How do I declare a variable for use in a PostgreSQL 8.3 query?
In MS SQL Server I can do this:
DECLARE #myvar INT
SET #myvar = 5
SELECT *
FROM somewhere
WHERE something = #myvar
How do I do the same in PostgreSQL? According to the documentation variables are declared simply as "name type;", but this gives me a syntax error:
myvar INTEGER;
Could someone give me an example of the correct syntax?
I accomplished the same goal by using a WITH clause, it's nowhere near as elegant but can do the same thing. Though for this example it's really overkill. I also don't particularly recommend this.
WITH myconstants (var1, var2) as (
values (5, 'foo')
)
SELECT *
FROM somewhere, myconstants
WHERE something = var1
OR something_else = var2;
There is no such feature in PostgreSQL. You can do it only in pl/PgSQL (or other pl/*), but not in plain SQL.
An exception is WITH () query which can work as a variable, or even tuple of variables. It allows you to return a table of temporary values.
WITH master_user AS (
SELECT
login,
registration_date
FROM users
WHERE ...
)
SELECT *
FROM users
WHERE master_login = (SELECT login
FROM master_user)
AND (SELECT registration_date
FROM master_user) > ...;
You could also try this in PLPGSQL:
DO $$
DECLARE myvar integer;
BEGIN
SELECT 5 INTO myvar;
DROP TABLE IF EXISTS tmp_table;
CREATE TABLE tmp_table AS
SELECT * FROM yourtable WHERE id = myvar;
END $$;
SELECT * FROM tmp_table;
The above requires Postgres 9.0 or later.
Dynamic Config Settings
you can "abuse" dynamic config settings for this:
-- choose some prefix that is unlikely to be used by postgres
set session my.vars.id = '1';
select *
from person
where id = current_setting('my.vars.id')::int;
Config settings are always varchar values, so you need to cast them to the correct data type when using them. This works with any SQL client whereas \set only works in psql
The above requires Postgres 9.2 or later.
For previous versions, the variable had to be declared in postgresql.conf prior to being used, so it limited its usability somewhat. Actually not the variable completely, but the config "class" which is essentially the prefix. But once the prefix was defined, any variable could be used without changing postgresql.conf
It depends on your client.
However, if you're using the psql client, then you can use the following:
my_db=> \set myvar 5
my_db=> SELECT :myvar + 1 AS my_var_plus_1;
my_var_plus_1
---------------
6
If you are using text variables you need to quote.
\set myvar 'sometextvalue'
select * from sometable where name = :'myvar';
This solution is based on the one proposed by fei0x but it has the advantages that there is no need to join the value list of constants in the query and constants can be easily listed at the start of the query. It also works in recursive queries.
Basically, every constant is a single-value table declared in a WITH clause which can then be called anywhere in the remaining part of the query.
Basic example with two constants:
WITH
constant_1_str AS (VALUES ('Hello World')),
constant_2_int AS (VALUES (100))
SELECT *
FROM some_table
WHERE table_column = (table constant_1_str)
LIMIT (table constant_2_int)
Alternatively you can use SELECT * FROM constant_name instead of TABLE constant_name which might not be valid for other query languages different to postgresql.
Using a Temp Table outside of pl/PgSQL
Outside of using pl/pgsql or other pl/* language as suggested, this is the only other possibility I could think of.
begin;
select 5::int as var into temp table myvar;
select *
from somewhere s, myvar v
where s.something = v.var;
commit;
I want to propose an improvement to #DarioBarrionuevo's answer, to make it simpler leveraging temporary tables.
DO $$
DECLARE myvar integer = 5;
BEGIN
CREATE TEMP TABLE tmp_table ON COMMIT DROP AS
-- put here your query with variables:
SELECT *
FROM yourtable
WHERE id = myvar;
END $$;
SELECT * FROM tmp_table;
True, there is no vivid and unambiguous way to declare a single-value variable, what you can do is
with myVar as (select "any value really")
then, to get access to the value stored in this construction, you do
(select * from myVar)
for example
with var as (select 123)
... where id = (select * from var)
You may resort to tool special features. Like for DBeaver own proprietary syntax:
#set name = 'me'
SELECT :name;
SELECT ${name};
DELETE FROM book b
WHERE b.author_id IN (SELECT a.id FROM author AS a WHERE a.name = :name);
As you will have gathered from the other answers, PostgreSQL doesn’t have this mechanism in straight SQL, though you can now use an anonymous block. However, you can do something similar with a Common Table Expression (CTE):
WITH vars AS (
SELECT 5 AS myvar
)
SELECT *
FROM somewhere,vars
WHERE something = vars.myvar;
You can, of course, have as many variables as you like, and they can also be derived. For example:
WITH vars AS (
SELECT
'1980-01-01'::date AS start,
'1999-12-31'::date AS end,
(SELECT avg(height) FROM customers) AS avg_height
)
SELECT *
FROM customers,vars
WHERE (dob BETWEEN vars.start AND vars.end) AND height<vars.avg_height;
The process is:
Generate a one-row cte using SELECT without a table (in Oracle you will need to include FROM DUAL).
