case of condition in where clause

Hello

I am a beginner in SQL, but according to my reading of the manual, this syntax should be acceptable. I'm using application express but tested also in Developer SQL and the conclusion that it fails with an error "invalid relational operator", with the line number pointing to the ELSE clause of the condition, but a number of column pointing to white space.

This is a select statement nested, but that shouldn't matter. The fragment in question is:

Select columns
RES
where res.villaid =: P605_VILLAID
and INSTR (: P604_RES_STATES, res.states). = 0
and (CASE: P604_EXCLUDE_ZERO)
WHEN 'Y' THEN ' res.rate > 0'
ELSE ' res.rate > = 0'
END)
and res. DEPARTDATE >: P605_STARTDATE
and res. ARRIVEDATE < =: P605_ENDDATE.

I hope that this puts ok shaped during the validation. The code is looking for bookings within a range of dates (which works very well when the other conditions are discussed), and should only output offset lines (nights) if the user asks for it in the variable of liaison P604_EXCLUDE_ZERO. There are probably other ways to go about this, but I'm too stubborn to renounce it. Well, not yet anyway.

Thanks and greetings
CS

select some columns
  from res
 where res.villaid = :p605_villaid
       and instr (:p604_res_states, res.states) != 0
       and sign (res.rate) >= case when :p604_exclude_zero = 'Y' then 1 else 0 end
       and res.departdate > :p605_startdate
       and res.arrivedate <= :p605_enddate

Tags: Database

Similar Questions

Tuning - using the CASE statement in the WHERE clause of the query

Hi all

My request has been changed to use a CASE statement in the WHERE clause to examine the data to some columns based on a parameter value. This modified request is made a full table and constantly running scan. Please suggest what can be done to improve its performance:

Query:

SELECT LAST_DAY(TRUNC(TO_TIMESTAMP(os.requestdatetime, 'yyyymmddhh24:mi:ss.ff4'))) AS summary_date,
os.acctnum,
os.avieworigin_refid,
COUNT(1) cnt_articleview,
SUM(NVL(autocompletedterm,0)) cnt_autocompletedterm
FROM TABLE1 os
WHERE os.acctnum IS NOT NULL
AND os.avieworigin_refid IS NOT NULL
AND os.requestdatetime IS NOT NULL
AND UPPER(os.success_ind) = 'S'

AND CASE WHEN
Param_ValueToCheck  = 'FULL' AND get_date_timestamp(os.requestdatetime)  BETWEEN 
TO_DATE('01-MAY-2011 00:00:00','DD-MON-YYYY HH24:MI:SS') AND
TO_DATE('31-MAY-2011 00:00:00','DD-MON-YYYY HH24:MI:SS')
THEN 1
WHEN
Param_ValueToCheck  = 'INCR' AND os.entry_createddate  BETWEEN 
TO_DATE('01-MAY-2011 00:00:00','DD-MON-YYYY HH24:MI:SS') AND
TO_DATE('31-MAY-2011 00:00:00','DD-MON-YYYY HH24:MI:SS')
THEN 1
END = 1
AND CASE WHEN
Param_ValueToCheck  = 'FULL' AND os.entry_CreatedDate BETWEEN 
TO_DATE('01-APR-2011 00:00:00','DD-MON-YYYY HH24:MI:SS') AND
TO_DATE('07-JUN-2011 00:00:00','DD-MON-YYYY HH24:MI:SS')
THEN 1
WHEN
Param_ValueToCheck  = 'INCR' THEN 1
END = 1
GROUP BY LAST_DAY(TRUNC(TO_TIMESTAMP(os.requestdatetime, 'yyyymmddhh24:mi:ss.ff4'))), os.acctnum,os.avieworigin_refid;

Description of the table:
(Number of lines: approx. > amount 600 000 000)

Name                           Null     Type       
------------------------------ -------- ------------
ARTICLEID                      NOT NULL NUMBER(20) 
USERKEY                                 NUMBER(10) 
AVIEWORIGIN_REFID                       VARCHAR2(10)
SUCCESS_IND                             VARCHAR2(2)
ENTRY_CREATEDDATE                       DATE       
CREATED_BY                              VARCHAR2(10)
FILENUMBER                              NUMBER(10) 
LINENUMBER                              NUMBER(10) 
ACCTNUM                                 VARCHAR2(10)
AUTOCOMPLETEDTERM                       NUMBER(2)  
REQUESTDATETIME                         VARCHAR2(19)

