Question

For a real estate website I need to implement a search mechanism that allows to search both for text and distance.
Doing distance calculations on a MySQL table when the lat & lon are recorded in separate columns is easy but a house tends to have a LOT of true/false properties.

I will need to store all those fields in a database because they need to be editable so I intend to use a simple table like | houseID | property | in which I store all properties that are true (set).

That will save me from making a ridiculously wide table with hundreds of columns, but searching this database will not be very feasible.

I have thought about adding a column of type text to each house’s main record that contains the fieldnames of all true properties. I would then search both the human text description and that text column but I feel that’s still not the best approach.

How could I solve this in a clean way?

Thanks in advance!

Answer 1

I would reccomend the Entity Attribute Value or EAV model of storing data. http://en.wikipedia.org/wiki/Entity%E2%80%93attribute%E2%80%93value_model (this is how WordPress Posts and Post Meta work).

so assuming tables like:

ENTITY_TABLE: (id,title,author,content,date_created,date_modified)
ATTRIBUTE_TABLE: (id, entity_id, akey, avalue)

Use a query like this:

SELECT e.*, 
    MAX( IF(m.akey= 'has_ac', m.avalue, 0) ) as 'has_ac', 
    MAX( IF(m.akey= 'has_garage', m.avalue, 0) ) as 'has_garage', 
    MAX( IF(m.akey= 'has_fridge', m.avalue, 0) ) as 'has_fridge', 
    MAX( IF(m.akey= 'latitude', m.avalue, 0) ) as 'latitude', 
    MAX( IF(m.akey= 'longitude', m.avalue, 0) ) as 'longitude'
FROM ENTITY_TABLE e
JOIN ATTRIBUTE_TABLE m ON e.id = m.entity_id
WHERE has_ac=1

that will select entities and their related attributes (has_ac, has_garage, has_fridge, latitude, and longitude) and require that all entities selected will have has_ac equal to 1 (true)

Now for the geo stuff:

SELECT e.*, 
    MAX( IF(m.akey= 'has_ac', m.avalue, 0) ) as 'has_ac', 
    MAX( IF(m.akey= 'has_garage', m.avalue, 0) ) as 'has_garage', 
    MAX( IF(m.akey= 'has_fridge', m.avalue, 0) ) as 'has_fridge', 
    MAX( IF(m.akey= 'latitude', m.avalue, 0) ) as 'latitude', 
    MAX( IF(m.akey= 'longitude', m.avalue, 0) ) as 'longitude',
    (
        3959 * 
        acos(
            cos( radians( MAX( IF(m.akey= 'latitude', m.avalue, 0) ) ) ) * 
            cos( radians( CUSTOMER_LAT ) ) * 
            cos( radians( CUSTOMER_LONG ) - radians( MAX( IF(m.akey= 'longitude', m.avalue, 0) ) ) ) + 
            sin( radians( MAX( IF(m.akey= 'latitude', m.avalue, 0) ) ) ) * 
            sin( radians( CUSTOMER_LAT ) ) 
        ) 
    ) AS distance
FROM ENTITY_TABLE e
JOIN ATTRIBUTE_TABLE m ON e.id = m.entity_id
WHERE has_ac=1
ORDER BY distance ASC