: getYYLiveUrlsPlugin,
+    getRoomInfoFn: getYYRoomInfoPlugin
+  },
 
-  'web.immomo.com': getMomoLiveUrlsPlugin,
+  'www.huya.com': {
+    getLiveUrlsFn: getHuyaLiveUrlsPlugin,
+    getRoomInfoFn: getHuyaRoomInfoPlugin
+  },
+  'huya.com': {
+    getLiveUrlsFn: getHuyaLiveUrlsPlugin,
+    getRoomInfoFn: getHuyaRoomInfoPlugin
+  },
 
-  'www.rengzu.com': getShiGuangLiveUrlsPlugin,
+  'lives.jd.com': {
+    getLiveUrlsFn: getJDLiveUrlsPlugin,
+    getRoomInfoFn: getJDRoomInfoPlugin
+  },
 
-  'h5webcdn-pro.vvxqiu.com': getVvLiveUrlsPlugin,
+  'web.immomo.com': {
+    getLiveUrlsFn: getMomoLiveUrlsPlugin,
+    getRoomInfoFn: getMomoRoomInfoPlugin
+  },
 
-  'www.tlclw.com': getChangLiaoLiveUrlsPlugin,
-  'tlclw.com': getChangLiaoLiveUrlsPlugin,
+  'www.rengzu.com': {
+    getLiveUrlsFn: getShiGuangLiveUrlsPlugin,
+    getRoomInfoFn: getShiGuangRoomInfoPlugin
+  },
 
-  'www.17.live': get17LiveUrlsPlugin,
-  '17.live': get17LiveUrlsPlugin,
+  'h5webcdn-pro.vvxqiu.com': {
+    getLiveUrlsFn: getVvLiveUrlsPlugin,
+    getRoomInfoFn: getVvRoomInfoPlugin
+  },
 
-  'www.xiaohongshu.com': getXhsUrlsPlugin,
-  'xiaohongshu.com': getXhsUrlsPlugin,
+  'www.tlclw.com': {
+    getLiveUrlsFn: getChangLiaoLiveUrlsPlugin,
+    getRoomInfoFn: getChangLiaoRoomInfoPlugin
+  },
+  'tlclw.com': {
+    getLiveUrlsFn: getChangLiaoLiveUrlsPlugin,
+    getRoomInfoFn: getChangLiaoRoomInfoPlugin
+  },
 
-  'live.kilakila.cn': getKilakilaLiveUrlsPlugin,
+  'www.17.live': {
+    getLiveUrlsFn: get17LiveUrlsPlugin,
+    getRoomInfoFn: get17LiveRoomInfoPlugin
+  },
+  '17.live': {
+    getLiveUrlsFn: get17LiveUrlsPlugin,
+    getRoomInfoFn: get17LiveRoomInfoPlugin
+  },
 
-  'live.acfun.cn': getAcFunLiveUrlsPlugin
+  'www.xiaohongshu.com': {
+    getLiveUrlsFn: getXhsUrlsPlugin,
+    getRoomInfoFn: getXhsRoomInfoPlugin
+  },
+  'xiaohongshu.com': {
+    getLiveUrlsFn: getXhsUrlsPlugin,
+    getRoomInfoFn: getXhsRoomInfoPlugin
+  },
+
+  'live.kilakila.cn': {
+    getLiveUrlsFn: getKilakilaLiveUrlsPlugin,
+    getRoomInfoFn: getKilakilaRoomInfoPlugin
+  },
+
+  'live.acfun.cn': {
+    getLiveUrlsFn: getAcFunLiveUrlsPlugin,
+    getRoomInfoFn: getAcFunRoomInfoPlugin
+  }
 }
 
 // eslint-disable-next-line @typescript-eslint/no-unused-vars
@@ -124,7 +234,7 @@ export async function getLiveUrls(info, writeLog) {
     }
   }
 
-  const getLiveUrlsFn = hostnameToPlatformCrawlerFnMap[host]
+  const getLiveUrlsFn = hostnameToPlatformCrawlerFnMap[host]?.getLiveUrlsFn
 