Explain Plan

SQL> select * from table(dbms_xplan.display);

PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Plan hash value: 2224314832

----------------------------------------------------------------------------------------------------------
| Id  | Operation            | Name              | Rows  | Bytes | Cost (%CPU)| Time     | Pstart| Pstop |
----------------------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT     |                   |   590 | 33040 |  2501K  (1)| 08:20:15 |       |       |
|   1 |  HASH GROUP BY       |                   |   590 | 33040 |  2501K  (1)| 08:20:15 |       |       |
|   2 |   PARTITION RANGE ALL|                   |   590 | 33040 |  2501K  (1)| 08:20:15 |     1 |1048575|
|*  3 |    TABLE ACCESS FULL | TABLE1 |   590 | 33040 |  2501K  (1)| 08:20:15 |     1 |1048575|
----------------------------------------------------------------------------------------------------------


PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------

   3 - filter(UPPER("OS"."SUCCESS_IND")='S' AND CASE  WHEN ('FULL'='FULL' AND
              "OS"."ENTRY_CREATEDDATE">=TO_DATE(' 2011-04-01 00:00:00', 'syyyy-mm-dd hh24:mi:ss') AND
              "OS"."ENTRY_CREATEDDATE"<=TO_DATE(' 2011-06-07 00:00:00', 'syyyy-mm-dd hh24:mi:ss')) THEN 1 WHEN
              'FULL'='INCR' THEN 1 END =1 AND "OS"."REQUESTDATETIME" IS NOT NULL AND CASE  WHEN ('FULL'='FULL'
              AND "ODS"."GET_DATE_TIMESTAMP"("REQUESTDATETIME")>=TO_DATE(' 2011-05-01 00:00:00', 'syyyy-mm-dd
              hh24:mi:ss') AND "ODS"."GET_DATE_TIMESTAMP"("REQUESTDATETIME")<=TO_DATE(' 2011-05-31 00:00:00',
              'syyyy-mm-dd hh24:mi:ss')) THEN 1 WHEN ('FULL'='INCR' AND "OS"."ENTRY_CREATEDDATE">=TO_DATE('
              2011-05-01 00:00:00', 'syyyy-mm-dd hh24:mi:ss') AND "OS"."ENTRY_CREATEDDATE"<=TO_DATE('

PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
              2011-05-31 00:00:00', 'syyyy-mm-dd hh24:mi:ss')) THEN 1 END =1 AND "OS"."ACCTNUM" IS NOT NULL AND
              "OS"."AVIEWORIGIN_REFID" IS NOT NULL)

Published by: Chaitanya on June 9, 2011 02:44

Published by: Chaitanya on June 9, 2011 02:47

When it is executed individually, MERGE LOGIC 1 and LOGIC of MERGE 2 take about ten for a daterange of 30 days data.

BUT FULL is to get the scores of APR - JUN, INCR becomes partitions for MAY so not both 30 days.
Are compare us like with like?

Case statement in a WHERE clause
Gurus,

Im trying to avoid Union of several select statements using a CASE inside a WHERE clause. When the parameter is defined as "What's new" it is advisable to use a code and for "Update Items" another condition.

Having a problem with the following... I know that BOX in WHEREs instructions are allowed only can not make it work for this one.
```
and case  when 'New Items' = 'xxx' --p_item_status 
         then  msi.creation_date = msi.last_update_date
      when 'Updated Items' = 'bbb' --p_item_status  
         then  msi.creation_date != msi.last_update_date
      else 1=1
end;  
```
Points for correct and helpful answers!

