I'm looking for the best strategy to reset a sequence which is used to generate unique keys. These keys are build using a prefix value, also generated by a combination of month and year values, like the following example:
2020060100000001 - Year(4d)Month(2d)Sequence(8d)
The sequence value must be restarted at the first second of each new month.
Is there any Oracle event that allows calling a function or procedure based in this situation to do this job?
Can anyone help me with opinions and experiences?
Thanks a lot!
Rodrigo
If you really want to set the value of a sequence you can use something like the following:
PROCEDURE SET_SEQUENCE(pinSequence_owner IN VARCHAR2,
pinSequence_name IN VARCHAR2,
pinNew_next_value IN NUMBER,
pinDebug IN BOOLEAN := FALSE)
IS
strSQL VARCHAR2(4000);
nNext_number NUMBER;
nOriginal_increment NUMBER;
nNew_nextval NUMBER;
nNew_last_number NUMBER;
BEGIN
strSQL := 'SELECT s.LAST_NUMBER, INCREMENT_BY ' ||
'FROM DBA_SEQUENCES s ' ||
'WHERE s.SEQUENCE_OWNER = ''' || pinSequence_owner || ''' AND ' ||
's.SEQUENCE_NAME = ''' || pinSequence_name || '''';
EXECUTE IMMEDIATE strSQL INTO nNext_number, nOriginal_increment;
-- Note that DBA_SEQUENCES.LAST_NUMBER represents the *next* number which will be
-- returned by a call to NEXTVAL.
IF pinNew_next_value NOT IN (nNext_number-1, nNext_number)
THEN
strSQL := 'ALTER SEQUENCE ' || pinSequence_owner || '.' || pinSequence_name ||
' INCREMENT BY ' || TO_CHAR(pinNew_next_value - nNext_number) || ' NOCACHE';
EXECUTE IMMEDIATE strSQL;
strSQL := 'SELECT ' || pinSequence_owner || '.' || pinSequence_name || '.NEXTVAL FROM DUAL';
EXECUTE IMMEDIATE strSQL INTO nNew_nextval;
strSQL := 'ALTER SEQUENCE ' || pinSequence_owner || '.' || pinSequence_name ||
' INCREMENT BY ' || nOriginal_increment || ' NOCACHE';
EXECUTE IMMEDIATE strSQL;
strSQL := 'SELECT s.LAST_NUMBER FROM DBA_SEQUENCES s WHERE s.SEQUENCE_OWNER = ''' || pinSequence_owner ||
''' AND s.SEQUENCE_NAME = ''' || pinSequence_name || '''';
EXECUTE IMMEDIATE strSQL INTO nNew_last_number;
END IF;
END SET_SEQUENCE;
Resetting sequence every month is not a valid approach. Basically you are violating the utilization of sequence. From your question i could understand, you wanted to append year and month to your column. In this case you can simple concatenate while inserting,
year||month||sequence_name.nextval
Eg: 2020||02||sequence_name.nextval
If you still wanted to reset, you can create a trigger to reset the sequence every month,
Related
I want to delete rows with an "Execute immediate" because the table name is in a variable.
How can I count the number of lines deleted?
I tried this, but it does not work with the INTO v_LINE_REMOVE;
v_sql := '
DELETE /*+parallel(t,4)*/
FROM "' || v_owner || '"."' || v_table_name ||'" t
where t."'|| v_column_name ||'" in (
select /*+parallel(rem,4)*/
rem.' || v_type_data || '
from ' || v_table_listeremove || ' rem
WHERE rem.dt_vact = '''|| v_dt_vact ||'''
)
';EXECUTE IMMEDIATE v_sql;--INTO v_LINE_REMOVE;
Thanks a lot
You should be able to use SQL%ROWCOUNT after running your DML statement
EXECUTE IMMEDIATE v_sql;
v_line_remove := SQL%ROWCOUNT;
I am trying to use insert into statement, but getting the error:
ORA-01628: max # extents (32765) reached for rollback segment _SYSSMU134_1882489978$
Increasing the UNDO tablespace is not an option, so I would like a way to insert those data by chunks (for example with 1 million rows at a time). Can someone help to rewrite this procedure in that way?
