Taking a Single column of data and splitting it into several columns based on the subtitles.

[MVanderkloot]

I have an Excel file with this list - except it is 10,000 rows

----sas-eg-cca-full----
wlo
knak
votte
khu
sdza
pdeb
bmc
chnel
jell
mghe
alleb
----sas-eg-hcs-chaps-full----
cmat
obri
----sas-eg-dca-phe-read----
hswe
----sas-eg-fnihbabr-ship-read----
iagm
grwn

Final dataset would be:

wlo                  sas-eg-cca-full
knak                sas-eg-cca-full
votte                sas-eg-cca-full
khu                  sas-eg-cca-full
sdza                sas-eg-cca-full
pdeb                sas-eg-cca-full
bmc                 sas-eg-cca-full
chnel               sas-eg-cca-full
jell                   sas-eg-cca-full
mghe              sas-eg-cca-full
alleb               sas-eg-cca-full
cmat               sas-eg-hcs-chaps-full
obri                 sas-eg-hcs-chaps-full
hswe               sas-eg-dca-phe-read
iagm               sas-eg-fnihbabr-ship-read
grwn               sas-eg-fnihbabr-ship-read 

Thanks MARK

[tomrvincent]

What have you tried so far?

[MVanderkloot]

I have checked the community for Ideas - but couldn't find anything - so I don't know what to try.

[tomrvincent]

You need to identify the pattern and logic you're trying to achieve. Maybe keep the field as the 2nd column if it starts with '--'...

[MVanderkloot]

I was hoping for a SAS Program that can determine the cells with titles
that need to be repeated in the second column for each user in that part of
the list
Then keep going through the list - as there is no pattern - it is just the
names of the users within each Shared Directory on the SAS Server.
The data is pumped out of our system in this manner - we need to get a
system that we can look at the access of each user.

[Tom · Accepted solution]

First let's get your sample into an actual SAS dataset we can work with.  Let's call it HAVE and the variable USERNAME.

data have;
  input username :$80.;
cards;
----sas-eg-cca-full----
wlo
knak
votte
khu
sdza
pdeb
bmc
chnel
jell
mghe
alleb
----sas-eg-hcs-chaps-full----
cmat
obri
----sas-eg-dca-phe-read----
hswe
----sas-eg-fnihbabr-ship-read----
iagm
grwn
;

Then you just need to run a data step that creates a new retained variable, GROUP, that gets assigned the value of USERNAME when it looks like a group name instead of a username.

data want;
  set have;
  if username=:'----' then group=username;
  else output;
  retain group;
run;

Results:

Obs    username                  group

  1     wlo        ----sas-eg-cca-full----
  2     knak       ----sas-eg-cca-full----
  3     votte      ----sas-eg-cca-full----
  4     khu        ----sas-eg-cca-full----
  5     sdza       ----sas-eg-cca-full----
  6     pdeb       ----sas-eg-cca-full----
  7     bmc        ----sas-eg-cca-full----
  8     chnel      ----sas-eg-cca-full----
  9     jell       ----sas-eg-cca-full----
 10     mghe       ----sas-eg-cca-full----
 11     alleb      ----sas-eg-cca-full----
 12     cmat       ----sas-eg-hcs-chaps-full----
 13     obri       ----sas-eg-hcs-chaps-full----
 14     hswe       ----sas-eg-dca-phe-read----
 15     iagm       ----sas-eg-fnihbabr-ship-read----
 16     grwn       ----sas-eg-fnihbabr-ship-read----

If you really want to remove those quadruple dashes you might update the code that sets GROUP.

group=transtrn(username,'----',trimn(' '));

[MVanderkloot]

Thank you for your help - it works perfectly...

[Tom]

If you are reading it directly from a file (or a piped output of another operating system command) then you can take advantage or the DLMSTR= option on the INFILE statement.

data have;
  infile cards dlmstr='----';
  input username :$80.;
  if _infile_=:'----' then group=username;
  else output;
  retain group;
cards;
----sas-eg-cca-full----
wlo
knak
votte
khu
sdza
pdeb
bmc
chnel
jell
mghe
alleb
----sas-eg-hcs-chaps-full----
cmat
obri
----sas-eg-dca-phe-read----
hswe
----sas-eg-fnihbabr-ship-read----
iagm
grwn
;