Published by: sreese on March 27, 2012 17:51
Hello

The great thing about CASE expressions, is that they allow you to use a login IF-THEN-ELSE in the clause SELECT, the ORDER BY clause or anywhere else. The WHERE clause has its own way to IF-THEN-ELSE logic, so even if you can use a CASE expression in a clause WHERRE, it usually does not help anything. You can do somehting like this:
```
WHERE     (     :p_item_status        = 'New Items'
     AND     msi.creation_date  = msi.last_update_date
     )
OR     (     :p_item_status        = 'Updated Items'
     AND     msi.creation_date  != msi.last_update_date
     )
OR     (     :p_item_status        NOT IN ( 'New Items'
                           , 'Updated Items'
                           )
     )
```
This guess: p_item_status is not NULL.
If: p_item_status can be NULL, and you want to include all the lines where it is, and then change the last condition to
```
OR     ( NVL (:p_item_status, 'OK') NOT IN ( 'New Items'
                             , 'Updated Items'
                             )
     )
```
I hope that answers your question.
If not, post a small example of data (CREATE TABLE and INSERT, only relevant columns instructions), some values of the parameter and the expected results of these data for each value of the parameter.
Explain, using specific examples, how you get these results from these data.
Always tell what version of Oracle you are using.

A CASE statement with additional where clause
Hi all

I need assistance in which this obligation clause

Table:-balance_table

Columns:-balance_type, balance_amount, balance_month, budget_name

balance_type column had given 'Real' OR 'Budget '.

Now I am trying to extract data such as: if: entered_month is Mar-2009 then before and including Mar-2009 will show real balance sheet Date and after Mar-2009 will show budget data.

It works fine, I need to add a condition more restrict the budget name, there are several budget_name in the table for balance_type = 'Budget '.

SELECT SUM (balance_amount), balance_type, balance_month
OF balance_table
WHERE balance_type =
(CASE
WHEN balance_month < =: entered_month
THEN 'real '.
ANOTHER 'Budget '.
END
)
AND budget_name = 'BUDGET1.
Balance_type GROUP, balance_month

Stated above is erroneous because ' AND budget_name = ' BUDGET1 "clause restricts the set of data."

Please help in contrcuting where clause 1) to sort the data according to Budget/actual and 2) for specific budgets, so budget_type = Budget

Thank you
Bobin
That should do it.
```
SELECT SUM (balance_amount),
       balance_type,
       balance_month
FROM balance_table
WHERE (balance_type = 'Actual'
       and balance_month <= :entered_month
      )
or (balance_type = 'Budget'
    and balance_month > :entered_month
    and budget_name = :entered_budget_name
   )
GROUP BY balance_type,
         balance_month
ORDER BY 2
```
See you soon
Sarma.

Using a CASE statement in 'IN' where clause clause

Hello
I have a form that users see after they connect with a few items and they type or select values based on the element type. All reports in the application use these values of PAGE 1 in where clause to filter the lines.
One of these question is "Quarter". I have a group of radio buttons on the PAGE 1 on the P1_QTR point.
I need to provide users a way to take the last 4, last 3, last 2 or current quarter and I need to dynamically build a clause WHERE the interactive report based on the selection in the form.
I am trying to use the CASE as shown below and get the syntax error.
AND QTR to (when BOX: P1_QTR = 'CURRENT' then ("T4"))
When: P1_QTR = "LAST" then ("Q3")
When: P1_QTR = 'LAST TWO' then ('Q3, "Q4")
When: P1_QTR = 'LAST THREE' then ("Q2", "Q3,' T4 ')
another ("T1", "T2", "Q3,' T4 ')
END)
How can I achieve this?

Following a simple logic, try

and QTR to

(

Select 'T4' double where: P1_QTR <> 'LAST '.

Union of all the

Select 'Q3' to double where: P1_QTR <> "in PROCESS".

Union of all the

Select "Q2" double where: P1_QTR not in ('CURRENT', 'LAST', "LAST TWO")

Union of all the

Select "Q1" double where: P1_QTR not in ('CURRENT', 'LAST', 'LAST TWO', 'THREE LATEST')

)

Case instructions within a Where clause clause

Hello group,
I know this has been asked several times, but I do not enter simply to a tuition assistance box in a WHERE clause. So I need help:
My current WHERE the clause reads:
A16. FULL_DATE between (SELECT (TRUNC (SysDate - 8)) From Dual) and (select (TRUNC (SysDate - 2)) double)
However, I need to 'automate' a bit, based on the current time/day of the month. I need my instruction box to say:
(Case when to_char (sysdate-3, 'mm') <>to_char (sysdate, 'mm')
then the a16. FULL_DATE (between SELECT (trunc (ADD_MONTHS ((LAST_DAY(SysDate-3)),-1)+1) and (SELECT (add_months (trunc(SysDate-2)-1)) of double)))
Of another a16. FULL_DATE between (SELECT trunc (ADD_MONTHS ((LAST_DAY (SysDate)),-1)+1) From Dual) and (select (TRUNC (SysDate - 2)) From Dual)
However, I have an error on "Else". Can someone explain what is the problem and how to fix it? I tried both 'then' and 'Else' syntax and both will run individually. So I'm quite sure that he does not like my CASE statement in general.
In advance, thank you for your help.
Don

