C-ask a big file statistics question about massive data

There is a single row in a large file over 20 GB (csv format, separated by commas) the order information mainly includes the user id, the merchant id, and the order id. you need to find "every merchant", "cumulative payment user per day", and "number of new payment users per day". PS: it is best to paste the code. it is best to use C or C ++ and php, another reason is that the memory is not very large. I need to consider the memory reason. the uid of "accumulative payment users" and "new payment users" that I need to find is the final display effect that needs to be removed from duplicates. it is similar to this! [Image description] (img. ask. csdn. netupload20... cphpc ++ multi-dimensional query of massive files

There is a single row in a large file over 20 GB (csv format, separated by commas) the order information mainly includes the user id, the merchant id, and the order id. you need to find "every merchant", "cumulative payment user per day", and "number of new payment users per day"
PS:
It is best to paste the code,
It is best to use C, C ++, and php,
Another reason is that memory is not very large.
The uid of the "accumulative payment user" and "New Payment user" that I need to find must be removed from duplicates.

The final display effect is similar to this