Pivoting Rows into Columns Dynamically in Oracle

Oracle Dynamic Pivoting

Using dynamic sql for a result where the columns are unknown at the time of executing is a bit of a hassle in Oracle compared to certain other RDMBS.

Because the record type for the output is yet unknown, it can't be defined beforehand.

In Oracle 11g, one way is to use a nameless procedure that generates a temporary table with the pivoted result.

Then select the results from that temporary table.

declare
  v_sqlqry clob;
  v_cols clob;
begin
  -- Generating a string with a list of the unique names
  select listagg(''''||CCL||''' as "'||CCL||'"', ', ') within group (order by CCL)
  into v_cols
  from 
  (
    select distinct CCL
    from tableA
  );

  -- drop the temporary table if it exists
  EXECUTE IMMEDIATE 'DROP TABLE tmpPivotTableA';
  EXCEPTION WHEN OTHERS THEN IF SQLCODE != -942 THEN RAISE; END IF;

  -- A dynamic SQL to create a temporary table 
  -- based on the results of the pivot
  v_sqlqry := '
    CREATE GLOBAL TEMPORARY TABLE tmpPivotTableA
    ON COMMIT PRESERVE ROWS AS
    SELECT * 
    FROM (SELECT ID, CCL, Flag FROM TableA) src 
    PIVOT (MAX(Flag) FOR (CCL) IN ('||v_cols||')) pvt';

  -- dbms_output.Put_line(v_sqlqry); -- just to check how the sql looks like
  execute immediate v_sqlqry;
  
end;
/

select * from tmpPivotTableA;

Returns:

ID  adam john rob terry
--  ---- ---- --- -----
1   x    x    x
2        x      

You can find a test on db<>fiddle here

In Oracle 11g, another cool trick (created by Anton Scheffer) to be used can be found in this blog. But you'll have to add the pivot function for it.

The source code can be found in this zip

After that the SQL can be as simple as this:

select * from 
table(pivot('SELECT ID, CCL, Flag FROM TableA'));

You'll find a test on db<>fiddle here

Pivot with Dynamic Columns in Oracle

SYS_REFCURSOR within a stored function(eg. PL/SQL used instead of using SQL directly) might be used in order to get dynamically generated result set(eg. Dynamic Pivot). In this case, a string for Conditional Aggregation is generated :

CREATE OR REPLACE FUNCTION get_passengers_rs RETURN SYS_REFCURSOR IS
  v_recordset SYS_REFCURSOR;
  v_sql       VARCHAR2(32767);
  v_str       VARCHAR2(32767);
BEGIN
  SELECT LISTAGG('MAX(CASE WHEN rn = '||lvl||' THEN age||''(''||passengers||'')'' END) 
                   AS "Age'||lvl||'"' ,',') WITHIN GROUP (ORDER BY 0)
    INTO v_str
    FROM ( SELECT level AS lvl  
             FROM dual
          CONNECT BY level <= (SELECT MAX(COUNT(*)) FROM t GROUP BY ID ) ) t;  

  v_sql :=
  'SELECT ID, '|| v_str ||'
     FROM 
     (
      SELECT t.*, 
             ROW_NUMBER() OVER (PARTITION BY ID ORDER BY 0) AS rn
        FROM t     
      )
    GROUP BY ID';

  OPEN v_recordset FOR v_sql;
  RETURN v_recordset;
END;

I also added the names of passengers in order to distinguish each data well.

Then run the below code :

VAR rc REFCURSOR
EXEC :rc := get_passengers_rs;
PRINT rc

from SQL Developer's Command Line in order to see the expected result set.

Above code generates this SQL string(v_sql) for the currently existing data

SELECT ID, MAX(CASE WHEN rn = 1 THEN age||'('||passengers||')' END) AS "Age1",
           MAX(CASE WHEN rn = 2 THEN age||'('||passengers||')' END) AS "Age2",
           MAX(CASE WHEN rn = 3 THEN age||'('||passengers||')' END) AS "Age3"
  FROM 
     (
      SELECT t.*, 
             ROW_NUMBER() OVER (PARTITION BY ID ORDER BY 0) AS rn
        FROM t     
      )
 GROUP BY ID

which yields

ID      Age1        Age2            Age3
123456  58(Marie)   65(Ben) 
123458  32(Aaron)   18(Caroline)    37(Stephanie)

as result set.

