I'm trying to use MERGE, in HANA, to insert, or update a table without a second table as a source. This must be done in one command, no store procedures. Also, UPSERT won't work in this case.
I found this answer for SQL, but HANA doesn't seem to like VALUES in the USING clause.
SQL Server MERGE without a source table
Here is the answer for SQL from the post above:
MERGE TARGET_TABLE AS I
USING (VALUES ('VALUE1','VALUE2')) as s(COL1,COL2)
ON I.COL1 = s.COL1
WHEN MATCHED THEN
...
WHEN NOT MATCHED THEN
...
Thanks.
The MERGE INTO command is specifically designed for ETL-type use cases where data from one table should be merged into another table and its data in there. A single tuple insert is possible via a subquery like so:
select * from t; -- single integer column 'C'
insert into t values (2);
c
-
2
merge command, inserting 4 or updating matches to 100
merge into t
using (select 4 c from dummy) s
on t.c = s.c
when matched then
update set t.c = 100
when not matched then
insert values (s.c);
c
--
2
4
run the merge command again
c
--
2
100
So, that works just fine.
As for the UPSERT/REPLACE command using VALUES is perfectly possible and even explained in the command examples in the reference documentation.
Related
PostgreSQL DB: v 9.4.24
create table my_a_b_data ... // with a_uuid, b_uuid, and c columns
NOTE: the my_a_b_data keeps the references to a and b table. So it keeps the uuids of a and b.
where: the primary key (a_uuid, b_uuid)
there is also an index:
create unique index my_a_b_data_pkey
on my_a_b_data (a_uuid, b_uuid);
In the Java jdbc-alike code, in the scope one single transaction: (start() -> [code (delete, insert)] ->commit()]) (org.postgresql:postgresql:42.2.5 driver)
delete from my_a_b_data where b_uuid = 'bbb';
insert into my_a_b_data (a_uuid, b_uuid, c) values ('aaa', 'bbb', null);
I found that the insert fails, because the delete is not yet deleted. So it fails because it can not be duplicated.
Q: Is it is some kind of limitation in PostgreSQL that DB can’t do a delete and insert in one transaction because PostgreSQL doesn’t update its indexes until the commit for the delete is executed, therefore the insert will fail since the id or key (whatever we may be using) already exists in the index?
What would be possible solution? Splitting in two transactions?
UPDATE: the order is exactly the same. When I test the sql alone in the SQL console. It works fine. We use JDBI library v 5.29.
there it looks like this:
#Transaction
#SqlUpdate("insert into my_a_b_data (...; // similar for the delete
public abstract void addB() ..
So in the code:
this.begin();
this.deleteByB(b_id);
this.addB(a_id, b_id);
this.commit();
I had a similar problem to insert duplicated values and I resolved it by using Insert and Update instead of Delete. I created this process on Python but you might be able to reproduce it:
First, you create a temporary table like the target table where you want to insert values, the difference is that this table is dropped after commit.
CREATE TEMP TABLE temp_my_a_b_data
(LIKE public.my_a_b_data INCLUDING DEFAULTS)
ON COMMIT DROP;
I have created a CSV (I had to merge different data to input) with the values that I want to input/insert on my table and I used the COPY function to insert them to the temp_table (temp_my_a_b_data).
I found this code on this post related to Java and COPY PostgreSQL - \copy command:
String query ="COPY tmp from 'E://load.csv' delimiter ','";
Use the INSERT INTO but with the ON_CONFLICT clause which you can decide to do an action when the insert cannot be done because of specified constrains, on the case below we do the update:
INSERT INTO public.my_a_b_data
SELECT *
FROM temp_my_a_b_data
ON CONFLICT (a_uuid, b_uuid,c) DO UPDATE
SET a_uuid = EXCLUDED.a_uuid,
b_uuid = EXCLUDED. c = EXCLUDED.c;`
Considerations:
I am not sure but you might be able to perform the third step without using the previous steps, temp table or copy from. You can just a loop over the values:
INSERT INTO public.my_a_b_data VALUES(value1, value2, null)
ON CONFLICT (a_uuid, b_uuid,c) DO UPDATE
SET a_uuid = EXCLUDED.a_uuid,
b_uuid = EXCLUDED.b_uuid, c = EXCLUDED.c;
Does PostgreSQL support computed / calculated columns, like MS SQL Server? I can't find anything in the docs, but as this feature is included in many other DBMSs I thought I might be missing something.
