.NET Core 查询Oracle BLOB字段报错：TTC Error，转换成base64

.NET Core NuGet程序包：Oracle.ManagedDataAccess.Core 查询Oracle BLOB字段报错：TTC Error
查询blob的SQL语句在.NET FrameWork下并无异常，但用.NET Core NuGet程序包：Oracle.ManagedDataAccess.Core访问Oracle数据库时报：TTC Error 异常
解决思路：查询结果会把BLOB字段拆分为多行，每行2000个字节，程序按字段 OFFSET 顺序拼接起来即可
BLOB字段拆分的SQL代码：

SELECT DBMS_LOB.SUBSTR(O.FILE_CONTENT, 2000, O.OFFSET1) AS CONTENT,
       O.OFFSET1,
       O.MOD,
       O.FILE_CONTENT_LENGTH,
       O.INTERATION_COUNT
  FROM (SELECT (2000 * (ROWNUM - 1)) + 1 AS OFFSET1,
               I.MOD,
               I.FILE_CONTENT_LENGTH,
               I.INTERATION_COUNT,
               i.FILE_CONTENT
          FROM (SELECT dbms_lob.getlength(a.照片) AS FILE_CONTENT_LENGTH,
                       MOD(dbms_lob.getlength(a.照片), 2000) AS MOD,
                       CASE
                         WHEN MOD(dbms_lob.getlength(a.照片), 2000) > 0 THEN
                          TRUNC((dbms_lob.getlength(a.照片) / 2000) + 1)
                         ELSE
                          TRUNC(dbms_lob.getlength(a.照片) / 2000)
                       END INTERATION_COUNT,
                       a.照片 as FILE_CONTENT
                  FROM 人员照片 A
                 WHERE A.人员Id = 576) I
        CONNECT BY LEVEL <= I.INTERATION_COUNT) o
 order by OFFSET1

程序拼接，并转成Base64的C#代码

//取数据到databale
DataTable dtx = db.queryDatatable(sql);
                    List<byte> byteSource = new List<byte>();
                    byte[] arr = new byte[Convert.ToInt64(dtx.Rows[0]["FILE_CONTENT_LENGTH"].ToString())];

                    if (dtx.Rows.Count > 2)
                    {
                        for (int xx = 0; xx < dtx.Rows.Count; xx++)
                        {
                            byteSource.AddRange((byte[])dtx.Rows[xx]["CONTENT"]);
                        }
                        arr = byteSource.ToArray();
                    }
                    //转为Base64
                    string pic = Convert.ToBase64String(arr);