Let's say we have the following table:
CREATE TABLE foo (
column_1 bigint,
column_2 bytea DEFAULT gen_random_bytes(2),
PRIMARY KEY (column_1, column_2)
);
Note: We want column_2 to be random & cryptographically strong.
How do we insert a row without causing a primary key conflict?
I guess we'd have to do a loop until gen_random_bytes(2) returns a unique result? If so, can we do this loop with pure SQL, maybe with recursive CTE, instead of with plpgsql?
insert into t (col1, col2)
select 1, ('\x' || right('000' || to_hex(i), 4))::bytea
from (
select generate_series(0, 65535) i
except
select get_byte(col2, 0) * 256 + get_byte(col2, 1)
from t
where col1 = 1
) s
order by random()
limit 1
Related
I have the following database schema (oversimplified):
create sequence partners_partner_id_seq;
create table partners
(
partner_id integer default nextval('partners_partner_id_seq'::regclass) not null primary key,
name varchar(255) default NULL::character varying,
company_id varchar(20) default NULL::character varying,
vat_id varchar(50) default NULL::character varying,
is_deleted boolean default false not null
);
INSERT INTO partners(name, company_id, vat_id) VALUES('test1','1010109191191', 'BG1010109191192');
INSERT INTO partners(name, company_id, vat_id) VALUES('test2','1010109191191', 'BG1010109191192');
INSERT INTO partners(name, company_id, vat_id) VALUES('test3','3214567890102', 'BG1010109191192');
INSERT INTO partners(name, company_id, vat_id) VALUES('test4','9999999999999', 'GE9999999999999');
I am trying to figure out how to return test1, test2 (because the company_id column value duplicates vertically) and test3 (because the vat_id column value duplicates vertically as well).
To put it in other words - I need to find duplicating company_id and vat_id records and group them together, so that test1, test2 and test3 would be together, because they duplicate by company_id and vat_id.
So far I have the following query:
SELECT *
FROM (
SELECT *, LEAD(row, 1) OVER () AS nextrow
FROM (
SELECT *, ROW_NUMBER() OVER (w) AS row
FROM partners
WHERE is_deleted = false
AND ((company_id != '' AND company_id IS NOT null) OR (vat_id != '' AND vat_id IS NOT NULL))
WINDOW w AS (PARTITION BY company_id, vat_id ORDER BY partner_id DESC)
) x
) y
WHERE (row > 1 OR nextrow > 1)
AND is_deleted = false
This successfully shows all company_id duplicates, but does not appear to show vat_id ones - test3 row is missing. Is this possible to be done within one query?
Here is a db-fiddle with the schema, data and predefined query reproducing my result.
You can do this with recursion, but depending on the size of your data you may want to iterate, instead.
The trick is to make the name just another match key instead of treating it differently than the company_id and vat_id:
create table partners (
partner_id integer generated always as identity primary key,
name text,
company_id text,
vat_id text,
is_deleted boolean not null default false
);
insert into partners (name, company_id, vat_id) values
('test1','1010109191191', 'BG1010109191192'),
('test2','1010109191191', 'BG1010109191192'),
('test3','3214567890102', 'BG1010109191192'),
('test4','9999999999999', 'GE9999999999999'),
('test5','3214567890102', 'BG8888888888888'),
('test6','2983489023408', 'BG8888888888888')
;
I added a couple of test cases and left in the lone partner.
with recursive keys as (
select partner_id,
array['n_'||name, 'c_'||company_id, 'v_'||vat_id] as matcher,
array[partner_id] as matchlist,
1 as size
from partners
), matchers as (
select *
from keys
union all
select p.partner_id, c.matcher,
p.matchlist||c.partner_id as matchlist,
p.size + 1
from matchers p
join keys c
on c.matcher && p.matcher
and not p.matchlist #> array[c.partner_id]
), largest as (
select distinct sort(matchlist) as matchlist
from matchers m
where not exists (select 1
from matchers
where matchlist #> m.matchlist
and size > m.size)
-- and size > 1
)
select *
from largest
;
matchlist
{1,2,3,5,6}
{4}
fiddle
EDIT UPDATE
Since recursion did not perform, here is an iterative example in plpgsql that uses a temporary table:
create temporary table match1 (
partner_id int not null,
group_id int not null,
matchkey uuid not null
);
create index on match1 (matchkey);
create index on match1 (group_id);
insert into match1
select partner_id, partner_id, md5('n_'||name)::uuid from partners
union all
select partner_id, partner_id, md5('c_'||company_id)::uuid from partners
union all
select partner_id, partner_id, md5('v_'||vat_id)::uuid from partners;
do $$
declare _cnt bigint;
begin
loop
with consolidate as (
select group_id,
min(group_id) over (partition by matchkey) as new_group_id
from match1
), minimize as (
select group_id, min(new_group_id) as new_group_id
from consolidate
group by group_id
), doupdate as (
update match1
set group_id = m.new_group_id
from minimize m
where m.group_id = match1.group_id
and m.new_group_id != match1.group_id
returning *
)
select count(*) into _cnt from doupdate;
if _cnt = 0 then
exit;
end if;
end loop;
end;
$$;
updated fiddle
the issue:
i need to do something like this
drop table if exists tt_t;
create temp table tt_t(id serial primary key, main_id int, external_id int);
insert into tt_t(main_id, external_id)
select currval(pg_get_serial_sequence('tt_t', 'id')), 1
where not exists (select from tt_t where external_id = 1);
but execution raises an error
SQL Error [55000]: ERROR: currval of sequence "tt_t_id_seq" is not yet defined in this session
solution:
there is a way to solve this with anonymous code block
do
$$
begin
if not exists(select from tt_t where external_id = 1)
then
insert into tt_t(external_id, main_id)
values(1, currval(pg_get_serial_sequence('tt_t', 'id')));
end if;
end;
$$
;
but anonymous blocks has some restrictions e.g. Dapper parameters not working with PostgreSQL through npgsql connection, is postgres anonymous function parameterization supported?
