I have a table.
create table tb(
userid int,
nodes jsonb)
nodes sample:
{
"weight": [{"date":"<date>", value: 50}, {"date":"<date>", value: 60}],
"height": [{"date":"<date>", value: 170}, {"date":"<date>", value: 172}],
}
I want to change the weight of a date in the list of weights.
Or enter a new value in the height list on a specific date. What should I do?
I wrote this. but it is add a new item only.
INSERT INTO tb (UserId, Nodes)
values (1, '{"weight": [{"value":"50","date":"2021-07-24 18:17:33.000"}]}')
ON CONFLICT (UserId) DO UPDATE
SET Nodes = jsonb_set(tb.Nodes, '{"weight"}',
tb.Nodes->'weight' || '{"value":"50","date":"2021-07-24 18:17:43.000"}');
I need to edit a value for a specific date.
I self wrote this.
BEGIN;
LOCK TABLE "UserFitnessData" IN EXCLUSIVE MODE;
do $$
declare today date;
declare r jsonb; -- RECORD
declare idx int :=0;
begin
INSERT INTO "UserFitnessData" (UserId, Nodes)
values (1, '{"FrontImage": [{"value":"50","date":"2021-07-28 18:17:33.000"}]}');
EXCEPTION WHEN unique_violation THEN
today := '2021-07-25 18:17:33.000'::date;
raise notice 'exep >>> %', today;
FOR r IN
SELECT jsonb_array_elements(Nodes->'FrontImage') arr FROM "UserFitnessData" WHERE userid = 1
LOOP
if (r->>'date')::date = today then
UPDATE "UserFitnessData"
SET Nodes = Nodes #- FORMAT('{"FrontImage", %s}', idx)::text[] WHERE userid = 1;
idx := idx-1;
UPDATE "UserFitnessData"
SET Nodes = jsonb_set("UserFitnessData".Nodes, '{"FrontImage"}',
"UserFitnessData".Nodes->'FrontImage' || '{"value":"1245","date":"2021-07-27 18:17:43.000"}');
end if;
idx := idx+1;
END LOOP;
RETURN;
end $$;
COMMIT;
This is what started me right now. If anyone has a shorter solution, tell me.
Related
i need the values of a json_array. I tried this:
DECLARE
l_stuff json_array_t;
BEGIN
l_stuff := json_array_t ('["Stirfry", "Yogurt", "Apple"] ');
FOR indx IN 0 .. l_stuff.get_size - 1
LOOP
INSERT INTO t_taböe (name, type)
VALUES(l_stuff.get(i), 'TEXT');
END LOOP;
END;
You are passing the position as i instead of indx; but you need a string so use get_string(indx) as #Sayan said.
But if you try to use that directly in an insert you'll get "ORA-40573: Invalid use of PL/SQL JSON object type" because of a still-outstanding (as far as I know) bug.
To work around that you can assign the string to a variable first:
l_name := l_stuff.get_string(indx);
INSERT INTO t_taböe (name, type)
VALUES(l_name, 'TEXT');
db<>fiddle
You do not need PL/SQL and can do it in a single SQL statement:
INSERT INTO t_taböe (name, type)
SELECT value,
'TEXT'
FROM JSON_TABLE(
'["Stirfry","Yogurt","Apple"]',
'$[*]'
COLUMNS (
value VARCHAR2(50) PATH '$'
)
);
db<>fiddle here
First convert the JSON array into an ordinary PL/SQL array, then use a bulk insert.
Here is a reproducible example:
create table tab (name varchar2 (8), type varchar2 (8))
/
declare
type namelist is table of varchar2(8) index by pls_integer;
names namelist;
arr json_array_t := json_array_t ('["Stirfry", "Yogurt", "Apple"]');
begin
for idx in 1..arr.get_size loop
names(idx) := arr.get_string(idx-1);
end loop;
forall idx in indices of names
insert into tab (name, type) values (names(idx), 'TEXT');
end;
/
The query and outcomes:
select * from tab
/
NAME TYPE
-------- --------
Stirfry TEXT
Yogurt TEXT
Apple TEXT
Just use get_string:
DECLARE
l_stuff json_array_t;
BEGIN
l_stuff := json_array_t ('["Stirfry", "Yogurt", "Apple"] ');
FOR indx IN 0 .. l_stuff.get_size - 1
LOOP
--INSERT INTO t_taböe (name, type)
-- VALUES(l_stuff.get_string(indx), 'TEXT');
dbms_output.put_line(l_stuff.get_string(indx));
END LOOP;
END;
I need to store a string value in a field in a table, specifically in its Subcategory VarChar(50) column.
