It seems like it should be simple but as of yet I havent found a way to save the value stored in an SSIS string variable to a text file. I've looked at using the flat file destination inside of a data flow but that requires a data flow source.
Any ideas on how to do this?
Use a script task.
I just tried this. I created a File connection manager, with the connection string pointing to the file I wanted to write to. I then created a string variable containing the text to write.
I added a Script Task, specified my string variable in the Read Only Variables list, then clicked Edit Script. The script was as follows:
public void Main()
{
ConnectionManager cm = Dts.Connections["File.tmp"];
var path = cm.ConnectionString;
var textToWrite = (string)Dts.Variables["User::StringVariable"].Value;
System.IO.File.WriteAllText(path, textToWrite);
Dts.TaskResult = (int)ScriptResults.Success;
}
This worked with no problems.
Here's a little sample of some code that worked in a SQL CLR in C#. You'll need to use VB if you're on 2005 I believe. The script task also needs the read variable property set to MyVariable to make the value of your variable available to it.
// create a writer and open the file
TextWriter tw = new StreamWriter("\\\\server\\share$\\myfile.txt");
// write a line of text to the file
tw.WriteLine(Dts.Variables["MyVariable"].Value);
// close the stream
tw.Close();
All it takes is one line of code in a simple Script task. No other dependencies, such as a connection manager, are needed.
Here's what it would look like in C#:
public void Main()
{
string variableValue = Dts.Variables["TheVariable"].Value.ToString();
string outputFile = Dts.Variables["Path"].Value.ToString();
System.IO.File.WriteAllText(outputFile, variableValue);
Dts.TaskResult = (int)ScriptResults.Success;
}
Obviously the most important line here is the one containing the WriteAllText function call.
The Path variable should contain a full path + filename for the output file.
Ok, I have an answer that doesn't involve use of script task. Pick some oledb sql source you have that's simple and you have a lot of control over. Make a query that returns only one row. Then put this query in a string variable: "select vara, ' var =: " + @[User:varIWantToSee] + "' as myvar from tablea where vara = 1" Then in OLEDB source pick "SQL command from a variable" For varIWantToSee make sure you initialize it with a lot characters or ssis makes a very small length for that column that it doesn't let you override. At run time varIWantToSee will get set and you can see it. Pump this all into a flat file destination and you are in business. Why do some people have to do this? Because some people need to know the value of the variables in the runtime environment, their laptop development doesn't show the variable values they need. In my case I was running this on an Azure environment that had the database accesses I needed to test. If I were microsoft I would create a task that shows the runtime variable value at that stage of the job by writing it to the ssis log file created when the package runs. If someone knows how to do that, please enlighten us.
It's possible to use a Derived Column transformation to write the value of a variable into a column. The problem is that it needs a source to drive it, and there's no stock data source you can use that just spits out a null row onto the pipeline.
So, either you repurpose a single-row source to drive the derived column transformation, or you do what another answer suggests, and do it with a Script source.
I did it the way you described. I already had a oledb connection manager defined so I used an OLE DB Source and used the SQL Command data access mode. I used a simple query:
select getdate() as dt
...just to get it out of the way. Now I know the date of my variable pull. Then I used a Derived Column Transform to make my package variables available and wrote it out to a flat file.
Elegant? No, but it gets the job done.
Lets say you don't want to mess with Script tasks and you don't have a database you can connect to just to issue a data source command like:
SELECT 'Some arbitrary text'
There are still several ways to use a Process task for something as simple as writing a line of text to a file. For example you can use PowerShell with an input variable built using the following expression:
"'"+REPLACE(@[User::Text],"'","''")+"' > '"+REPLACE(@[User::Filename],"'","''")+"'"
Notice I escaped the filename because single quotes are legal there. Also note I used '>' for redirecting which overwrites the file if it exists. If I wanted to append I'd use '>>'.
Initially I had trouble with this method when User::Text contained multiple lines. It turns out you need some extra EOL characters after your filename when a command spans lines. Like this:
"'"+REPLACE(@[User::Text],"'","''")+"' > '"+REPLACE(@[User::Filename],"'","''")+"'\r\n\r\n"
Using cmd.exe with echo is a bit more precarious but can also work in certain circumstances and has much less overhead.
P.S. I've noticed with some versions of PowerShell that StandardInputVariable content is ignored without this:
-Command -
in the Arguments box. A lone minus sign as a Command argument is 'magic' and documented at https://learn.microsoft.com/en-us/powershell/scripting/powershell.exe-command-line-help. I believe all versions of PowerShell accept this param so even if it's not required for your version you may want to include it since it shouldn't break anything and may keep your code from breaking if PowerShell is updated to a version that requires it.