Eg: http://msdn.microsoft.com/en-us/library/ms191250.aspx
Postgres 12 or newer
STORED generated columns are introduced with Postgres 12 - as defined in the SQL standard and implemented by some RDBMS including DB2, MySQL, and Oracle. Or the similar "computed columns" of SQL Server.
Trivial example:
CREATE TABLE tbl (
int1 int
, int2 int
, product bigint GENERATED ALWAYS AS (int1 * int2) STORED
);
fiddle
VIRTUAL generated columns may come with one of the next iterations. (Not in Postgres 15, yet).
Related:
Attribute notation for function call gives error
Postgres 11 or older
Up to Postgres 11 "generated columns" are not supported.
You can emulate VIRTUAL generated columns with a function using attribute notation (tbl.col) that looks and works much like a virtual generated column. That's a bit of a syntax oddity which exists in Postgres for historic reasons and happens to fit the case. This related answer has code examples:
Store common query as column?
The expression (looking like a column) is not included in a SELECT * FROM tbl, though. You always have to list it explicitly.
Can also be supported with a matching expression index - provided the function is IMMUTABLE. Like:
CREATE FUNCTION col(tbl) ... AS ... -- your computed expression here
CREATE INDEX ON tbl(col(tbl));
Alternatives
Alternatively, you can implement similar functionality with a VIEW, optionally coupled with expression indexes. Then SELECT * can include the generated column.
"Persisted" (STORED) computed columns can be implemented with triggers in a functionally equivalent way.
Materialized views are a related concept, implemented since Postgres 9.3.
In earlier versions one can manage MVs manually.
YES you can!! The solution should be easy, safe, and performant...
I'm new to postgresql, but it seems you can create computed columns by using an expression index, paired with a view (the view is optional, but makes makes life a bit easier).
Suppose my computation is md5(some_string_field), then I create the index as:
CREATE INDEX some_string_field_md5_index ON some_table(MD5(some_string_field));
Now, any queries that act on MD5(some_string_field) will use the index rather than computing it from scratch. For example:
SELECT MAX(some_field) FROM some_table GROUP BY MD5(some_string_field);
You can check this with explain.
However at this point you are relying on users of the table knowing exactly how to construct the column. To make life easier, you can create a VIEW onto an augmented version of the original table, adding in the computed value as a new column:
CREATE VIEW some_table_augmented AS
SELECT *, MD5(some_string_field) as some_string_field_md5 from some_table;
Now any queries using some_table_augmented will be able to use some_string_field_md5 without worrying about how it works..they just get good performance. The view doesn't copy any data from the original table, so it is good memory-wise as well as performance-wise. Note however that you can't update/insert into a view, only into the source table, but if you really want, I believe you can redirect inserts and updates to the source table using rules (I could be wrong on that last point as I've never tried it myself).
Edit: it seems if the query involves competing indices, the planner engine may sometimes not use the expression-index at all. The choice seems to be data dependant.
One way to do this is with a trigger!
CREATE TABLE computed(
one SERIAL,
two INT NOT NULL
);
CREATE OR REPLACE FUNCTION computed_two_trg()
RETURNS trigger
LANGUAGE plpgsql
SECURITY DEFINER
AS $BODY$
BEGIN
NEW.two = NEW.one * 2;
RETURN NEW;
END
$BODY$;
CREATE TRIGGER computed_500
BEFORE INSERT OR UPDATE
ON computed
FOR EACH ROW
EXECUTE PROCEDURE computed_two_trg();
The trigger is fired before the row is updated or inserted. It changes the field that we want to compute of NEW record and then it returns that record.
PostgreSQL 12 supports generated columns:
PostgreSQL 12 Beta 1 Released!
