Oracle Sequence wastes/reserves values (in INSERT SELECT) - sql

I've been struggling with sequences for a few days. I have this Origin data table called "datos" with the next columns:
CENTRO
CODV
TEXT
INCIDENCY
And a Destiny data table called "anda" with the following:
TIPO = 31 (for all rows)
DESCRI = 'Site' (for all rows)
SECU = sequence number generated with Myseq.NEXTVAL
CENTRO
CODV
TEXT
The last three columns must be filled in with data from "datos" table.
When I execute my query, it all works fine, my table is filled and the sequence generates its values. But, in the INSERT INTO SELECT, I have the following conditions:
Every row in origin "datos" must not already be in the destiny "anda", so it won't be duplicated, and every row in "datos" must have the INCIDENCY flag value to 'N' or NULL.
If each row matches the conditions, it should be filled.
The thing is, that the query works fine and I have been trying with many different values. Here comes the problem:
When a row has its INCIDENCY value set to 'Y' (so it must not be copied into destiny table), it doesn't appear, but the sequence DOES consumes one value, and when I check Myseq.NEXTVAL its value is higher.
How can I prevent the sequence to add any value when it doesn't match the conditions? I've read that Oracle first reserves all the possible values returning from the SELECT query, but I can't find how to prevent it.
Here's the SQL:
INSERT INTO anda (TIPO, DESCRI, SECU, CENTRO, CODV, TEXT)
SELECT( 31 TIPO,
'Site' DESCRI,
Myseq.NEXTVAL,
datos.CENTRO,
datos.CODV,
datos.TEXT
FROM datos
WHERE (CENTRO, CODV) NOT IN
(SELECT CENTRO, CODV
FROM anda)
AND (datos.INCIDENCY = 'N' OR datos.INCIDENCY IS NULL)
)
Thanks in advance!!
Definition of MySeq
CREATE SEQUENCE CREATE SEQUENCE "BBDD"."MySeq" MINVALUE 800000000000
MAXVALUE 899999999999 INCREMENT BY 1 START WITH 800000000000 CACHE 20 ORDER NOCYCLE ;

You might be able to trick Oracle into doing this with a CTE:
INSERT INTO anda (TIPO, DESCRI, SECU, CENTRO, CODV, TEXT)
WITH toinsert as (
SELECT d.*
FROM datos d
WHERE (CENTRO, CODV) NOT IN (SELECT CENTRO, CODV FROM anda) AND
(d.INCIDENCY = 'N' OR d.INCIDENCY IS NULL)
)
SELECT 31 as TIPO, 'Site' as DESCRI, Myseq.NEXTVAL,
d.CENTRO, d.CODV, d.TEXT
FROM toinsert d;
I'm not quite sure if that will work. A more guaranteed approach is to use a before insert trigger (or identity column if you are using 12c+). You would increment the value in the trigger.
However, I do agree with Hugh Jones. You should be confident using the sequence to add a unique value to each row and this value will be increasing. Gaps can appear for other reasons, such as deletes. Also, I know that SQL Server can create gaps when doing parallel inserts; I'm not sure if that also happens with Oracle.

I don't believe you have a real problem(the gaps are not really an issue) but you can put a before insert (at row level) trigger on anda table and set sequ there with your sequence generated value.
But keep in mind that this will keep consecutive only the sequ number in a statement. You'll get gaps anyway for other reasons.
UPDATE: as Alex Poole has commented, the insert itself does not generate gaps.
See a test below:
> drop sequence tst_fgg_seq;
sequence TST_FGG_SEQ dropped.
> drop table tst_fgg;
table TST_FGG dropped.
> drop table tst_insert_fgg;
table TST_INSERT_FGG dropped.
> create sequence tst_fgg_seq start with 1 nocycle;
sequence TST_FGG_SEQ created.
> create table tst_fgg as select level l from dual connect by level < 11;
table TST_FGG created.
> create table tst_insert_fgg as
select tst_fgg_seq.nextval
from tst_fgg
where l between 3 and 5;
table TST_INSERT_FGG created.
> select * from tst_insert_fgg;
NEXTVAL
----------
1
2
3
> insert into tst_insert_fgg
select tst_fgg_seq.nextval
from tst_fgg
where l between 3 and 5;
3 rows inserted.
> select * from tst_insert_fgg;
NEXTVAL
----------
1
2
3
4
5
6
6 rows selected

