DECLARE
CURSOR contacts
IS
SELECT SUM (budget) AS budget
FROM et_bp_gl_account a, et_bp_fact f
WHERE f.gl_account_id = a.gl_account_id
AND total_flag = 0
GROUP BY month_id, org_unit_id;
BEGIN
FOR r IN contacts
LOOP
UPDATE et_bp_fact
SET budget = r.budget
WHERE gl_account_id IN (SELECT total_element
FROM et_bp_gl_account g, et_bp_fact f
WHERE f.gl_account_id = g.gl_account_id);
END LOOP;
END;
I want to update the table ET_BP_FACT by many values example(25,50,75)
returned from Cursor but when i execute the table updated by (25,25,25)
I think there an issue in the loop
That's because the UPDATE's WHERE clause doesn't "reference" cursor's values. Something like this:
DECLARE
CURSOR contacts
IS
SELECT month_id, org_unit_id, --> include additional columns here ...
SUM (budget) AS budget
FROM et_bp_gl_account a, et_bp_fact f
WHERE f.gl_account_id = a.gl_account_id
AND total_flag = 0
GROUP BY month_id, org_unit_id;
BEGIN
FOR r IN contacts
LOOP
UPDATE et_bp_fact
SET budget = r.budget
WHERE gl_account_id IN
(SELECT total_element
FROM et_bp_gl_account g, et_bp_fact f
WHERE f.gl_account_id = g.gl_account_id
--
AND f.org_unit_id = r.org_unit_id --> ... and reference them here ...
AND g.month_id = r.month_id);
--> ... or, possibly, here
END LOOP;
END;
I'd suggest you to always precede column names with table aliases, because - looking at your code, there's no way to guess which table(s) MONTH_ID and ORG_UNIT_ID belong to, so my code might (or might not) work, but I hope you got the general idea.
Related
I have a simplified code like below:
PROCEDURE MY_PROC (d1 IN DATE) IS
CURSOR curs IS SELECT * FROM TAB1 WHERE date1 = d1;
result BOOLEAN;
rTab1 TABLE1%ROWTYPE;
BEGIN
OPEN curs;
FETCH curs INTO rTab1 ;
result := curs%FOUND;
CLOSE curs;
IF result = FALSE THEN
rTab1.NUM_ID := 0;
END IF;
SELECT * FROM TAB2 WHERE NVL(NUM_ID, 0) = rTab1.NUM_ID;
END;
Is it possible to write everything in one query? I mean without having to chceck if rTab1.NUM_ID exists. Maybe with join? And get same results at the end like above?
I mean something like:
SELECT * FROM TAB1, TAB2
WHERE NVL(TAB1.NUM_ID, 0) = TAB2.NUM_ID
AND date1 = d1;
The code is looking for tab1 rows where the date matches the given date d1.
If there is one matching row, then its num_id is used.
If there is more than one matching row, then one of their num_ids is used arbitrarily.
If there is no matching row, the value 0 is used.
Then with that value the code selects all matching tab2 rows.
This can be done with a single query. To get only one tab1.num_id, use MIN, MAX or ANY_VALUE. To get a zero when there is no date match use NVL or COALESCE.
SELECT *
FROM tab2
WHERE NVL(num_id, 0) = (SELECT NVL(MIN(num_id), 0) FROM tab1 WHERE date1 = d1);
declare
type employee_t is table of employee%rowtype index by pls_integer;
l_employee_data employee_t;
begin
select *
bulk collect into l_employee_data from employee;
forall indx in 1 .. l_employee_data.count
update roster r
set r.job = l_employee_data(indx).job,
r.position = l_employee_data(indx).position,
r.organisation = l_employee_data(indx).organisation
where r.employee_code = l_employee_data(indx).employee_code;
commit;
end;
In the above example, query updates all the data where it matches employee code with idx value. This is fine. My requirement here is, is it possible to update also, the other rows which doesn't matches employee code with idx value? if yes, then please let me know.
Update with what, if employee code doesn't match? If there's no match, then there's no match. You could update those columns with some value, but the question is which.
Besides, instead of what you used, consider a single MERGE statement:
merge into roster r
using employee e
on (e.employee_code = r.employee_code)
when matched then update set
r.job = e.job,
r.position = e.position,
r.organisation = e.organisation;
So I have this assignment where I have to create a stored procedure to search for movies in an Oracle DataBase.
The search string follow the following logic :
It looks for the year of the movie in parenthesis
Ex. (1992)
It looks for a range of year in brackets
Ex. [1992,2000]
It looks for a word contained in the title, country, realisator, genre, actors or scenarists.
Any of the above can be combined multiple times.
Ex. : The Lord Of the Ring Ian McKellen Christopher Lee [1992,2000]
The logic used to solve this problem was to make a giant query to group all the required data then use a cursor to loop through the result set with a cursor to check if every word of the search string is valid.
