SQL2005CLR函数扩展-数据导出的实现详解

SQLServer数据导出到excel有很多种方法，比如dts、ssis、还可以用sql语句调用openrowset。我们这里开拓思路，用CLR来生成Excel文件，并且会考虑一些方便操作的细节。

下面我先演示一下我实现的效果，先看测试语句
--------------------------------------------------------------------------------
exec BulkCopyToXls 'select * from testTable' , 'd:/test' , 'testTable' ,- 1
/*
开始导出数据
文件 d:/test/testTable.0.xls, 共65534条 , 大小20 ,450,868 字节
文件 d:/test/testTable.1.xls, 共65534条 , 大小 20 ,101,773 字节
文件 d:/test/testTable.2.xls, 共65534条 , 大小 20 ,040,589 字节
文件 d:/test/testTable.3.xls, 共65534条 , 大小 19 ,948,925 字节
文件 d:/test/testTable.4.xls, 共65534条 , 大小 20 ,080,974 字节
文件 d:/test/testTable.5.xls, 共65534条 , 大小 20 ,056,737 字节
文件 d:/test/testTable.6.xls, 共65534条 , 大小 20 ,590,933 字节
文件 d:/test/testTable.7.xls, 共26002条 , 大小 8,419,533 字节
导出数据完成
-------
共484740条数据，耗时 23812ms
*/
--------------------------------------------------------------------------------
上面的BulkCopyToXls存储过程是自定的CLR存储过程。他有四个参数：
第一个是sql语句用来获取数据集
第二个是文件保存的路径
第三个是结果集的名字，我们用它来给文件命名
第四个是限制单个文件可以保存多少条记录,小于等于0表示最多65534条。

前三个参数没有什么特别，最后一个参数的设置可以让一个数据集分多个excel文件保存。比如传统excel的最大容量是65535条数据。我们这里参数设置为-1就表示导出达到这个数字之后自动写下一个文件。如果你设置了比如100，那么每导出100条就会自动写下一个文件。

另外每个文件都可以输出字段名作为表头，所以单个文件最多容纳65534条数据。

用微软公开的biff8格式通过二进制流生成excel，服务器无需安装excel组件，而且性能上不会比sql自带的功能差，48万多条数据，150M，用了24秒完成。
--------------------------------------------------------------------------------
下面我们来看下CLR代码。通过sql语句获取DataReader，然后分批用biff格式来写xls文件。
--------------------------------------------------------------------------------

using System;
using System.Data;
using System.Data.SqlClient;
using System.Data.SqlTypes;
using Microsoft.SqlServer.Server;
public partial class StoredProcedures
{
    /// <summary>
    /// 导出数据
    /// </summary>
    /// <param name="sql"></param>
    /// <param name="savePath"></param>
    /// <param name="tableName"></param>
    /// <param name="maxRecordCount"></param>
    [Microsoft.SqlServer.Server.SqlProcedure ]
    public static void BulkCopyToXls(SqlString sql, SqlString savePath, SqlString tableName, SqlInt32 maxRecordCount)
    {
         if (sql.IsNull || savePath.IsNull || tableName.IsNull)
        {
            SqlContext .Pipe.Send(" 输入信息不完整！" );
        }
        ushort _maxRecordCount = ushort .MaxValue-1;

        if (maxRecordCount.IsNull == false && maxRecordCount.Value < ushort .MaxValue&&maxRecordCount.Value>0)
            _maxRecordCount = (ushort )maxRecordCount.Value;

        ExportXls(sql.Value, savePath.Value, tableName.Value, _maxRecordCount);
    }

    /// <summary>
    /// 查询数据，生成文件
    /// </summary>
    /// <param name="sql"></param>
    /// <param name="savePath"></param>
    /// <param name="tableName"></param>
    /// <param name="maxRecordCount"></param>
    private static void ExportXls(string sql, string savePath, string tableName, System.UInt16 maxRecordCount)
    {

        if (System.IO.Directory .Exists(savePath) == false )
        {
            System.IO.Directory .CreateDirectory(savePath);
        }

        using (SqlConnection conn = new SqlConnection ("context connection=true" ))
        {
            conn.Open();
            using (SqlCommand command = conn.CreateCommand())
            {
                command.CommandText = sql;
                using (SqlDataReader reader = command.ExecuteReader())
                {
                    int i = 0;
                    int totalCount = 0;
                    int tick = System.Environment .TickCount;
                    SqlContext .Pipe.Send(" 开始导出数据" );
                    while (true )
                    {
                        string fileName = string .Format(@"{0}/{1}.{2}.xls" , savePath, tableName, i++);
                        int iExp = Write(reader, maxRecordCount, fileName);
                        long size = new System.IO.FileInfo (fileName).Length;
                        totalCount += iExp;
                        SqlContext .Pipe.Send(string .Format(" 文件{0}, 共{1} 条, 大小{2} 字节" , fileName, iExp, size.ToString("###,###" )));
                        if (iExp < maxRecordCount) break ;
                    }
                    tick = System.Environment .TickCount - tick;
                     SqlContext .Pipe.Send(" 导出数据完成" );

