When I run the this select '$(''test'')' in SQL Management Studio 2008 it returns $('test')
When I run sqlcmd -S SERVER -E -d DATABASE -q "select $(''test'')" on the command line it returns Sqlcmd: Error: Syntax error at line 1 near command '''.
If I remove the dollar sign it works. Is the "$" a special character?
Is this a sqlcmd bug? How can I change the script to get the desired result.
Yes, $(id) has special semantics in SQLCMD: it is a variable substitution. You can run commands like:
sqlcmd /E /S . /v variable=MyTable /Q "select * from $(variable)"
and this will select from MyTable. As you guess, the /v is the switch to define a variable. SSMS on the other hand does not, by default, interpret the SQL for variable substitution. SSMS can be made to do this, by checking the option 'Open query widows in SQLCMD mode'.
For mode details see Using sqlcmd with Scripting Variables.
the command you are trying to run:
select $('test')
is not valid. As you note, when you remove the "$" it works:
select ('test')
I'm not sure what you are really trying to do, you have three " double quote characters, you could try using this command:
select '$(test)'
which would be:
sqlcmd -S SERVER -E -d DATABASE -q "select ''$(test)''"
Related
I'm getting the following error message with execution T-SQL script with this construction "$(", in sqlcmd utility:
C:\Users\Admin>sqlcmd -S DESKTOP-5AA9JIA\SQLEXPRESS -q "select '$(' " -R
Sqlcmd: Error: Syntax error at line 1 near command '''.
1>
if run this script via SSMS then everything works smoothly.
Getting the same error, when using INSERT INTO statement or any other statement.
Any suggestion on how to resolve it?
You should use -x when launch sqlcmd to ignore scripting variables.
-x
Causes sqlcmd to ignore scripting variables. This is useful when a
script contains many INSERT statements that may contain strings that
have the same format as regular variables, such as $(variable_name).
sqlcmd Utility
When sqlcmd parser sees $(it expects scripting variable that is not provided so it throws the error.
Here is my test:
I want to run a weekly extract from a SQL Server database using SQLCMD under SQL Agent. Because I need to save multiple extracts in the same share, I want to use the current date as part of the extract's file name. When doing this from the command line, I use:
sqlcmd -S POC -i "\\org-data\data\dept\share\registry\SQLCMD\extractdata.sql" -s "|" -W -h-1 -o "\\org-data\data\dept\share\registry\Extracts\extractdata.%date:~-4,4%%date:~-10,2%%date:~-7,2%.txt"
and it works perfectly.
When I place the same statement into a CmdExec under SQL Agent, my date becomes a syntax error -- ("The filename, directory name, or volume label syntax is incorrect")
How do others handle this? Thanks.
Try using the SQL Server Agent tokens. They are described in MSDN article, "Use Tokens in Job Steps". The DATE token provides the current date in YYYYMMDD format. For your example use:
"...\Extracts\extractdata.$(ESCAPE_DQUOTE(DATE)).txt"
This isn't working for me
echo off
sqlcmd -m 1 -S 10.108.96.210\QA832 -U Exception -P Password1 -i E:\KCM_UAT\Exception.sql -o C:\Test_$(ESCAPE_DQUOTE(DATE)).txt -W -h-1 -s " "
set /p delExit=Press the ENTER key to exit...:
The file is written out like this
Test_$(ESCAPE_DQUOTE(DATE)).txt
What is incorrect about this syntax?
bcp transitschedule in calendar_dates.txt -T -f calenar_dates.fmt -F 2
I have tried this through sqlcmd and SSMS to a database on Azure. When I run the command I get:
Msg 156, Level 15, State 1, Line 1
Incorrect syntax near the keyword 'in'.
Yet, the examples here are not very much different syntactically. If I add database name or schema name error shifts towards the period.
Like SQLCMD, BCP is a stand-alone command-prompt utility that must be invoked from a command prompt. If the source text and format files reside on your client, you'll need to add the -S -U and -P parameters like you do with SQLCMD. For example:
bcp databasename.schemaname.transitschedule in calendar_dates.txt -f calenar_dates.fmt -F 2 /S azure-database-server /U azure-database_login /P azure-database-password
I'm trying to create a script just to run other scripts and do some extra stuff in case of successful or failure.
You have the full code on this link just to try to be clear what I'm trying to achieve.
Basically I want to:
-- DO SOME STUFF HERE
r: "$(fileName)"
-- MORE STUFF HERE
and call it from sqlcmd this way:
sqlcmd -i "RunScript.sql" -v fileName="someFileName.sql" -s server -d database
But I can't, I'm getting the following error:
Msg 102, Level 15, State 1, Server SERVER, Line 19 Incorrect syntax
near 'someFileName.sql'.
So, it seems that the little r: couldn't be used with a parameter on his side.
Just to clarify, someFileName.sql isn't in the SQL Server, but in my machine, so I couldn't use this way to read the file. In fact, I just tried it later.
Is there a workaround to archive this? Any ideas to solve it?
You could break your current RunScript.sql script into separate header and footer SQL scripts and concatenate them together with a SQL script in the middle that is denoted by an input parameter to a CMD script. For example:
RunSQL.CMD consists of:
#ECHO OFF
COPY /V /Y RunScriptHeader.sql + %1 + RunScriptFooter.sql RunScriptTemp.sql
SQLCMD -i "RunScriptTemp.sql" -s server -d database
The /V does a verify on the copy
The /Y suppresses prompting to confirm overwriting an existing file
You would run it as follows:
RunSQL.CMD someFileName.sql
i have looked all over the internet and cant seem to find a solution to this problem.
i am trying to output query results as a CSV through using a combination of sqlcmd and windows batch. here is what i have so far:
sqlcmd.exe -S %DBSERVER% -U %DBUSER% -P %DBPASS% -d %USERPREFIX% -Q "SELECT Username, UserDOB, UserGender FROM TABLE" -o %USERDATA%\%USERPREFIX%\FACT_BP.CSV -h-1 -s","
is there something i'm missing here? some setting that only looks at the first column of the query results?
any advice at all would be a huge help - i'm lost.
Here is the reference page from MSDN on SQLCMD.
http://technet.microsoft.com/en-us/library/ms162773.aspx
I placed this command in a batch file in C:\temp as go.bat.
sqlcmd -S(local) -E -dmaster
-Q"select cast(name as varchar(16)), str(database_id,1,0), create_date from sys.databases"
-oc:\temp\sys.databases.csv -h-1 -s,
Notice I hard coded the file name and removed the "" around the field delimiter.
I get the expected output below.
Either the command does not like the system variables or something else is wrong. Please try my code as a base line test. It works for SQL 2012.
Also, the number of lines is always dumped to file. You must clear this out of the file. That is why I do not use SQLCMD for ETL.
Why not use BCP instead?
I have writing several articles on my website.
http://craftydba.com/?p=1584