SQL Advance, creating fill new column with coditional value - sql

Question

Cound you please support with question below, i trying to fill a new colmn in my SQL query based in a conditional value.
pseudo code:
IF ap_id exists in visit_table and visit_type = 'first'
then firs_tvisit_id = visit_table.visit_id and first_visit_user = visit_table.Username
The same logic for second and third visit.
This are visit_table.
And this are ap_table (where are the key values):
And below is the expected table.
SQL is able to make these kind of data manipulation, if there are how can i get this.
I alrady tryed left joins, inner joins, and full outher join. however i were not able to create a new column and fill this based in a conditional.

Answer 1

One method is multiple joins:
select a.*,
v1.visit_id, v1.user_name,
v2.visit_id, v2.user_name,
v3.visit_id, v3.user_name
from ap_table a left join
visit_table v1
on v1.ap_id = a.ap_id and v1.type = 'first' left join
visit_table v2
on v2.ap_id = a.ap_id and v2.type = 'second' left join
visit_table v3
on v3.ap_id = a.ap_id and v3.type = 'third'

Answer 2

You can join and do conditional aggregation:
select
a.ap_id,
a.location,
max(case when visit_type = 'first' then visit_id end) first_visit_id,
max(case when visit_type = 'second' then visit_id end) second_visit_id,
...
max(case when visit_type = 'first' then username end) first_visit_username,
max(case when visit_type = 'second' then username end) second_visit_username,
...
from ap_table a
inner join visit_table v on v.ap_id = a.ap_id
group by a.ap_id, a.location

Answer 3

It looks like you have some potential data issues here that must be answered for a proper solution. Most specifically, what should happen if the visit_table has multiple records for the same ap_id and visit_type? Or should we assume that there is a unique key on those two columns?
The best answer to this question as posted is actually likely to be dependent on the flavor of SQL you are using and the nuance of what you are really trying to do. For example, PostgreSQL has the ability to "SELECT DISTINCT ON" columns without requiring grouping. Oracle, SQL Server, and others have the ability to simplify complex queries using the with clause. And if other methods fail you can always write nested queries to try and grab everything in one monster query.
If you are able to use a temp table or table variable you may be better off selecting information into a constructed table as needed. If you need more than 3 visits, you may have to write dynamic sql. And if this is information you want to have readily available for other queries to access you might consider writing it as a view instead.
One quick note about simply selecting min(visit_id) and min(username) for your query - if you have multiple grouped rows you are likely to get the id from one visit incorrectly paired with the username from another visit.
Here is an attempt at creating a monster query for just the 3 visit types included:
SELECT a.ap_id,
a.location,
v1.visit_id AS first_visit_id,
v1.username AS first_visit_username,
v2.visit_id AS second_visit_id,
v2.username AS second_visit_username,
v3.visit_id AS third_visit_id,
v3.username AS third_visit_username
FROM (
SELECT ai.ap_id,
(SELECT MIN(v1i.visit_id) FROM visit_table v1i ON v1i.ap_id = ai.ap_id AND v1i.type = 'first') AS v1_id,
(SELECT MIN(v2i.visit_id) FROM visit_table v2i ON v2i.ap_id = ai.ap_id AND v2i.type = 'second') AS v2_id,
(SELECT MIN(v3i.visit_id) FROM visit_table v3i ON v3i.ap_id = ai.ap_id AND v3i.type = 'third') AS v3_id
FROM ap_table ai
) x
JOIN ap_table a ON x.ap_id = a.ap_id
LEFT JOIN visit_table v1 ON x.v1_id = v1.visit_id
LEFT JOIN visit_table v3 ON x.v2_id = v2.visit_id
LEFT JOIN visit_table v3 ON x.v3_id = v3.visit_id
WHERE
x.v1_id IS NOT NULL
OR x.v2_id IS NOT NULL
OR x.v3_id IS NOT NULL