How To Return a Result Set from a SQL Server 2005 CLR Stored Procedure

By: Ray Barley | Comments (3) | Related: > Stored Procedures

Problem
We occasionally come up with a requirement that would be a good fit for a CLR function or stored procedure. For instance we would like to call a stored procedure to get the list of files in a particular folder. How can we return the list of files as a standard result set (i.e. rows and columns) using the CLR?

Solution
SQL Server 2005 and later integrates the CLR (Common Language Runtime) which allows us to code stored procedures, triggers, user-defined functions, user-defined aggregates, and user-defined types using Microsoft .NET code; e.g. Visual Basic .NET or C#. The various T-SQL CREATE commands have been enhanced to allow us to define a database object (e.g. a stored procedure) and connect our .NET code to it. When you execute the stored procedure, the .NET code is executed.

When writing a stored procedure or function, the most natural way to return data is a result set. In this tip we will implement a stored procedure to retrieve the list of files from a folder and return the list as a standard result set with rows and a single column. We will walk through the following steps:

Enable the CLR
Review the C# code sample
Deploy the CLR stored procedure

Enabling the CLR

By default the CLR is not enabled in SQL Server 2005. You can enable the CLR either by running the Surface Area Configuration utility or sp_configure. In the Surface Area Configuration drill down to CLR Integration then click the checkbox to enable it as shown below:

To enable the CLR by using sp_configure, execute the following script:

sp_configure 'clr enabled', 1
GO
reconfigure
GO

Code Sample

The following C# function will get the list of files from a folder and return the list as a result set. The function's parameters are as follows:

path - folder name to retrieve the files from; e.g. C:\TEMP
pattern - wildcards like *.* (all files), *.dat, etc.
recursive - include files in subfolders; 1 if true, 0 otherwise

    public static void GetListOfFiles(
        SqlString path, 
        SqlString pattern, 
        SqlBoolean recursive)
    {
        SqlPipe pipe = SqlContext.Pipe;
        SqlMetaData[] cols = new SqlMetaData[1];
        cols[0] = new SqlMetaData(
            "FILE_NAME", SqlDbType.NVarChar, 1024);
        SearchOption searchOption;
        if (recursive == true)
            searchOption = SearchOption.AllDirectories;
        else
            searchOption = SearchOption.TopDirectoryOnly;
        string dir = path.ToString();
        if (Directory.Exists(dir) == false)
        {
            pipe.Send("Directory does not exist");
            return;
        }
        string[] files = Directory.GetFiles(
            dir, pattern.ToString(), searchOption);
        if (files.Length > 0)
        {
            SqlDataRecord rec = new SqlDataRecord(cols);
            pipe.SendResultsStart(rec);
            foreach (string file in files)
            {
                rec.SetSqlString(0, new SqlString(file));
                pipe.SendResultsRow(rec);
            }
            pipe.SendResultsEnd();
        }
        else
        {
            pipe.Send("No files");
        }
    }

Directory is a class in the .NET Framework. It is used to check whether the path exists and to get the list of files. The main points in the above code from the standpoint of CLR integration are:

The SqlPipe object is used to send the results back to the caller. The SqlContext object is automatically available and provides the SqlPipe object.
The SqlMetaData class is used to specify a column in the result set. We specify the column name, type and size. We are only returning a single column in this case but you can return multiple columns.
The SqlDataRecord class is used to populate a single row in the result set. It is initialized with the array of SqlMetaData objects (i.e. the columns). The SetSqlString method is called to assign a value to each column based on the ordinal number (i.e. index in the array of SqlMetaData objects).
The SendResultsRow method of SqlPipe sends a single row back to the caller via the SqlDataRecord object.
The SendResultsEnd method of SqlPipe is used to indicate the end of the result set.
The Send method of SqlPipe is used to send back a message to the caller.

Deploy the CLR Stored Procedure

The above sample code needs to be compiled in order to be called via a stored procedure. Execute the following from the command line to compile the code and create the class library DLL MSSQLTipsCLRLib.dll:

CSC /target:library StoredProcedures.cs /out:MSSQLTipsCLRLib.dll

Assuming version 2 of the Microsoft .NET framework, you can find CSC.EXE in the folder C:\WINDOWS\Microsoft.NET\Framework\v2.0.50727.

