Convert mixed case value in colum into upper case values

Question

Hi All,
I have got a table call emp and has the following colums:
id number 10
suename_name varchar2(20)
given_name varchar2(20)
position varchar2(30)
date_hired date
Now the values in colums are mixed case and I need to change them to upper case. I think I need to use upper function to do it.
Could anyone help me by providing me sql/pl sql script to through each colum in emp table and change them to the upper case.
Many Thanks
Michael

Answer 1

Hi,
If you really want to try this using PL/SQL, then you'll have to use dynamic SQL, something like this untested code, since the table and column names are vaiable:
SET   SERVEROUTPUT  ON  SIZE 50000
DECLARE
     sql_txt     VARCHAR2 (4000);
BEGIN
     FOR  r  IN ( SELECT  table_name
               ,       column_name
               FROM    user_tab_columns
               WHERE   data_type = 'VARCHAR2'
          --   AND     ...  -- if you don't want all columns in all tables
     LOOP
          sql_txt := 'UPDATE "'
               || r.table_name
               || '" SET "'
               || r.column_name
               || '" = UPPER ("'
               || r.column_name
               || '")';
          dbms_output.put_line (sql_txt || ' = sql_txt');     -- For debugging
          -- EXECUTE IMMEDIATE sql_txt;               -- For real
     END LOOP;
END;Dynamic SQL (here) means creating a string that contains the SQL statement to be done, then using EXECUTE IMMEDIATE to run it.
Whenever you write dynamic SQL, it's a good idea to only display the SQL statement during the early stages of debugging. When it's displaying correctly, then you can try un-commenting the EXECUTE IMMEDIATE line.
Edited by: Frank Kulash on Jul 19, 2009 4:40 PM
A little more complicated, but a lot more efficient:
SET   SERVEROUTPUT  ON  SIZE 50000
DECLARE
     sql_txt          VARCHAR2 (4000);
BEGIN
     FOR  r  IN ( SELECT    table_name
               ,         column_name
               ,         ROW_NUMBER () OVER ( PARTITION BY  table_name
                                  ORDER BY       column_name
                                )     AS column_num
               ,            COUNT (1)     OVER ( PARTITION BY  table_name
                                )     AS total_column_cnt
               FROM      user_tab_columns
               WHERE     data_type = 'VARCHAR2'
          --   AND       ...  -- if you don't want all columns in all tables
               ORDER BY  table_name
               ,            column_name
     LOOP
          IF  r.column_num = 1
          THEN
               sql_txt := 'UPDATE "'
               || r.table_name
               || '" SET "';
          ELSE
               sql_txt := sql_txt || ', "';
          END IF;
          sql_txt := sql_txt ||
               || r.column_name
               || '" = UPPER ("'
               || r.column_name
               || '")';
          IF  r.column_num = r.total_column_cnt
          THEN     -- This is the last row for this table; run it
               dbms_output.put_line (sql_txt || ' = sql_txt');     -- For debugging
               -- EXECUTE IMMEDIATE sql_txt;               -- For real
          END IF;
     END LOOP;
END;
{code}
The difference is that the first solution produces and executes a separate UPDATE statement for each column, like this:
{code}
UPDATE "TABLE_1" SET "COLUMN_A" = UPPER ("COLUMN_A");
UPDATE "TABLE_1" SET "COLUMN_B" = UPPER ("COLUMN_B");
UPDATE "TABLE_1" SET "COLUMN_C" = UPPER ("COLUMN_C");
{code}
but it's much more efficient to do change all the columns at once, as long as you have the row in hand.  So the second solution only creates one SQL statement per table, like this:
{code}
UPDATE "TABLE_1" SET "COLUMN_A" = UPPER ("COLUMN_A")
           ,   "COLUMN_B" = UPPER ("COLUMN_B")
           ,   "COLUMN_C" = UPPER ("COLUMN_C");
{code} where every line above corresponds to a row fom the query.  The first line for every table will start with
{code}
UPDATE "TABLE_1" SET "but all the others will start with
{code}
and only on the last column for a given table will the statement be executed.