I can't tell what your business logic, so here's a simple example of an instruction box in the where clause:

SQL > select *.
2 double
3 where dummy = case when extracted (sysdate months) = 9 then 'W '.
4 Once extracted (sysdate months) = 10 then 'X '.
When 5 extract (month from sysdate) = 11 then 'Y '.
6 else 'Z'
7 end
8;

D
-
X

No output for report of XML editor using CASE/DECODE in a Where Clause

Hello

I have a requirement of the company to modify an existing report that has two input parameters,
-> p_statcode (closed status) which may have values "Y" or "n".
-> p_overdue (flag late), which can have values of "Y" or "n".

The flag late is a column evaluated with O/N values and it is evaluated as follows,

ONTF_MOD_VAL(NVL (
                                     (TRUNC (SYSDATE)
                                      - (TO_DATE (oe_order_lines.attribute18,
                                                  'DD-MON-RRRR')
                                         + TO_NUMBER (fnd_lookup_values.meaning))),
                                     0
                                  ))
                        overdue_flag

The requirement of user now is that they must be a third option for setting called p_overdue ALL,
which by the way the output should include records with
p_statcode is p_statcode ELSE is N AND Y OR p_overdue p_overdue is N

In other words records with raising the Y and N for flag late must be returned regardless of the value given to the closed state.

Original where clause contained in the data definition file is as follows,

WHERE Closed_Status = nvl(:p_statcode,Closed_Status)
                   AND overdue_flag = nvl(:p_overdue,overdue_flag)

My modified code is the following,

WHERE   Closed_Status = NVL (:p_statcode, Closed_Status)
         AND overdue_flag = (CASE
         WHEN :p_overdue = 'Y' THEN 'Y'
         WHEN :p_overdue = 'N' THEN 'N'
         ELSE overdue_flag
         END)

WHERE   Closed_Status = NVL (:p_statcode, Closed_Status)
         AND overdue_flag = DECODE (:p_overdue, 'Y', 'Y', 'N', 'N',overdue_flag)

Both approaches have the same problem.
The output is in EXCEL format. The modified query works very well for p_overdue as Y or N but when p_overdue is passed as ALL he returned a blank EXCEL sheet with just exit report column headings.
Any help regarding, why is this the case? What's not in my approach?

Kind regards
Vishal

unclear on p_overdue = ALL
P_overdue = ALL on what necessary values?

try sth like

WHERE   Closed_Status = NVL (:p_statcode, Closed_Status)
AND (
   overdue_flag = DECODE (:p_overdue, 'Y', 'Y', 'N', 'N',overdue_flag) and :p_overdue != 'ALL'
   or
  :p_overdue = 'ALL' and (overdue_flag = 'Y' or overdue_flag = 'N')
)

for overdue_flag, which has more then 'Y' values of "n".
If overdue_flag that in ('Y', ' don't) then

WHERE   Closed_Status = NVL (:p_statcode, Closed_Status)
AND (
   overdue_flag = DECODE (:p_overdue, 'Y', 'Y', 'N', 'N',overdue_flag) and :p_overdue != 'ALL'
   or
  :p_overdue = 'ALL'
)

