first commit

历届学生/fontendsix/每日课程/2021-11-17/src/config.js

@@ -0,0 +1,3 @@
+module.exports = {
+    url: 'http://www.netbian.com/'
+}

历届学生/fontendsix/每日课程/2021-11-17/src/http.js

@@ -0,0 +1,33 @@
+const axios = require('axios');
+const config = require("./config");
+const iconv = require('iconv-lite');
+
+class http {
+    constructor() {
+        this.http = axios.create({
+            baseURL: config.url,
+            timeout: 100000,
+            headers: { 
+                "Cookie": "__yjs_duid=1_54d7d9761ff38fd500742b8c976378851637137358248; xygkqecookieclassrecord=%2C1%2C; Hm_lvt_14b14198b6e26157b7eba06b390ab763=1637137357,1637138134,1637138153,1637138272; xygkqecookieinforecord=%2C7-24028%2C7-24022%2C1-1%2C19-23857%2C11-23937%2C12-24000%2C; Hm_lpvt_14b14198b6e26157b7eba06b390ab763=1637144818",
+                "Host": "www.netbian.com",
+                "Referer": "http://www.netbian.com/index.htm",
+                "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/95.0.4638.69 Safari/537.36"
+            }
+        });
+    }
+
+    async get(params) {
+        var res = (await this.http.get(params.url, {
+            params: params.data,
+            responseType: params.type ? "arraybuffer" : "stream"
+        })).data;
+        if (params.type == "1") {
+            return iconv.decode(res,'gbk');
+        } else {
+            return res;
+        }
+
+    }
+}
+
+module.exports = new http();

历届学生/fontendsix/每日课程/2021-11-17/src/pc.js

@@ -0,0 +1,87 @@
+var http = require("./http");
+const cheerio = require('cheerio');
+const fs = require('fs');
+const path = require('path');
+
+var pagePicNums = 20;
+
+var urlArr = [] // 存放每一页的网址
+var urlIndex = 0;
+
+var pagePicArr = [] // 存放当前页有多少列表需要爬取
+var picIndex = 0;
+
+for (var i = 1; i <= 10; i++) {
+    if (i == 1) {
+        urlArr.push("")
+    } else {
+        urlArr.push(`/index_${i}.htm`)
+    }
+}
+
+GetPageHtml();
+
+async function GetPageHtml() {
+    console.log("GetPageHtml: ",urlArr[urlIndex]);
+    var res = await http.get({
+        url: urlArr[urlIndex],
+        data: {},
+        type: "1"
+    });
+    const $ = cheerio.load(res);
+    $(".list ul li").each(function() {
+        if ($(this).attr("class") != "nextpage") {
+            if (! ($(this).find("a").attr("href").includes("pic")) ) {
+                pagePicArr.push({
+                    title: $(this).find("a img").attr("alt"),
+                    url: $(this).find("a").attr("href")
+                })
+            }
+        }
+    });
+
+    console.log(pagePicArr);
+    GetPageInfo();
+}
+
+
+async function GetPageInfo() {
+    var res = await http.get({
+        url: pagePicArr[picIndex].url,
+        data: {},
+        type: "1"
+    });
+    const $ = cheerio.load(res);
+
+    createdImg($(".endpage .pic img").attr("src"))
+
+
+    if (picIndex <= pagePicArr.length - 1) {
+        
+        picIndex += 1;
+
+        // 这一页爬结束了
+        if (picIndex > pagePicArr.length - 1) {
+            console.log(`当前第 ${urlIndex} 页结束，开始下一页 ${urlIndex + 1}，下一页网址是：${urlArr[urlIndex]}`);
+            pagePicArr = [];
+            picIndex = 0;
+            urlIndex+=1;
+            GetPageHtml();
+        } else {
+            
+            GetPageInfo();
+        }
+    }
+}
+
+async function createdImg(url) {
+    console.log(url);
+    var res = await http.get({
+        url: url,
+        data: {},
+        type: "2"
+    });
+
+    fs.writeFileSync(path.join(__dirname, `./img/${pagePicArr[picIndex]?.title}.jpg`), res);
+   
+}