CREATE OR REPLACE PROCEDURE create_chunks (
p_source_table IN VARCHAR2,
p_table_name_chunk IN VARCHAR2,
p_chunks IN VARCHAR2
) AS
v_insert_sql CLOB;
BEGIN
v_insert_sql := 'INSERT INTO ' || p_table_name_chunk ||
' (rid, chunk_number) ' ||
'SELECT /*+ parallel(64) */ rowid rid,' ||
'mod( ora_hash(rowid), :p_chunks ) as chunk_number '
'FROM ' || p_source_table;
EXECUTE IMMEDIATE v_insert_sql USING p_chunks;
COMMIT;
END;
This v_insert_sql is failing with above mentioned error. I have working solution using the cursor fetching like that:
DECLARE
CURSOR v_cur IS SELECT /*+ parallel(64) */
rowid rid, mod( ora_hash(rowid), 20000 ) AS chunk_number
-- I need this table to be parametric name
FROM some_table;
TYPE t_sample IS TABLE OF v_cur%ROWTYPE;
v_sample t_sample;
v_row_limit CONSTANT NUMBER := 1000000;
BEGIN
OPEN v_cur;
LOOP
FETCH v_cur BULK COLLECT INTO v_sample LIMIT v_row_limit;
FORALL i IN v_sample.first .. v_sample.last
INSERT INTO chunk_table VALUES v_sample(i);
COMMIT;
EXIT WHEN v_cur%NOTFOUND;
END LOOP;
CLOSE v_cur;
END;
I can't move this cursor straight into the procedure as the table name is varying and I need it to be parametric as with cursor approach I have to repeat the same code for different tables. So the question is how to deal with this?
Basically, I was able to just past the whole chunks approach inside procedure as dynamic query, like that:
CREATE OR REPLACE PROCEDURE create_chunks (
p_source_table IN VARCHAR2,
p_table_name_chunk IN VARCHAR2,
p_chunks IN VARCHAR2
) AS
v_insert_sql CLOB;
BEGIN
v_insert_sql := '' ||
' DECLARE ' || CHR(10) ||
' CURSOR cur1 IS SELECT ' || CHR(10) ||
' /*+ parallel(64) full(tbn)*/ rowid rid,' || CHR(10) ||
' mod( ora_hash(rowid), :p_chunks ) AS chunk_number' || CHR(10) ||
' FROM ' || p_source_table || ' tbn;' || CHR(10) ||
' TYPE t_sample IS TABLE OF cur1%ROWTYPE;' || CHR(10) ||
' v_sample t_sample;' || CHR(10) ||
' v_row_limit CONSTANT NUMBER := 1000000;' || CHR(10) ||
' BEGIN' || CHR(10) ||
' OPEN cur1;' || CHR(10) ||
' LOOP' || CHR(10) ||
' FETCH cur1 BULK COLLECT INTO v_sample LIMIT v_row_limit;' || CHR(10) ||
' FORALL i IN v_sample.first .. v_sample.last' || CHR(10) ||
' INSERT INTO ' || p_table_name_chunk || ' VALUES v_sample(i);' || CHR(10) ||
' COMMIT;' || CHR(10) ||
' EXIT WHEN cur1%NOTFOUND;' || CHR(10) ||
' END LOOP;' || CHR(10) ||
' CLOSE cur1;' || CHR(10) ||
' END;';
EXECUTE IMMEDIATE v_insert_sql USING p_chunks;
COMMIT;
END;
I am attempting to add an AND statement to a dynamic WHERE clause using a bind variable and I am receiving the following Oracle error:
ORA-01830: date format picture ends before converting entire input string bind variable
Here is the offending code:
FUNCTION WhereClause RETURN VARCHAR2
IS
BEGIN
where_sql := 'WHERE TRUNC( ' || parm_rec.SRC_DATE_COLUMN || ' ) < ADD_MONTHS( ' ||
'ADD_MONTHS ( TRUNC ( NVL ( :SYS_OFFSET, SYSDATE ) - ( :DAY_OFFSET )), ' ||
'( :MON_OFFSET * :kNEGATIVE ) ), ' ||
'( :YR_OFFSET * ( :kANNUM * :kNEGATIVE ) ) ) ';
RETURN where_sql;
END WhereClause;
PROCEDURE ArchiveSrcDateFilter
IS
BEGIN
DBMS_OUTPUT.PUT_LINE('ArchiveDynamic - ENTER');
IF parm_rec.SRC_DATE_COLUMN IS NULL THEN parm_rec.SRC_DATE_COLUMN := 'NULL';
END IF;
FOR i in tbl_cur
LOOP
where_sql := WhereClause; -- defines the WHERE clause (where_sql) via function, Spec will not return variable to body?
