Go Premium for a chance to win a PS4. Enter to Win

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 1023
  • Last Modified:

Int and VarChar types in where clause

hi,

I have the below Stored proc which passes in the field name and values to filter on. It works great. However, some of the columns are of type int and others are varchar. The varchar columns need the values in quotes, but the int columns do not. Therefore the below only works for varchars.

Is there any easy fix? Thanks
CREATE PROC test
@columName  varchar(30),
@value  varchar(30)
AS
BEGIN
    EXECUTE(
    'SELECT     D.id_desk,
                D.id_region           
    FROM        DESK D,
                INSTRUMENT I
    WHERE '      + @columName + ' LIKE ''' + @value +'''
    AND         I.id_desk = D.id_desk' )
END
go

Open in new window

0
bowemc
Asked:
bowemc
  • 4
  • 2
2 Solutions
 
IncisiveOneCommented:
Hang on, that does not "work great".  This sort of thing is completely incorrect.  First an explanation.  Actually, two explanations.

1  SQL is sometimes called a loosely-typed language, because there are few enforcements of the Datatypes in the language itself.  That is incorrect.  The whole idea is that SQL is a DATABASE manipulation language, and the DATABASE is not loosely-typed, it is fixed (the Datatypes are set once, and they do not change.  The correct understanding is, SQL is actually a tightly-typed language, the Typing is in the database.  Therefore it is incumbent upon you, all coders, to always be aware of the Datatype of the column, and to treat it accordingly; using the correct Datatype wherever the column is referenced.

Note that Datatype mismatches are probably the most common mistake, and they always lead to poor performance (eg. Sybase cannot use the index on an indexe column, when the wrong Datatype is used, and thus has to table scan).  This is easily confirmed in the showplan, which every coder should produce and examine for every code segment.  Datatypes mismatches are also the easiest mistakes to avoid; by having good simple standards:

Always use the correct Datatype for the column.
1.1  Therefore you cannot do what you are trying to do (write a generic SQL statement [forget the proc for now] that can perform a LIKE on any column).

1.2  Like is for char and varchar only, you cannot perform like on int, smallint, etc.

2  I appreciate that you are learning SQL, stored procs, and how to go about creating the various objects you need; that this is test proc.  However, in order for you to learn the correct methods (and to identify the incorrect methods), I will treat this proc as real (not test), and identify the issues.

2.1  The use of execute here is superfluous, and can be removed.  The code will  perform better, and it would lead you to a better understanding of the problem/solution.  Stated another way: take the execute out, get the code working without it, and if you need to, after you have working code, then put it back in.

2.2  Thing about this.  Regardless of what you write, immediate code, ordinary code, dynamic SQL, stored procs, the Optimiser has to (a) parse it (b) compile it and then (c) execute it.  Therefore the simple straight SQL is the best, to write and test.  Once you have that working correctly, then place it in dynamic or immediate code.  That means you will have specific code segments for each table or table combination; generic SQL in an application is not a reasonable goal.  Generic SQL is qiuite reasonable for utilities, but that is a different story, not addressed here; by definition, your application code is not an utility.

2.3  You do not have any error checking or column Datatype checking, which you must have

2.4  Remember the Datatype of the column in the database is fixed, and that Datatype is the final truth about the column, so that is what our code should be based on.  It should not be based on "what we want to do", it should be based on "what the column is", and therefore what is possible on the column.

Cheers



CREATE PROC
    DeskInstrument_search
        @columnName  varchar(30) = NULL,
        @searchStr   varchar(30) = NULL
AS
    IF @columnName = NULL
        BEGIN
        PRINT "You must provide a @columnName"
        -- cannot search a column without knowing which col to search
        GOTO EXIT_ERR
        END
 
    -- check if column is known var/char column
    -- exclude int, numeric, date, etc
    IF  @columnName != "deskName" AND 
        @columnName != "deskDescription" AND
        @columnName != "instrName" AND
        @columnName != "instrDescription" AND
        @columnName != "instrAddress1" AND
        @columnName != "instrAddress2"
        BEGIN
        PRINT "@columnName must be a var/char column in DESK or INSTRUMENT"
        GOTO EXIT_ERR
        END
 
    IF CHARINDEX( "%", @searchStr ) = 0  -- et cetera
        BEGIN
        PRINT "@searchStr must be a valid LIKE operand"
        -- otherwise we cannot perform LIKE with it
        GOTO EXIT_ERR
        END
 
    IF @searchStr = NULL
        SELECT  D.id_desk,
                D.id_region           
            FROM  DESK D,
                  INSTRUMENT I
            WHERE I.id_desk = D.id_desk
    ELSE
        SELECT  D.id_desk,
                D.id_region           
            FROM  DESK D,
                  INSTRUMENT I
            WHERE I.id_desk = D.id_desk
            AND   @columnName LIKE @searchStr
 
    IF @@rowcount = 0
        BEGIN
        PRINT "Search string %1! not found in column %1!", @searchStr, @columnName
        END
 
EXIT_OK:
    RETURN 0
 
EXIT_ERR:
    RETURN -1

Open in new window

0
 
IncisiveOneCommented:
Er, that's for discussion/progress, obviously the code will not work without the execute.
0
 
alpmoonCommented:
What about this? It would convert numeric columns into char and then search the value:

CREATE PROC test
@columName  varchar(30),
@value  varchar(30)
AS
BEGIN
  IF exists(select * from syscolumns
                  where id in (object_id('DESK), object_id(' INSTRUMENT'))
                       and type in (39, 47) -- char or varchar
    EXECUTE(
    'SELECT     D.id_desk,
                D.id_region          
    FROM        DESK D,
                INSTRUMENT I
    WHERE '      + @columName + ' LIKE ''' + @value +'''
    AND         I.id_desk = D.id_desk' )

  ELSE

    EXECUTE(
    'SELECT     D.id_desk,
                D.id_region          
    FROM        DESK D,
                INSTRUMENT I
    WHERE convert(char,'      + @columName + ') LIKE ''' + @value +'''
    AND         I.id_desk = D.id_desk' )

END
go
0
Vote for the Most Valuable Expert

It’s time to recognize experts that go above and beyond with helpful solutions and engagement on site. Choose from the top experts in the Hall of Fame or on the right rail of your favorite topic page. Look for the blue “Nominate” button on their profile to vote.

 
alpmoonCommented:
Have you tried it? Is it useful for your end?
0
 
alpmoonCommented:
I think my solution is more flexible and efficient. If table structure is changed, the procedure doesn't need to be modified.

I suggest split between me and IncisiveOne
0
 
alpmoonCommented:
I think my solution is more flexible and efficient. If table structure is changed, the procedure doesn't need to be modified.

I suggest split between me and IncisiveOne
0

Featured Post

How to Use the Help Bell

Need to boost the visibility of your question for solutions? Use the Experts Exchange Help Bell to confirm priority levels and contact subject-matter experts for question attention.  Check out this how-to article for more information.

  • 4
  • 2
Tackle projects and never again get stuck behind a technical roadblock.
Join Now