The value of Subcategory prior to this post processing is either 0 or 1; I need to change that to a more human-friendly value.
I haven't created a database trigger in decades and need some help with the code. This is my pseudo SQL (a hodgepodge of SQL and VB):
CREATE OR REPLACE TRIGGER tr_CustomerCategoryLog_BeforeInsert
BEFORE INSERT ON CustomerCategoryLog FOR EACH ROW
DECLARE _Category = :new.Category;
DECLARE _Subcategory = :new.Subcategory;
BEGIN
If _Category = "New"
If _Subcategory = 0
:new.Subcategory := 'New';
End If
Else If _Subcategory = 1
:new.Subcategory := 'Assumed';
End If
End If
If _Category = "Existing"
If _Subcategory = 0
:new.Subcategory := 'Existing';
End If
Else If _Subcategory = 1
:new.Subcategory := 'Organic'
End If
End If
Return "Unknown"
End Function
END;
If the logic isn't clear, in semi-plain English it is:
If the value of the Category field is "New", set the Subcategory field value also to "New" if the value of Subcategory is currently 0; otherwise, set it to "Assumed"
If the value of the Category field is "Existing", set the Subcategory field value also to "Existing" if the value of Subcategory is currently 0; otherwise, set it to "Organic"
Maybe I need to give Steely Dan's album "Trigger Logic" a listen.
UPDATE
I think the answer will work, but it's not complete enough for me.
Since I apparently have Oracle code mixed up in the pseudoSQL above, what would the complete code need to look like (to create a BeforeInsert trigger on the CustomerCategoryLog table)?
Is this more like it:
CREATE TRIGGER tr_CustomerCategoryLog_BeforeInsert
ON CustomerCategoryLog
INSTEAD OF INSERT
AS
BEGIN
SELECT
CASE
WHEN #Category = 'New'
THEN CHOOSE(#Subcategory + 1, 'New', 'Assumed')
WHEN #Category = 'Existing'
THEN CHOOSE(#Subcategory + 1, 'Existing', 'Organic')
ELSE 'Unknown'
END
END
?
I tend to avoid triggers (perhaps a character flaw... I also don't like mashed potatoes), but the following illustration could simplify your logic
Declare #Category varchar(50) = 'Existing'
Declare #Subcategory int = 1 -- works if BIT
Select case when #Category = 'New' then choose(#Subcategory+1,'New','Assumed')
when #Category = 'Existing' then choose(#Subcategory+1,'Existing','Organic')
else 'Unknown' end
Returns
Organic
I have a problem that i've spent about 3 days on.
I have a table(CDKEY) with 6 columns: CDKEYSEQ, Userseq,Banned, Communityseq, cdkey, Email.
Banned is always 0 (at this point), Userseq is NULL unless someone logged on/registered with the cdkey and email is NULL until the cdkey is registered.
Basically Userseq doesn't get filled in until a user logs in. So there will always be an email value before a user sequence value.
NOW The issue:
I'm trying to create a stored procedure that gets called when someone wants a cdkey (which they provide an email for).
The procedure first checks a table called community to make sure the Community exists.
Then if the Community exists The procedure is supposed to check the CDKEY Table for a key that has the correct community sequence AND AlSO has a NULL Value for both USERSEQ and EMAIL.
Obviously using just a select query doesnt work because there are multiple rows that are returned that match those conditions.
I tried using cursors, which got me a little further.
The problem with the cursors is that when I had two conditions after the WHERE clause, it didnt return anything.
