How to pass parameters to SQL script via Powershell - sql

Question

I have a Power-shell script that calls a SQL script. This is currently working, but inside my sql script I have some hard coded parameters that I would like to pass to the SQL script via the powershell.
So this is the snip-it from the Power-shell script
function ExecSqlScript([string] $scriptName)
{
$scriptFile = $script:currentDir + $scriptName
$sqlLog = $script:logFileDir + $scriptName + "_{0:yyyyMMdd_HHmmss}.log" -f (Get-Date)
$result = sqlcmd -S uk-ldn-dt270 -U sa -P passwordhere3! -i $scriptFile -b | Tee-Object - filepath $sqlLog
if ($result -like "*Msg *, Level *, State *" -Or $result -like "*Sqlcmd: Error:*")
{
throw "SQL script " + $scriptFile + " failed: " + $result
}
}
try
{
ExecSqlScript "restoreDatabase.sql"
}
catch
{
//Some Error handling here
}
And this is from the SQL
USE MASTER
GO
DECLARE #dbName varchar(255)
SET #dbName = 'HardCodedDatabaseName'
So I want to pass the value for dbName, any ideas?

Answer 1

You could take advantage of sqlcmd's scripting variables. Those can be used in script file and are marked with $(). Like so,
-- Sql script file
use $(db);
select someting from somewhere;
When calling sqlcmd, use the -v parameter to assign variables. Like so,
sqlcmd -S server\instance -E -v db ="MyDatabase" -i s.sql
Edit
Mind the Sql syntax when setting variables. Consider the following script:
DECLARE #dbName varchar(255)
SET #dbName = $(db)
select 'val' = #dbName
As passed to the Sql Server, it looks like so (Profiler helps here):
use master;
DECLARE #dbName varchar(255)
SET #dbName = foo
select 'val' = #dbName
This is, obviously invalid a syntax, as SET #dbName = foo won't make much sense. The value ought to be within single quotes like so,
sqlcmd -S server\instance -E -v db ="'foo'" -i s.sql

Answer 2

Just in case someone else needs to do this... here is a working example.
Power Shell Script:
sqlcmd -S uk-ldn-dt270 -U sa -P 1NetNasdf£! -v db = "'DatabaseNameHere'" -i $scriptFile -b | Tee-Object -filepath $sqlLog
Note the -v switch to assign the variables
And here is the MS SQL:
USE MASTER
GO
if db_id($(db)) is null
BEGIN
EXEC('
RESTORE DATABASE ' + $(db) + '
FROM DISK = ''D:\DB Backup\EmptyLiveV5.bak''
WITH MOVE ''LiveV5_Data'' TO ''C:\Program Files (x86)\Microsoft SQL Server\MSSQL11.MSSQLSERVER\MSSQL\DATA\LiveV5_' + $(db) + '.MDF'',
MOVE ''LiveV5_Log'' To ''C:\Program Files (x86)\Microsoft SQL Server\MSSQL11.MSSQLSERVER\MSSQL\DATA\LiveV5_' + $(db) + '_log.LDF'', REPLACE,
STATS =10')
END
Note: You do not have to assign the scripting varible to a normal sql varible like this.
SET #dbName = $(db)
you can just use it in your sql code. - Happy coding.

Answer 3

Here's a full example using a different PowerShell approach. Here I'm using a specific script to reset local databases for a clean development environment.
## Reset the Database
$resetScript= "C:\ResetSite\resetDatabases.sql"
Write-Host "Resetting the DB - Running $($resetScript)"
$connectionString = "Server = localhost; Database = 'master'; UID = myusername; PWD = mypassword"
# Create variables & params
$sqlCmdVariables = #(
"Database=$($siteConfig.db_name)",
"UserName=$($siteConfig.db_username)",
"UserPassword=$($siteConfig.db_user_password)"
)
$sqlCmdParameters = #{
InputFile = $resetScript
QueryTimeout = 1800
ConnectionString = $connectionString
Variable = $sqlCmdVariables
}
# Invoke
Invoke-SqlCmd #sqlCmdParameters
The .sql file then uses the parameters passed in, the same way #nmbell mentions.
-- Declare the vars
DECLARE #Database nvarchar(100), #UserName nvarchar(100), #UserPassword nvarchar(100)
-- Set the vars
SET #Database = '$(Database)' -- database name
SET #UserName = '$(UserName)' -- SQL login and database username
SET #UserPassword = '$(UserPassword)' -- login password
... more stuff here.. use the vars like normal
This is partly derived from this blog post but modified slightly to use a file rather than an inline query.

Answer 4

Adjusting vonPryz's answer to use:
SET #dbName = '$(db)'
means you can pass in the parameter from the command line in a more natural form as
sqlcmd -S server\instance -E -v db ="foo" -i s.sql
The SqlCmd variable still substitutes correctly.

Answer 5

I know this is an old answer but I do have a better way that is much easier if you only need a small amount of data from Powershell (or even a large amount as long as all you want is text), and you work mainly in SQL for your scripting like I do:
1: Start the PowerShell from SQL using xp_cmdshell, and insert the results to a one-column table which allows NULLs e.g:
DECLARE #Results (Line varchar(1000) NULL)
INSERT #Results
EXEC master.dbo.xp_cmdshell '"powershell.exe C:\PowershellScripts\MyScript.ps1 MyParams"'
2: During your PowerShell script, for anything you want to pass back to SQL, simply use "Write-Output", e.g:
Write-Output $returned_data
You can do this as many times as you want. If you have 10,000 values to pass back to SQL, then you could use write-output 10,000 times.
So in the above example once the "MyScript.ps1" PowerShell script finishes running, all of the output will be in the #Results table variable, ready to be used, queried, imported into individual variables, whatever you want really.