Generated Columns
PostgreSQL 12 allows the creation of generated columns that compute their values with an expression using the contents of other columns. This feature provides stored generated columns, which are computed on inserts and updates and are saved on disk. Virtual generated columns, which are computed only when a column is read as part of a query, are not implemented yet.
Generated Columns
A generated column is a special column that is always computed from other columns. Thus, it is for columns what a view is for tables.
CREATE TABLE people (
...,
height_cm numeric,
height_in numeric GENERATED ALWAYS AS (height_cm * 2.54) STORED
);
db<>fiddle demo
Well, not sure if this is what You mean but Posgres normally support "dummy" ETL syntax.
I created one empty column in table and then needed to fill it by calculated records depending on values in row.
UPDATE table01
SET column03 = column01*column02; /*e.g. for multiplication of 2 values*/
It is so dummy I suspect it is not what You are looking for.
Obviously it is not dynamic, you run it once. But no obstacle to get it into trigger.
Example on creating an empty virtual column
,(SELECT *
From (values (''))
A("virtual_col"))
Example on creating two virtual columns with values
SELECT *
From (values (45,'Completed')
, (1,'In Progress')
, (1,'Waiting')
, (1,'Loading')
) A("Count","Status")
order by "Count" desc
I have a code that works and use the term calculated, I'm not on postgresSQL pure tho we run on PADB
here is how it's used
create table some_table as
select category,
txn_type,
indiv_id,
accum_trip_flag,
max(first_true_origin) as true_origin,
max(first_true_dest ) as true_destination,
max(id) as id,
count(id) as tkts_cnt,
(case when calculated tkts_cnt=1 then 1 else 0 end) as one_way
from some_rando_table
group by 1,2,3,4 ;
A lightweight solution with Check constraint:
CREATE TABLE example (
discriminator INTEGER DEFAULT 0 NOT NULL CHECK (discriminator = 0)
);
Does PostgreSQL support computed / calculated columns, like MS SQL Server? I can't find anything in the docs, but as this feature is included in many other DBMSs I thought I might be missing something.
Eg: http://msdn.microsoft.com/en-us/library/ms191250.aspx
Postgres 12 or newer
STORED generated columns are introduced with Postgres 12 - as defined in the SQL standard and implemented by some RDBMS including DB2, MySQL, and Oracle. Or the similar "computed columns" of SQL Server.
Trivial example:
CREATE TABLE tbl (
int1 int
, int2 int
, product bigint GENERATED ALWAYS AS (int1 * int2) STORED
);
fiddle
VIRTUAL generated columns may come with one of the next iterations. (Not in Postgres 15, yet).
Related:
Attribute notation for function call gives error
Postgres 11 or older
Up to Postgres 11 "generated columns" are not supported.
You can emulate VIRTUAL generated columns with a function using attribute notation (tbl.col) that looks and works much like a virtual generated column. That's a bit of a syntax oddity which exists in Postgres for historic reasons and happens to fit the case. This related answer has code examples:
Store common query as column?
The expression (looking like a column) is not included in a SELECT * FROM tbl, though. You always have to list it explicitly.
Can also be supported with a matching expression index - provided the function is IMMUTABLE. Like:
CREATE FUNCTION col(tbl) ... AS ... -- your computed expression here
CREATE INDEX ON tbl(col(tbl));
Alternatives
Alternatively, you can implement similar functionality with a VIEW, optionally coupled with expression indexes. Then SELECT * can include the generated column.
"Persisted" (STORED) computed columns can be implemented with triggers in a functionally equivalent way.
Materialized views are a related concept, implemented since Postgres 9.3.
In earlier versions one can manage MVs manually.
YES you can!! The solution should be easy, safe, and performant...
I'm new to postgresql, but it seems you can create computed columns by using an expression index, paired with a view (the view is optional, but makes makes life a bit easier).
Suppose my computation is md5(some_string_field), then I create the index as:
CREATE INDEX some_string_field_md5_index ON some_table(MD5(some_string_field));
Now, any queries that act on MD5(some_string_field) will use the index rather than computing it from scratch. For example:
SELECT MAX(some_field) FROM some_table GROUP BY MD5(some_string_field);
You can check this with explain.
