单个MySQL select语句中的“不同计数”和“如果存在则不同计数”

Question

I have a table of emails that I have sent along with the date I sent the email.

-----------------------------
|   email   |   date_sent   |
-----------------------------
|  a@a.com  |  2013-01-10   | 
|  a@a.com  |  2013-01-10   |
|  a@a.com  |  2013-01-10   |
|  b@b.com  |  2013-01-10   |
|  a@a.com  |  2013-01-11   |
|  c@c.com  |  2013-01-11   |
-----------------------------

For a given date I need to get a count of the total emails sent, the unique addresses for today and the unique addresses that haven't appeared before.

The first part is easy:

SELECT count(email) emails_sent, count(distinct email) unique_users
from sent_emails where date_sent = '2013-01-10';

I am having trouble getting the emails that haven't appeared before. Is it possible?

I have tried many different ways with little to no success. eg:

SELECT count(email) emails_sent, count(distinct email) unique_users,
sum( (select count(email) from sent_emails se
where se.date_sent < date_sent 
and se.email = email)) new_emailed_users
from sent_emails where date_sent = '2013-01-10';

I'm sure this is possible.

This is what I expect the result to be for 2013-01-10:

--------------------------------------------------
| emails_sent | unique_users | new_emailed_users |
--------------------------------------------------
|      4      |      2       |         2         |
--------------------------------------------------

This is what I expect the result to be for 2013-01-11:

--------------------------------------------------
| emails_sent | unique_users | new_emailed_users |
--------------------------------------------------
|      2      |      2       |         1         |
--------------------------------------------------

Thanks for any help

Answer 1

without a subquery, improving speed

select
count(if(date_sent='2013-01-10',email, null)) as emails,
count(distinct if(date_sent='2013-01-10',email, null)) as uniques,
count(distinct email) as total_unique_email
from sent_emails

Answer 2

Here is my solution. Not sure it is the most efficiant:

SELECT count(email) emails_sent, count(distinct email) unique_users,
count(distinct if(not exists(select 1 from sent_emails where  date_sent < se.date_sent and se.email = email ),email,null)) new_emailed_users
from sent_emails se where date_sent = '2013-01-10';