在Golang中查找模式的字节偏移

Question

We can find the byte offset of a pattern from file by "grep -ob pattern filename"; However, grep is not utf8 safe. How do I find byte offset of a pattern in Go? The file is process log, which can be in TB.

This is what I want to get in Go:

$ cat fname
hello world
findme
hello 世界
findme again

...

$ grep -ob findme fname

12:findme
32:findme

Answer 1

FindAllStringIndex(s string, n int) returns byte start/finish indexes (i.e., slices) of all successive matches of the expression:

package main

import "fmt"
import "io/ioutil"
import "regexp"

func main() {
    fname := "C:\\Users\\UserName\\go\\src\\so56798431\\fname"
    b, err := ioutil.ReadFile(fname)
    if err != nil {
      panic(err)
    }

    re, err := regexp.Compile("findme")
    if err != nil {
      // handle error
    }
    fmt.Println(re.FindAllStringIndex(string(b), -1))
}

Output:

[[12 18] [32 38]]

Note: I did this on Microsoft Windows, but saved the file in UNIX format (linefeed); if input file saved in Windows format (carriage return & linefeed) the byte offsets would increment to 13 and 35, respectively.

UPDATE: for large files, use bufio.Scanner; for example:

package main

import (
    "bufio"
    "fmt"
    "log"
    "os"
    "regexp"
)

func main() {
    fname, err := os.Open("C:\\Users\\UserName\\go\\src\\so56798431\\fname")
    if err != nil {
        log.Fatal(err)
    }
    defer fname.Close()

    re, err := regexp.Compile("findme")
    if err != nil {
      // handle error
    }

    scanner := bufio.NewScanner(fname)
    bytesRead := 0
    for scanner.Scan() {
        b := scanner.Text()
        //fmt.Println(b)
        results := re.FindAllStringIndex(b, -1)
        for _, result := range results {
            fmt.Println(bytesRead + result[0])
        }
        // account for UNIX EOL marker
        bytesRead += len(b) + 1
    }

    if err := scanner.Err(); err != nil {
        log.Fatal(err)
    }
}

Output:

12

32