However at this point you are relying on users of the table knowing exactly how to construct the column. To make life easier, you can create a VIEW onto an augmented version of the original table, adding in the computed value as a new column:
CREATE VIEW some_table_augmented AS
SELECT *, MD5(some_string_field) as some_string_field_md5 from some_table;
Now any queries using some_table_augmented will be able to use some_string_field_md5 without worrying about how it works..they just get good performance. The view doesn't copy any data from the original table, so it is good memory-wise as well as performance-wise. Note however that you can't update/insert into a view, only into the source table, but if you really want, I believe you can redirect inserts and updates to the source table using rules (I could be wrong on that last point as I've never tried it myself).
Edit: it seems if the query involves competing indices, the planner engine may sometimes not use the expression-index at all. The choice seems to be data dependant.
One way to do this is with a trigger!
CREATE TABLE computed(
one SERIAL,
two INT NOT NULL
);
CREATE OR REPLACE FUNCTION computed_two_trg()
RETURNS trigger
LANGUAGE plpgsql
SECURITY DEFINER
AS $BODY$
BEGIN
NEW.two = NEW.one * 2;
RETURN NEW;
END
$BODY$;
CREATE TRIGGER computed_500
BEFORE INSERT OR UPDATE
ON computed
FOR EACH ROW
EXECUTE PROCEDURE computed_two_trg();
The trigger is fired before the row is updated or inserted. It changes the field that we want to compute of NEW record and then it returns that record.
PostgreSQL 12 supports generated columns:
PostgreSQL 12 Beta 1 Released!
Generated Columns
PostgreSQL 12 allows the creation of generated columns that compute their values with an expression using the contents of other columns. This feature provides stored generated columns, which are computed on inserts and updates and are saved on disk. Virtual generated columns, which are computed only when a column is read as part of a query, are not implemented yet.
Generated Columns
A generated column is a special column that is always computed from other columns. Thus, it is for columns what a view is for tables.
CREATE TABLE people (
...,
height_cm numeric,
height_in numeric GENERATED ALWAYS AS (height_cm * 2.54) STORED
);
db<>fiddle demo
Well, not sure if this is what You mean but Posgres normally support "dummy" ETL syntax.
I created one empty column in table and then needed to fill it by calculated records depending on values in row.
UPDATE table01
SET column03 = column01*column02; /*e.g. for multiplication of 2 values*/
It is so dummy I suspect it is not what You are looking for.
Obviously it is not dynamic, you run it once. But no obstacle to get it into trigger.
Example on creating an empty virtual column
,(SELECT *
From (values (''))
A("virtual_col"))
Example on creating two virtual columns with values
SELECT *
From (values (45,'Completed')
, (1,'In Progress')
, (1,'Waiting')
, (1,'Loading')
) A("Count","Status")
order by "Count" desc
I have a code that works and use the term calculated, I'm not on postgresSQL pure tho we run on PADB
here is how it's used
create table some_table as
select category,
txn_type,
indiv_id,
accum_trip_flag,
max(first_true_origin) as true_origin,
max(first_true_dest ) as true_destination,
max(id) as id,
count(id) as tkts_cnt,
(case when calculated tkts_cnt=1 then 1 else 0 end) as one_way
from some_rando_table
group by 1,2,3,4 ;
A lightweight solution with Check constraint:
CREATE TABLE example (
discriminator INTEGER DEFAULT 0 NOT NULL CHECK (discriminator = 0)
);
I have a stored procedure which creates and works with a temporary #table
Some of the queries would be tremendously optimized if that temporary #table would have an index created on it.
However, creating an index within the stored procedure fails:
create procedure test1 as
SELECT f1, f2, f3
INTO #table1
FROM main_table
WHERE 1 = 2
-- insert rows into #table1
create index my_idx on #table1 (f1)
SELECT f1, f2, f3 FROM #table1 (index my_idx) WHERE f1 = 11 -- "QUERY X"
When I call the above, the query plan for "QUERY X" shows a table scan.
If I simply run the code above outside the stored procedure, the messages show the following warning:
Index 'my_idx' specified as optimizer hint in the FROM clause of table '#table1' does not exist. Optimizer will choose another index instead.
