if I want to duplicate a specific teacher so that they appear twice in this select statement without creating a new a table, how can I do this?
Teacher | Grade | District |
---|---|---|
Mr. Smith | 3rd | West |
Mrs. John | 4th | South |
Mr. Cameron | 2nd | North |
Mr. Cameron | 2nd | North |
Kirk Horn | 1st | West |
And I want it to look like this for Mr. Cameron:
Teacher | Grade | District |
---|---|---|
Mr. Smith | 3rd | West |
Mrs. John | 4th | South |
Mr. Cameron | 2nd | North |
Mr. Cameron | 2nd | North |
Mr. Cameron | 2nd | North |
Mr. Cameron | 2nd | North |
Kirk Horn | 1st | West |
What would that select statement look like?
I think for reference I want to do something like this but without the insert: https://dba.stackexchange.com/questions/142414/easiest-way-to-duplicate-rows
CodePudding user response:
If the table is big, and the filter is not selective and backed by an index, then this "trick" avoids a second sequential scan over the table - using PostgreSQL:
SELECT t.* -- or your list of columns
FROM test t
, generate_series(1, CASE WHEN t.teacher = 'Mr. Cameron' THEN 2 ELSE 1 END);
It's short syntax for a LATERAL
join. See:
If you need more copies, just replace '2' above.
Postgres has a hard time estimating the number of rows to expect with this construct, which may confuse query planning. Stu's variant (doing the same) is slightly more expensive, but easier to estimate for query planning. Syntax needs to be adapted for Postgres:
SELECT teacher, grade, district
FROM test t
JOIN LATERAL (VALUES (1),(2)) x(v) ON v = 1 OR teacher = 'Mr. Cameron';
CodePudding user response:
You could use UNION ALL
select Teacher,Grade,District
from test
union all
select Teacher,Grade,District
from test
where Teacher='Mr. Cameron'
order by Teacher;
If you want to force/predict how many duplicate values for Mr. Cameron
you will add ,try below query which add only one duplicate value limit 1
(select Teacher,Grade,District
from test
)
union all
(
select Teacher,Grade,District
from test
where Teacher='Mr. Cameron'
limit 1
);
CodePudding user response:
You can make use of a rows constructor here with a lateral join.
You've tagged two different RDBMS engines, both support the following, a MySql solution looks like
select Teacher, Grade, District
from t
join lateral (values row(1),row(2))x(v)
where v = 1 or Teacher = 'Mr. Cameron';