dandanzan的电影直接更新到movie表

package.json

@@ -21,6 +21,7 @@
     "cheerio": "^1.0.0-rc.3",
     "dateformat": "^3.0.3",
     "file-stream-rotator": "^0.4.1",
+    "free-proxy": "^0.1.5",
     "fs": "0.0.1-security",
     "fs-extra": "^7.0.1",
     "glob": "^7.1.3",
@@ -28,8 +29,11 @@
     "mkdirp": "^0.5.1",
     "mongoose": "^5.2.15",
     "multi-progress": "^2.0.0",
+    "proxy-lists": "^1.16.0",
+    "random-useragent": "^0.3.1",
     "request": "^2.88.0",
     "request-promise": "^4.2.4",
+    "socks5-http-client": "^1.0.4",
     "ws": "^6.1.2"
   },
   "devDependencies": {

src/models/Movies.js

@@ -3,7 +3,7 @@ import mongoose from 'mongoose';
 const Schema = mongoose.Schema;
 
 const Movies = new Schema({
-  gid: {type: Number, required: true},
+  gid: {type: Number, required: true, index: true},
   // 电影名
   name: {type: String},
   nameAlias: {type: String},
@@ -35,9 +35,15 @@ const Movies = new Schema({
   // 显示的播放数量
   showCount: {type: Number, default: 1000},
   // 排序字段, 倒序
-  sortIdx: {type: Number},
+  sortIdx: {type: Number, default: 0},
+  // jc排序字段， 倒序，查询电影时候优先此字段倒序
+  sortJc: {type: Number, default: 0},
+  // 是否公开
+  open: {type: Boolean, default: false},
   // 购买价格
-  price: {type: Number},
+  price: {type: Number, default: 10},
+  // 视频分类，movie: 电影；tv: 电视剧； show: 综艺节目
+  category: {type: String},
   // 是否已删除
   deleted: {type: Boolean, default: false},
   delete_time: {type: Date},
@@ -53,4 +59,9 @@ MovieModel.findByGid = function(gid) {
   return MovieModel.findOne({gid: gid, deleted: false});
 };
 
+MovieModel.updateOne = async function (gid, record) {
+  const query = {gid: gid};
+  const options = {upsert: true, setDefaultsOnInsert:true};
+  await MovieModel.update(query, record, options);
+}
 export default MovieModel;

src/sites/dandanzan.js

@@ -1,57 +1,67 @@
 import netUtil from "../utils/net.util";
 import cheerio from "cheerio";
 import stringUtil from '../utils/string.util';
-import SpiderData from "../models/SpiderData";
 import Movie from '../models/Movies';
 import generalQueue from '../utils/general.queue';
 
