Fix fetchFile API file save is lost

Author: coder-hxl

package.json

@@ -1,7 +1,7 @@
 {
   "private": true,
   "name": "x-crawl",
-  "version": "0.1.4",
+  "version": "0.1.5",
   "author": "CoderHxl",
   "description": "XCrawl is a Nodejs multifunctional crawler library. Crawl HTML, JSON, file resources, etc. through simple configuration.",
   "license": "MIT",

publish/package.json

@@ -1,6 +1,6 @@
 {
   "name": "x-crawl",
-  "version": "0.1.4",
+  "version": "0.1.5",
   "author": "CoderHxl",
   "description": "XCrawl is a Nodejs multifunctional crawler library. Crawl HTML, JSON, file resources, etc. through simple configuration.",
   "license": "MIT",

src/index.ts

@@ -3,7 +3,15 @@ import path from 'node:path'
 import { JSDOM } from 'jsdom'
 
 import { batchRequest, syncBatchRequest, request } from './request'
-import { isArray, isString, isUndefined, log, logError } from './utils'
+import {
+  isArray,
+  isString,
+  isUndefined,
+  log,
+  logError,
+  logNumber,
+  logSuccess
+} from './utils'
 
 import {
   IXCrawlBaseConifg,
@@ -135,39 +143,40 @@ export default class XCrawl {
       intervalTime
     )
 
-    return new Promise((resolve) => {
-      const container: IFetchCommon<IFileInfo> = []
-
-      requestRes.forEach((requestResItem, index) => {
-        const { id, statusCode, headers, data } = requestResItem
-
-        const mimeType = headers['content-type'] ?? ''
-        const suffix = mimeType.split('/').pop()
-        const fileName = new Date().getTime().toString()
-        const filePath = path.resolve(
-          fileConfig.storeDir,
-          `${fileName}.${suffix}`
-        )
-
-        fs.createWriteStream(filePath, 'binary').write(data, (err) => {
-          if (err) {
-            log(logError(`File save error at id ${id}: ${err.message}`))
-          } else {
-            const fileInfo: IFileInfo = {
-              fileName,
-              mimeType,
-              size: data.length,
-              filePath
-            }
-
-            container.push({ id, statusCode, headers, data: fileInfo })
-          }
-
-          if (index === requestRes.length - 1) {
-            resolve(container)
-          }
+    const container: IFetchCommon<IFileInfo> = []
+
+    requestRes.forEach((requestResItem) => {
+      const { id, headers, data } = requestResItem
+
+      const mimeType = headers['content-type'] ?? ''
+      const suffix = mimeType.split('/').pop()
+      const fileName = new Date().getTime().toString()
+      const filePath = path.resolve(
+        fileConfig.storeDir,
+        `${fileName}.${suffix}`
+      )
+
+      try {
+        fs.writeFileSync(filePath, data)
+
+        container.push({
+          ...requestResItem,
+          data: { fileName, mimeType, size: data.length, filePath }
         })
-      })
+      } catch (error: any) {
+        log(logError(`File save error at id ${id}: ${error.message}`))
+      }
     })
+
+    const saveTotal = requestRes.length
+    const success = container.length
+    const error = requestRes.length - container.length
+    log(
+      `saveTotal: ${logNumber(saveTotal)}, success: ${logSuccess(
+        success
+      )}, error: ${logError(error)}`
+    )
+
+    return container
   }
 }

src/request.ts

@@ -208,7 +208,7 @@ export async function batchRequest(
   error.forEach((message) => log(logError(message)))
 
   log(
-    `total: ${logNumber(requestConifgs.length)}, success: ${logSuccess(
+    `requestsTotal: ${logNumber(requestConifgs.length)}, success: ${logSuccess(
       success.length
     )}, error: ${logError(error.length)}`
   )
@@ -255,7 +255,7 @@ export async function syncBatchRequest(
   log(logSuccess('All requests are over!'))
 
