大文件上传会消耗大量的时间,而且中途有可能上传失败。这时我们需要前端和后端配合来解决这个问题。
解决步骤:
- 文件分片,减少每次请求消耗的时间,如果某次请求失败可以单独上传,而不是从头开始
- 通知服务端合并文件分片
- 控制并发的请求数量,避免浏览器内存溢出
- 当因为网络或者其他原因导致某次的请求失败,我们重新发送请求
文件的分片与合并
在JavaScript中,FIle对象是' Blob '对象的子类,该对象包含一个重要的方法slice,通过该方法我们可以这样分割二进制文件:
Document
并发控制
如果文件很大,这样切分的分片会很多,浏览器短时间内就会发起大量的请求,可能会导致内存耗尽,所以要进行并发控制。
这里我们结合Promise.race()方法 控制并发请求的数量,避免浏览器内存溢出。
// 加入并发控制
async function SliceUpload() {
const file = document.getElementById('fileInput').files[0]
if (!file) return
// 文件分片
let size = 1024 * 50; //50KB 50KB Section size
let fileChunks = [];
let index = 0; //Section num
for (let cur = 0; cur < file.size; cur += size) {
fileChunks.push({
hash: index++,
chunk: file.slice(cur, cur + size),
});
}
let pool = []; //Concurrent pool
let max = 3; //Maximum concurrency
for (let i = 0; i < fileChunks.length; i++) {
let item = fileChunks[i];
let formData = new FormData();
formData.append("filename", file.name);
formData.append("hash", item.hash);
formData.append("chunk", item.chunk);
// 上传分片
let task = axios({
method: "post",
url: "/upload",
data: formData,
});
task.then(() => {
// 从并发池中移除已经完成的请求
let index = pool.findIndex((t) => t === task);
pool.splice(index);
});
// 把请求放入并发池中,如果已经达到最大并发量
pool.push(task);
if (pool.length === max) {
//All requests are requested complete
await Promise.race(pool);
}
}
// 所有分片上传完成,通知服务器合并分片
await axios({
method: "get",
url: "/merge",
params: {
filename: file.name,
},
});
console.log("Upload to complete");
}
使代码可复用
function SliceUpload() {
const file = document.getElementById('fileInput').files[0]
if (!file) return
// 文件分片
let size = 1024 * 50; // 分片大小设置
let fileChunks = [];
let index = 0; // 分片序号
for (let cur = 0; cur < file.size; cur += size) {
fileChunks.push({
hash: index++,
chunk: file.slice(cur, cur + size),
});
}
const uploadFileChunks = async function(list){
if(list.length === 0){
// 所有分片上传完成,通知如无
await axios({
method: 'get',
url: '/merge',
params: {
filename: file.name
}
});
console.log('Upload to complete')
return
}
let pool = [] // 并发池
let max = 3 // 最大并发数
let finish = 0 // 完成数量
let failList = [] // 失败列表
for(let i=0;i{
// 从并发池中移除已经完成的请求
let index = pool.findIndex(t=> t===task)
pool.splice(index)
}).catch(()=>{
failList.push(item)
}).finally(()=>{
finish++
// 如果有失败的重新上传
if(finish===list.length){
uploadFileChunks(failList)
}
})
pool.push(task)
if(pool.length === max){
await Promise.race(pool)
}
}
}
uploadFileChunks(fileChunks)
}
服务端接口实现
const express = require('express')
const multiparty = require('multiparty')
const fs = require('fs')
const path = require('path')
const { Buffer } = require('buffer')
// file path
const STATIC_FILES = path.join(__dirname, './static/files')
// Temporary path to upload files
const STATIC_TEMPORARY = path.join(__dirname, './static/temporary')
const server = express()
// Static file hosting
server.use(express.static(path.join(__dirname, './dist')))
// Interface for uploading slices
server.post('/upload', (req, res) => {
const form = new multiparty.Form();
form.parse(req, function(err, fields, files) {
let filename = fields.filename[0]
let hash = fields.hash[0]
let chunk = files.chunk[0]
let dir = `${STATIC_TEMPORARY}/${filename}`
// console.log(filename, hash, chunk)
try {
if (!fs.existsSync(dir)) fs.mkdirSync(dir)
const buffer = fs.readFileSync(chunk.path)
const ws = fs.createWriteStream(`${dir}/${hash}`)
ws.write(buffer)
ws.close()
res.send(`${filename}-${hash} Section uploaded successfully`)
} catch (error) {
console.error(error)
res.status(500).send(`${filename}-${hash} Section uploading failed`)
}
})
})
//Merged slice interface
server.get('/merge', async (req, res) => {
const { filename } = req.query
try {
let len = 0
const bufferList = fs.readdirSync(`${STATIC_TEMPORARY}/${filename}`).map((hash,index) => {
const buffer = fs.readFileSync(`${STATIC_TEMPORARY}/${filename}/${index}`)
len += buffer.length
return buffer
});
//Merge files
const buffer = Buffer.concat(bufferList, len);
const ws = fs.createWriteStream(`${STATIC_FILES}/${filename}`)
ws.write(buffer);
ws.close();
res.send(`Section merge completed`);
} catch (error) {
console.error(error);
}
})
server.listen(3000, _ => {
console.log('http://localhost:3000/')
})