-const parseOnePage = async (subLink) => {
-  // const url = `https://www.dandanzan.com/dianying/${idx}.html`
-  const url = `https://www.dandanzan.com${subLink}`
+const URL_BASE = 'https://www.dandanzan.com'
+/**
+ * 处理一个页面
+ * @param {String} subLink
+ * @param {String} category, 种类， movie: 电影， tv: 电视剧， show: 综艺
+ * */
+const parseOnePage = async (subLink, category) => {
+  const url = `${URL_BASE}${subLink}`
   try {
     let html = await netUtil.getData(url, {})
     const $ = cheerio.load(html);
     if ($('.error404').text()) {
       console.log(`>>>>>>>>>>>> ${url} not found`);
     } else {
-      let link = stringUtil.getContentByReg(html,/links='(.+?)\|'/);
-      let idx = parseInt(subLink.replace('/dianying/', '').replace('.html', ''));
-      let obj = {
-        id: idx,
-        daoYan: $('meta[property="og:video:director"]').attr('content'),
-        zhuYan: $('meta[property="og:video:actor"]').attr('content'),
-        grade: $('meta[property="og:video:score"]').attr('content'),
+      let resourceStr = stringUtil.getContentByReg(html,/links='(.+?)\|'/);
+      let idx = parseInt(subLink.replace('/dianying/', '').replace('.html', '')
+        .replace('/dongman/', '')
+        .replace('/dianshiju/', '')
+        .replace('/zongyi/', ''));
+      const type = $('meta[property="og:video:class"]').attr('content');
+      let typeArr = type ? type.split(',') : [];
+      let arr = resourceStr.replace(/\|/g, '#').split('#');
+      let resourceArr = [];
+      for (let str of arr) {
+        resourceArr.push({
+          title: str.substring(0, str.indexOf('$')),
+          link: str.substring(str.indexOf('$') + 1)
+        })
+      }
+      let daoYan = $('meta[property="og:video:director"]').attr('content');
+      let zhuYan = $('meta[property="og:video:actor"]').attr('content').replace(/ \/ /g, ',');
+      let region = $('meta[property="og:video:area"]').attr('content');
+      let record = {
+        gid: idx,
+        name: $('meta[property="og:title"]').attr('content'),
+        type: typeArr,
+        resources: resourceArr,
+        daoYan: daoYan ? daoYan.split(',') : [],
+        zhuYan: zhuYan ? zhuYan.split(',') : [],
+        score: Number($('meta[property="og:video:score"]').attr('content')),
         img: $('meta[property="og:image"]').attr('content'),
         introduce: $('meta[property="og:description"]').attr('content'),
-        language: '',
-        name: $('meta[property="og:title"]').attr('content'),
-        name_alias: $('meta[property="og:video:alias"]').attr('content'),
-        playResourceUrl: link,
-        region: $('meta[property="og:video:area"]').attr('content'),
-        releaseDate: $('meta[property="og:video:release_date"]').attr('content'),
-        type: $('meta[property="og:video:class"]').attr('content')
+        nameAlias: $('meta[property="og:video:alias"]').attr('content'),
+        region: region ? region.split(',') : [],
+        year: Number($('meta[property="og:video:release_date"]').attr('content')),
+        category: category,
       }
-      let sdata = new SpiderData({
-        type: 'movie',
-        data: obj,
-        status: 0
-      });
-      await sdata.save();
-      console.log(`@@@@@ ${subLink} @ ${obj.name} saved`);
+      await Movie.updateOne(idx, record);
+      console.log(`@@@@@ ${subLink} @ ${record.name} saved`);
     }
   } catch (err) {
     console.log(err);
   }
 }
 
-const parseListPage = async (idx) => {
-  let url;
-  if (idx === 1) {
-    url = `https://www.dandanzan.com/dianying/index.html`;
-  } else {
-    url = `https://www.dandanzan.com/dianying/index_${idx}.html`;
-  }
-  console.log(`begin parse page: ${idx}`);
+const parseListPage = async (subPage, category) => {
+  const url = `${URL_BASE}${subPage}`
+  console.log(`begin parse category: ${category} page: ${subPage}`);
   let html;
   try {
     html = await netUtil.getData(url, {})
@@ -69,71 +79,79 @@ const parseListPage = async (idx) => {
       try {
         generalQueue.addQueue({
           run: async function () {
-            await parseOnePage(page);
+            await parseOnePage(page, category);
           }
         })
       } catch (err) {
         console.log(err);
       }
     }
-  }
-  console.log(`end parse page: ${idx}`);
-}
-const parseAllMovie = async () => {
-  const maxPage = 939;
-  // const maxPage = 10;
-  for (let i = 1; i < maxPage; i++) {
-    try {
-      await parseListPage(i);
-    } catch (err) {
-      console.log(err);
-    }
-  }
-}
-// 将下载的电影导入到正式表中
-const parseAllMovieToDb = async () => {
-  try {
-    let list = await SpiderData.find({type: 'movie'});
-    for (let obj of list) {
-      let data = obj.data;
-      let typeArr = data.type.split(',');
-      let resourceStr = data.playResourceUrl;
-      let arr = resourceStr.replace(/\|/g, '#').split('#');
-      let resourceArr = [];
-      for (let str of arr) {
-        resourceArr.push({
-          title: str.substring(0, str.indexOf('$')),
-          link: str.substring(str.indexOf('$') + 1)
-        })
+    if ($('.next-page')) {
+      let nextStr = $('.next-page a').attr('href');
+      console.log('has next page: ', nextStr);
+      try {
+        await parseListPage(nextStr, category);
+      } catch (err) {
+        console.log(err);
       }
-      let record = new Movie({
-        gid: data.id,
-        name: data.name,
-        nameAlias: data.name_alias,
-        type: typeArr,
-        score: data.grade,
-        introduce: data.introduce,
-        language: data.language,
-        img: data.img,
-        daoYan: data.daoYan.split(','),
-        zhuYan: data.zhuYan.split(','),
-        region: data.region.split(','),
-        year: data.releaseDate,
-        sortIdx: 0,
-        price: 10,
-        resources: resourceArr,
-        published: true,
-      });
-      await record.save();
+    } else {
+      console.log('########################### ALL LIST PAGE END ###########################');
     }
-  } catch (err) {
-    console.log(err);
   }
+
+  console.log(`end parse category: ${category} page: ${subPage}`);
 }
+const parseAllMovie = async (category) => {
+  console.time('all');
+  let subName = 'dianying';
+  switch (category) {
+    case 'movie':
+      subName = 'dianying';
+      break;
+    case 'tv':
+      subName = 'dianshiju'
+      break;
+    case 'show':
+      subName = 'zongyi';
+      break;
+    case 'cartoon':
+      subName = 'dongman';
+      break;
+  }
+  const subPage = `/${subName}/index.html`
+  await parseListPage(subPage, category);
+}
+
 export default {
   run: async () => {
-    // await parseAllMovie();
-    await parseAllMovieToDb();
-    console.log('all done');
+    // await parseAllMovie('movie');
+    // await parseAllMovie('tv');
+    // await parseAllMovie('show');
+    // await parseAllMovie('cartoon');
+    // console.log('all done');
+    let html = await netUtil.getData('https://wechat-test.kingsome.cn/', {})
+    console.log(html);
+    // var ProxyLists = require('proxy-lists');
+    //
+    // var options = {
+    //   countries: ['cn'],
+    //   protocols: ['https'],
+    // };
+    //
+    // var gettingProxies = ProxyLists.getProxies(options);
+    // gettingProxies.on('data', function(proxies) {
+    //   // Received some proxies.
+    //   console.log(proxies);
+    // });
+    //
+    // gettingProxies.on('error', function(error) {
+    //   // Some error has occurred.
+    //   // console.error(error);
+    // });
+    //
+    // gettingProxies.once('end', function() {
+    //   // Done getting proxies.
+    //   console.log('finish get proxy');
+    // });
   }
 }