case in PL/SQL where clause statement
- Hello
- I have a research procedure where the user can search on the city, street, zip code.
- I need meet the 6 different combinations as I don't know what value the user will pass.
- It should be checked if the user passes the value or not, and then look on values passed.
- But the procedure below is out.it the slightest mistake can be transformed with if-then-else, but case statement seems much cleaner in the code.
- procedure searchaddress (pc_town IN SEARCHADDRESS. CITY % TYPE,
pc_street IN SEARCHADDRESS. TYPICAL STREET %,
pc_postcode IN SEARCHADDRESS. POSTCODE % TYPE,
pResultSet to ref_cursor) as
Start
Open the pResultset for
Select searchaddress_pk from searchaddress
where
case
When pc_town is null then ((street = pc_street) and (ZipCode = pc_postcode))
When pc_street is null then ((town = pc_town) and (ZipCode = pc_postcode))
When pc_postcode is null then ((town = pc_town) and (street = pc_street))
When pc_town is null and pc_street is null then (postal code = pc_postcode)
When pc_street is null and pc_postcode is null then (city = pc_town)
When pc_town is null and pc_postcode is null then (street = pc_street)
end
end searchaddress;
Thank you
CenterB

A basic concept that is important to understand - sliders aren't 'sets of results' (sets of data in memory). A cursor is a series of executable steps (see plans run to cursor). It's like a program. An extraction run the program and output data. In general, which is repeated until the cursor (aka program) is no longer found the corresponding lines.

A ref cursor is a pointer or a handle for a such slider, passed to a caller, allowing the appellant to interface directly with this slider/program.

The slider more effective to create a search would be one without superfluous code and predicates and filters. Consider the following approach - a beefier PL/SQL procedure, do not try and play little tricks with SQL predicates in an attempt to create one (and probably not optimal) SQL.

(not tested/compiled code)
```
create or replace procedure SearchAddress(
  town     IN SEARCHADDRESS.TOWN%TYPE,
  street   IN SEARCHADDRESS.STREET%TYPE,
  postcode IN SEARCHADDRESS.POSTCODE%TYPE,
  refCur out ref_cursor
) is
  curHandle integer;
  res integer;
  searchSQL varchar2(1000);
begin
  -- create base query
  searchSQL := 'select searchaddress_pk from searchaddress where 1 = 1 ';

  -- add dynamic predicates as needed
  if town is not null then
    searchSQL := searchSQL || 'and town = :town';
  end if;

  if street is not null then
    searchSQL := searchSQL || ' and street = :street';
  end if;

  if postcode is not null then
    searchSQL := searchSQL || ' and postcode = :postcode';
  end if;

  -- create a DBMS_SQL cursor
  curHandle := DBMS_SQL.open_cursor;
  DBMS_SQL.parse( curHandle, searchSQL, DBMS_SQL.NATIVE );

   -- bind the dynamic bind variables added as predicate values
  if town is not null then
    DBMS_SQL.Bind_Variable( curHandle, 'town', town );
  end if;

  if street is not null then
    DBMS_SQL.Bind_Variable( curHandle, 'street', street );
  end if;

  if postcode is not null then
    DBMS_SQL.Bind_Variable( curHandle, 'postcode', postcode );
  end if;

  -- execute cursor (to create the cursor program) and pass it back as a ref cursor
  res := DBMS_SQL.Execute( curHandle );
  refCur := DBMS_SQL.To_RefCursor( curHandle );
end;
```

Simple conditional SQL WHERE Clause with two Subselects

Hi SQL Experts,
I need some quick help with a query.
It is a ' select ' for a data source object repository phyiscal table OBI.
So, I can't put in PL/SQL or in stored procedures. I just need
have two conditions / selects different according to the user's role (this information comes
of the session).
The above does not work:
-ORA00905. 00000 - 'lack the key word'
SELECT OFFICENO FROM orgunit
WHERE

BOX WHEN "SESSION_VARIABLE' LIKE '% globalmanager %' THEN
ASSIGNEDOFFICES =
(
SELECT DISTINCT ASSIGNEDOFFICES
FROM USERSTABLE INNER JOIN orgunit WE
OrgUnit. KEY = USERSTABLE. OFFICE
WHERE USERSTABLE. USERNAME ='VALUEOF (NQ_SESSION. THE USER)"
)
ON THE OTHER
OFFICENO =
(
SELECT DISTINCT USERSTABLE OFFICENO
INNER JOIN orgunit WE
OrgUnit. KEY = USERSTABLE. OFFICE
WHERE USERSTABLE. USERNAME ='VALUEOF (NQ_SESSION. THE USER)"
)
END;
Can anyone help?