I managed to make a procedure that works as expected, but the only way I found to return the results was to use DBMS_OUTPUT. Now the problem is that when I plug this is Hibernate the DBMS_OUTPUT is not sent over to the client. I've read some way to force the output by setting DBMS_OUTPUT.enable, but I feel that this is not the proper way to do it.
So here are my questions :
Is my logic flawed? Is there a simpler way to archive this with a single select or something?
Is there a way to dynamically push data inside a cursor and return it?
Am I really supposed to trick the DBMS_OUTPUT so it is sent over to hibernate?
Here's my code :
CREATE OR REPLACE PROCEDURE p_SearchFilm(searchString IN VARCHAR2) IS
IsValid BOOLEAN;
y1 INTEGER;
y2 INTEGER;
subStrArray apex_application_global.vc_arr2;
term VARCHAR(100);
CURSOR films IS
Select FilmId, Titre, real.Prenom||' '||real.nom as Realisateur, anneeSortie, ListPays, ListGenres,
ListScenaristes, ListActeurs, langueOrigine
from Film
natural left join
(select FilmId, listagg(p.Nom, ',') within group (Order By p.nom) ListPays from Film
natural join Film_vs_pays
natural join Pays p
Group by FilmId)
natural left join
(select FilmId, listagg(g.Nom, ',') within group (Order By g.nom) ListGenres from Film
natural join Film_vs_Genre
natural join Genre g
Group by FilmId)
natural left join
(select FilmId, listagg(p.Prenom||' '||p.Nom, ',') within group (Order By p.nom) ListScenaristes from Film
natural join Scenariste s
join Personne p on s.personneId = p.personneId
Group by FilmId)
natural left join
(select FilmId, listagg(p.Prenom||' '||p.Nom, ',') within group (Order By p.nom) ListActeurs from Film
natural join Personnage perso
join Personne p on perso.personneId = p.personneId
Group by FilmId)
left join Personne real on real.personneId = realisateurId;
BEGIN
<<FILM_LOOP>>
FOR film IN films LOOP
subStrArray := apex_util.string_to_table(searchString, ' ');
FOR i in 1..subStrArray.count LOOP
IsValid:= FALSE;
term:= subStrArray(i);
IF REGEXP_LIKE(term, '\(\d{4}\)') THEN
IF film.anneeSortie = TO_NUMBER(regexp_substr(term, '\d{4}')) THEN
IsValid:= TRUE;
END IF;
ELSIF REGEXP_LIKE(term, '\[\d{4},\d{4}\]') THEN
y1:= regexp_substr(term, '\d{4}', 1, 1);
y2:= regexp_substr(term, '\d{4}', 1, 2);
IF film.anneeSortie BETWEEN y1 AND y2 THEN
IsValid:= TRUE;
END IF;
ELSE
IF UPPER(film.Titre||film.Realisateur||film.ListActeurs||film.ListScenaristes||film.ListGenres||film.ListPays||film.langueOrigine)
LIKE '%'||UPPER(term)||'%' THEN
IsValid:= TRUE;
END IF;
END IF;
IF NOT IsValid THEN
CONTINUE FILM_LOOP;
END IF;
END LOOP;
DBMS_OUTPUT.put_line(film.FilmId||'|'||film.Titre);
END LOOP;
END;
A small disclaimer here :
I saw some similar questions that addressed this issue, but the ones that used cursors were returning a complete select, not hand picked rows.
The question about DBMS_OUTPUT and Hibernate stated that it should be avoided.
The questions using piped rows seamed to work only with functions (Changing the procedure for a function called by the procedure could be a valid work around, I'd like to know if something else is possible before tho).
The use of the DBMS_OUTPUT package is pretty much bounded to developer executions of anonymous blocks, and hence not suited for your intended communication with the Hibernate framework.
If you already have a stored procedure to apply your filter and determine your positive results, a solution could be populate a temporary table with those positives, and then return an open cursor which will only have the data from that temporary table, something like:
create global temporary table movie_results( movie varchar2(200) ) on commit preserve rows;
Of course your temp table can have more columns, but let me leave it like this, just to illustrate my solution.
