I'm trying to execute a rather simple Oracle query against a table with a XMLTYPE column:
Select POB_CODPOL, CODSIS From (
Select T1.POB_CODPOL, EXTRACTVALUE(T1.POB_XMLPOL, '/Polbas/polfun[nomfun="filterBySystem"]/extpar[codele="codsis"]/valele/text()') CODSIS
From TDTX_POLITICA_CLOB T1
Where T1.POB_CODEMP = '001840'
)
Group By POB_CODPOL, CODSIS
This throws a ORA-00979 Not a GROUP BY Expression, which I don't really understand.
Even worse: when I execute the exact same query, but with a simplified XPATH query does work:
Select POB_CODPOL, CODSIS From (
Select T1.POB_CODPOL, EXTRACTVALUE(T1.POB_XMLPOL, 'Polbas/codpol/text()') CODSIS
From TDTX_POLITICA_CLOB T1
Where T1.POB_CODEMP = '001840'
)
Group By POB_CODPOL, CODSIS
It looks like Oracle doesn't like conditions like [nomfun="filterBySystem"] when using a GROUP BY (without the grouping clause, everything works fine).
Any idea on why this can be happening?
Edit: the result of the inner query is rather simple:
EXTRACTVALUE is deprecated.
Oracle recommends to use XMLQUERY, XMLTABLE for it.
This one should work:
WITH t as
(SELECT T1.POB_CODPOL, x.CODSIS
FROM TDTX_POLITICA_CLOB T1
NATURAL JOIN XMLTABLE('/Polbas/polfun[nomfun="filterBySystem"]/extpar[codele="codsis"]/valele'
PASSING POB_XMLPOL COLUMNS
CODSIS VARCHAR2(50) PATH '/') x
Where T1.POB_CODEMP = '001840')
SELECT POB_CODPOL, CODSIS
FROM t
GROUP BY POB_CODPOL, CODSIS;
There's a Bug 28588011 : "ORA-00979: NOT A GROUP BY EXPRESSION" WHEN TABLE FORMAT IS BINARY XML.
Severity 2 - Severe Loss of Service
Created 02-Sep-2018
Status 11 - Code/Hardware Bug (Response/Resolution)
WORKAROUND No
I guess they won't fix it. But the hint NO_XML_QUERY_REWRITE (that I use for other XML related bugs, too) worked for me.
Related
I have two table which I want to select Name from t2.
The situation is following.
I have t1 Policy which containt EmployeeID
And table t2 which containt Name
Now I want to select which Employee release policy.
So in t1(Policy- AUTO.POL) I have column: SIFRA_RAD
and t2(Employee-AUTO.SIFRAD) I have colum: SIFRA_R, Ime
I try something like this:
select auto.pol.sifra_rad, auto.sifrad.ime
from auto.sifrad
inner join auto.pol on auto.sifrad.ime = auto.pol.sifra_rad;
After that I get error
ORA-00933: SQL command not properly ended
I have no idea what is wrong here. Any suggestion?
The problem comes from query, and we fix it
select p.sifra_rad, s.ime from auto.sifrad s, auto.pol p where s.ime = p.sifra_rad
I want to retrieve users name and there responsibility_key where there end_date is null and i want to convert it to (sysdate+1) using nvl but i am only able to retrieve the responsibility_key not the name please help.
The error in the image says "column ambiguously defined". Take a close look. Your last END_DATE could refer to either the u alias or the table from the subquery. Change it to match the rest of your subquery (FIND_USER_GROUPS_DIRECT.END_DATE)
EDIT
Your query is
select u.USER_NAME, d.responsibility_key from FND_USER u,FND_RESPONSIBILITY_VL d
where responsibility_id in(
select responsibility_id from
FND_USER_RESP_GROUPS_DIRECT WHERE END_USER_RESP_GROUPS_DIRECT.END_DATE=nvl(END_DATE,sysdate+1)) and
u.END_DATE=nvl(END_DATE,SYSDATE + 1)
;
The query isn't formatted, which makes it hard to read.
Not all columns are qualified with table name (or aliases), as mentioned in the comments.
The query currently uses an implicit join.
The query is impossible to understand without seeing the table definitions (desc [table_name]).
For points 1 and 2, a properly formatted query will look something like
select u.user_name, d.responsibility_key
from
fnd_user u,
fnd_responsibility_vl d
where
d.responsibility_id in (
select urgd.responsibility_id
from
fnd_user_resp_groups_direct urgd
where
urgd.end_date = nvl(u.end_date, sysdate+1)
) and
u.end_date = nvl(urgd.end_date, sysdate + 1)
;
This makes it easier to read and in addition to this, you can see that without table definitions I guessed (see point 4) as to which tables the end_date column belongs in your query. If I had to guess, so does Oracle. That means you have an ambiguity problem. To fix it, take a close look at the end_date column as it appears in your original query and where you do not prefix it with anything, you need to prefix it with the appropriate alias (after you have aliased all your tables).
