Unicode string normalization example

normalize.go

package main

import (
	"fmt"
	"regexp"
	"strings"

	"golang.org/x/text/unicode/norm"
)

var (
	controlCodeRegexp = regexp.MustCompile(`\{p}`)
	lineFeedRegexp    = regexp.MustCompile(`\r\n|\r`)
)

func Normalize(value string) string {
	values := strings.Split(lineFeedRegexp.ReplaceAllString(value, "\n"), "\n")
	lines := make([]string, len(values))
	for index, line := range values {
		lines[index] = controlCodeRegexp.ReplaceAllString(line, "")
	}

	return strings.TrimSpace(norm.NFKC.String(strings.Join(lines, "")))
}

func main() {
	message := "Ｆｉｇｈｔ　ｄａｙｏ！"

	fmt.Println(message)
	fmt.Println(Normalize(message))
}