How to combine results from multiple tables with different columns? - php

Question

I have several tables with different numbers and types of columns, and a single column in common.
+--------+---------+------------+-------------+
| person | beardID | beardStyle | beardLength |
+--------+---------+------------+-------------+
+--------+-------------+----------------+
| person | moustacheID | moustacheStyle |
+--------+-------------+----------------+
I want to fetch all the results that match a given value of the shared column. I can do it using multiple select statements like this:
SELECT * FROM beards WHERE person = "bob"
and
SELECT * FROM moustaches WHERE person = "bob"
But this requires multiple mysql API calls, which seems inefficient. I was hoping I could use UNION ALL to get all the results in a single API call, but UNION requires that the tables have the same number and similar type of columns. I could write a SELECT statement that would manually pad the results from each table by adding columns with NULL values, but that would quickly get unmanageable for a few more tables with a few more columns.
I'm looking for a result set roughly like this:
+--------+---------+------------+-------------+-------------+----------------+
| person | beardID | beardStyle | beardLength | moustacheID | moustacheStyle |
+--------+---------+------------+-------------+-------------+----------------+
| bob | 1 | rasputin | 1 | | |
+--------+---------+------------+-------------+-------------+----------------+
| bob | 2 | samson | 12 | | |
+--------+---------+------------+-------------+-------------+----------------+
| bob | | | | 1 | fu manchu |
+--------+---------+------------+-------------+-------------+----------------+
Is there a way to achieve this that's fast and maintainable? Or am I better off running a separate query for each table?
Clarification:
I'm not looking for a cartesian product. I don't want a row for every combination of beard-and-moustache, I want a row for every beard and a row for every moustache.
So if there are 3 matching beards and 2 matching moustaches I should get 5 rows, not 6.

Answer 1

this should be working fine:
SELECT * FROM `beards` b LEFT OUTER JOIN `mustaches` ON (0) WHERE person = "bob"
UNION ALL
SELECT * FROM `beards` b RIGHT OUTER JOIN `mustaches` ON (0) WHERE person = "bob"
you don't have to handle the columns by yourself. the left and right outer join do this job.
unfortunately mysql doesn't have a full join. that's why you have to do it this way with a union
SELECT * FROM `customer` b LEFT OUTER JOIN `charges` ON (0) LEFT OUTER JOIN `day` ON (0)
UNION
SELECT * FROM `customer` b RIGHT OUTER JOIN `charges` ON (0) LEFT OUTER JOIN `day` ON (0)
UNION
SELECT * FROM `customer` b LEFT OUTER JOIN `charges` ON (0) RIGHT OUTER JOIN `day` ON (0)
this is a local test i made

Answer 2

Join on person....
I.e.
Select
t1.(asterix), t2.(asterix)
FROM
beards t1
INNER JOIN
moustaches t2 On t2.person = t1.person

Answer 3

SELECT *
FROM beards
JOIN moustaches
ON moustaches.person = beards.person
WHERE person = "bob"

Answer 4

I had fun with this, not sure it's entirely manageable with what more you have to add, but it accomplished the goal.
create table beard (
person varchar(20)
,beardID int
,beardStyle varchar(20)
,beardLength int )
create table moustache(
person varchar(20)
,moustacheID int
,moustacheStyle varchar(20))
insert into beard
select 'bob', 1, 'rasputin', 1
union select 'bob', 2, 'samson', 12
insert into moustache
select 'bob', 1, 'fu manchu'
declare #facialhair table (
person varchar(20)
,beardID int
,beardStyle varchar(20)
,beardLength int
,moustacheID int
,moustacheStyle varchar(20))
declare #i int
declare #name varchar(20)
set #name = 'bob'
set #i = (select COUNT(*) from beard where person = #name)
+ (select COUNT(*) from moustache where person = #name)
print #i
while #i > 0
begin
insert into #facialhair (person, beardID, beardStyle, beardLength)
select person, beardID, beardStyle, beardLength
from beard
where person = #name
set #i = #i-##ROWCOUNT
insert into #facialhair (person, moustacheID, moustacheStyle)
select person, moustacheID, moustacheStyle
from moustache
where person = #name
set #i = #i-##ROWCOUNT
end
select *
from #facialhair

Answer 5

I think you would be better by making queries for data in each table.
One of other possibilities is to concatenate data from all columns into one big string (you could choose some sign to separete column's values), then you should be able to use union all clause to combine results from each query - but then you will have to parse each row.. And data types will be lost.