Hello

I can't tell what you want to do by looking at the code that do not do. Maybe:

SELECT officeno

For orgunit

WHERE (session_variable LIKE '% globalmanager %'

AND IN assignedofficies

(

SELECT assignedoffices

UserTable U1

JOIN orgunit o1 ON o1.officekey = u1.officekey

WHERE u1.username = ' VALUEOF (NQ_SESSION. THE USER)"

)

)

OR (NVL (session_variable

, '?'

) NOT LIKE '% globalmanager % '.

AND IN officeno

(

SELECT officeno

Of u2 usertable

JOIN orgunit o2 ON o2.officekey = u2.officekey

WHERE u2.username = ' VALUEOF (NQ_SESSION. THE USER)"

)

)

;

There might be a simpler and more efficient way in function your tables and your limit.

The condition u2.username = ' VALUEOF (NQ_SESSION. The USER) "almost certainly not what you really want.

.

Whenever you have any questions, post a small example data (CREATE TABLE and only relevant columns, INSERT statements) for all of the tables involved and also publish outcomes from these data.
Explain, using specific examples, how you get these results from these data.
Always say what version of Oracle you are using (for example, 11.2.0.2.0).

See the FAQ forum: https://forums.oracle.com/message/9362002#9362002

Case statement in a where clause clause
Hi all
This is a query that works perfectly except when: P_value don't have any value the two result Y and N should show as a result.
I tried with a statement of NVL but it gives me an error. Someone can help me.

-----------------------------------
Select
ID
CASE
WHEN (colonne_1) > (column_2)
THEN 'Y '.
ANOTHER "N".
END 'Test '.
OF tabel_1
WHERE
(CASE
WHEN (colonne_1) > (column_2)
THEN 'Y '.
ANOTHER "N".
END
) =: P_value
---------------------------------
When: P_value have no any value the two result Y and N should show as a result.

It is not fair
```
SELECT id, CASE WHEN (column_1) > (column_2) THEN 'Y' ELSE 'N' END "Test"
  FROM tabel_1
 WHERE (CASE WHEN (column_1) > (column_2) THEN 'Y' ELSE 'N' END = :p_value
     OR :p_value IS NULL)
```
??

How to check this condition in where clause

WITH t1 AS
    (SELECT 1 empno, 'Sun' name,100 deptno, 'M' Gender  FROM dual
    UNION ALL
    SELECT 2,'Gun',100 deptno, 'F' FROM dual
    UNION ALL
    SELECT 3, 'Run',50,'M' FROM dual
    UNION ALL
    SELECT 4, 'Nun',75,'F' FROM dual
    UNION ALL
    SELECT 5, 'Tun',25,NULL FROM dual)
    SELECT empno,name,deptno,gender FROM t1;

I want to display those dept. data which is having Genders both 'M' and 'F',

You can use two ANDed 'exists' conditions...

...
SELECT empno,name,deptno,gender FROM t1 t1o;
where
exists (select 'x' from t1 t1i where t1o.deptno = t1i.deptno and t1.gender = 'M')
and
exists (select 'x' from t1 t1i where t1o.deptno = t1i.deptno and t1.gender = 'F')

I need account condition in where clause.

Hi all
I have a data
Product ID AMT
1 100 A
1 100 B
1 100 C
1 100 D
2 25 A
2 40 E
2 55 F
3 50 B
3 50 A
3 50 C
3 50 G
4 12 H
4 13 E
4 14 A
4 15 I have
I need following output
ID AMT Product
1 100 A
1 100 B
1 100 C
1 100 D
3 50 B
3 50 A
3 50 C
3 50
G
I tried after query, but it does not work
Select
ID,
AMT,
Product
t
Product ID, AMT Group,
having count (distinct ID) = count (distinct AMT)
It returns all the rows.
Help, please

with t as)

SELECT id,

AMT,

product,

County (separate amt) over (partition by id) cnt

of sample_data

)

SELECT id,

AMT,

product

t

where cnt = 1

order by id

/

AMT ID P
---------- ---------- -
1 100 A
1 100 B
1 100 C
1 100 D
3 50
3 50
3 50
3 50

8 selected lines.

SQL >

SY.