how do i fix it without anonymous code block (UPD: and without any DDL changes)?
probable solution:
insert into tt_t(id, main_id, external_id)
select nextval(pg_get_serial_sequence('tt_t', 'id')), currval(pg_get_serial_sequence('tt_t', 'id')), 1
where not exists (select from tt_t where external_id = 1);
shorter code has been proposed to me
insert into tt_t(id, main_id, external_id)
select nextval(pg_get_serial_sequence('tt_t', 'id')), lastval(), 1
where not exists (select from tt_t where external_id = 1);
but i'm not sure if nextval will be calculated first
What about using a default value:
drop table if exists tt_t;
create temp table tt_t(id serial primary key, main_id int default lastval(), external_id int);
insert into tt_t(external_id)
select 1
where not exists (select * from tt_t where external_id = 1);
In theory it shouldn't be possible that another nextval() is called between the one for the id and the lastval(). However I am not 100% sure if there are some corner cases that I don't know of.
The following works as well (even if one or more of the external_id values already exist).
insert into tt_t(external_id)
select *
from (values (1),(2),(3)) x (external_id)
where not exists (select *
from tt_t
where external_id = x.external_id);
In Postgres 8.4 or higher, what is the most efficient way to get a row of data populated by defaults without actually creating the row. Eg, as a transaction (pseudocode):
create table "mytable"
(
id serial PRIMARY KEY NOT NULL,
parent_id integer NOT NULL DEFAULT 1,
random_id integer NOT NULL DEFAULT random(),
)
begin transaction
fake_row = insert into mytable (id) values (0) returning *;
delete from mytable where id=0;
return fake_row;
end transaction
Basically I'd expect a query with a single row where parent_id is 1 and random_id is a random number (or other function return value) but I don't want this record to persist in the table or impact on the primary key sequence serial_id_seq.
My options seem to be using a transaction like above or creating views which are copies of the table with the fake row added but I don't know all the pros and cons of each or whether a better way exists.
I'm looking for an answer that assumes no prior knowledge of the datatypes or default values of any column except id or the number or ordering of the columns. Only the table name will be known and that a record with id 0 should not exist in the table.
In the past I created the fake record 0 as a permanent record but I've come to consider this record a type of pollution (since I typically have to filter it out of future queries).
You can copy the table definition and defaults to the temp table with:
CREATE TEMP TABLE table_name_rt (LIKE table_name INCLUDING DEFAULTS);
And use this temp table to generate dummy rows. Such table will be dropped at the end of the session (or transaction) and will only be visible to current session.
You can query the catalog and build a dynamic query
Say we have this table:
create table test10(
id serial primary key,
first_name varchar( 100 ),
last_name varchar( 100 ) default 'Tom',
age int not null default 38,
salary float default 100.22
);
When you run following query:
SELECT string_agg( txt, ' ' order by id )
FROM (
select 1 id, 'SELECT ' txt
union all
select 2, -9999 || ' as id '
union all
select 3, ', '
|| coalesce( column_default, 'null'||'::'||c.data_type )
|| ' as ' || c.column_name
from information_schema.columns c
where table_schema = 'public'
and table_name = 'test10'
and ordinal_position > 1
) xx
;
you will get this sting as a result:
"SELECT -9999 as id , null::character varying as first_name ,
'Tom'::character varying as last_name , 38 as age , 100.22 as salary"
then execute this query and you will get the "phantom row".
