Return situation-dependent messages (exceptions) - postgresql

Question

I've got stored procedures in my PostgreSQL environment. One of these procedures stores user defined data.
In plpgsql-language, I would like to realize the following processing:
Because there are unique constraints in the destination-table, I first check, if the affected attributes are already in this table. If so, I want the stored procedure to return these affected attributes, so the user knows: "Uh, some of my data already exists, let's try something different!"
For example a part of the affected table:
CREATE TABLE XYZ (
Name varchar UNIQUE
...
)
Now I want to add a new row to XYZ. If the function notices that the name already exists it should return a message with the offending name. There can be multiple duplicates and the function should return all of them. How can this be implemented?
I first thought to check every attribute individually, but that's very slow:
-- Check, if abbreviation or name already exist.
SELECT EXISTS(
SELECT
1
FROM
"sample-scheme"."sample-table" AS tableName
WHERE
tableName.abbreviation = argAbbreviation
OR
tableName.name = argName
)
INTO
varEntryExists;
-- If abbreviation or name exists, return error message.
IF varEntryExists THEN
RETURN "Already exists.";
END IF;
This sample also doesn't return the offending attribute.

Answer 1

All you need is this simple query:
SELECT name
FROM XYZ
JOIN unnest ('{name1, name2, name3}'::varchar[]) AS n(name) USING (name);
Or use the "is contained by" operator <# (faster for small arrays):
SELECT name
FROM XYZ
WHERE name <# '{name1, name2, name3}'::varchar[]
Wrapped into a function, to return multiple finds, you can either return an array or RETURNS SETOF ... or RETURNS TABLE(...) to return a set. Demonstrating a plpgsql function (could also just be SQL), with a convenient VARIADIC parameter:
CREATE OR REPLACE FUNCTION f_name_dupes(VARIADIC _names varchar[])
RETURNS SETOF varchar AS
$func$
BEGIN
SELECT name
FROM XYZ x
WHERE name <# $1;
END
$func$ LANGUAGE sql;
Call:
SELECT * FROM f_name_dupes('name1', 'name2', 'name3');
Returns all names that already exist in table XYZ. More details for VARIADIC:
How to do WHERE x IN (val1, val2,…) in plpgsql