Result of conditional function where clause
create the table datatest
(number of primaryid,
Identification number,
power varchar2 (4).
alias varchar2 (15))

insert into values(1,100,'CNN','NEWS') datatest.
insert into values(2,100,'BBC','SPORTS') datatest.
insert into values(3,101,'CNN','INV') datatest
insert into values(4,101,'ABC','MON') datatest
insert into values(5,102,'BBC','REAL') datatest
insert into datatest values (6 103, 'NEC', "TVL")

I am looking for a query which returns lines with power CNN
If the CNN feed isn't here then return the line with power BBC
This is the result I'm looking for.

1 100, CNN, NEWS
3 101, CNN, INV
5 102, BBC, REAL

Thnx in advance
```
create table datatest
(
  primaryid number,
  secondaryid number,
  feed varchar2(4),
  the_alias varchar2(15)
);
begin
insert into datatest values(1,100,'CNN','NEWS');
insert into datatest values(2,100,'BBC','SPORTS');
insert into datatest values(3,101,'CNN','INV');
insert into datatest values(4,101,'ABC','MON');
insert into datatest values(5,102,'BBC','REAL');
insert into datatest values (6,103,'NEC','TVL');
end;

select *
from
(
  select
    primaryid,
    secondaryid,
    feed,
    the_alias,
    row_number() over (partition by secondaryid order by decode(feed, 'CNN', 1, 'BBC', 2) asc) as rn
  from datatest
  where feed in ('CNN', 'BBC')
)
where rn = 1
```
Would be a solution.

CASE of Condition causing Plan difference

Version

SQL> SELECT * FROM v$version;

BANNER
--------------------------------------------------------------------------------
Oracle Database 11g Enterprise Edition Release 11.2.0.2.0 - 64bit Production
PL/SQL Release 11.2.0.2.0 - Production
CORE    11.2.0.2.0      Production
TNS for Linux: Version 11.2.0.2.0 - Production
NLSRTL Version 11.2.0.2.0 - Production

5 rows selected.

Optimizer

SQL> show parameter optimizer

NAME                                 TYPE        VALUE
------------------------------------ ----------- ------------------------------
_optimizer_autostats_job             boolean     FALSE
_optimizer_join_elimination_enabled  boolean     FALSE
optimizer_capture_sql_plan_baselines boolean     FALSE
optimizer_dynamic_sampling           integer     2
optimizer_features_enable            string      11.2.0.2
optimizer_index_caching              integer     0
optimizer_index_cost_adj             integer     100
optimizer_mode                       string      ALL_ROWS
optimizer_secure_view_merging        boolean     FALSE
optimizer_use_invisible_indexes      boolean     FALSE
optimizer_use_pending_statistics     boolean     FALSE
optimizer_use_sql_plan_baselines     boolean     TRUE

Problem

We have a system of providers that built the SQL to be sent to the database. These queries have the following generic structure for a condition in WHERE clause:

CASE WHEN <condition> THEN <table>.<column> END = '<value>'

We have noticed different execution plans when the optimizer compares the numbers instead of varchar2. The examples below show that. In our case, this difference is an impact on the execution of a longer query plans effectively doubling buffer gets and leading to the execution times that are 2 to 4 times as worse.

Query 1

SQL> SELECT /*+gather_plan_statistics*/ * FROM DUAL WHERE CASE WHEN 1=1 THEN 'X' END = dummy;

D
-
X

1 row selected.

SQL> SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY_CURSOR(null,null,'ALLSTATS LAST'));

PLAN_TABLE_OUTPUT
----------------------------------------------------------------------------------------------
SQL_ID  6ghjubgpwpr61, child number 0
-------------------------------------
SELECT /*+gather_plan_statistics*/ * FROM DUAL WHERE CASE WHEN 1=1 THEN
'X' END = dummy

Plan hash value: 272002086

---------------------------------------------------------------------------------------------
| Id  | Operation         | Name | Starts | E-Rows | A-Rows |   A-Time   | Buffers | Reads  |
---------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT  |      |      1 |        |      1 |00:00:00.02 |       2 |      2 |
|*  1 |  TABLE ACCESS FULL| DUAL |      1 |      1 |      1 |00:00:00.02 |       2 |      2 |
---------------------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   1 - filter("DUMMY"='X')

Query 2

SQL> SELECT /*+gather_plan_statistics*/ * FROM DUAL WHERE CASE WHEN 'A'='A' THEN 'X' END = dummy;

D
-
X

1 row selected.

