[游戏开发]Unity分享一个文本乱码编码转换工具代码

关于编码转换

工作多年，好像没搞通过编码转换

JAVA做过，C#做过，就是碰到问题，随便一通转换就过了，没搞清楚过

很多年了，没深究，惭愧

就好像正则表达式，虽然做过一段时间前端，但是就是没做过爬虫，页面分析之类的，所以也没深究

今天刚好需要，搞了一下转换

一开始，网上找的方法都不好用，根本实现不了

测试了一下，发现甚至网上的方法都不能转换，难道直接Ctrl+C也能出错？

关键在于，必须先用本机编码，先获得“文本源”，原理还是没深究

        byte[] temp;
        //将 textBox1.Text 转换为 byte 数组
        temp = Encoding.Default.GetBytes(currText);
        
        temp = Encoding.Convert(code1,code2,temp);

整个代码：

/* 转码还原（多对比）参考，语法Java
 * https://blog.csdn.net/weixin_30715523/article/details/99492142
 * c# 字串码合集
 * https://blog.csdn.net/luanpeng825485697/article/details/77622243
 * 
 */

using System;
using System.Collections.Generic;
using System.Text;
using I18N.CJK;
using UnityEngine;
using UnityEditor;
public class TextCodeWnd:EditorWindow
{
    public enum TransferType
    {
        Test,
        Src2Dest,
    }

    public enum FormatType
    {
        GBK,
        utf_8,
        GB2312,
    }
    [MenuItem("Tools/TextCodeRecover")]
    public static void ShowWindow()
    {
        GetWindow<TextCodeWnd>();
    }

    private string currText;
    
    //private List<Encoding> charsetAttr;
    private List<string> charsetAttr;
    private FormatType srcFormat;
    private FormatType destFormat;
    private TransferType transferType;
    private void Awake()
    {
        titleContent = new GUIContent("TextCodeRecover");
        Debug.LogError("完美");
        //charsetAttr = new List<string>{"GBK", "utf-8","utf-16","utf-16BE","GB18030","GB2312","GBK","Windows-1252","ISO8859-1"};
        if (string.IsNullOrEmpty(currText))
        {
            currText = "閽╃埅";
        }
    }

    private void OnGUI()
    {
        transferType = (TransferType)EditorGUILayout.EnumPopup("转换方式",transferType);
        GUILayout.Space(5);
        GUILayout.BeginHorizontal();
        srcFormat = (FormatType)EditorGUILayout.EnumPopup("转换从", srcFormat);
        GUILayout.Label("->");
        destFormat = (FormatType) EditorGUILayout.EnumPopup("到", destFormat);
        GUILayout.EndHorizontal();
        currText = EditorGUILayout.TextField(currText);
        string keySrc = srcFormat.ToString();
        string keyDest = destFormat.ToString();
        if (keySrc != keyDest)
        {
            if (GUILayout.Button("ttt"))
            {
                if (transferType == TransferType.Test)
                {
                    var str = FinalTransferEncoding(keySrc, keyDest, currText);
                    Debug.LogError(str);
                }
                else
                {

                    //var str = TransferEncoding(new GB18030Encoding(), new UTF8Encoding(), currText);
                    Debug.LogError(srcFormat + "-》utf-8");

                    keySrc = keySrc.Replace("_", "-");
                    keyDest = keyDest.Replace("_", "-");
                    var str = TransferEncoding(keySrc, keyDest, currText);
                    Debug.LogError(str);
                }

                //RecoverTextAllCheck(currText);
            }
        }
    }

    string FinalTransferEncoding(string keySrc,string keyDest,string currText)
    {
        byte[] temp;
        //将 textBox1.Text 转换为 byte 数组
        temp = Encoding.Default.GetBytes(currText);
        keySrc = keySrc.Replace("_", "-");
        keyDest = keyDest.Replace("_", "-");
        //利用 Encoding 类的 Convert 方法，将 temp 的编码由 gb2312 转换为 big5 编码
        //Encoding.GetEncoding("gb2312")
        var code1 = Encoding.GetEncoding(keySrc);
        //Encoding.GetEncoding("big5")
        var code2 = Encoding.GetEncoding(keyDest);
        temp = Encoding.Convert(code1,code2,temp);
        return Encoding.Default.GetString(temp);
    }

    /// <summary>
    /// 网上抄的穷举对比方法，没什么卵用
    /// </summary>
    /// <param name="str"></param>
    void RecoverTextAllCheck(string str)
    {
        //GB18030Encoding text = new GB18030Encoding();
        //byte[] btArr = text.GetBytes(str);
        foreach (var key in charsetAttr)
        {
            var charset = Encoding.GetEncoding(key);
            byte[] srcBytes = charset.GetBytes(str);
            foreach (var originCharsetKey in charsetAttr)
            {
                var originCharset = Encoding.GetEncoding(originCharsetKey);
                if (originCharset.Equals(charset)) {
                    continue;
                }

                var bytes = Encoding.Convert(charset, originCharset, srcBytes);
                var encodeText = originCharset.GetString(bytes);
                Debug.LogError(string.Format("{0}=>({1}){2};结果：{3}",charset, originCharsetKey, originCharset, encodeText));
            }
        }

    }
    /// <summary>
    /// 本来重载下面的方法，不行；换了一个实现 测试，还是不行；几乎确实是不行的，换方法；最终 FinalTransferEncoding()，完成。
    /// </summary>
    /// <param name="src"></param>
    /// <param name="dest"></param>
    /// <param name="str"></param>
    /// <returns></returns>
    public static string TransferEncoding(string src, string dest, string str)
    {
        var srcFormat = Encoding.GetEncoding(src);
        var bytes = Encoding.GetEncoding(src).GetBytes(str);
        return srcFormat.GetString(bytes);
        //return TransferEncoding(Encoding.GetEncoding(src), Encoding.GetEncoding(dest), str);
    }

    /// <summary>
    /// 字符串编码转换（其实没用的方法,也可能原代码是Java和c#油茶）
    /// </summary>
    /// <param name="srcEncoding">原编码</param>
    /// <param name="dstEncoding">目标编码</param>
    /// <param name="srcBytes">原字符串</param>
    /// <returns>字符串</returns>
    public static string TransferEncoding(Encoding srcEncoding, Encoding dstEncoding, string srcStr)
    {
        byte[] srcBytes = srcEncoding.GetBytes(srcStr);
        byte[] bytes = Encoding.Convert(srcEncoding, dstEncoding, srcBytes);
        return dstEncoding.GetString(bytes);
    }
}