Related

I don't understand how make task on SQL

There is a table with two fields: Id and Timestamp.
Id is an increasing sequence. Each insertion of a new record into the table leads to the generation of ID(n)=ID(n-1) + 1. Timestamp is a timestamp that, when inserted retroactively, can take any values less than the maximum time of all previous records.
Retroactive insertion is the operation of inserting a record into a table in which
ID(n) > ID(n-1)
Timestamp(n) < max(timestamp(1):timestamp(n-1))
Example of a table:
ID
Timestamp
1
2016.09.11
2
2016.09.12
3
2016.09.13
4
2016.09.14
5
2016.09.09
6
2016.09.12
7
2016.09.15
IDs 5 and 6 were inserted retroactively (their timestamps are lower than later records).
I need a query that will return a list of all ids that fit the definition of insertion retroactively. How can I do this?
It can be rephrased to :
Find every entries for which, in the same table, there is an entry with a lesser id (a previous entry) having a greater timestamp
It can be achieved using a WHERE EXISTS clause :
SELECT t.id, t.timestamp
FROM tbl t
WHERE EXISTS (
SELECT 1
FROM tbl t2
WHERE t.id > t2.id
AND t.timestamp < t2.timestamp
);
Fiddle for MySQL It should work with any DBMS, since it's a standard SQL syntax.

Reduce the table fragmentation Postgresql

I have a table with n elements with ids from 1 to n. Then I delete nearly all the data from the table, leaving only m elements, m << n, including element with id=n. Before deleting the difference between ids of adjacent rows was 1, now it could be much larger. I need to update the table such a way that remaining element ids were from 1 to m and difference between ids of adjacent rows was 1. How I can do this? I use PostgreSQL.
You may find a better option to not update the primary key (especially in heavy multiuser einvironment) but to delegate the removal of the gaps to the accessing view.
Simple Example
create table test as
select id from generate_series(1,5) id;
delete from test
where id in (2,3,4);
The view assigns teh new continuous sequence using the row_number function.
create view test_gapless as
select
id,
row_number() over (order by id) gapless_id
from test;
select * from test_gapless;
id, gapless_id
1 1
5 2

SQL - Split container row conditionally

In the process of migrating containers, if we have two tables;
TABLE_MAPPING (old_value, new_value)
TABLE_USING (value, data...)
TABLE_USING is referencing (FK) a container in a irrelevant table.
TABLE_MAPPING is temporarily used for a migration, the goal is to move contents from deprecated to new containers.
The problem here is that sometimes the container is not only replaced, but split into multiple new containers, for example TABLE_MAPPING could contain:
OLD_VALUE
NEW VALUE
1
10
1
11
2
20
And the query would result in an "update" of one row with value '1' to two rows with values '10' and '11'.
Is there a plain SQL way to do that? Or should I use PL/SQL?
EDIT: as requested, here is an example of before/after using the TABLE_MAPPING above
Before:
VALUE
IRRELEVANT_COLUMNS ...
1
...
2
...
After:
VALUE
IRRELEVANT_COLUMNS ...
10
...
12
...
20
...
You need two steps. Below I first insert all new rows, then I delete all old rows.
-- insert rows with new values
insert into table_using (value, data ...)
select m.new_value, u.data ...
from table_using u
join table_mapping m on m.old_value = u.value;
-- delete rows with old values
delete from table_using where value in (select old_value from table_mapping);
-- commit the transaction
commit;

Update SQL column with sequenced number

I have a table SL_PROD which has the following columns, NUMBER, DEPTCODE, DISP_SEQ AND SL_PROD_ID.
SL_PROD_ID is an identity column which incrementally increases with each row.
I need to write a query which updates the DISP_SEQ column with sequential numbers (1-X) for the rows which have a DEPTCODE of '725'. I've tried several things with no luck, any ideas?
Try this:
A common table expression can be used in updates. This is extremely usefull, if you want to use the values of window functions (with OVER) as update values.
Attention: Look carefully what you are ordering for. I used NUMBER but you might need some other sort column (maybe your IDENTITY column)
CREATE TABLE #SL_PROD(NUMBER INT,DEPT_CODE INT,DISP_SEQ INT,SL_PROD_ID INT IDENTITY);
INSERT INTO #SL_PROD(NUMBER,DEPT_CODE,DISP_SEQ) VALUES
(1,123,0)
,(2,725,0)
,(3,725,0)
,(4,123,0)
,(5,725,0);
WITH UpdateableCTE AS
(
SELECT ROW_NUMBER() OVER(ORDER BY NUMBER) AS NewDispSeq
,DISP_SEQ
FROM #SL_PROD
WHERE DEPT_CODE=725
)
UPDATE UpdateableCTE SET DISP_SEQ=NewDispSeq;
SELECT * FROM #SL_PROD;
GO
--Clean up
--DROP TABLE #SL_PROD;
The result (look at the lines with 725)
1 123 0 1
2 725 1 2
3 725 2 3
4 123 0 4
5 725 3 5

Need a SQL statement focus on combination of tables but entries always with unique ID