SQL> SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY_CURSOR(null,null,'ALLSTATS LAST'));

PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------
SQL_ID  gcpwzksqr2w9n, child number 0
-------------------------------------
SELECT /*+gather_plan_statistics*/ * FROM DUAL WHERE CASE WHEN 'A'='A'
THEN 'X' END = dummy

Plan hash value: 272002086

------------------------------------------------------------------------------------
| Id  | Operation         | Name | Starts | E-Rows | A-Rows |   A-Time   | Buffers |
------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT  |      |      1 |        |      1 |00:00:00.01 |       2 |
|*  1 |  TABLE ACCESS FULL| DUAL |      1 |      1 |      1 |00:00:00.01 |       2 |
------------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   1 - filter("DUMMY"=CASE  WHEN ('A'='A') THEN 'X' END )

To my eyes the status 'A' = 'A' is equivalent to the condition 1 = 1. How did in the case of the comparison digital Oracle allows to eliminate the case expression, but compared to the character he can't?

Thank you!

Centinul wrote:
To my eyes the status 'A' = 'A' is equivalent to the condition 1 = 1.

Well, question is Oracle looks like 'A' = 'A' as the string comparison. And the result of a string comparison is NLS charging. This is why it can not evaluate CASES at compile time.

SY.

several instructions box in where clause

Hello
I'm trying to create report filters using substitution variables in a case statement in a where clause clause. I have several instructions box running on the same column. I don't really know how to write it. Here's what I have so far but I know I'm going in the wrong direction.
where
i.compute_zone =
case
When: P14_zone is NOT NULL
then: P14_zone
of other i.compute_zone
end
AND
i.compute_zone =
case
When: P14_zone_2 is NOT NULL
then: P14_zone_2
other: P14_zone
end
AND
i.compute_zone =
case
When: P14_zone_3 is NOT NULL
then: P14_zone_3
of other i.compute_zone
end
AND
i.compute_zone =
case
When: P14_zone_4 is NOT NULL
then: P14_zone_4
of other i.compute_zone
end
any suggestions?
Thank you
MICAH

Hi, Micah,

User12611868-Oracle wrote:

Hello

I'm trying to create report filters using substitution variables in a case statement in a where clause clause. I have several instructions box running on the same column. I don't really know how to write it. Here's what I have so far but I know I'm going in the wrong direction.

where

i.compute_zone =

case

When: P14_zone is NOT NULL

then: P14_zone

of other i.compute_zone

end

AND

i.compute_zone =

case

When: P14_zone_2 is NOT NULL

then: P14_zone_2

other: P14_zone

end

AND

i.compute_zone =

case

When: P14_zone_3 is NOT NULL

then: P14_zone_3

of other i.compute_zone

end

AND

i.compute_zone =

case

When: P14_zone_4 is NOT NULL

then: P14_zone_4

of other i.compute_zone

end

any suggestions?

Thank you

MICAH

How get the desired results of your data depends on the desired results and your data. What do you do? (I can't tell just by looking at the aat code that do not do). After CREATE TABLE and INSERT statements for some sample data and a couple of sets of parameters (variables such as p14_zone) and the exact results that your choices in each set of parameters, given the same sample data.

Check out the Forum FAQ: Re: 2. How can I ask a question on the forums?

CASE expressions are generally not useful in a WHERE clause. CASE expressions are a convenient way to IF-THEN-ELSE logic of the places where you can't do anything other than (the SELECT clause), but WHERE the clauses allow IF-THEN-ELSE logic in any case

Maybe you want something like

WHERE MERGE (: p14_zone,: p14_zone_2,: p14_zone_3,: p14_zone_4) IS NULL

OR IN i.compute_zone (: p14_zone,: p14_zone_2,: p14_zone_3,: p14_zone_4)

It returns TRUE if all 4 parameters are set to NULL. If 1 or more of them are not NULL, then it will return TRUE if i.compute_zone is equal to one of them.

1	100	A
1	100	B
1	100	C
1	100	D
2	25	A
2	40	E
2	55	F
3	50	B
3	50	A
3	50	C
3	50	G
4	12	H
4	13	E
4	14	A
4	15	I have

case of condition in where clause

Similar Questions

Maybe you are looking for