Here is my current Procedure Code:
create or replace PROCEDURE KEYREGISTRATION(
PRODUCT_IN IN VARCHAR2 ,
in_CPUID IN LONG ,
in_MACID IN LONG ,
in_MACID2 IN LONG ,
in_HDID IN LONG ,
in_PCCores IN LONG ,
in_PCName IN VARCHAR2 ,
in_Email IN VARCHAR2 ,
out_cdkey OUT VARCHAR2 ,
returncode OUT NUMBER )
AS
CodeSuccess CONSTANT NUMBER := 0;
CoreError CONSTANT NUMBER := 2;
CodeAlreadyExists CONSTANT NUMBER := 3;
CodeBadProduct CONSTANT NUMBER := 4;
new_cdkey VARCHAR2(50);
old_cdkey VARCHAR2(50);
acommunitySeq NUMBER;
BEGIN
acommunitySeq := 0;
new_cdkey := '';
old_cdkey := '';
SELECT COMMUNITYSEQ INTO acommunityseq FROM COMMUNITY WHERE NAME = PRODUCT_IN;
returncode := CodeSuccess;
/*EXCEPTION
WHEN NO_DATA_FOUND THEN
returncode := CodeBadProduct; */
IF returncode = CodeSuccess THEN
BEGIN
SELECT CDKEY INTO old_cdkey FROM CDKEY WHERE EMAIL = in_email;
returncode := CodeBadProduct;
out_cdkey := old_cdkey;
RETURN;
EXCEPTION
WHEN NO_DATA_FOUND THEN
returncode := CodeSuccess;
END;
END IF;
IF returncode = CodeSuccess THEN
/*SELECT CDKEY into new_cdkey FROM CDKEY WHERE EMAIL = NULL AND COMMUNITYSEQ = acommunityseq; */
DECLARE
CURSOR c1
IS
SELECT CDKEY FROM CDKEY WHERE COMMUNITYSEQ = acommunityseq AND EMAIL = NULL;
BEGIN
OPEN c1;
FETCH c1 INTO new_cdkey;
IF ( c1%notfound ) THEN
returncode := CoreError;
END IF;
UPDATE cdkey SET EMAIL = in_email WHERE CDKEY = new_cdkey;
INSERT INTO user_hw VALUES( EMAIL = in_email, CPUID = in_cpuid,
MACID = in_macid, MACID2 = in_macid2, CPUCORES = in_pccores, PCNAME = in_pcname;
out_cdkey := new_cdkey;
returncode := CodeSuccess;
COMMIT;
END;
ELSE
returncode := CoreError;
ROLLBACK;
END IF;
END KEYREGISTRATION;
You think a query will not work - and your reasoning is "because a query will return too many rows." That is incorrect. Add a WHERE clause (or add to the filters you already have), with the condition ROWNUM = 1 - this will return the first row that meets all the other conditions, the processing will end, and you will get just this row and nothing else.
I want to store the content of a cursor in an associative array (Table index by binary_integer). But in the same array I also want to store an additional variable, say a boolean.
My cursor has n elements per row and the array is defined to have n+1 elements (n with the same %type as the cursorelements), the last one being the boolean.
What I whant is something like this
for cursorrow in cursor(...)
loop
array(row i) := cursorrow, boolean_variable;
end loop;
|1|2|...|n|n+1| := |1|2|...|n|, |1|
Unfortunately I can't get it to work.
Anybody knows how to do it?
How about defining a composite record type consisting of a %rowtype record and a Boolean?
Test setup:
create table demo_table
( some_id integer primary key
, some_name varchar2(30) not null unique
, some_type varchar2(10) not null );
insert all
into demo_table values (1, 'One', 'X' )
into demo_table values (2, 'Two', 'Y' )
into demo_table values (3, 'Three', 'Z' )
select * from dual;
Test:
(Edit: added dbms_output messages within loop)
declare
subtype demo_rectype is demo_table%rowtype;
type demo_rec is record
( details demo_rectype
, somecheck boolean );
type demo_tt is table of demo_rec index by pls_integer;
t_demo demo_tt;
cursor c_demo is select * from demo_table;
begin
for r in c_demo
loop
t_demo(c_demo%rowcount).details := r;
t_demo(c_demo%rowcount).somecheck := dbms_random.value < 0.5;
dbms_output.put_line
( t_demo(c_demo%rowcount).details.some_name || ': ' ||
case t_demo(c_demo%rowcount).somecheck
when true then 'TRUE'
when false then 'FALSE'
else 'NULL'
end );
end loop;
dbms_output.new_line();
dbms_output.put_line('Array t_demo contains ' || t_demo.count || ' items.');
end;
Output:
One: FALSE
Two: FALSE
Three: TRUE
Array t_demo contains 3 items.