src/utils/general.queue.js

@@ -13,6 +13,7 @@ let q = async.queue( async (reqObj, cb) => {
 }, 10);
 q.drain = function(){
   console.info('all queue done');
+  console.timeEnd('all');
 };
 module.exports = {
   addQueue(obj) {

src/utils/net.util.js

@@ -1,5 +1,8 @@
 import request from 'request';
 import Promise from 'bluebird';
+import proxy from './proxys';
+import random_useragent from 'random-useragent';
+let agent = require('socks5-http-client/lib/Agent')
 
 
 const iconv = require('iconv-lite');
@@ -26,6 +29,7 @@ export default {
       'Cache-Control': 'no-cache',
       'Content-Type': 'application/x-www-form-urlencoded;charset=UTF-8',
     }
+    header['User-Agent'] = random_useragent.getRandom();
     const options = {
       method: 'POST',
       url: url,
@@ -39,10 +43,17 @@ export default {
     header = header || {
       'Cache-Control': 'no-cache',
     }
+    header['User-Agent'] = random_useragent.getRandom();
     const options = {
       method: 'GET',
       url: url,
       headers: header,
+      proxy: proxy.randomProxy(),
+      // agentClass: agent,
+      // agentOptions: {
+      //   socksHost: '101.71.41.169',
+      //   socksPort: 43,
+      // }
     };
     if (encoding) {
       options.encoding = null;

src/utils/proxys.js

@@ -0,0 +1,12 @@
+import stringUtil from './string.util';
+const proxys = [
+    'http://101.71.41.169:443',
+    'http://116.196.81.58:3128',
+    'http://113.200.56.13:8010'
+    ];
+export default {
+
+  randomProxy() {
+    return proxys[stringUtil.randomNum(0, proxys.length - 1)];
+  }
+}

src/utils/string.util.js

@@ -17,7 +17,9 @@ export default {
       return content.replace(/<.+?>/g, '').replace(/\s/g, '');
     }
   },
-
+  randomNum(minNum, maxNum) {
+    return parseInt(Math.random()*(maxNum-minNum+1)+minNum, 10);
+  },
   getContentByReg(str, reg) {
     const contents = str.match(reg);
     if (contents && contents.length > 1) {