We can build a function that build and excecutes the query and return our row as a result:
CREATE OR REPLACE FUNCTION get_phantom_rec (p_i test10.id%type )
returns test10 as $$
DECLARE
v_sql text;
myrow test10%rowtype;
begin
SELECT string_agg( txt, ' ' order by id )
INTO v_sql
FROM (
select 1 id, 'SELECT ' txt
union all
select 2, p_i || ' as id '
union all
select 3, ', '
|| coalesce( column_default, 'null'||'::'||c.data_type )
|| ' as ' || c.column_name
from information_schema.columns c
where table_schema = 'public'
and table_name = 'test10'
and ordinal_position > 1
) xx
;
EXECUTE v_sql INTO myrow;
RETURN myrow;
END$$ LANGUAGE plpgsql ;
and then this simple query gives you what you want:
select * from get_phantom_rec ( -9999 );
id | first_name | last_name | age | salary
-------+------------+-----------+-----+--------
-9999 | | Tom | 38 | 100.22
I would just select the fake values as literals:
select 1 id, 1 parent_id, 1 user_id
The returned row will be (virtually) indistinguishable from a real row.
To get the values from the catalog:
select
0 as id, -- special case for serial type, just return 0
(select column_default::int -- Cast to int, because we know the column is int
from INFORMATION_SCHEMA.COLUMNS
where table_name = 'mytable'
and column_name = 'parent_id') as parent_id,
(select column_default::int -- Cast to int, because we know the column is int
from INFORMATION_SCHEMA.COLUMNS
where table_name = 'mytable'
and column_name = 'user_id') as user_id;
Note that you must know what the columns are and their type, but this is reasonable. If you change the table schema (except default value), you would need to tweak the query.
See the above as a SQLFiddle.
How would I do the following TSQL query in DB2? I'm having problems creating a temp table based on the results from a query.
SELECT
COLUMN_1, COLUMN_2, COLUMN_3
INTO #TEMP_A
FROM TABLE_A
WHERE COLUMN_1 = 1 AND COLUMN_2 = 2
The error message is:
"Error: SQL0104N An unexpected token "#TEMP_A" was found following "". Expected tokens may include: ":". SQLSTATE=42601"
You have to declare a temp table in DB2 before you can use it. Either with the same query you are running:
DECLARE GLOBAL TEMPORARY TABLE SESSION.YOUR_TEMP_TABLE_NAME AS (
SELECT COLUMN_1, COLUMN_2, COLUMN_3
FROM TABLE_A
) DEFINITION ONLY
Or "manually" define the columns:
DECLARE GLOBAL TEMPORARY TABLE SESSION.YOUR_TEMP_TABLE_NAME (
COLUMN_1 CHAR(10)
,COLUMN_2 TIMESTAMP
,COLUMN_3 INTEGER
)
Then populate it:
INSERT INTO SESSION.YOUR_TEMP_TABLE_NAME
SELECT COLUMN_1, COLUMN_2, COLUMN_3
FROM TABLE_A
WHERE COLUMN_1 = 1
AND COLUMN_2 = 2
It's not quite as straight-forward as in SQL Server. :)
And even though it's called a "global" temporary table, it only exists for the current session. Note that all temp tables should be prefixed with the SESSION schema. If you do not provide a schema name, then SESSION will be implied.
maybe the "with" clause is what you look for:
with TEMP_A as (
SELECT COLUMN_1, COLUMN_2, COLUMN_3
FROM TABLE_A
WHERE COLUMN_1 = 1 AND COLUMN_2 = 2
)
-- now use TEMP_A
select * from TEMP_A
As it turned out, I did not have permissions to create temp tables.
I'm using PostgreSQL 9.0 and I have a table with just an artificial key (auto-incrementing sequence) and another unique key. (Yes, there is a reason for this table. :)) I want to look up an ID by the other key or, if it doesn't exist, insert it:
SELECT id
FROM mytable
WHERE other_key = 'SOMETHING'
Then, if no match:
INSERT INTO mytable (other_key)
VALUES ('SOMETHING')
RETURNING id
The question: is it possible to save a round-trip to the DB by doing both of these in one statement? I can insert the row if it doesn't exist like this:
INSERT INTO mytable (other_key)
SELECT 'SOMETHING'
WHERE NOT EXISTS (SELECT * FROM mytable WHERE other_key = 'SOMETHING')
RETURNING id
... but that doesn't give the ID of an existing row. Any ideas? There is a unique constraint on other_key, if that helps.
Have you tried to union it?
