Error in unreachable SQL Code - sql

The following tsql fails:
IF OBJECT_ID('FDSCorp.XLFILES') IS NOT NULL
BEGIN
DELETE FROM FDSCorp.XLFILES;
INSERT INTO FDSCorp.XLFILES
SELECT DISTINCT * FROM dbo.XLFILES;
END
ELSE
exec sp_changeobjectowner XLFILES, FDSCorp;
Error:
The image data type cannot be selected as DISTINCT because it is not comparable.
Yes XLFilES has an image column, but in this case FDSCorp.XLFILES doesn't exist so that distinct code would never get to run.
This code is generated for each table in the database and I know that this section of the code will never be run on a table where it could fail due to the distinct issue.
I really don't want to have to overcomplicate the code checking for types which I can't use distinct with if that scenario could never happen in a real situation.
Is there some way I can bypass this check?

The only way to avoid the error is for you to prevent the server from "seeing" the code you don't want it to compile. Each batch is compiled entirely (including every statement, ignoring control flow) before execution starts:
IF OBJECT_ID('FDSCorp.XLFILES') IS NOT NULL
BEGIN
DELETE FROM FDSCorp.XLFILES;
exec sp_executesql N'INSERT INTO FDSCorp.XLFILES
SELECT DISTINCT * FROM dbo.XLFILES;';
END
ELSE
exec sp_changeobjectowner XLFILES, FDSCorp;
Now, when this batch is compiled, it won't attempt to compile the INSERT, since so far as this batch is concerned, it's just a string literal.

Related

Why is a query under a IF statement that is false running?

I have a application that uses a lot of string interpolation for SQL queries. I know it is a SQL injection threat, this is something that the customer and us know about and is hopefully something we can focus on next big refactor. I say that to make sense of the {Root Container.property} things that come from a GUI.
I have this query
IF ({Root Container.UserSelectedProduct}=1)
begin
DECLARE #TestNumbers {Root Container.SQLProductType};
INSERT INTO #TestNumbers SELECT * FROM {Root Container.DBTable};
SELECT *
FROM {Root Container.SQLProductFunction} (#TestNumbers)
WHERE [ID] = {Root Container.Level};
end
else
Select 0
Before a user selects a product it looks like this
IF (0=1)     
BEGIN
DECLARE #TestNumbers myDataType;
INSERT INTO #TestNumbers SELECT * FROM [MySchema].[TheWrongTable];     
SELECT * FROM [dbo].[myfunction] (#TestNumbers)
WHERE [ID] = 1;
END
ELSE
SELECT 0
Which is giving me the error:
Column name or number of supplied values does not match table definition.
I am aware why this error shows up, the table I am selecting from is not made for that data type.
However, why is it even attempting to run the first IF clause when I have IF (0=1) - how come this part is not just skipped and the SELECT 0 is only run? I would have thought that is how it was supposed to work, but I keep getting the error regarding column name/number not matching the table definition. When the user does select a Product and I get IF (1=1) and I have the appropriate table/function/datatype, it all works smoothly. I just don't know why it throws me an error prior when IF(1=0). Why does this happen/how can I get my intended behavior that everything inside my BEGIN\END under my first IF statement does not run unless the expression is true.
T-SQL is not interpreted. It must make sense regardless of what the runtime conditions are. It doesn't even do short-circuiting, in fact. Your code is invalid, and it doesn't matter that it's unreachable - T-SQL isn't going to ignore a piece of invalid code just because it could be eliminated, that's a thing that is a common source of bugs (e.g. in C++ where it's pretty common with templates).
Just make sure you still get valid SQL for the case where no product is selected; use the wrong table (or a helper table) if you have to.
The answer is simple: SQL code is fully compiled by the server before being executed, so this is basically a compile error. It's a bit like trying to compile the following in C#
if(someBoolWhichIsFalse)
intValue = "hello";
It's simply not valid.
The runtime code has not even been executed, it's still in the parsing and lexing stage. Nothing is being skipped, it just needs to be fully valid code, irrespective of runtime conditions.
This happens in every scope, i.e. on every call to a procedure or ad-hoc batch, that code must be compilable.