/*** DYNAMIC SQL DECLARATIONS ***/
arc_sql := 'DECLARE ' ||
/*** DYNAMIC %ROWTYPE SELECT ***/
'CURSOR arc_cur IS ' ||
'SELECT * '||
'FROM ' || i.ARC_TABLE_NAME || '; '|| --obtain ARCHIVE ARC_SCHEMA_NAME.ARC_TABLE_NAME
'TYPE arc_cur_type IS TABLE OF arc_cur%ROWTYPE; ' || -- dynamically set archive record cursor %ROWTYPE for BULK COLECT as table collection
'arc_rec arc_cur_type; ' || -- define archive record as TABLE OF cursor.%ROWTYPE
/*** ARCHIVE PARAMETERS CURSOR - SRC_CREATE_DATE IS NOT NULL***/
'CURSOR parm_cur IS '||
'SELECT :seq_val AS ARCHIVE_ID, '||
'A.*, ' ||
'SYSDATE AS ARCHIVE_DATE ' ||
'FROM ' || srcSchemaTable || ' A ' || -- archive SRC_SCHEMA_NAME.SRC_TABLE_NAME (source table not archive table)
where_sql || ' || :ADD_FILTER ' || ' ; ' ||
/*** DYNAMIC SQL STATEMENT BODY ***/
'BEGIN '||
'IF parm_cur%ISOPEN THEN CLOSE parm_cur; ' ||
'END IF; ' ||
'OPEN parm_cur; ' ||
'LOOP ' ||
'FETCH parm_cur ' ||
'BULK COLLECT INTO arc_rec LIMIT 500; ' ||
'EXIT WHEN arc_rec.COUNT = 0; ' ||
'FORALL i IN 1..arc_rec.COUNT ' ||
'INSERT INTO ' || arcTable ||
' VALUES arc_rec( i );' ||
'DBMS_OUTPUT.PUT_LINE( ''ARC_REC_COUNT: '' || arc_rec.COUNT ); ' ||
'END LOOP; ' ||
'CLOSE parm_cur; ' ||
'DBMS_OUTPUT.PUT_LINE(''SUCCESS...''); '||
'END; ';
DBMS_OUTPUT.PUT_LINE('ArchiveDynamic - INSIDE LOOP: ' || arc_sql );
EXECUTE IMMEDIATE arc_sql
USING seq_val, parm_rec.SYS_OFFSET, parm_rec.DAY_OFFSET, parm_rec.MON_OFFSET, kNEGATIVE, parm_rec.YR_OFFSET, kANNUM, parm_rec.ADD_FILTER;
END LOOP;
END ArchiveSrcDateFilter;
This is the specific piece of code in the ArchiveSrcFilter procedure where the :ADD_FILTER bind variable is located (Note: I have attempted different concatenation iterations for the bind variable without success this was simply my last attempt before posting the issue here) :
'CURSOR parm_cur IS '||
'SELECT :seq_val AS ARCHIVE_ID, '||
'A.*, ' ||
'SYSDATE AS ARCHIVE_DATE ' ||
'FROM ' || srcSchemaTable || ' A ' ||
where_sql || ' || :ADD_FILTER ' || ' ; ' ||
And the EXECUTE IMMEDIATE USING with the last parameter as the bind:
EXECUTE IMMEDIATE arc_sql
USING seq_val, parm_rec.SYS_OFFSET, parm_rec.DAY_OFFSET, parm_rec.MON_OFFSET, kNEGATIVE, parm_rec.YR_OFFSET, kANNUM, parm_rec.ADD_FILTER;
the parm_rec.ADD_FILTER = AND STATUS = 1062
Is it possible to do what I am attempting by concatenating the bind to the where?
I don't understand the odd error message I am receiving given that the code executes without exception if I concatenate the parm_rec.ADD_FILTER object variable or hard code the AND STATUS = 1062.
I can concatenate the parm_rec.ADD_FILTER in place of the bind variable and the code executes without exception, but I have been unsuccessful in attempts to get the bind variable to work.
I'm grateful for any suggestions and/or insight.
Thanks!
No. Column names (i.e. STATUS) and operators (i.e. AND, =) can never be resolved from bind variables.
I have already asked almost the same question before:
Finding non-numeric values in varchar column
And I received some great answer's and idea's to make not only fix the issue in my query but also make it much more efficient.
Old Query:
Select * From Table_Name
Where
(NOT REGEXP_LIKE(COLUMN_NAME, '^-?[0-9.]+$')
OR
LENGTH(Function_To_Fetch_Left_Component(COLUMN_NAME)) > (Precision-Scale)
OR
LENGTH(Column_Name) - LENGTH(REPLACE(Column_Name,'.','')) > 1)
New Query:
Select * From Table_Name
WHERE (Translate(column_name,'x0123456789-.','x') is not null
OR column_name = '.'