CREATE OR REPLACE PROCEDURE p_SearchFilm(searchString IN VARCHAR2, movies out SYS_REFCURSOR) IS
IsValid BOOLEAN;
y1 INTEGER;
y2 INTEGER;
subStrArray apex_application_global.vc_arr2;
term VARCHAR(100);
CURSOR films IS
Select FilmId, Titre, real.Prenom||' '||real.nom as Realisateur, anneeSortie, ListPays, ListGenres,
ListScenaristes, ListActeurs, langueOrigine
from Film
natural left join
(select FilmId, listagg(p.Nom, ',') within group (Order By p.nom) ListPays from Film
natural join Film_vs_pays
natural join Pays p
Group by FilmId)
natural left join
(select FilmId, listagg(g.Nom, ',') within group (Order By g.nom) ListGenres from Film
natural join Film_vs_Genre
natural join Genre g
Group by FilmId)
natural left join
(select FilmId, listagg(p.Prenom||' '||p.Nom, ',') within group (Order By p.nom) ListScenaristes from Film
natural join Scenariste s
join Personne p on s.personneId = p.personneId
Group by FilmId)
natural left join
(select FilmId, listagg(p.Prenom||' '||p.Nom, ',') within group (Order By p.nom) ListActeurs from Film
natural join Personnage perso
join Personne p on perso.personneId = p.personneId
Group by FilmId)
left join Personne real on real.personneId = realisateurId;
BEGIN
<<FILM_LOOP>>
FOR film IN films LOOP
subStrArray := apex_util.string_to_table(searchString, ' ');
FOR i in 1..subStrArray.count LOOP
IsValid:= FALSE;
term:= subStrArray(i);
IF REGEXP_LIKE(term, '\(\d{4}\)') THEN
IF film.anneeSortie = TO_NUMBER(regexp_substr(term, '\d{4}')) THEN
IsValid:= TRUE;
END IF;
ELSIF REGEXP_LIKE(term, '\[\d{4},\d{4}\]') THEN
y1:= regexp_substr(term, '\d{4}', 1, 1);
y2:= regexp_substr(term, '\d{4}', 1, 2);
IF film.anneeSortie BETWEEN y1 AND y2 THEN
IsValid:= TRUE;
END IF;
ELSE
IF UPPER(film.Titre||film.Realisateur||film.ListActeurs||film.ListScenaristes||film.ListGenres||film.ListPays||film.langueOrigine)
LIKE '%'||UPPER(term)||'%' THEN
IsValid:= TRUE;
END IF;
END IF;
IF NOT IsValid THEN
CONTINUE FILM_LOOP;
END IF;
END LOOP;
--DBMS_OUTPUT.put_line(film.FilmId||'|'||film.Titre);
insert into movie_results( movie )
values film.FilmId||'|'||film.Titre;
commit;
END LOOP;
open movies for
select *
from movie_results;
END;
Now your parameter 'movies' has all the positive results derived from your procedure, and all you'll have to do is read the cursor in Hibernate.
Note that, once you close your connection, the temporary table looses all the data and will be ready to use again when another session begins (always remember to close your Cursors/Connections).
The first part of this can probably be accomplished with just a query. You can define your search terms like this: How to declare variable and use it in the same SQL script? (Oracle SQL)
I'll leave the basic query writing to you (e.g. joining appropriate data), but your script will in effect look something like this:
DEFINE var_year1 = 1992;
DEFINE var_year2 = 1994;
DEFINE var_country = null;
DEFINE var_title = 'Lord Of The Rings';
DEFINE var_realisator = null;
DEFINE var_genre = null;
DEFINE var_actors = 'Ian';
DEFINE var_scenarists = 'Lee';
SELECT film_id, title, year
FROM ...
WHERE year BETWEEN &var_year1 AND &var_year2
AND UPPER(title) LIKE UPPER('%&var_title%')
AND UPPER(country) LIKE UPPER('%&var_country%')
AND UPPER(realisator) LIKE UPPER('%&var_realisator%')
AND UPPER(genre) LIKE UPPER('%&var_genre%')
AND UPPER(actors) LIKE UPPER('%&var_actors%')
AND UPPER(scenarists) LIKE UPPER('%&var_scenarists%');
/
The reason I only select film_id, title, and year is because the title and year will be distinct values to the film_id. There are a couple of ways to skin this cat, but it all revolves around using a LIKE statement with variables that you have defined. If you do not want to explicitly define the values in the script, you can use the ACCEPT command. Or, you can interface it with something else that applies those values into the variables.
With this approach, it was kind of a shot in the dark on how you really have your data laid out. But keep in mind good relational database management practices, and you should be able to turn this into a useable query.
Additional Method:
You can also use a series of subqueries to find the same data above using the like statements, then INNER JOIN them together to figure out which ones are in common. Example:
WITH sub_title AS (SELECT film_id FROM Film WHERE UPPER(title) LIKE UPPER('%&var_title%')),
...
sub_actor AS (SELECT film_id FROM (...) WHERE UPPER(actor) LIKE UPPER('%&var_actor%'))
SELECT film_id
FROM sub_title t
INNER JOIN sub_actor a ON a.film_id = t.film_id;
/
If you had searched for 'star war' in that, you would have returned for example
1 Star Wars A New Hope
2 Star Wars The Empire Strikes Back
3 Star Wars Return of the Jedi
Then in the second subquery if you had searched 'Donald Glover' you would have gotten
2 Star Wars The Empire Strikes Back
3 Star Wars Return of the Jedi
Then, the INNER JOIN would have given you ID numbers 2 and 3 because those were the only ones where all of the criteria were met. Repeat the process for your other search terms. I hope this helps.