This can be resolved when running ad-hoc (outside the stored procedure) by splitting the code above in two batches by addding "go" after index creation:
create index my_idx on #table1 (f1)
go
Now, "QUERY X" query plan shows the use of index "my_idx".
QUESTION: How do I mimique running the "create index" in a separate batch when it's inside the stored procedure? I can't insert a "go" there like I do with the ad-hoc copy above. Please note that I'm aware of the solution of "split up the 'QUERY X' into a separate stored procedure" and am looking for a solution that will avoid that.
P.S. If it matters, this is on Sybase 12 (ASE 12.5.4)
UPDATE:
I have been seeing several references to "schema bumping" during my Googling before posing the question. But that doesn't seem to happen in my case.
You can create a table, populate it, create an index on it and select values
from it in the same porc and have the optimizer fully cost it based on
accurate information. This is called 'schema bumping' and has been in place
since 11.5.1.
The Sybase documentation says that you create and use a temporary index in the same stored procedure:
http://infocenter.sybase.com/help/index.jsp?topic=/com.sybase.dc20023_1251/html/optimizer/X26029.htm
I think to get around this you will need to split your stored procedure into at least two parts, one to create and populate the table then build the index, and then a second one to run the select query.
I am not sure how you are getting this problem, might be in older version of Sybase, however with version 12.5.4 I tried executing the same thing as suggested by you but in my case the optimizer correctly suggested the use of index created in the stored procedure. Usually in a stored procedure we do not need to break sql into batches because else we would have been required to have a seperate batch for create table command as well.
In case we try to create index within a same batch (not in a stored procedure) we will do get the same error as specified by you above because we are trying to create an index on a table and then trying to use it within the same batch. Usually the Sybase server will compile the whole batch in one go and hence the problem. But as far as stored procedure is concerned in Sybase 12.5.4 there will be no problem.
So I'm coming from MySQL where I could do INSERT on DUPLICATE UPDATE:
INSERT INTO table (a,b,c)
VALUES (1,2,3)
ON DUPLICATE KEY UPDATE c=c+1;
But now I'm using PostgreSQL and there are efforts to add the UPSERT functionality, looks like MERGE might work for what I would like but wanted to see if this is the most optimal syntax. Example Syntax 1, I've also seen this but don't understand how to implement. I haven't tried this yet because I thought MERGE was used for merging data from table1 to Table2 or would something like this work?
MERGE
INTO table
USING table
ON c = 1
WHEN MATCHED THEN
UPDATE
SET c=c+1
WHEN NOT MATCHED THEN
INSERT (a,b,c)
VALUES (1,2,3)
Any other suggestions?
Until MERGE is available, use this robust approach: Insert, on duplicate update in PostgreSQL?
Until merge is supported the simplest way IMO is to just break it up into two queries:
BEGIN;
INSERT INTO t (a,b,c) VALUES (1,2,3) WHERE id != 1;
UPDATE t SET c=c+1 WHERE id = 1;
END;
where id would be changed to the appropriate condition.
MERGE INTO table
USING (VALUES (1, 2, 3)) AS newvalues (a, b, c)
ON table.c = newvalues.c -- or whatever the PK is
WHEN MATCHED THEN UPDATE SET c = c + 1
WHEN NOT MATCHED THEN INSERT (a, b, c)
VALUES (newvalues.a, newvalues.b, newvalues.c)
The key here is that instead of merging in another table you create a constant table source using the VALUES construct in the USING clause. The exact merging rules you can obviously tailor to taste.
See also http://petereisentraut.blogspot.com/2010/05/merge-syntax.html.
I think "MERGE" is not yet in Postgres but is suposed to be in 9.1.
I like to use RULEs instead
CREATE OR REPLACE RULE "insert_ignore"
AS ON INSERT TO "table" WHERE
NEW.id = OLD.id --whatever your conditions are
DO INSTEAD NOTHING;
What you have linked to ("Insert, on duplicate update (postgresql)") is basically some pgsql that you feed the data. I think the RULE is more elegant since you don't need to call them explicitly and they work transparent in the background without the need to call a procedure within your actual INSERT.