Execute the following T-SQL script to create the stored procedure:

ALTER DATABASE mssqltips
SET TRUSTWORTHY ON
GO
USE mssqltips
GO
CREATE ASSEMBLY MSSQLTipsCLRLib 
FROM 'C:\mssqltips\MSSQLTipsCLRLib.dll'
WITH PERMISSION_SET = EXTERNAL_ACCESS
GO
CREATE PROCEDURE dbo.GetListOfFiles
  @path NVARCHAR(256)
, @pattern NVARCHAR(64)
, @recursive BIT
AS
EXTERNAL NAME MSSQLTipsCLRLib.StoredProcedures.GetListOfFiles

After compiling the code into a class library (.dll) the CREATE ASSEMBLY command is executed to associate the DLL with the ASSEMBLY database object. The FROM clause must point to the actual path of the DLL. PERMISSION_SET must be set to EXTERNAL_ACCESS because the .NET code is accessing the file system which is outside of SQL Server. The TRUSTWORTHY option is set on to allow the external access. Finally the EXTERNAME NAME of the CREATE PROCEDURE command associates the assembly, class and function to the stored procedure name.

To execute the stored procedure, execute the following script:

EXEC dbo.GetListOfFiles 'C:\mssqltips', '*.*', 0

You will see output that looks something like this, depending on the contents of the folder you choose; i.e. a result set with a single column and a row for each file:

Next Steps

Take a look at our earlier tips on CLR String Sort Function in SQL Server 2005 and CLR function to delete older backup and log files in SQL Server for additional details on CLR functions in SQL Server.
Download the sample script here and experiment with returning result sets from a CLR stored procedure.
Keep in mind that when there is a function in the Microsoft .NET framework that does what you need, using the CLR in SQL Server may be a good solution.
Visual Studio 2005 has a SQL Server project template which simplifies the creation and deployment of the CLR integration code.

About the author

Ray Barley is a Principal Architect at IT Resource Partners and a MSSQLTips.com BI Expert.

This author pledges the content of this article is based on professional experience and not AI generated.

View all my tips

Comments For This Article

Tuesday, March 1, 2016 - 10:36:21 PM - yogesh mehla

Sir can you tell.me how to return multiple recordset or dataset wig mutilple table back with clr storedprocedure or function. I want to use threading to get data parallel. But how to return it.

Monday, June 30, 2008 - 2:57:54 PM - raybarley

As far as I know you have to do it this way. You can also return a result set from a function (i.e. a table-valued function) using the CLR. It is a bit more elegant (well maybe) but you still have to manually assign the values to the result set. I have done a tip on that; it should be published soon.

What you do is create a function (let's call it A) that retrieves the data you want to return as a result set and returns it as a collection class that implements the IEnumerable interface. Then you have a companion function (let's call it B) that has an output parameter for each column in the result set. When the CLR function is called it invokes A once then iterates over the collection calling B one time for each element in the collection.

Here are the 2 functions:

/* THIS IS 'A' */

    [SqlFunction(SystemDataAccess = SystemDataAccessKind.Read, FillRowMethodName = "GetListInfo")]
    public static IEnumerable GetListCollection(SqlString url)
    {
        DataTable t = new DataTable();
        WindowsImpersonationContext ctx = null;

        WindowsIdentity id = SqlContext.WindowsIdentity;

        try
        {
            ctx = id.Impersonate();

            WSS.Lists svc = new WSS.Lists();
            svc.Url = url.ToString();

            //svc.Credentials = new NetworkCredential("barley", "pass@word1", "VS");
            svc.Credentials = CredentialCache.DefaultNetworkCredentials;
            XmlNode node = svc.GetListCollection();
            XmlTextReader rdr = new XmlTextReader(node.OuterXml,
                                        XmlNodeType.Element, null);
            DataSet ds = new DataSet();
            ds.ReadXml(rdr);
            t = ds.Tables[0];
        }
        finally
        {
            if (ctx != null)
                ctx.Undo();
        }

        return t.Rows;
    }

/* THIS IS 'B' */
    public static void GetListInfo(
                        object obj,
                        out SqlString name,
                        out SqlString title,
                        out SqlString url )
    {
        DataRow r = (DataRow)obj;
        name = new SqlString(r["Name"].ToString());
        title = new SqlString(r["Title"].ToString());
        url = new SqlString(r["DefaultViewUrl"].ToString());
    }

Monday, June 30, 2008 - 10:00:22 AM - AndreQ

Its a very handy utility but the part that caught my eye is assigning each column for each row of a result set. Is this just one of those unfortunate requirements when translatng from a collection to a recordset?

I remember having to do this in VBA way back when and although it always worked, its slow work. Are there any functions that would do this automatically, cycling through a collection's data elements and automatically creating recordset and writing to it?

THAT would be an incredibly useful library function.

Regards

AndreQ