Since the record type differs by 1 field from the table structure (i.e. the boolean), you can not assign the cursor row to the record type in 1 assignment statement. You need to do it for every table column individually:
for cursorrow in cursor(...)
loop
array(row i).col1 := cursorrow.col1;
array(row i).col2 := cursorrow.col2;
...
array(row i).coln := cursorrow.coln;
array(row i).boolean_variable := some_boolean_value;
end loop;
As mentioned in comments,you can create a record and do it. You can use the below procedure to achieve your requirement.
create or replace procedure proc_test
as
cursor cur_tab is
select a --have selected only 1 column..you can choose many
from test;
TYPE var_temp IS TABLE OF cur_tab%ROWTYPE INDEX BY PLS_INTEGER;
v_var var_temp;
/**You can add columns selected in you cursor here in your record****/
TYPE abc IS RECORD
(
id varchar2(100),
orig_name boolean
);
TYPE xx IS TABLE OF abc ;
-- initialization of record
v_xx xx := xx() ;
t boolean:=TRUE;
begin
open cur_tab;
fetch cur_tab bulk collect into v_var;
close cur_tab;
for i in 1..v_var.count
loop
v_xx.extend;
v_xx(i).id := v_var(i).a;
v_xx(i).orig_name := t;
dbms_output.put_line (v_xx(i).id ||'----'||sys.diutil.bool_to_int(v_xx(i).orig_name));
---OR
dbms_output.put_line (v_xx(i).id ||'----'||case when v_xx(i).orig_name = true then 'TRUE' ELSE 'FALSE' end );
end loop;
exception
when others then
null;
end;
Call:
execute proc_test;
I am testing out Absolute Database by ComponentAce
I have on my Form a TABSTable, TABSDatabase and a TDataSource and the data is being displayed in a TDBAdvListView, MultiSelect and RowSelect are True. I have only one Table.
When either one or more of the Items in the TDBAdvListView are selected I want to have the Database Delete the selected Records.
I have tried this way in the code below:
procedure TMain.DeleteEntry2Click(Sender: TObject);
var
i: Integer;
begin
with DBAdvListView1.DataSource.DataSet do
begin
for i := DBAdvListView1.Items.Count - 1 downto 0 do begin
if DBAdvListView1.Items[i].Selected then
begin
DBAdvListView1.DataSource.DataSet.GotoBookmark(Pointer(DBAdvListView1.Items[i]));
DBAdvListView1.DataSource.DataSet.Delete;
end;
end;
end;
end;
This always results in an Error Message:
Cannot retrieve record - Native error: 10026
I have very little experience with database programming, what am I doing wrong?
Edit:
I have added a new field into the database named ID as an integer starting from 0 in the hopes that I can reference them with the Locate method and tried with the code below. This produces no error but will only delete the top record in the ListView and if I select more than one it will delete different records than selected.
My new code:
procedure TMain.DeleteEntry2Click(Sender: TObject);
var
i: Integer;
begin
with DBAdvListView1.DataSource.DataSet do
begin
DBAdvListView1.BeginUpdate;
First;
for i := DBAdvListView1.Items.Count - 1 downto 0 do begin
if DBAdvListView1.Items[i].Selected then
begin
if dbTable.Locate('ID',DBAdvListView1.Items[i].Selected,[]) then
dbTable.Delete;
Next;
end;
end;
dbTable.Close;
dbTable.Open;
DBAdvListView1.EndUpdate;
end;
end;
The dbTable has to be Closed and Opened to see changes for some strange reason - I have tried Refresh to no avail...
Edit:
// To include Table Structure as requested...
ID integer 0
Title string 200
Author string 100
Date string 20
Location string 60
Category string 100
ISBN-13 string 20
ISBN-10 string 20
In the Absolute Database Utils directory there is a DatabaseManager.exe which I used to create the actual table with and in here I have also now set a Primary Key of the type:
Type - Primary
Name - ID
The fields for the Index:
ColumnName - ID
CaseInsensitive - False
ASC - True
MaxIndexSize - 20
If you know the primary keys of all the records to be deleted, then you can use one SQL query statement in order to delete all the selected records in one go -
delete from table
where id in (1, 7, 15, 23, 45);
You would have to build this query manually, i.e. create the string which holds the id numbers.
Answer to own question... I remove the selected records with the code below:
procedure TMain.Button5Click(Sender: TObject);
var
i: Integer;
begin
with DBAdvListView1 do
for i := 0 to Items.Count - 1 do
if Items[i].Selected then
begin
Memo1.Lines.Add(Items[i].Caption + ' - Selected!'); //Test for Correct ID's!
if dbTable.Locate('ID', Items[i].Caption, []) then
DBAdvListView1.Datasource.DataSet.Delete;
end;
dbTable.Close;
dbTable.Open;
end;
second code will work if you just remove the Next; from the if section. Since u're already inside a 'for' loop, there's no need to use the Database.Next.