C Consult a large-file statistical problem on massive data

Cphpc++ multi-Dimension query large files

There is a large file of more than 20 g inside a row of (CSV format, comma-separated type of the kind) is the order information is the main user ID merchant ID and the order ID must be required to find out "each merchant" "Daily cumulative Payment users" and "the number of new payment users per day"
Ps:
It's best to post the code,
It's better to use C or c++,php and OK,
And the memory is not very big need to consider memory reasons
I need to ask for the UID of "cumulative payment user" and "number of new payment users" to remove duplicate

The final effect is similar to this