I am trying to iterate through a cursor which stores the value of the table. I use a FOR Loop to iterate and IF one of the conditions is met, I store the output in another table. I am not sure of the approach I am following and also getting error(ORA-00933: SQL command not ended properly). Stats_Queries is my reference table where I iterate my cursor through. STATS_RESULT_CARD is my output table where I have to store the results. Please help.
DECLARE
CURSOR c1 IS
select Stats_Queries.OBJECTTYPE, Stats_Queries.CATEGORY, Stats_Queries.QUERY
from Stats_Queries;
r1 c1%ROWTYPE;
BEGIN
FOR r1 IN c1 LOOP
If (r1.OBJECTTYPE = 'CARD') THEN
INSERT INTO STATS_RESULTS_CARD (NODETYPENAME, NODEDEFNAME , CARDTYPENAME, PROVISIONSTATUSNAME, STATDATE, CARDCOUNT)
select nt.name, nd.name, ct.name, ps.name, sysdate, count(c.cardid)
from cardtype ct, card c, node n, nodetype nt, nodedef nd, provisionstatus ps
where ct.name in ('SRA AMP', 'XLA AMP', 'SAM', 'ESAM')
and ct.cardtypeid = c.card2cardtype
and c.card2node = n.nodeid
and n.node2nodetype = nt.nodetypeid
and n.node2nodedef = nd.nodedefid
and c.card2provisionstatus = ps.provisionstatusid
group by nt.name, nd.name, ct.name, ps.name
END If;
END LOOP;
END;
As an aside from the answer that Finbarr has provided (which is perfectly correct; add in the missing semi-colon and your procedure should work), why do you need to loop through the cursor at all? That's the slow way of doing it.
You could just do a single insert statement instead, such as:
insert into stats_results_card (nodetypename,
nodedefname,
cardtypename,
provisionstatusname,
statdate,
cardcount)
select x.nt_name,
x.nd_name,
x.ct_name,
x.ps_name,
x.statdate,
x.cnt_cardid
from (select nt.name nt_name,
nd.name nd_name,
ct.name ct_name,
ps.name ps_name,
sysdate statdate,
count (c.cardid) cnt_cardid
from cardtype ct,
card c,
node n,
nodetype nt,
nodedef nd,
provisionstatus ps
where ct.name in ('SRA AMP',
'XLA AMP',
'SAM',
'ESAM')
and ct.cardtypeid = c.card2cardtype
and c.card2node = n.nodeid
and n.node2nodetype = nt.nodetypeid
and n.node2nodedef = nd.nodedefid
and c.card2provisionstatus = ps.provisionstatusid
group by nt.name,
nd.name,
ct.name,
ps.name) x
cross join (select stats_queries.objecttype,
stats_queries.category,
stats_queries.query
from stats_queries
where objecttype = 'CARD');
N.B. This assumes that there really isn't any link between the original cursor and the select statement that was inside the loop; we do a cross join to replicate the rows the required number of times.
If there was an actual join between the two queries, you would put that in place of the cross join.
ORA-00933: SQL command not ended properly
Probably occurring because you missed a semicolon after
group by nt.name, nd.name, ct.name, ps.name
quite the same question as here : Copy data from one existing row to another existing row in SQL?
but in Oracle, where update ... from and update t1, t2 are not supported.
I'll repeat it here in my own words ;
I have a table T, which looks like this :
and as the arrow shows it, I want to copy everything from r where c = 1 to e where c = 2, with t matching.
I have the select statement to get what I want to copy :
select
told.t,
told.r
from
T told
inner join
T tnew
on
told.t= tnew.t
where
told.c = 1
and
tnew.c = 2
I just don't know how to put this together in an update. An Oracle update, specifically.
try this:
update T tnew
set tnew.e = (select told.r from T told where told.c = 2 and told.t = tnew.t)
where tnew.c = 1
Sounds like the time for a bulk collects! Not as pretty as AB Cade's solution but more efficient.
declare
c_data is
select t1.rowid as rid, t2.r
from my_table t1
join my_table t2
on t1.t = t2.t
where t1.c = 2
and t2.c = 1
;
type t__data is table of c_data index by binary_integer;
t_data t__data;
begin
open c_data;
loop
fetch c_data bulk collect into t_data limit 25000;
exit when t_data.count = 0;
forall i in t_data.first .. t_data.last loop
update my_table
set e = t_data(i).r
where rowid = t_data(i).rid
;
commit;
end loop;
close c_data;
end;
/