From ad9bae788ede47a1f781d23073c3f0946d4f2e2f Mon Sep 17 00:00:00 2001
From: starainrt <i@b612.me>
Date: Thu, 15 Jul 2021 11:09:32 +0800
Subject: [PATCH] first commit

---
 gbk.go      | 90 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 gbk_test.go | 31 ++++++++++++++++++
 2 files changed, 121 insertions(+)
 create mode 100644 gbk.go
 create mode 100644 gbk_test.go

diff --git a/gbk.go b/gbk.go
new file mode 100644
index 0000000..8e68f79
--- /dev/null
+++ b/gbk.go
@@ -0,0 +1,90 @@
+package startext
+
+import "golang.org/x/text/encoding/simplifiedchinese"
+
+func IsUtf8(data []byte) bool {
+	return isUtf8(data)
+}
+
+func IsGBK(data []byte) bool {
+	return (!isUtf8(data)) && isGBK(data)
+}
+
+func isGBK(data []byte) bool {
+	length := len(data)
+	var i int = 0
+	for i < length {
+		if data[i] <= 0x7f {
+			//编码0~127,只有一个字节的编码，兼容ASCII码
+			i++
+			continue
+		} else {
+			//大于127的使用双字节编码，落在gbk编码范围内的字符
+			if data[i] >= 0x81 &&
+				data[i] <= 0xfe &&
+				data[i+1] >= 0x40 &&
+				data[i+1] <= 0xfe &&
+				data[i+1] != 0x7f {
+				i += 2
+				continue
+			} else {
+				return false
+			}
+		}
+	}
+	return true
+}
+
+//UTF-8编码格式的判断
+
+func preNUm(data byte) int {
+	var mask byte = 0x80
+	var num int = 0
+	//8bit中首个0bit前有多少个1bits
+	for i := 0; i < 8; i++ {
+		if (data & mask) == mask {
+			num++
+			mask = mask >> 1
+		} else {
+			break
+		}
+	}
+	return num
+}
+func isUtf8(data []byte) bool {
+	i := 0
+	for i < len(data) {
+		if (data[i] & 0x80) == 0x00 {
+			// 0XXX_XXXX
+			i++
+			continue
+		} else if num := preNUm(data[i]); num > 2 {
+			// 110X_XXXX 10XX_XXXX
+			// 1110_XXXX 10XX_XXXX 10XX_XXXX
+			// 1111_0XXX 10XX_XXXX 10XX_XXXX 10XX_XXXX
+			// 1111_10XX 10XX_XXXX 10XX_XXXX 10XX_XXXX 10XX_XXXX
+			// 1111_110X 10XX_XXXX 10XX_XXXX 10XX_XXXX 10XX_XXXX 10XX_XXXX
+			// preNUm() 返回首个字节的8个bits中首个0bit前面1bit的个数，该数量也是该字符所使用的字节数
+			i++
+			for j := 0; j < num-1; j++ {
+				//判断后面的 num - 1 个字节是不是都是10开头
+				if (data[i] & 0xc0) != 0x80 {
+					return false
+				}
+				i++
+			}
+		} else {
+			//其他情况说明不是utf-8
+			return false
+		}
+	}
+	return true
+}
+
+func GBK2UTF8(data []byte) ([]byte,error) {
+	return simplifiedchinese.GBK.NewDecoder().Bytes(data)
+}
+
+func UTF82GBK(data []byte) ([]byte,error) {
+	return simplifiedchinese.GBK.NewEncoder().Bytes(data)
+}
\ No newline at end of file
diff --git a/gbk_test.go b/gbk_test.go
new file mode 100644
index 0000000..22fd6e8
--- /dev/null
+++ b/gbk_test.go
@@ -0,0 +1,31 @@
+package startext
+
+import (
+	"fmt"
+	"testing"
+)
+
+func Test_Gbk(t *testing.T) {
+	str:="你好"
+	if IsGBK([]byte(str)) {
+		t.Fail()
+	}
+	if !IsUtf8([]byte(str)) {
+		t.Fail()
+	}
+	gbk,err:=UTF82GBK([]byte(str))
+	if err!=nil{
+		t.Fatal(err)
+	}
+	if !IsGBK(gbk) {
+		t.Fail()
+	}
+	if IsUtf8(gbk) {
+		t.Fail()
+	}
+	utf8,err:=GBK2UTF8(gbk)
+	if err!=nil{
+		t.Fatal(err)
+	}
+	fmt.Println(string(utf8))
+}