How to search husband and wife data with one MySQL query? - mysql

Question

I'm trying to find the most efficient and optimized way of querying husband and wife data for a search function in a finance application. The clients can be single or married.
Currently, when the data is created, there is a table for the household that shares information such as username, password, address, location, etc...
There is a separate table that stores individual information about the husband and wife in separate rows including birth dates and income.
The app has a search function where a user can search using criteria such as location, husband age range and income range and wife age range and income range and should return individual household results.
For instance, a user can search for clients that are located within 20 miles where the husband is between 50 and 60 years old and the wife is between 40 and 50 years old with an income range of $30,000 to $40,000.
The result would produce all results for singles and couples.
Here is just an idea of what the tables and results may look like. Keep in mind that the location data would actually use lat and long but for the purpose of this example, we are just using actual miles to keep it simple.
Table Users:
ID | Username | Location | Password | Email | Status
-------------------------------------------
1 | singleclient | 5 miles | 24##$#dls | user1#email.com | Single
2 | marriedclient | 7 miles | $#$sls33 | user2#email.com | Married
Table UserDetails
ID | User_ID | Gender | Name | Age | Income
----------------------------------
1 | 1 | Male | John Smith | 55 | 32000
2 | 2 | Male | Mike Jones | 53 | 37000
3 | 2 | Female | Cindy Jones | 47 | 31000
Result:
ID | Username | Distance | Status | Male Name | Female Name | Male Age | Female Age | Male Income | Female Income
----------------------------------------
1 | singleclient | 5 miles | Single | John Smith | null | 55 | null | 32000 | null
2 | marriedclient | 7 miles | Married | Mike Jones | Cindy Jones | 53 | 47 | 37000 | 31000

Answer 1

First, in many countries, the assumption that a marital unit consists of a single male and a single female is not true. I would try to avoid building this assumption into the data model or application.
I think you can approach this question using aggregation with a having clause:
select ud.user_id
from UserDetails ud
group by ud.user_id
having sum(case when ud.gender = 'Male' and ud.age between 50 and 60) = 1 and
sum(case when ud.gender = 'Female' and ud.age between 40 and 50 and ud.income between 30000 and 40000) = 1;
This gives you the user_ids that match. You can then format it however you like.
The above is quite generic. You might find that this version works faster:
select ud1.*, ud2.*
from UserDetails ud1 join
UserDetails ud2
on ud1.user_id = ud2.user_id
where ud1.gender = 'Male' and ud1.age between 50 and 60 and
ud2.gender = 'Female' and ud2.age between 40 and 50 and ud2.income between 30000 and 40000;
Which is faster depends on the size of your data and how indexes are set up.

Answer 2

You can join the same table twice under different names, and use this to populate all the fields.
You could do this by selecting the male person and the female person, but this will obviously get you in trouble when you need to deal with parties where both members are of the same sex. It might be better then to just pick the lowest ID and the highest ID in the database, or the youngest and oldest person, or whatever.
Query would look something like this (untested)
SELECT u.id, u.username, u.distance, u.status, p1.name, p2.name, p1.age, p2.age, p1.income, p2.income
FROM Users u
INNER JOIN UserDetails p1 ON u.id = p1.user_id AND p1.id = (SELECT MIN(id) FROM UserDetails WHERE user_id = u.id)
RIGHT JOIN UserDetails p2 ON u.status == 'married' AND u.id = p2.user_id AND p2.id = (SELECT MAX(id) FROM UserDetails WHERE user_id = u.id)
Adding in the "status == married" in the second (right) join will make sure that the second query will not show the same person twice, but will instead just return a row of nulls.
You'll probably need to do the query twice (so that each person can be searched as p1 and as p2) if one of the two spouses should be "50-60" and the other "40-50" or one should make $10.000 and the other $20.000, because you don't know in which order they'll come out.