Edit - this requires Postgres 9.1:
create table mytable (id serial primary key, other_key varchar not null unique);
WITH new_row AS (
INSERT INTO mytable (other_key)
SELECT 'SOMETHING'
WHERE NOT EXISTS (SELECT * FROM mytable WHERE other_key = 'SOMETHING')
RETURNING *
)
SELECT * FROM new_row
UNION
SELECT * FROM mytable WHERE other_key = 'SOMETHING';
results in:
id | other_key
----+-----------
1 | SOMETHING
(1 row)
No, there is no special SQL syntax that allows you to do select or insert. You can do what Ilia mentions and create a sproc, which means it will not do a round trip fromt he client to server, but it will still result in two queries (three actually, if you count the sproc itself).
using 9.5 i successfully tried this
based on Denis de Bernardy's answer
only 1 parameter
no union
no stored procedure
atomic, thus no concurrency problems (i think...)
The Query:
WITH neworexisting AS (
INSERT INTO mytable(other_key) VALUES('hello 2')
ON CONFLICT(other_key) DO UPDATE SET existed=true -- need some update to return sth
RETURNING *
)
SELECT * FROM neworexisting
first call:
id|other_key|created |existed|
--|---------|-------------------|-------|
6|hello 1 |2019-09-11 11:39:29|false |
second call:
id|other_key|created |existed|
--|---------|-------------------|-------|
6|hello 1 |2019-09-11 11:39:29|true |
First create your table ;-)
CREATE TABLE mytable (
id serial NOT NULL,
other_key text NOT NULL,
created timestamptz NOT NULL DEFAULT now(),
existed bool NOT NULL DEFAULT false,
CONSTRAINT mytable_pk PRIMARY KEY (id),
CONSTRAINT mytable_uniq UNIQUE (other_key) --needed for on conflict
);
you can use a stored procedure
IF (SELECT id FROM mytable WHERE other_key = 'SOMETHING' LIMIT 1) < 0 THEN
INSERT INTO mytable (other_key) VALUES ('SOMETHING')
END IF
I have an alternative to Denis answer, that I think is less database-intensive, although a bit more complex:
create table mytable (id serial primary key, other_key varchar not null unique);
WITH table_sel AS (
SELECT id
FROM mytable
WHERE other_key = 'test'
UNION
SELECT NULL AS id
ORDER BY id NULLS LAST
LIMIT 1
), table_ins AS (
INSERT INTO mytable (id, other_key)
SELECT
COALESCE(id, NEXTVAL('mytable_id_seq'::REGCLASS)),
'test'
FROM table_sel
ON CONFLICT (id) DO NOTHING
RETURNING id
)
SELECT * FROM table_ins
UNION ALL
SELECT * FROM table_sel
WHERE id IS NOT NULL;
In table_sel CTE I'm looking for the right row. If I don't find it, I assure that table_sel returns at least one row, with a union with a SELECT NULL.
In table_ins CTE I try to insert the same row I was looking for earlier. COALESCE(id, NEXTVAL('mytable_id_seq'::REGCLASS)) is saying: id could be defined, if so, use it; whereas if id is null, increment the sequence on id and use this new value to insert a row. The ON CONFLICT clause assure
that if id is already in mytable I don't insert anything.
At the end I put everything together with a UNION between table_ins and table_sel, so that I'm sure to take my sweet id value and execute both CTE.
This query needs to search for the value other_key only once, and is a "search this value" not a "check if this value not exists in the table", that is very heavy; in Denis alternative you use other_key in both types of searches. In my query you "check if a value not exists" only on id that is a integer primary key, that, for construction, is fast.
Minor tweak a decade late to Denis's excellent answer:
-- Create the table with a unique constraint
CREATE TABLE mytable (
id serial PRIMARY KEY
, other_key varchar NOT NULL UNIQUE
);
WITH new_row AS (
-- Only insert when we don't find anything, avoiding a table lock if
-- possible.
INSERT INTO mytable ( other_key )
SELECT 'SOMETHING'
WHERE NOT EXISTS (
SELECT *
FROM mytable
WHERE other_key = 'SOMETHING'
)
RETURNING *
)
(
-- This comes first in the UNION ALL since it'll almost certainly be
-- in the query cache. Marginally slower for the insert case, but also
-- marginally faster for the much more common read-only case.
SELECT *
FROM mytable
WHERE other_key = 'SOMETHING'
-- Don't check for duplicates to be removed
UNION ALL
-- If we reach this point in iteration, we needed to do the INSERT and
-- lock after all.
SELECT *
FROM new_row
) LIMIT 1 -- Just return whatever comes first in the results and allow
-- the query engine to cut processing short for the INSERT
-- calculation.
;
The UNION ALL tells the planner it doesn't have to collect results for de-duplication. The LIMIT 1 at the end allows the planner to short-circuit further processing/iteration once it knows there's an answer available.
NOTE: There is a race condition present here and in the original answer. If the entry does not already exist, the INSERT will fail with a unique constraint violation. The error can be suppressed with ON CONFLICT DO NOTHING, but the query will return an empty set instead of the new row. This is a difficult problem because getting that info from another transaction would violate the I in ACID.