   log(
-    `total: ${logNumber(requestConifgs.length)}, success: ${logSuccess(
+    `requestsTotal: ${logNumber(requestConifgs.length)}, success: ${logSuccess(
       successTotal
     )}, error: ${logError(errorTotal)}`
   )

test/start/index.js

@@ -3,54 +3,77 @@ import XCrawl from '../../src'
 
 const testXCrawl = new XCrawl({
   timeout: 10000,
-  intervalTime: {
-    max: 3000,
-    min: 2000
-  },
+  intervalTime: { max: 2000, min: 1000 },
   mode: 'async'
 })
 
-testXCrawl
-  .fetchData({
-    requestConifg: [
-      { url: 'http://localhost:3001/home' },
-      { url: 'http://localhost:9001/api/home/wonderfulplace' },
-      { url: 'http://localhost:9001/api/home/goodprice' },
-      { url: 'http://localhost:3001/home' },
-      { url: 'http://localhost:9001/ai/home/goodprice' }
-    ]
-  })
-  .then((res) => {
-    // console.log(res)
+// testXCrawl
+//   .fetchData({
+//     requestConifg: [
+//       { url: 'http://localhost:3001/home' },
+//       { url: 'http://localhost:9001/api/home/wonderfulplace' },
+//       { url: 'http://localhost:9001/api/home/goodprice' },
+//       { url: 'http://localhost:3001/home' },
+//       { url: 'http://localhost:9001/ai/home/goodprice' }
+//     ]
+//   })
+//   .then((res) => {
+//     // console.log(res)
+//   })
+
+testXCrawl.fetchHTML({ url: 'https://www.bilibili.com/' }).then((res) => {
+  const { jsdom } = res.data
+
+  const document = jsdom.window.document
+  const imgBoxEl = document.querySelectorAll('.bili-video-card__cover')
+
+  const imgUrls: string[] = []
+  imgBoxEl.forEach((item, index) => {
+    const img = item.lastChild as HTMLImageElement
+
+    if (index % 2) {
+      imgUrls.push('https:' + img.src)
+    } else {
+      imgUrls.push(img.src)
+    }
   })
 
-// testXCrawl.fetchHTML({ url: 'https://www.bilibili.com/' }).then((res) => {
-//   const { jsdom } = res.data
+  console.log(imgUrls)
 
-//   const document = jsdom.window.document
-//   const imgBoxEl = document.querySelectorAll('.bili-video-card__cover')
+  const requestConifg = imgUrls.map((url) => ({ url }))
 
-//   const imgUrls: string[] = []
-//   imgBoxEl.forEach((item, index) => {
-//     const img = item.lastChild as HTMLImageElement
+  testXCrawl
+    .fetchFile({
+      requestConifg,
+      fileConfig: { storeDir: path.resolve(__dirname, './upload') }
+    })
+    .then((res) => {
+      // console.log(res)
+    })
+})
 
-//     if (index % 2) {
-//       imgUrls.push('https:' + img.src)
-//     } else {
-//       imgUrls.push(img.src)
+// testXCrawl
+//   .fetchData({
+//     requestConifg: {
+//       url: 'http://localhost:9001/api/area/阳江市',
+//       method: 'POST',
+//       data: {
+//         type: 'plus',
+//         offset: 0,
+//         size: 20
+//       }
 //     }
 //   })
+//   .then((res) => {
+//     const room = res[0].data.data.list[0]
+//     const requestConifg = room.pictureUrls.map((item: any) => ({
+//       url: item
+//     }))
 
-//   console.log(imgUrls)
-
-//   const requestConifg = imgUrls.map((url) => ({ url }))
-
-//   testXCrawl
-//     .fetchFile({
+//     testXCrawl.fetchFile({
 //       requestConifg,
-//       fileConfig: { storeDir: path.resolve(__dirname, './upload') }
-//     })
-//     .then((res) => {
-//       console.log(res)
+//       fileConfig: {
+//         storeDir: path.resolve(__dirname, './upload')
+//       }
 //     })
-// })
+//   })