CROSS JOIN the cte with the other table. Although there is a CROSS JOIN syntax, the older comma syntax is slightly more readable.
Note that I have cast the dates to avoid possible issues in the SELECT clause. I used PostgreSQL’s shorter syntax, but you could have used the more formal CAST('1980-01-01' AS date) for cross-dialect compatibility.
Normally, you want to avoid cross joins, but since you’re only cross joining a single row, this has the effect of simply widening the table with the variable data.
In many cases, you don’t need to include the vars. prefix if the names don’t clash with the names in the other table. I include it here to make the point clear.
Also, you can go on to add more CTEs.
This also works in all current versions of MSSQL and MySQL, which do support variables, as well as SQLite which doesn’t, and Oracle which sort of does and sort of doesn’t.
Here is an example using PREPARE statements. You still can't use ?, but you can use $n notation:
PREPARE foo(integer) AS
SELECT *
FROM somewhere
WHERE something = $1;
EXECUTE foo(5);
DEALLOCATE foo;
In DBeaver you can use parameters in queries just like you can from code, so this will work:
SELECT *
FROM somewhere
WHERE something = :myvar
When you run the query DBeaver will ask you for the value for :myvar and run the query.
Here is a code segment using plain variable in postges terminal. I have used it a few times. But need to figure a better way. Here I am working with string variable. Working with integer variable, you don't need the triple quote. Triple quote becomes single quote at query time; otherwise you got syntax error. There might be a way to eliminate the need of triple quote when working with string variables. Please update if you find a way to improve.
\set strainname '''B.1.1.7'''
select *
from covid19strain
where name = :strainname ;
In psql, you can use these 'variables' as macros. Note that they get "evaluated" every time they are used, rather than at the time that they are "set".
Simple example:
\set my_random '(SELECT random())'
select :my_random; -- gives 0.23330629315990592
select :my_random; -- gives 0.67458399344433542
this gives two different answers each time.
However, you can still use these as a valuable shorthand to avoid repeating lots of subselects.
\set the_id '(SELECT id FROM table_1 WHERE name = ''xxx'' LIMIT 1)'
and then use it in your queries later as
:the_id
e.g.
INSERT INTO table2 (table1_id,x,y,z) VALUES (:the_id, 1,2,3)
Note you have to double-quote the strings in the variables, because the whole thing is then string-interpolated (i.e. macro-expanded) into your query.
I need to insert either set A or set B of records into a #temptable, depending on certain condition
My pseudo-code:
IF OBJECT_ID('tempdb..#t1') IS NOT NULL DROP TABLE #t1;
IF {some-condition}
SELECT {columns}
INTO #t1
FROM {some-big-table}
WHERE {some-filter}
ELSE
SELECT {columns}
INTO #t1
FROM {some-other-big-table}
WHERE {some-other-filter}
The two SELECTs above are exclusive (guaranteed by the ELSE operator). However, SQL compiler tries to outsmart me and throws the following message:
There is already an object named '#t1' in the database.
My idea of "fixing" this is to create #t1 upfront and then executing a simple INSERT INTO (instead of SELECT... INTO). But I like minimalism and am wondering whether this can be achieved in an easier way i.e. without explicit CREATE TABLE #t1 upfront.
Btw why is it NOT giving me an error on a conditional DROP TABLE in the first line? Just wondering.
You can't have 2 temp tables with the same name in a single SQL batch. One of the MSDN article says "If more than one temporary table is created inside a single stored procedure or batch, they must have different names". You can have this logic with 2 different temp tables or table variable/temp table declared outside the IF-Else block.
Using a Dyamic sql we can handle this situation. As a developoer its not a good practice. Best to use table variable or temp table.
IF 1=2
BEGIN
EXEC ('SELECT 1 ID INTO #TEMP1
SELECT * FROM #TEMP1
')
END
ELSE
EXEC ('SELECT 2 ID INTO #TEMP1
SELECT * FROM #TEMP1
')
insert into
#resultSet
SELECT TOP (#topN)
field1,
field2
FROM
dbo.table1 DataLog
WHERE
DataLog.SelectedForProcessing is null
I'm passing 300 into #topN in the above sql, a value I've got configured in my app.config file, but this query running on 2 different servers has returned 304 rows in one instance and 307 rows in another instance.
I cant find anywhere that may be interfering with the 300, to turn it into 304 or 307, so I'm beginning to wonder whether SQL Server will just return a few extra rows sometimes? (Same code on another server IS returning the expected 300 rows)
Is this expected behaviour?
Test this
declare #topN int = 100;
select #topN ;
delete * from #resultSet;
insert into
#resultSet
SELECT TOP (#topN)
field1,
field2
FROM
dbo.table1 DataLog
WHERE
DataLog.SelectedForProcessing is null;
select count(*)
FROM
dbo.table1 DataLog
WHERE
DataLog.SelectedForProcessing is null;
select count(*) from #resultSet;
SQL Server will consistently return TOP N rows when N is a constant value - no wiggle room there.