I need SQL code to solve the tables combination problem, described on below:
Table old data: table old
name version status lastupdate ID
A 0.1 on 6/8/2010 1
B 0.1 on 6/8/2010 2
C 0.1 on 6/8/2010 3
D 0.1 on 6/8/2010 4
E 0.1 on 6/8/2010 5
F 0.1 on 6/8/2010 6
G 0.1 on 6/8/2010 7
Table new data: table new
name version status lastupdate ID
A 0.1 on 6/18/2010
#B entry deleted
C 0.3 on 6/18/2010 #version_updated
C1 0.1 on 6/18/2010 #new_added
D 0.1 on 6/18/2010
E 0.1 off 6/18/2010 #status_updated
F 0.1 on 6/18/2010
G 0.1 on 6/18/2010
H 0.1 on 6/18/2010 #new_added
H1 0.1 on 6/18/2010 #new_added
the difference of new data and old date:
B entry deleted
C entry version updated
E entry status updated
C1/H/H1 entry new added
What I want is always keeping the ID - name mapping relationship in old data table no matter how data changed later, a.k.a the name always has an unique ID number bind with it.
If entry has update, then update the data, if entry is new added, insert to the table then give a new assigned unique ID. If the entry was deleted, delete the entry and do not reuse that ID later.
However, I can only use SQL with simple select or update statement then it may too hard for me to write such code, then I hope someone with expertise can give direction, no details needed on the different of SQL variant, a standard sql code as sample is enough.
Thanks in advance!
Rgs
KC
========
I listed my draft sql here, but not sure if it works, some one with expertise pls comment, thanks!
1.duplicate old table as tmp for store updates
create table tmp as
select * from old
2.update into tmp where the "name" is same in old and new table
update tmp
where name in (select name from new)
3.insert different "name" (old vs new) into tmp and assign new ID
insert into tmp (name version status lastupdate ID)
set idvar = max(select max(id) from tmp) + 1
select * from
(select new.name new.version new.status new.lastupdate new.ID
from old, new
where old.name <> new.name)
4. delete the deleted entries from tmp table (such as B)
delete from tmp
where
(select ???)
You never mentioned what DBMS you are using but if you are using SQL Server, one really good one is the SQL MERGE statement. See: http://www.mssqltips.com/tip.asp?tip=1704
The MERGE statement basically works as
separate insert, update, and delete
statements all within the same
statement. You specify a "Source"
record set and a "Target" table, and
the join between the two. You then
specify the type of data modification
that is to occur when the records
between the two data are matched or
are not matched. MERGE is very useful,
especially when it comes to loading
data warehouse tables, which can be
very large and require specific
actions to be taken when rows are or
are not present.
Example:
MERGE Products AS TARGET
USING UpdatedProducts AS SOURCE
ON (TARGET.ProductID = SOURCE.ProductID)
--When records are matched, update
--the records if there is any change
WHEN MATCHED AND TARGET.ProductName <> SOURCE.ProductName
OR TARGET.Rate <> SOURCE.Rate THEN
UPDATE SET TARGET.ProductName = SOURCE.ProductName,
TARGET.Rate = SOURCE.Rate
--When no records are matched, insert
--the incoming records from source
--table to target table
WHEN NOT MATCHED BY TARGET THEN
INSERT (ProductID, ProductName, Rate)
VALUES (SOURCE.ProductID, SOURCE.ProductName, SOURCE.Rate)
--When there is a row that exists in target table and
--same record does not exist in source table
--then delete this record from target table
WHEN NOT MATCHED BY SOURCE THEN
DELETE
--$action specifies a column of type nvarchar(10)
--in the OUTPUT clause that returns one of three
--values for each row: 'INSERT', 'UPDATE', or 'DELETE',
--according to the action that was performed on that row
OUTPUT $action,
DELETED.ProductID AS TargetProductID,
DELETED.ProductName AS TargetProductName,
DELETED.Rate AS TargetRate,
INSERTED.ProductID AS SourceProductID,
INSERTED.ProductName AS SourceProductName,
INSERTED.Rate AS SourceRate;
SELECT ##ROWCOUNT;
GO
Let me start from the end:
In #4 you would delete all rows in tmp; what you wanted to say there is WHERE tmp.name NOT IN (SELECT name FROM new); similarly #3 is not correct syntax, but if it was it would try to insert all rows.
Regarding #2, why not use auto increment on the ID?
Regarding #1, if your tmp table is the same as new the queries #2-#4 make no sense, unless you change (update, insert, delete) new table in some way.
But (!), if you do update the table new and it has an auto increment field on ID and if you are properly updating the table (using ID) from the application then your whole procedure is unnecessary (!).
So, the important thing is that you should not design the system to work like above.
