Grouping multiple columns

[Kysol]

Is there a way to group over multiple columns and still retain the unique count over those columns. Sorry I can't explain it better 🙁
For instance if I had the structure like this, I'm wanting to group all numbers in column 'c1' and 'c2' and get the overall count. But the tricky part is in 'id' #3 where the same number spans two columns, I want it to count that as 2 and not 1 instance.

+----+----+----+
| id | c1 | c2 |
+----+----+----+
|  1 |  1 |  3 |
|  2 |  1 |  1 |
|  3 |  2 |  3 |
|  4 |  4 |  1 |
|  5 |  3 |  1 |
+----+----+----+

What I'm looking to get is:

[1] = 5
[3] = 3
[2] = 1
[4] = 1

Any idea's ?

[Drakla]

Can you explain how the first table relates to the results? I have no idea how they relate.

[Kysol]

Sorry, basically you can read it like this (using classic apples oranges pairs style example):

1 - Apple, 2 - Orange, 3 - Pair, 4 - Banana

2 people choose random fruit each row. I'm wanting to find the results of how many times each fruit was chosen, but do this in one query if possible.

I have it working at the moment by creating a temp table, feeding the data from column 1 in then feeding column 2 in. After which doing a count + group by on the temp table, but this involves 4 seperate query strings (including creation and drop of temp table).

If there's no easier way this will have to do, but I was just checking if there was something I was missing after browsing through google for abit.

Thanks

[Lars_Berg]

select f,count(*)
from (select c1 as f from t union all select c2 from t) dt
group by f