Restrict many - many results in SQL join - sql

Question

The SQL below contains some DDL and a simple query.
The result I am getting is
a1|b1|c1
a1|b2|c3
a3|b3|c2
a3|b3|c3
a3|b3|c4
a3|b3|c5
a3|b5|c6
a3|b5|c7
The result I want is
a1 |b1 |c1
a1 |b2 |c3
a3 |b3 |c2
null |null |c4
null |null |c5
a3 |b5 |c6
null |null |c7
I tried using MAX, MIN, rownums and what not. I am at my wit's end. I am including only the base query I started with and not all the options I tried because they don't work at all. Any help is appreciated!
BEGIN TRANSACTION;
drop table if exists table_A;
drop table if exists table_B;
drop table if exists table_C;
/* Create a table called NAMES */
CREATE TABLE table_A(a_Id text PRIMARY KEY, val_a text);
CREATE TABLE table_B(a_Id text, b_Id text, val_b text);
CREATE TABLE table_C(b_Id text, c_Id text, val_c text);
/* Create few records in this table */
INSERT INTO table_A VALUES('a1','va1');
INSERT INTO table_A VALUES('a2','va2');
INSERT INTO table_A VALUES('a3','va3');
INSERT INTO table_B VALUES('a1', 'b1','vb1');
INSERT INTO table_B VALUES('a1', 'b2','vb2');
INSERT INTO table_B VALUES('a3', 'b3','vb31');
INSERT INTO table_B VALUES('a2', 'b4','vb4');
INSERT INTO table_B VALUES('a3', 'b5','vb31');
INSERT INTO table_C VALUES('b1', 'c1','vc1');
INSERT INTO table_C VALUES('b3', 'c2','vc2');
INSERT INTO table_C VALUES('b3', 'c3','vc3');
INSERT INTO table_C VALUES('b2', 'c3','vc3');
INSERT INTO table_C VALUES('b3', 'c4','vc2');
INSERT INTO table_C VALUES('b3', 'c5','vc3');
INSERT INTO table_C VALUES('b5', 'c6','vc3');
INSERT INTO table_C VALUES('b5', 'c7','vc3');
COMMIT;
select
a.a_Id, b.b_Id, c.c_Id
from
table_A as a
join
table_B as b
on a.a_Id = b.a_Id
join
table_C as c
on b.b_Id = c.b_Id;

Answer 1

something like this should work (I have tested it on PostgreSql, should work on Oracle too)
SELECT
case when row_number = 1 then a_id end as a_id,
case when row_number = 1 then b_id end as b_id,
c_id
FROM (
SELECT
a.a_Id,
b.b_Id,
c.c_Id,
row_number() OVER (partition by a.a_id, b.b_id order by c.c_id) as row_number, --for a_id, b_id
row_number() OVER (partition by c.c_id order by c.c_id) as row_number2 --to avoid c_id duplicates
FROM
table_A a
join
table_B b on a.a_Id = b.a_Id
join table_C c on b.b_Id = c.b_Id
) innerquery
WHERE
row_number2 = 1 --this is to avoid c_id duplicates
SQLFIDDLE

Answer 2

I'd first recommend that this sounds better to handle in your presentation logic. However, it is possible to accomplish with SQL alone.
You can take advantage of Oracle's LAG() function along with CASE to check if the previous row had the same a and b id values.
Here's an example using a common table expression:
with cte as (
select
a.a_Id, b.b_Id, c.c_Id,
lag (a.a_Id,1) over (order by a.a_Id, b.b_Id) prev_a_Id,
lag (b.b_Id,1) over (order by a.a_Id, b.b_Id) prev_b_Id
from table_A a
join table_B b
on a.a_Id = b.a_Id
join table_C c
on b.b_Id = c.b_Id
order by
a.a_id, b.b_id
)
select
case
when prev_a_Id is null or
prev_a_Id <> a_Id or
prev_b_Id <> b_Id
then a_id
end new_a_Id,
case
when prev_a_Id is null or
prev_a_Id = a_Id or
prev_b_Id = b_Id
then b_id
end new_b_Id, c_Id
from cte;
SQL Fiddle Demo

Answer 3

select t1.a_id, t1.b_id, table_c.c_id
from table_c
left join
(
select a_Id, b_Id, c_Id
from
(
select a.a_Id as a_id, b.b_Id as b_id, c.c_Id as c_id,
ROW_NUMBER() OVER (PARTITION BY a.a_ID, b.b_id ORDER BY C_ID) as aNum
from table_A as a
join table_B as b on a.a_Id = b.a_Id
join table_C as c on b.b_Id = c.b_Id
) t2
where aNum = 1
) t1 on table_c.c_id = t1.c_id
order by table_c.c_id
fiddle:
http://sqlfiddle.com/#!3/6049b/1

Answer 4

I don't think i fully understood that you are trying to achieve, but correct me please if i'm wrong.
First, you use inner join to join tables (at least it will be if you use sql server, but should be the same in oracle). That means that for example you will get row from first table only if it has a correspondent row in the second table, And if now rows in first table for corresponding rows in second table, that rows from second table never appear in results.
According to description of result that you want to achieve you need is a outer join . Which one exactly left/right/full outer join depends on that you are trying to achieve (looks like you need left outer join or full outer join). I'm not quite sure that you exact aim because you explain how your data in this concrete example should looks like not general case.
So please have a look at description of different join types and choose sql join types
And also one importation remark: text type is probably last type from the list that i would consider as primary key.