   // TODO: 判断是否是支持的平台但是url不对，提供更好的错误提示
   if (!getLiveUrlsFn) {
@@ -137,3 +247,29 @@ export async function getLiveUrls(info, writeLog) {
   log('res:', res)
   return res
 }
+
+export async function getRoomInfo(info, writeLog) {
+  const { roomUrl, proxy, cookie } = info
+  let host
+  try {
+    host = new URL(roomUrl).host
+  } catch (e) {
+    console.error(e)
+    return {
+      code: CRAWLER_ERROR_CODE.INVALID_URL
+    }
+  }
+
+  const getRoomInfoFn = hostnameToPlatformCrawlerFnMap[host]?.getRoomInfoFn
+
+  // TODO: 判断是否是支持的平台但是url不对，提供更好的错误提示
+  if (!getRoomInfoFn) {
+    return {
+      code: CRAWLER_ERROR_CODE.NOT_SUPPORT
+    }
+  }
+
+  const res = await getRoomInfoFn(roomUrl, { proxy, cookie }, writeLog)
+  log('res:', res)
+  return res
+}
diff --git a/src/main/crawler/plugins/kuaishou.js b/src/main/crawler/plugins/kuaishou.js
@@ -86,4 +86,10 @@ async function baseGetKuaishouLiveUrlsPlugin(roomUrl, others = {}) {
   }
 }
 
+async function baseGetKuaishouRoomInfoPlugin(roomUrl, others = {}) {
+  const roomId = getRoomIdByUrl(roomUrl)
+  const { proxy, cookie } = others
+}
+
 export const getKuaishouLiveUrlsPlugin = captureError(baseGetKuaishouLiveUrlsPlugin)
+export const getKuaishouRoomInfoPlugin = captureError(baseGetKuaishouRoomInfoPlugin)
diff --git a/src/main/crawler/plugins/momo.js b/src/main/crawler/plugins/momo.js
@@ -56,4 +56,10 @@ async function baseGetMomoLiveUrlsPlugin(roomUrl, others = {}) {
   }
 }
 
+async function baseGetMomoRoomInfoPlugin(roomUrl, others = {}) {
+  const roomId = getRoomIdByUrl(roomUrl)
+  const { proxy, cookie } = others
+}
+
 export const getMomoLiveUrlsPlugin = captureError(baseGetMomoLiveUrlsPlugin)
+export const getMomoRoomInfoPlugin = captureError(baseGetMomoRoomInfoPlugin)
diff --git a/src/main/crawler/plugins/shiguang.js b/src/main/crawler/plugins/shiguang.js
@@ -87,4 +87,10 @@ async function baseGetShiGuangLiveUrlsPlugin(roomUrl, others = {}) {
   }
 }
 
+async function baseGetShiGuangRoomInfoPlugin(roomUrl, others = {}) {
+  const roomId = getRoomIdByUrl(roomUrl)
+  const { proxy, cookie } = others
+}
+
 export const getShiGuangLiveUrlsPlugin = captureError(baseGetShiGuangLiveUrlsPlugin)
+export const getShiGuangRoomInfoPlugin = captureError(baseGetShiGuangRoomInfoPlugin)
diff --git a/src/main/crawler/plugins/taobao.js b/src/main/crawler/plugins/taobao.js
@@ -336,4 +336,10 @@ async function baseGetTaobaoLiveUrlsPlugin(roomUrl, others = {}) {
   }
 }
 
+async function baseGetTaobaoRoomInfoPlugin(roomUrl, others = {}) {
+  const roomId = getRoomIdByUrl(roomUrl)
+  const { proxy, cookie } = others
+}
+
 export const getTaobaoLiveUrlsPlugin = captureError(baseGetTaobaoLiveUrlsPlugin)
+export const getTaobaoRoomInfoPlugin = captureError(baseGetTaobaoRoomInfoPlugin)
diff --git a/src/main/crawler/plugins/tiktok.js b/src/main/crawler/plugins/tiktok.js
@@ -87,4 +87,10 @@ async function baseGetTiktokLiveUrlsPlugin(roomUrl, others = {}) {
   }
 }
 
+async function baseGetTiktokRoomInfoPlugin(roomUrl, others = {}) {
+  const roomId = getRoomIdByUrl(roomUrl)
+  const { proxy, cookie } = others
+}
+
 export const getTiktokLiveUrlsPlugin = captureError(baseGetTiktokLiveUrlsPlugin)
+export const getTiktokRoomInfoPlugin = captureError(baseGetTiktokRoomInfoPlugin)
diff --git a/src/main/crawler/plugins/twitch.js b/src/main/crawler/plugins/twitch.js
@@ -106,4 +106,10 @@ async function baseGetTwitchLiveUrlsPlugin(roomUrl, others = {}) {
   }
 }
 
+async function baseGetTwitchRoomInfoPlugin(roomUrl, others = {}) {
+  const roomId = getRoomIdByUrl(roomUrl)
+  const { proxy, cookie } = others
+}
+
 export const getTwitchLiveUrlsPlugin = captureError(baseGetTwitchLiveUrlsPlugin)
+export const getTwitchRoomInfoPlugin = captureError(baseGetTwitchRoomInfoPlugin)