为什么逐行读取文件需要更多内存?

问题描述 投票:0回答:2

我试图读取这种格式的大文件:

a string key, 200 values separated by comma

并将其写入地图。

我写了这段代码:

package main

import (
    "bufio"
    "unsafe"
    "fmt"
    "log"
    "os"
    "runtime"
    "strings"
)

func main() {

    file, err := os.Open("file_address.txt")
    if err != nil {
        log.Fatal(err)
    }
    defer file.Close()

    mp := make(map[string]float32)
    var total_size int64 = 0
    scanner := bufio.NewScanner(file)
    var counter int64 = 0

    for scanner.Scan() {
        counter++
        sliced := strings.Split(scanner.Text(), ",")
        mp[sliced[0]] = 2.2
    }

    if err := scanner.Err(); err != nil {
        log.Fatal(err)
    }
    fmt.Printf("loaded: %d. Took %d Mb of memory.", counter, total_size/1024.0/1024.0)
    fmt.Println("Loading finished. Now waiting...")

    var ms runtime.MemStats
    runtime.ReadMemStats(&ms)

    fmt.Printf("\n")
    fmt.Printf("Alloc: %d MB, TotalAlloc: %d MB, Sys: %d MB\n",
        ms.Alloc/1024/1024, ms.TotalAlloc/1024/1024, ms.Sys/1024/1024)
    fmt.Printf("Mallocs: %d, Frees: %d\n",
        ms.Mallocs, ms.Frees)
    fmt.Printf("HeapAlloc: %d MB, HeapSys: %d MB, HeapIdle: %d MB\n",
        ms.HeapAlloc/1024/1024, ms.HeapSys/1024/1024, ms.HeapIdle/1024/1024)
    fmt.Printf("HeapObjects: %d\n", ms.HeapObjects)
    fmt.Printf("\n")
}

这里是输出:

loaded: 544594. Took 8 Mb of memory.Loading finished. Now waiting...

Alloc: 2667 MB, TotalAlloc: 3973 MB, Sys: 2831 MB
Mallocs: 1108463, Frees: 401665
HeapAlloc: 2667 MB, HeapSys: 2687 MB, HeapIdle: 11 MB
HeapObjects: 706798

Done!

尽管密钥仅占用8Mb,但该程序占用了大约2.7Gb的内存!似乎sliced从未从堆中删除。我尝试在sliced=nil的末尾设置for,但这没有帮助。我已经读过,如果我将整个文件加载到内存中然后进行拆分,则可以避免此问题,但是我必须逐行读取文件,因为我没有足够的内存来加载一些较大的文件文件

为什么内存被占用?处理完每一行后如何释放它?

go memory-management
2个回答
0
投票

我想我发现了问题!我将大文件的每一行切成薄片。返回的[]string是一个切片,其中包含原始字符串(文件行)的子字符串。现在的问题是,每个子字符串都不是新字符串。 Is只是slice,它保留对未切片的字符串(文件行!)的引用。我保留每一行的sliced[0],因此,我保留对文件每一行的引用。垃圾收集器不会触摸读取行,因为我仍然对其进行引用。从技术上讲,我会读取文件的所有行并将其保留在内存中。

解决方案是将我想要的部分(sliced[0])复制到新字符串中,从而有效地丢失了对整行的引用。我这样做是这样的:

    sliced := strings.Split(scanner.Text(), ",")
    key_rune_arr := []rune(sliced[0])
    key := string(key_rune_arr) // now key is a copy of sliced[0] without reference to line
    mp[key] = 2.2 //instead of mp[sliced[0]] = 2.2

程序现在变成:

package main

import (
    "bufio"
    "unsafe"
    "fmt"
    "log"
    "os"
    "runtime"
    "strings"
)

func main() {

    file, err := os.Open("file_address.txt")
    if err != nil {
        log.Fatal(err)
    }
    defer file.Close()

    mp := make(map[string]float32)
    var total_size int64 = 0
    scanner := bufio.NewScanner(file)
    var counter int64 = 0

    for scanner.Scan() {
        counter++
        sliced := strings.Split(scanner.Text(), ",")
        key_rune_arr := []rune(sliced[0])
        key := string(key_rune_arr) // now key is a copy of sliced[0] without reference to line
        mp[key] = 2.2 //instead of mp[sliced[0]] = 2.2
    }

    if err := scanner.Err(); err != nil {
        log.Fatal(err)
    }
    fmt.Printf("loaded: %d. Took %d Mb of memory.", counter, total_size/1024.0/1024.0)
    fmt.Println("Loading finished. Now waiting...")

    var ms runtime.MemStats
    runtime.ReadMemStats(&ms)

    fmt.Printf("\n")
    fmt.Printf("Alloc: %d MB, TotalAlloc: %d MB, Sys: %d MB\n",
        ms.Alloc/1024/1024, ms.TotalAlloc/1024/1024, ms.Sys/1024/1024)
    fmt.Printf("Mallocs: %d, Frees: %d\n",
        ms.Mallocs, ms.Frees)
    fmt.Printf("HeapAlloc: %d MB, HeapSys: %d MB, HeapIdle: %d MB\n",
        ms.HeapAlloc/1024/1024, ms.HeapSys/1024/1024, ms.HeapIdle/1024/1024)
    fmt.Printf("HeapObjects: %d\n", ms.HeapObjects)
    fmt.Printf("\n")
}

结果如我所愿:

loaded: 544594. Took 8 Mb id memory.Loading finished. Now waiting...

Alloc: 94 MB, TotalAlloc: 3986 MB, Sys: 135 MB
Mallocs: 1653590, Frees: 1108129
HeapAlloc: 94 MB, HeapSys: 127 MB, HeapIdle: 32 MB
HeapObjects: 545461

Done!

0
投票

为了有效使用CPU和内存,

key := string(bytes.SplitN(scanner.Bytes(), []byte(","), 2)[0])
mp[key] = 2.2
© www.soinside.com 2019 - 2024. All rights reserved.