assign site name and number sequence to patid in SAS - sas

Question

I have a dataset with many sites (like in excerpt below) where the 3 participants at the end with XXX are additional dummy participants. What I want is to modify these XXX patids so that they are in the format I have in the "want" dataset below, as opposed to having XXX at the end, while keeping the group assignment from have dataset unchanged.
data have;
input site $ patid $ group $;
datalines;
ABC ABCPROT01001 A
ABC ABCPROT01002 B
ABC ABCPROT01003 A
ABC ABCPROT01005 A
ABC ABCPROT01006 A
ABC ABCPROT01XXX B
ABC ABCPROT01XXX A
ABC ABCPROT01XXX B
CDF CDFPROT01004 A
CDF CDFPROT01005 A
CDF CDFPROT01006 A
CDF CDFPROT01007 B
CDF CDFPROT01008 A
CDF CDFPROT01009 B
CDF CDFPROT01010 A
CDF CDFPROT01012 A
CDF CDFPROT01013 B
CDF CDFPROT01XXX B
CDF CDFPROT01XXX B
CDF CDFPROT01XXX A
AMD AMDPROT01001 A
AMD AMDPROT01002 B
AMD AMDPROT01003 A
AMD AMDPROT01XXX B
AMD AMDPROT01XXX A
AMD AMDPROT01XXX A
;
run;
data want;
input site $ patid $ group $;
datalines;
ABC ABCPROT01001 A
ABC ABCPROT01002 B
ABC ABCPROT01003 A
ABC ABCPROT01005 A
ABC ABCPROT01006 A
ABC ABCPROT01007 B
ABC ABCPROT01008 A
ABC ABCPROT01009 B
CDF CDFPROT01004 A
CDF CDFPROT01005 A
CDF CDFPROT01006 A
CDF CDFPROT01007 B
CDF CDFPROT01008 A
CDF CDFPROT01009 B
CDF CDFPROT01010 A
CDF CDFPROT01012 A
CDF CDFPROT01013 B
CDF CDFPROT01014 B
CDF CDFPROT01015 B
CDF CDFPROT01016 A
AMD AMDPROT01001 A
AMD AMDPROT01002 B
AMD AMDPROT01003 A
AMD AMDPROT01004 B
AMD AMDPROT01005 A
AMD AMDPROT01006 A
;
run;

Answer 1

This answer takes care of both problems:
Your XXX observations are already in the Data
It handles group with only XXX observations
code:
data have;
input site $ patid :$12. group $;
datalines;
ABC ABCPROT01001 A
ABC ABCPROT01002 B
ABC ABCPROT01003 A
ABC ABCPROT01005 A
ABC ABCPROT01006 A
ABC ABCPROT01XXX B
ABC ABCPROT01XXX A
ABC ABCPROT01XXX B
CDF CDFPROT01004 A
CDF CDFPROT01005 A
CDF CDFPROT01006 A
CDF CDFPROT01007 B
CDF CDFPROT01008 A
CDF CDFPROT01009 B
CDF CDFPROT01010 A
CDF CDFPROT01012 A
CDF CDFPROT01013 B
CDF CDFPROT01XXX B
CDF CDFPROT01XXX B
CDF CDFPROT01XXX A
QQL QQLPROT01XXX A
QQL QQLPROT01XXX B
QQL QQLPROT01XXX A
AMD AMDPROT01001 A
AMD AMDPROT01002 B
AMD AMDPROT01003 A
AMD AMDPROT01XXX B
AMD AMDPROT01XXX A
AMD AMDPROT01XXX A
;
run;
data want (drop = n l);
set have;
by site notsorted;
if first.site then n = 0;
l = substr(patid, length(patid) - 2);
if not find(l, 'x', 'i') then n = input(l, 3.);
else n = sum(n, 1);
substr(patid, 10, 3) = put(n, z3.);
retain n;
run;

Answer 2

I assume that the three extra obs are not there in your have data set.
Try this
data have;
input site $ patid $12.;
datalines;
ABC ABCPROT01001
ABC ABCPROT01002
ABC ABCPROT01003
ABC ABCPROT01005
ABC ABCPROT01006
CDF CDFPROT01004
CDF CDFPROT01005
CDF CDFPROT01006
CDF CDFPROT01007
CDF CDFPROT01008
CDF CDFPROT01009
CDF CDFPROT01010
CDF CDFPROT01012
CDF CDFPROT01013
AMD AMDPROT01001
AMD AMDPROT01002
AMD AMDPROT01003
;
data want;
do _N_ = 1 by 1 until (last.site);
set have;
by site notsorted;
l = input(substr(patid, length(patid) - 2), 3.);
output;
end;
do l = l + 1 to l + 3;
substr(patid, 10, 3) = put(l, z3.);
output;
end;
run;