To get the concept of updating data in the database from the application side take a look at examples here (php/mysql).
Also, to get the syntax correct on your queries go through the basic version of SET, INSERT, DELETE and SELECT commands (no way around this).
Note - if you are concerned about performance you can skip this whole answer :-)
If you can redesign have 2 tables - one with the data and other with the name - ID linkage. Something like
table_original
name version status lastupdate
A 0.1 on 6/8/2010
B 0.1 on 6/8/2010
C 0.1 on 6/8/2010
D 0.1 on 6/8/2010
E 0.1 on 6/8/2010
F 0.1 on 6/8/2010
G 0.1 on 6/8/2010
and name_id
name ID
A 1
B 2
C 3
D 4
E 5
F 6
G 7
When you get the table_new with the new set of data
TRUNCATE table_original
INSERT INTO name_id (names from table_new not in name_id)
copy table_new to table_original
Note : I think there's a bit of ambiguity about the deletion here
If the entry was deleted, delete the
entry and do not reuse that ID later.
If name A gets deleted, and it turns up again in a later set of updates do you want to a. reuse the original ID tagged to A, or b. generate a new ID?
If it's b. you need a column Deleted? in name_id and a last step
4 . set Deleted? = Y where name not in table_original
and 2. would exclude Deleted? = Y records.
You could also do the same thing without the name_id table based on the logic that the only thing you need from table_old is the name - ID links. Everything else you need is in table_new,
This works in Informix and gives exactly the display you require. Same or similar should work in MySQL, one would think. The trick here is to get the union of all names into a temp table and left join on that so that the values from the other two can be compared.
SELECT DISTINCT name FROM old
UNION
SELECT DISTINCT name FROM new
INTO TEMP _tmp;
SELECT
CASE WHEN b.name IS NULL THEN ''
ELSE aa.name
END AS name,
CASE WHEN b.version IS NULL THEN ''
WHEN a.version = b.version THEN a.version
ELSE b.version
END AS version,
CASE WHEN a.status = b.status THEN a.status
WHEN b.status IS NULL THEN ''
ELSE b.status
END AS status,
CASE WHEN a.lastupdate = b.lastupdate THEN a.lastupdate
WHEN b.lastupdate IS NULL THEN null
ELSE b.lastupdate
END AS lastupdate,
CASE WHEN a.name IS NULL THEN '#new_added'
WHEN b.name IS NULL THEN '#' || aa.name || ' entry deleted'
WHEN a.version b.version THEN '#version_updated'
WHEN a.status b.status THEN '#status_updated'
ELSE ''
END AS change
FROM _tmp aa
LEFT JOIN old a
ON a.name = aa.name
LEFT JOIN new b
ON b.name = aa.name;
a drafted approach, I have no idea if it works fine......
CREATE TRIGGER auto_next_id
AFTER INSERT ON table FOR EACH ROW
BEGIN
UPDATE table SET uid = max(uid) + 1 ;
END;
If I understood well what you need based on the comments in the two tables, I think you can simplify a lot your problem if you don't merge or update the old table because what you need is table new with the IDs in table old when they exist and new IDs when they do not exist, right?
New records: table new has the new records already - OK (but they need a new ID)
Deleted Records: they are not in table new - OK
Updated Records: already updated in table new - OK (need to copy ID from table old)
Unmodified records: already in table new - OK (need to copy ID from table old)
So the only thing you need to do is to:
(a) copy the IDs from table old to table new when they exist
(b) create new IDs in table new when they do not exist in table old
(c) copy table new to table old.
(a) UPDATE new SET ID = IFNULL((SELECT ID FROM old WHERE new.name = old.name),0);
(b) UPDATE new SET ID = FUNCTION_TO GENERATE_ID(new.name) WHERE ID = 0;
(c) Drop table old;
CREATE TABLE old (select * from new);
As I don't know which SQL database you are using, in (b) you can use an sql function to generate the unique id depending on the database. With SQL Server, newid(), With postgresql (not too old versions), now() seems a good choice as its precision looks sufficient (but not in other databases as MySQL for example as I think the precision is limited to seconds)
Edit: Sorry, I hadn't seen you're using sqlite and python. In this case you can use str(uuid.uuid4()) function (uuid module) in python to generate the uuid and fill the ID in new table where ID = 0 in step (b). This way you'll be able to join 2 independent databases if needed without conflicts on the IDs.
Why don't you use a UUID for this? Generate it once for a plug-in, and incorporate/keep it into the plug-in, not into the DB. Now that you mention python, here's how to generate it:
import uuid
UID = str(uuid.uuid4()) # this will yield new UUID string
Sure it does not guarantee global uniqueness, but chances you get the same string in your project is pretty low.