For point 3, you can write your query more clearly with an explicit join and by using aliases for all columns. As for the explicit join I have no idea what your tables look like but one possibility is something like
select u.user_name, d.responsibility_key
from fnd_user u
join fnd_responsibility_vl d
on u.id = d.user_id
where
d.responsibility_id in (
select responsibility_id
from fnd_user_resp_groups_direct urgd
where
urgd.end_date = nvl(u.end_date, sysdate+1)
) and
u.end_date = nvl(urgd.end_date, sysdate+1)
;
If you follow these points you will get to the root of the error.
I am just wondering if use = sign operator with sub-query instead of IN
Is it correct way ? and meet the oracle standard ?
Example
select column_name from my_table_1 where id = (select max(id) from my_table_2);
The Difference is related to the number of rows returned. If you have only one row returned from nested sql you may prefer both = or in operators. But if multiple rows returned from nested query, use in operator.
So, in your sql example you may prefer using any of the operators. Since, max functions returns only one row.
As you are fetching maximum value from subquery to compare with id, Both(= and IN )will work fine. But If you are trying to fetch more than one row then you have to use IN keyword.
If you have 1 result in sub query you are fine with using = sign, except when data type is wrong, for example , checking with same data type of dummy VARCHAR2(1)
select * from dual where 'X' = (select max(dual.dummy) from dual);
Is similar to using in (also same explain plain)
select * from dual where 'X' in (select max(dual.dummy) from dual);
But checking with different/wrong data type will result with exception ORA-01722 Invalid number
select * from dual where 1 =(select max(dual.dummy) from dual);
A long time ago in a database far, far away a developer wrote a query in which he/she was reliant on the order in which predicates were written.
For example,
select x
from a, b
where NOT REGEXP_LIKE (a.column, '[^[:digit:]]')
and a.char_column = b.numeric_column;
(explain plan suggests a to_number conversion will be applied to a.char_column)
I think by chance more than design this "just works" (Oracle 11g). However, the order of the predicates isn't adhered to when running in Oracle 12c, so this query breaks with an invalid number exception. I'm aware that I could try to force 12c to evaluate the predicates in order by using the ORDERED_PREDICATES hint as follows
select /*+ ORDERED_PREDICATES +*/ x
from a, b
where NOT REGEXP_LIKE (a.column, '[^[:digit:]]')
and a.char_column = b.numeric_column
.. Or cast one of the values using to_char for the comparison. The downside is that to_char could operate on say a million rows. I think the following inline view is probably a better solution. Am I guaranteed that the inline view will be evaluated first?
select x
from b
inner join (
select only_rows_with_numeric_values as numeric_column
from a
where NOT REGEXP_LIKE (a.column, '[^[:digit:]]')
) c
on c.numeric_column = b.numeric_column;
About predicate order - look at https://jonathanlewis.wordpress.com/2015/06/02/predicate-order-2/
You should rewrite your last query to next using rownum according to doc(https://docs.oracle.com/database/121/SQLRF/queries008.htm#SQLRF52358)
select x
from b
inner join (
select only_rows_with_numeric_values as numeric_column,
rownum
from a
where NOT REGEXP_LIKE (a.column, '[^[:digit:]]')
) c
on c.numeric_column = b.numeric_column;
to suppress query unnesting or simply using hint /*+ no_unnest*/
I'm facing a problem in Oracle.
I had a SQL where some values were fixed. Now I started replacing them with values from a parameter-table. Some of these fixed values where in a NVL().
Simply said my statement is like this.
SELECT NVL(MAX(t.datefield), to_date('01011900','DDMMYYYY'))
FROM table t;
That works fine.
Now I want to replace the fixed date to a date from my parameter-table with a subselect, which doesn't work at all.
// Works
SELECT NVL(MAX(NULL), 'hello') FROM DUAL;
// Doesn't work
SELECT NVL(MAX(NULL), (SELECT 'hello' FROM DUAL)) FROM DUAL;
The error is:
ORA-00937: .... "not a single-group group function"
I have no idea how to group by a subselect.
Any help is very appreciated! Thanks!
You can't group by a sub-select. However, in order to achieve this your sub-select is only going to be able to return one row. So, change it into a Cartesian join and group by that.
SELECT NVL(MAX(NULL), str)
FROM DUAL
CROSS JOIN ( SELECT 'hello' as str FROM DUAL )
GROUP BY STR
More generally every column that is not included in an aggregate function must be included in the GROUP BY. Plus NVL() is bad; use COALESCE() or CASE instead.