I see two possibilities:
#topN is getting a different value on occasion
#resultSet is somehow not empty before having new values inserted
If #resultSet is a variable declared elsewhere in your scripts, check to see that no other INSERT INTO statements might be leaving unnecessary rows.
One easy way to implement this in run-time would be to simply add another command before this INSERT INTO statement:
DELETE #resultSet;
INSERT INTO
#resultSet
SELECT TOP (#topN)
field1,
field2
FROM
dbo.table1 DataLog
WHERE
DataLog.SelectedForProcessing IS NULL
;
I'm looking to use SSIS to transform the data held from a single source table. One of the cells has a string of characters. For example:
##/\/\/\/\/\##HHHHHHBBBB##/\/\/\/\/\
There's also another cell on the same row which contains a date.
Basically I want a each character within that string to be transferred to a new table as a row on it's own. The first two characters represent the date given in the other cell. The next two characters represent the following day and so on. So as well as having each character on it's own I would also want to increment the data and store that too.
Any idea how I would go about doing this or even if SSIS is the correct tool to be using.
Many Thanks
I wonder if you'd be better running this through a split-string function in SQL first? That way you'l be getting rows for each character along-side the date, and then you can just output it straight to a destination.
I've created a function to facilitate this:
CREATE FUNCTION [dbo].[udf_SplitStringIntoRows](#text varchar(max))
RETURNS #tbl TABLE ([value] char(1) NOT NULL)
AS
BEGIN
WHILE len(#text) > 0
BEGIN
INSERT INTO #tbl
SELECT left(#text,1)
SET #text = RIGHT(#text,len(#text)-1)
END
RETURN
END
Then, to test the data i created a quick temp table with your data in:
DECLARE #source as TABLE([value] varchar(max), [date] datetime)
INSERT INTO #source
SELECT '##/\/\/\/\/\##HHHHHHBBBB##/\/\/\/\/\', getdate()
UNION
SELECT '##/\/\/\/\/\##HHHHHHBBBB##/\/\/\/\/\', getdate()+1
UNION
SELECT '##/\/\/\/\/\##HHHHHHBBBB##/\/\/\/\/\', getdate()+2
Then cross applied the function to this dataset:
SELECT d.[value], s.date
FROM #source s
CROSS APPLY dbo.[udf_SplitStringIntoRows](s.value) d
Which should give you the source dataset you require to further process in SSIS.
I have a table with some persistent data in it. Now when I query it, I also have a pretty complex CTE which computes the values required for the result and I need to insert missing rows into the persistent table. In the end I want to select the result consisting of all the rows identified by the CTE but with the data from the table if they were already in the table, and I need the information whether a row has been just inserted or not.
Simplified this works like this (the following code runs as a normal query if you like to try it):
-- Set-up of test data, this would be the persisted table
DECLARE #target TABLE (id int NOT NULL PRIMARY KEY) ;
INSERT INTO #target (id) SELECT v.id FROM (VALUES (1), (2)) v(id);
-- START OF THE CODE IN QUESTION
-- The result table variable (will be several columns in the end)
DECLARE #result TABLE (id int NOT NULL, new bit NOT NULL) ;
WITH Source AS (
-- Imagine a fairly expensive, recursive CTE here
SELECT * FROM (VALUES (1), (3)) AS Source (id)
)
MERGE INTO #target AS Target
USING Source
ON Target.id = Source.id
-- Perform a no-op on the match to get the output record
WHEN MATCHED THEN
UPDATE SET Target.id=Target.id
WHEN NOT MATCHED BY TARGET THEN
INSERT (id) VALUES (SOURCE.id)
-- select the data to be returned - will be more columns
OUTPUT source.id, CASE WHEN $action='INSERT' THEN CONVERT(bit, 1) ELSE CONVERT(bit, 0) END
INTO #result ;
-- Select the result
SELECT * FROM #result;
I don't like the WHEN MATCHED THEN UPDATE part, I'd rather leave the redundant update away but then I don't get the result row in the OUTPUT clause.
Is this the most efficient way to do this kind of completing and returning data?
Or would there be a more efficient solution without MERGE, for instance by pre-computing the result with a SELECT and then perform an INSERT of the rows which are new=0? I have difficulties interpreting the query plan since it basically boils down to a "Clustered Index Merge" which is pretty vague to me performance-wise compared to the separate SELECT followed by INSERT variant. And I wonder if SQL Server (2008 R2 with CU1) is actually smart enough to see that the UPDATE is a no-op (e.g. no write required).
You could declare a dummy variable and set its value in the WHEN MATCHED clause.
DECLARE #dummy int;
...
MERGE
...
WHEN MATCHED THEN
UPDATE SET #dummy = 0
...
I believe it should be less expensive than the actual table update.