去年的面試中我被問(wèn)到超大文件你怎么處理，這個(gè)問(wèn)題確實(shí)當(dāng)時(shí)沒(méi)多想，回來(lái)之后仔細(xì)研究和討論了下這個(gè)問(wèn)題，對(duì)大文件讀取做了一個(gè)分析
上面的代碼讀取幾兆的文件可以，但是如果大于你本身及其內(nèi)存，那就直接翻車了。因?yàn)樯厦娴拇a，是把文件所有的內(nèi)容全部都讀取到內(nèi)存之后返回，幾兆的文件，你內(nèi)存夠大可以處理，但是一旦上幾百兆的文件，就沒(méi)那么好處理了。
package main
import (
"bufio"
"fmt"
"io"
//"math"
"os"
"strings"
"sync"
"time"
)
func main() {
/*
文件數(shù)據(jù)樣例
{"remark": "來(lái)電時(shí)間： 2021/04/15 13:52:07客戶電話：13913xx39xx ", "no": "600020510132021101310210547639", "title": "b-ae0e-0242ac100907", "call_in_date": "2021-04-15 13:52:12", "name": "張三", "_date": "2021-06-15", "name": "張三", "meet": "1"}
1、我們?nèi)〕?call_in_date": "2021-04-15 13:52:1的數(shù)據(jù)寫入另一個(gè)文件
*/
var (
s time.Time //當(dāng)前時(shí)間
file *os.File
fileStat os.FileInfo
err error
lastLineSize int64
)
s = time.Now()
if file, err = os.Open("/Users/zhangsan/Downloads/log.txt");err != nil{
fmt.Println(err)
}
defer func() {
err = file.Close() //close after checking err
}()
//queryStartTime, err := time.Parse("2006-01-02T15:04:05.0000Z", startTimeArg)
//if err != nil {
// fmt.Println("Could not able to parse the start time", startTimeArg)
// return
//}
//
//queryFinishTime, err := time.Parse("2006-01-02T15:04:05.0000Z", finishTimeArg)
//if err != nil {
// fmt.Println("Could not able to parse the finish time", finishTimeArg)
// return
//}
/**
* {name:"log.log", size:911100961, mode:0x1a4,
modTime:time.Time{wall:0x656c25c, ext:63742660691,
loc:(*time.Location)(0x1192c80)}, sys:syscall.Stat_t{Dev:16777220,
Mode:0x81a4, Nlink:0x1, Ino:0x118cba7, Uid:0x1f5, Gid:0x14, Rdev:0,
Pad_cgo_0:[4]uint8{0x0, 0x0, 0x0, 0x0}, Atimespec:syscall.Timespec{Sec:1607063899, Nsec:977970393},
Mtimespec:syscall.Timespec{Sec:1607063891, Nsec:106349148}, Ctimespec:syscall.Timespec{Sec:1607063891,
Nsec:258847043}, Birthtimespec:syscall.Timespec{Sec:1607063883, Nsec:425808150},
Size:911100961, Blocks:1784104, Blksize:4096, Flags:0x0, Gen:0x0, Lspare:0, Qspare:[2]int64{0, 0}}
*
*/
if fileStat, err = file.Stat();err != nil {
return
}
fileSize := fileStat.Size()//72849354767
offset := fileSize - 1
//檢測(cè)是不是都是空行 只有\(zhòng)n
for {
var (
b []byte
n int
char string
)
b = make([]byte, 1)
//從指定位置讀取
if n, err = file.ReadAt(b, offset);err != nil {
fmt.Println("Error reading file ", err)
break
}
char = string(b[0])
if char == "\n" {
break
}
offset--
//獲取一行的大小
lastLineSize += int64(n)
}
var (
lastLine []byte
logSlice []string
logSlice1 []string
)
//初始化一行大小的空間
lastLine = make([]byte, lastLineSize)
_, err = file.ReadAt(lastLine, offset)
if err != nil {
fmt.Println("Could not able to read last line with offset", offset, "and lastline size", lastLineSize)
return
}
//根據(jù)條件進(jìn)行區(qū)分
logSlice = strings.Split(strings.Trim(string(lastLine),"\n"),"next_pay_date")
logSlice1 = strings.Split(logSlice[1],"\"")
if logSlice1[2] == "2021-06-15"{
Process(file)
}
fmt.Println("\nTime taken - ", time.Since(s))
fmt.Println(err)
}
func Process(f *os.File) error {
//讀取數(shù)據(jù)的key，減小gc壓力
linesPool := sync.Pool{New: func() interface{} {
lines := make([]byte, 250*1024)
return lines
}}
//讀取回來(lái)的數(shù)據(jù)池
stringPool := sync.Pool{New: func() interface{} {
lines := ""
return lines
}}
//一個(gè)文件對(duì)象本身是實(shí)現(xiàn)了io.Reader的 使用bufio.NewReader去初始化一個(gè)Reader對(duì)象，存在buffer中的，讀取一次就會(huì)被清空
r := bufio.NewReader(f) //
//設(shè)置讀取緩沖池大小 默認(rèn)16
r = bufio.NewReaderSize(r,250 *1024)
var wg sync.WaitGroup
for {
buf := linesPool.Get().([]byte)
//讀取Reader對(duì)象中的內(nèi)容到[]byte類型的buf中
n, err := r.Read(buf)
buf = buf[:n]
if n == 0 {
if err != nil {
fmt.Println(err)
break
}
if err == io.EOF {
break
}
return err
}
//補(bǔ)齊剩下沒(méi)滿足的剩余
nextUntillNewline, err := r.ReadBytes('\n')
//fmt.Println(string(nextUntillNewline))
if err != io.EOF {
buf = append(buf, nextUntillNewline...)
}
wg.Add(1)
go func() {
ProcessChunk(buf, linesPool, stringPool)
wg.Done()
}()
}
wg.Wait()
return nil
}
func ProcessChunk(chunk []byte, linesPool *sync.Pool,stringPool *sync.Pool) {
//做相應(yīng)的處理
}
