C# SHIFT-JISで指定バイトで文字列を分割、分割時に最後の文字が漢字の左半分なら文字化けしないように全角で分割を行う

ある意味、力業

UniCodeからshift-jisに文字変換して指定バイトで文字列を分割しますが

分割の最後が全角の判定を行い、文字化けしないようにする。

 

using System;
using System.Collections.Generic;
using System.Text;

             var vList = new List<string>();
            var wordList = text.Split(new[] { "\r\n", "\n", "\r" }, StringSplitOptions.None);
            int vLoop, vStart, vLength;
            int vSize = 101;
            byte[] btBytes = null;
            string strTemp = null;
            bool vZenkaku;

            Encoding hEncoding = Encoding.GetEncoding("Shift_JIS");

            foreach (var item in wordList) {

                btBytes = hEncoding.GetBytes(item);
                vLength = btBytes.Length;

                if (vLength > vSize) {
                    vLoop = (vLength + vSize) / vSize;
                    vStart = 0;
                    for (int j = 0; j < vLoop; j++) {

                        if ((vStart + vSize) < vLength) {

                            //最後の文字が漢字の左半分の場合、切出の先頭は泣き別れになる
                            vZenkaku = is_zenkaku(btBytes[vStart + vSize - 1]);
                            if (vZenkaku) {
                                strTemp = hEncoding.GetString(btBytes, vStart, (vSize - 1));
                                vStart = vStart + vSize - 1;
                            }
                            else {
                                strTemp = hEncoding.GetString(btBytes, vStart, vSize);
                                vStart = vStart + vSize;
                            }

                            vList.Add(strTemp);
                        }
                        else {
                            //vList.Add(item.Substring(vStart, (item.Length - vStart)));
                            vList.Add(hEncoding.GetString(btBytes, vStart, (vLength - vStart)));
                        }
                    }
                }
                else {
                    vList.Add(item);
                }
            }

            foreach (var rec in vList) {
                Console.WriteLine(rec);
            }
        }

        // 全角判定（シフトJIS）
        static private bool is_zenkaku(byte c) {
            if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc)) {    // 全角の第１バイト
                return true;
            }
            return false;
        }