Create BD structure for multiple stats - database

Question

my cuestion is simple but there are so many possibilities...
Imagine you have a car, the car has a state (available, rented, in workshop, sold,...) and each state has different attributes:
available: we want to keep the employee has established the state and date.
rented: employee, client(company or person), start date, end date and price
in workshop: workshop, date.
sold: client(company or person), employee, date, price.
and finally, we want to keep a record of states of each car.

Answer 1

Easy. This is one entity (car) that has a multi-valued attribute (state). Let's use status, the word state has other more obvious meanings. However, the entity has other associated attributes that only apply according to the status, so the status attribute can be treated as if it were an entity. This the status of a car is represented as a relationship with the status "entity." Your other requirement that the status be recorded is also easy. Be aware, though, that easy does not necessarily mean simple.
create table Cars(
ID int auto_generated primary key,
... ... -- other attributes
);
The status has a base-class, sub-class format
create table Status(
ID char( 1 ) not null primary key, -- 'A', 'R', 'W', 'S'
Label varchar not null - 'Available', 'Rented', 'In Workshop', 'Sold'
constraint UQ_StatusLabel unique( ID, Label )
);
Now the tables that will contain the actual status information for the cars.
create table AStatus(
A_ID char( 1 ) check( A_ID = 'A' ),
EstablishedBy int not null references Employees( ID ),
);
create table RStatus(
R_ID char( 1 ) check( R_ID = 'R' ),
RentedBy int not null references Employees( ID ),
RentedTo int not null references Clients( ID ),
Price currency not null
);
create table WStatus(
W_ID char( 1 ) check( W_ID = 'W' ),
WS_ID int not null references Workshops( ID )
);
create table SStatus(
S_ID char( 1 ) check( S_ID = 'S' ),
SoldBy int not null references Employees( ID ),
SoldTo int not null references Clients( ID ),
Price currency not null
);
Finally, the intersection table that ties these all together.
create table CarStatus(
CarID int not null references Cars( ID ),
EffDate date not null default Now(),
StatusID char( 1 ) not null references Status( ID ),
constraint PK_CarStatus primary key( CarID, EffDate )
);
Let's follow a car as it goes into the workshop, becomes available, is rented out, returned (becomes available again) and then sold. All these changes take place (conveniently enough) on the first date of successive months.
CarStatus xStatus
1 1/1/16 W W 17 -- Jan 1, Car #1 goes to workshop 17
1 2/1/16 A A 2 -- Feb 1, Employee #2 makes if available
1 3/1/16 R R 2 12 24 -- Mar 1, Employee #2 rents to client #12 for $24
1 4/1/16 A A 3 -- Apr 1, Employee #3 returns car to Available
1 5/1/16 S S 3 13 20000 -- May 1, Employee #3 sells car to client #13 for $20000
Note that the table rows do not replace each other. So the CarStatus table contains the five rows as shown, AStatus contains two rows, and each of the other xStatus tables contain one row each.
To read the history of what has happened to car # 1:
select c.id CarID, s.Label Status, cs.EffDate "Effective Date"
from CarStatus cs
join Cars c
on c.ID = cs.CarID
join Status s
on s.ID = cs.StatusID
where cs.CarID = 1
order by cs.EffDate;
CarID Status Effective Date
1 In Workshop 1/1/16
1 Available 2/1/16
1 Rented 3/1/16
1 Available 4/1/16
1 Sold 5/1/16
Note there is only one date associated with a rental. The return date is the date the car next becomes available. If there is some length of time between a return and when it can become available again, such as for cleaning, then add another status. Something like, oh, Cleaning. (Or you could add a "ReturnDate" field to the RStatus table, but that opens you up for the possibility of anomalous data so you have to be extra vigilant.)
To show the current state of a car is a little more involved. But I'm using Version Normal Form (search for it under my name for more details) which is a standard pattern I use for maintaining a history of data changes. So the query also fits a standard pattern which is quickly learned.
select c.id CarID, s.Label Status, cs.EffDate "Effective Date"
from CarStatus cs
join CarStatus cs2
on cs2.CarID = cs.CarID
and cs2.EffDate =(
select Max( cs3.EffDate )
from CarStatus cs3
where cs3.EffDate <= Now() )
join Cars c
on c.ID = cs.CarID
join Status s
on s.ID = cs.StatusID
where cs.CarID = 1
order by cs.EffDate;
This will show the last row of the result set above because that is the last status entered. You can remove the WHERE and ORDER BY clauses and use this query to create a view which shows the current status of all the cars.
Here's the cool part. Suppose you want to see the status of the car as of 15 Mar. Just change the "Now()" in the query above to "to_date('3/15/16')" and you will get the third row in the result set above: Rented. So you can look at the current status or the past status on any particular date using the same query.