save

package.json

@@ -4,6 +4,7 @@
   "main": "index.js",
   "scripts": {
     "dev": "npx ts-node src/index.ts",
+    "migrate": "npx ts-node src/mannual/migrate.ts",
     "test": "echo \"Error: no test specified\" && exit 1"
   },
   "keywords": [],

src/index.ts

@@ -7,13 +7,25 @@ export type FetchType = {
 const fetchList = [
   {
     name: 'AI写作工具',
-    typeId: '67908fc33de33b392c0330af',
+    typeId: '6790aae23de33b392c0330b2',
     url: 'https://ai-bot.cn/favorites/ai-writing-tools/'
   },
   {
-    name: '',
+    name: 'AI图像网站',
-    typeId: ''
+    typeId: '67908fc33de33b392c0330af',
-  }
+    url: 'https://ai-bot.cn/favorites/best-ai-image-tools/'
+  },
+  {
+    name: 'AI视频网站',
+    typeId: '67b6f0b7b139d1d6aa14cd06',
+    url: 'https://ai-bot.cn/favorites/ai-video-tools/'
+  },
+  {
+    name: 'AI音频网站',
+    typeId: '6791a98fc058e55ed0a094ca',
+    url: 'https://ai-bot.cn/favorites/ai-audio-tools/'
+  },
+
 ]
 function main() {
   console.log("Hello, this is the main function!");

src/lib/mongodb.ts

@@ -23,8 +23,8 @@ if (process.env.NODE_ENV === 'development') {
   clientPromise = client.connect();
 }
 
-export const getDb = async () => {
+export const getDb = async (dbName?: string) => {
-  return (await clientPromise).db('crawler');
+  return (await clientPromise).db(dbName || 'crawler');
 };
 export const getCollection = async (collection: string) => {
   const ins = await getDb();

src/link/index.ts

@@ -4,7 +4,6 @@ import { downloadImage } from "../share/tools"
 import { getCollection } from '../lib/mongodb';
 import { FetchType } from '..';
 // 要抓取的网页 URL
-const url = 'https://ai-bot.cn/favorites/ai-writing-tools/';
 
 async function getPageData(url: string, name: string) {
   try {
@@ -20,10 +19,13 @@ async function getPageData(url: string, name: string) {
   }
 }
 export async function queryListData(list: FetchType[]) {
-  const promiseList = list.map(item => fetchData(item.typeId))
+  const col = await getCollection('link');
-  await Promise.all(promiseList)
+  await col.deleteMany({})
+  for (const item of list) {
+    await fetchData(item.typeId, item.url)
   }
-export async function fetchData(typeName: string) {
+}
+export async function fetchData(typeName: string, url: string) {
   try {
     // 请求目标页面
     const { data } = await axios.get(url);
@@ -69,7 +71,10 @@ export async function fetchData(typeName: string) {
 
       toolsData.push(toolData);
       i++
-      console.log(`进度：${i}/${length}`);
+      console.clear()
+      console.log(`正在爬取${typeName}类别的数据，共${length}条数据`);
+      
+      console.log(`${typeName}：进度：${i}/${length}`);
 
     }
     console.log(toolsData);

src/mannual/migrate.ts

@@ -0,0 +1,12 @@
+import { getCollection, getDb } from "../lib/mongodb";
+
+async function migrateLink() {
+    const botDb = await getDb('ai-bot');
+    const botCol = botDb.collection('link');
+    const col = await getCollection('link')
+    const links = await col.find().toArray();
+    await botCol.deleteMany({});
+    await botCol.insertMany(links);
+    console.log('Migrate link success');
+}
+migrateLink();

src/share/tools.ts

@@ -38,7 +38,7 @@ export async function downloadImage(url: string) {
 
         // 写入文件
         await fs.writeFile(filePath, response.data);
-        return id;
+        return filename;
     } catch (error) {
         console.error('下载失败:', error);
         throw error;