Detect if SQL statement is correct

Question: Is there any way to detect if an SQL statement is syntactically correct?
Explanation:
I have a very complex application, which, at some point, need very specific (and different) processing for different cases.
The solution was to have a table where there is a record for each condition, and an SQL command that is to be executed.
That table is not accessible to normal users, only to system admins who define those cases when a new special case occurs. So far, a new record was added directly to the table.
However, from time to time there was typos, and the SQL was malformed, causing issues.
What I want to accomplish is to create a UI for managing that module, where to let admins to type the SQL command, and validate it before save.
My idea was to simply run the statement in a throw block and then capture the result (exception, if any), but I'm wondering of there is a more unobtrusive approach.
Any suggestion on this validation?
Thanks
PS. I'm aware of risk of SQL injection here, but it's not the case - the persons who have access to this are strictly controlled, and they are DBA or developers - so the risk of SQL injection here is the same as the risk to having access to Enterprise Manager
You can use SET PARSEONLY ON at the top of the query. Keep in mind that this will only check if the query is syntactically correct, and will not catch things like misspelled tables, insufficient permissions, etc.
Looking at the page here, you can modify the stored procedure to take a parameter:
CREATE PROC TestValid #stmt NVARCHAR(MAX)
AS
BEGIN
IF EXISTS (
SELECT 1 FROM sys.dm_exec_describe_first_result_set(#stmt, NULL, 0)
WHERE error_message IS NOT NULL
AND error_number IS NOT NULL
AND error_severity IS NOT NULL
AND error_state IS NOT NULL
AND error_type IS NOT NULL
AND error_type_desc IS NOT NULL )
BEGIN
SELECT error_message
FROM sys.dm_exec_describe_first_result_set(#stmt, NULL, 0)
WHERE column_ordinal = 0
END
END
GO
This will return an error if one exists and nothing otherwise.

USE DB that may not exist

I have a script that has a USE DATABASE statement.
The script runs perfectly fine if the database exists. If it doesn't exist, it fails with the message "the database doesn't exist", which makes perfect sense.
Now, I don't it to fail so I added a check to select if the DB exists on sys.databases (which I will represent here with a IF 1=2 check for the sake of simplicity), so, if the DB exists (1=1), then run the "use" statement.
To my surprise, the script kept failing. So I tried to add a TRY CATCH block. Same result.
It seems that the use statement is evaluated prior to anything else, which id quite annoying because now my script may break.
So my question is: how can I have an use statement on a script to a database that may not exist?
BEGIN TRY
IF (1=1) BEGIN --if DB exists
USE DB_THAT_MAY_NOT_EXIST
END
END TRY
BEGIN CATCH
END CATCH
I don't believe you can do what you want to do. The documentation specifies that use is executed at both compile time and execution time.
As such, use on a database that does not exist is going to create a compile time error. I am not aware of a way to bypass compile time errors.
As another answer suggests, use the database qualifier in all your names.
You can also check if a database exists, without switching to it. Here is one way:
begin try
exec('use dum');
print 'database exists'
end try
begin catch
print 'database does not exist'
end catch
How about this? May be you could check in this way.
if db_id('dbname') is not null
-- do stuff
or try this:
if not exists(select * from sys.databases where name = 'dbname')
-- do stuff
So for table:
if object_id('objectname', 'somename') is not null
or
sp_msforeachdb ‘select * from ?.sys.tables’
Reference
Off the top of my head, you could fully qualify all your references to avoid the USE statement.
I hope someone comes up with a solution that requires less PT.
After doing your check to see if the DB exists, instead of
SELECT Moo FROM MyTable
use
SELECT Moo FROM MyDB.MySchema.MyTable