                     SqlContext .Pipe.Send("-------" );
                     SqlContext .Pipe.Send(string .Format(" 共{0} 条数据，耗时{1}ms" ,totalCount,tick));
                }
            }
        }

 
    }
    /// <summary>
    /// 写单元格
    /// </summary>
    /// <param name="writer"></param>
    /// <param name="obj"></param>
    /// <param name="x"></param>
    /// <param name="y"></param>
    private static void WriteObject(ExcelWriter writer, object obj, System.UInt16 x, System.UInt16 y)
    {
        string type = obj.GetType().Name.ToString();
        switch (type)
        {
            case "SqlBoolean" :
            case "SqlByte" :
            case "SqlDecimal" :
            case "SqlDouble" :
            case "SqlInt16" :
            case "SqlInt32" :
            case "SqlInt64" :
            case "SqlMoney" :
            case "SqlSingle" :
                if (obj.ToString().ToLower() == "null" )
                    writer.WriteString(x, y, obj.ToString());
                else
                    writer.WriteNumber(x, y, Convert .ToDouble(obj.ToString()));
                break ;
            default :
                writer.WriteString(x, y, obj.ToString());
                break ;
        }
    }
    /// <summary>
    /// 写一批数据到一个excel 文件
    /// </summary>
    /// <param name="reader"></param>
    /// <param name="count"></param>
    /// <param name="fileName"></param>
    /// <returns></returns>
    private static int Write(SqlDataReader reader, System.UInt16 count, string fileName)
    {
        int iExp = count;
        ExcelWriter writer = new ExcelWriter (fileName);
        writer.BeginWrite();
        for (System.UInt16 j = 0; j < reader.FieldCount; j++)
        {
            writer.WriteString(0, j, reader.GetName(j));
        }
        for (System.UInt16 i = 1; i <= count; i++)
        {
            if (reader.Read() == false )
            {
                iExp = i-1;
                break ;
            }
            for (System.UInt16 j = 0; j < reader.FieldCount; j++)
            {
                WriteObject(writer, reader.GetSqlValue(j), i, j);
            }
        }
        writer.EndWrite();
        return iExp;
    }

    /// <summary>
    /// 写excel 的对象
    /// </summary>
    public class ExcelWriter
    {
        System.IO.FileStream _wirter;
        public ExcelWriter(string strPath)
        {
            _wirter = new System.IO.FileStream (strPath, System.IO.FileMode .OpenOrCreate);
        }
        /// <summary>
        /// 写入short 数组
        /// </summary>
        /// <param name="values"></param>
        private void _writeFile(System.UInt16 [] values)
        {
            foreach (System.UInt16 v in values)
            {
                byte [] b = System.BitConverter .GetBytes(v);
                _wirter.Write(b, 0, b.Length);
            }
        }
        /// <summary>
        /// 写文件头
        /// </summary>
        public void BeginWrite()
        {
            _writeFile(new System.UInt16 [] { 0x809, 8, 0, 0x10, 0, 0 });
        }
        /// <summary>
        /// 写文件尾
        /// </summary>
        public void EndWrite()
        {
            _writeFile(new System.UInt16 [] { 0xa, 0 });
            _wirter.Close();
        }
        /// <summary>
        /// 写一个数字到单元格x,y
        /// </summary>
        /// <param name="x"></param>
        /// <param name="y"></param>
        /// <param name="value"></param>
        public void WriteNumber(System.UInt16 x, System.UInt16 y, double value)
        {
            _writeFile(new System.UInt16 [] { 0x203, 14, x, y, 0 });
            byte [] b = System.BitConverter .GetBytes(value);
            _wirter.Write(b, 0, b.Length);
         }
        /// <summary>
        /// 写一个字符到单元格x,y
        /// </summary>
        /// <param name="x"></param>
        /// <param name="y"></param>
        /// <param name="value"></param>
        public void WriteString(System.UInt16 x, System.UInt16 y, string value)
        {
            byte [] b = System.Text.Encoding .Default.GetBytes(value);
            _writeFile(new System.UInt16 [] { 0x204, (System.UInt16 )(b.Length + 8), x, y, 0, (System.UInt16 )b.Length });
            _wirter.Write(b, 0, b.Length);
        }
    }
};

--------------------------------------------------------------------------------
把上面代码编译为TestExcel.dll，copy到服务器目录。然后通过如下SQL语句部署存储过程。
--------------------------------------------------------------------------------

CREATE ASSEMBLY TestExcelForSQLCLR FROM 'd:/sqlclr/TestExcel.dll' WITH PERMISSION_SET = UnSAFE;
--
go
CREATE proc dbo. BulkCopyToXls 
(  
    @sql nvarchar ( max ),
    @savePath nvarchar ( 1000),
    @tableName nvarchar ( 1000),
    @bathCount int
)    
AS EXTERNAL NAME TestExcelForSQLCLR. StoredProcedures. BulkCopyToXls

go

--------------------------------------------------------------------------------
当这项技术掌握在我们自己手中的时候，就可以随心所欲的来根据自己的需求定制。比如，我可以不要根据序号来分批写入excel，而是根据某个字段的值（比如一个表有200个城市的8万条记录）来划分为n个文件，而这个修改只要调整一下DataReader的循环里面的代码就行了。