OR column_name = '-'
OR column_name = '-.'
OR Instr(column_name,'-') > 1
OR column_name like '%-%-%'
OR column_name like '%.%.%'
OR Instr(column_name||'.','.')-1 > (precision-scale) ) ;
Known limitations which are okay:
1.) Doesn't handle scientific notation. Its by design as I want to reject those.
2.) Doesn't check total precision (decimal component) however as while inserting oracle itself round's it off I am fine with it.
3.) Doesn't check for leading zero's.
Please help to see if the above query cover's everything and I haven't missing anything.
Another great solution by one Stackoverflow member however increased execution time by 2.5 times:
Function:
CREATE OR REPLACE function temp_is_number(p_test_value varchar2,
p_scale NUMBER,
p_precision NUMBER) return varchar2 is
l_result varchar2(1);
begin
execute immediate 'DECLARE l_number NUMBER; ' ||
'l_test_number NUMBER(' || p_precision || ',' || p_scale || '); ' ||
'BEGIN ' ||
' l_number := cast(:b1 as number); ' ||
' l_test_number := cast(:b1 as number); ' ||
' IF (l_number = l_test_number) THEN ' ||
' :b2 := ''Y''; ' ||
' ELSE ' ||
' :b2 := ''N''; ' ||
' END IF; ' ||
'EXCEPTION ' ||
' WHEN OTHERS THEN ' ||
' :b2 := ''N''; ' ||
'END;'
using in p_test_value, out l_result;
return l_result;
end;
/
Select Count(1) from Table_Name Where temp_is_number (Column_Name,scale,precision) = 'N';
I have below query which gives an error as encounter an symbol ( in the line where loop is used. I am trying to develop a function which takes dynamic paramater as table_name,column_name,table_id and used for other tables as well.
FUNCTION get_encryp_pass( table_name IN varchar2,column_name IN varchar2,table_id IN varchar2) RETURN VARCHAR2
IS
BEGIN
EXECUTE IMMEDIATE 'for c1 in (select * from' || table_name ||) loop
EXECUTE IMMEDIATE 'update ' || table_name || ' set ' || column_name = encrypt_val(c1.column_name) || ' where ' || table_id || ' = ' || c1.table_id and column_name is not null;
end loop;
END get_encrypt_pass;
this should work:
CREATE PROCEDURE get_encryp_pass(table_name IN varchar2,
column_name IN varchar2,
table_id IN varchar2) IS
BEGIN
EXECUTE IMMEDIATE 'begin for c1 in (select * from ' || table_name ||
') loop update ' || table_name || ' set ' ||
column_name || ' = encrypt_val(c1.' || column_name ||
') where ' || table_id || ' = c1.'||table_id||' and ' || column_name ||
' is not null; end loop; end;'
;
END;
But why not simply call update FTP_SFTP_SERVER set PASSWORD=encrypt_val(PASSWORD) where PASSWORD is not null ?
keep care of what is a variable and what is a string-literal and must be single-quoted therefore ... and string-variables mus be double-quoted:
EXECUTE IMMEDIATE 'update ' || table_name || ' set ' || column_name || ' = ''' || encrypt_val(c1.column_name) || ''' where ' || table_id || ' = ' || c1.table_id || ' and column_name is not null';
Best practice is to concatenate the statement in a varchar2-variable first and inspect this. If the content of the variable is syntactical correct and executable, the EXECUTE IMMEDIATE should work as well
declare
stmt varchar2(4000);
begin
stmt := 'update ' || table_name || ' set ' || column_name || ' = ''' || encrypt_val(c1.column_name) || ''' where ' || table_id || ' = ' || c1.table_id || ' and column_name is not null';
EXECUTE IMMEDIATE stmt;
end;
I think i have one alternative for your question. MERGE can be used in this case. Please try it i dont have workspaceso dint test it but it should work Let me know if it helps.
FUNCTION get_encryp_pass(
table_name IN VARCHAR2,
column_name IN VARCHAR2,
table_id IN VARCHAR2)
RETURN VARCHAR2
IS
BEGIN
EXECUTE IMMEDIATE 'MERGE INTO '||table_name||' t1 USING
(
SELECT * FROM '||table_name|| ')t2
ON
(
t1.table_id = t2.table_id
)
WHEN MATCHED THEN
UPDATE SET t1.'||column_name||' = encrypt_val(t2.'||column_name||')'
||' WHERE and t1.'||column_name||' IS NOT NULL';
END;