Use pivot for dynamically changing column headers using sql in oracle

What you can do

SELECT * 
  FROM data  
 PIVOT  
 (
  MAX(value) FOR label IN ('A' AS "A", 'B' AS "B",'C' AS "C",'D' AS "D")
 )
 WHERE ID = 120 AND app_id = 1

as a static pivot statement might be converted to a function which contains two respective parameters

CREATE OR REPLACE FUNCTION Get_Pivoted_Labels( i_id data.id%type, i_app_id data.app_id%type ) 
RETURN SYS_REFCURSOR IS
  v_recordset SYS_REFCURSOR;
  v_sql       VARCHAR2(32767); 
  v_cols      VARCHAR2(32767);  
BEGIN
  SELECT LISTAGG( ''''||label||''' AS "'||label||'"' , ',' )
          WITHIN GROUP ( ORDER BY label ) 
    INTO v_cols
    FROM ( SELECT DISTINCT label 
             FROM data
            WHERE ID = i_id AND app_id = i_app_id );

  v_sql :=
      'SELECT * 
         FROM data
        PIVOT 
        (
          MAX(value) FOR label IN ( '|| v_cols ||' )
        )
       WHERE ID = :id AND app_id = :aid'; 

  OPEN v_recordset FOR v_sql USING i_id, i_app_id;
  RETURN v_recordset;
END;
/

in which an auxiliary query, in which the label columns are distinctly selected, is used to determine the string(v_cols for 'A' AS "A", 'B' AS "B",'C' AS "C",'D' AS "D") to be concatenated to the main SQL string in order to be used within the cursor which returns a value of type SYS_REFCURSOR.

and is invoked by

VAR rc REFCURSOR
VAR v_id NUMBER
VAR v_app_id NUMBER
EXEC :rc := Get_Pivoted_Labels(:v_id,:v_app_id);
PRINT rc

from SQL developer's console.

Demonstration with generated SQL statements

If order of columns in the SELECT list matters, then use the code below in order to create the function

CREATE OR REPLACE FUNCTION Get_Pivoted_Labels( i_id data.id%type, i_app_id data.app_id%type ) 
RETURN SYS_REFCURSOR IS
  v_recordset SYS_REFCURSOR;
  v_sql       VARCHAR2(32767); 
  v_cols_1    VARCHAR2(32767);    
  v_cols_2    VARCHAR2(32767);   
BEGIN
  SELECT LISTAGG( ''''||label||''' AS "'||label||'"' , ',' )
          WITHIN GROUP ( ORDER BY label ),
         LISTAGG( label , ',' )
          WITHIN GROUP ( ORDER BY label )  
    INTO v_cols_1, v_cols_2
    FROM ( SELECT DISTINCT label, value 
             FROM data
            WHERE ID = i_id AND app_id = i_app_id );

  v_sql :=
      'SELECT ID, '|| v_cols_2 ||', app_id
         FROM data
        PIVOT 
        (
          MAX(value) FOR label IN ( '|| v_cols_1 ||' )
        )
       WHERE ID = :id AND app_id = :aid'; 

  OPEN v_recordset FOR v_sql USING i_id, i_app_id;
  RETURN v_recordset;
END;
/

Pivoting rows into columns Oracle SQL

If you know the columns you want, you can use conditional aggregation:

select insertdate, id,
       max(case when PROPERTYNAME = 'ItemNumber' then propertyvalue end) as ItemNumber,
       max(case when PROPERTYNAME = 'ItemRev' then propertyvalue end) as ItemRev,
       max(case when PROPERTYNAME = 'BarCodeNumber' then propertyvalue end) as BarCodeNumber
from t
group by insertdate, id;

If you don't know all the properties up-front, then you need to construct the query dynamically as a string and use execute immediate.

Dynamic Pivot in Oracle's SQL

You can't put a non constant string in the IN clause of the pivot clause.

You can use Pivot XML for that.

From documentation:

subquery A subquery is used only in conjunction with the XML keyword.
When you specify a subquery, all values found by the subquery are used
for pivoting

It should look like this:

select xmlserialize(content t.B_XML) from t_aa
pivot xml(
sum(A) for B in(any)
) t;

You can also have a subquery instead of the ANY keyword:

select xmlserialize(content t.B_XML) from t_aa
pivot xml(
sum(A) for B in (select cl from t_bb)
) t;

Here is a sqlfiddle demo

---

Related Topics