Select Fails With Nonexisitent Columns

Executing the following statement with SQL Server 2005 (My tests are through SSMS) results in success upon first execution and failure upon subsequent executions.
IF OBJECT_ID('tempdb..#test') IS NULL
CREATE TABLE #test ( GoodColumn INT )
IF 1 = 0
SELECT BadColumn
FROM #test
What this means is that something is comparing the columns I am accessing in my select statement against the columns that exist on a table when the script is "compiled". For my purposes this is undesirable functionality. My question is if there is anything that can be done so that this code would execute successfully on every run, or if that is not possible perhaps someone could explain why the demonstrated functionality is desirable. The only solutions I have currently is to wrap the select with EXEC or select *, but I don't like either of those solution.
Thanks
If you put:
IF OBJECT_ID('tempdb..#test') IS NOT NULL
DROP TABLE #test
GO
At the start, then the problem will go away, as the batch will get parsed before the #test table exists.
What you're asking is for the system to recognise that "1=0" will always evaluate to false. If it were ever true (which could potentially be the case for most real-life conditions), then you'd probably want to know that you were about to run something that would cause failure.
If you drop the temporary table and then create a stored procedure that does the same:
CREATE PROC dbo.test
AS
BEGIN
IF OBJECT_ID('tempdb..#test') IS NULL
CREATE TABLE #test ( GoodColumn INT )
IF 1 = 0
SELECT BadColumn
FROM #test
END
Then this will happily be created, and you can run it as many times as you like.
Rob
Whether or not this behaviour is "desirable" from a programmer's point of view is debatable of course -- it basically comes down to the difference between statically typed and dynamically typed languages. From a performance point of view, it's desirable because SQL Server needs complete information in order to compile and optimize the execution plan (and also cache execution plans).
In a word, T-SQL is not an interpretted or dynamically typed language, and so you cannot write code like this. Your options are either to use EXEC, or to use another language and embed the SQL queries within it.
This problem is also visible in these situations:
IF 1 = 1
select dummy = GETDATE() into #tmp
ELSE
select dummy = GETDATE() into #tmp
Although the second statement is never executed the same error occurs.
It seems the query engine first level validation ignores all conditional statements.
You say you have problems with subsequent request and that is because the object already exits. It it recommended that you drop your temporary tables as soon as possible when you are done with it.
Read more about temporary table performance at:
SQL Server performance.com

Why does a T-SQL block give an error even if it shouldn't even be executed?

I was writing a (seemingly) straight-forward SQL snippet that drops a column after it makes sure the column exists.
The problem: if the column does NOT exist, the code inside the IF clause complains that it can't find the column! Well, doh, that's why it's inside the IF clause!
So my question is, why does a piece of code that shouldn't be executed give errors?
Here's the snippet:
IF exists (select * from syscolumns
WHERE id=object_id('Table_MD') and name='timeout')
BEGIN
ALTER TABLE [dbo].[Table_MD]
DROP COLUMN timeout
END
GO
...and here's the error:
Error executing SQL script [...]. Invalid column name 'timeout'
I'm using Microsoft SQL Server 2005 Express Edition.
IF exists (select * from syscolumns
WHERE id=object_id('Table_MD') and name='timeout')
BEGIN
DECLARE #SQL nvarchar(1000)
SET #SQL = N'ALTER TABLE [dbo].[Table_MD] DROP COLUMN timeout'
EXEC sp_executesql #SQL
END
GO
Reason:
When Sql server compiles the code, they check it for used objects ( if they exists ). This check procedure ignores any "IF", "WHILE", etc... constructs and simply check all used objects in code.
It may never be executed, but it's parsed for validity by Sql Server. The only way to "get around" this is to construct a block of dynamic sql and then selectively execute it
Here's how I got it to work:
Inside the IF clause, I changed the ALTER ... DROP ... command with exec ('ALTER ... DROP ...')
It seems the SQL server does a validity check on the code when parsing it, and sees that a non-existing column gets referenced somewhere (even if that piece of code will never be executed).
Using the exec(ute) command wraps the problematic code in a string, the parser doesn't complain, and the code only gets executed when necessary.
Here's the modified snippet:
IF exists (select * from syscolumns
WHERE id=object_id('Table_MD') and name='timeout')
BEGIN
exec ('ALTER TABLE [dbo].[Table_MD] DROP COLUMN timeout')
END
GO
By the way, there is a similar issue in Oracle, and a similar workaround using the "execute immediate" clause.