Prev: Selecting cases(data step) for non sequential IDs
Next: WHERE statement issue in data step
From: nina on 3 May 2010 13:33
a data dictionary defines a format value based on a wildcard

example:

name, value
A , M?H
A, GHH
B , ANN
C , K?N

since the values that are defined with a "?" can be any character, what is a
good method for parsing out the matches in order to come up with the Name
value ?

data test;
input testvalue $4.;
cards;
MAH
MOJ
MOH
ANU
ANN
KIN
K*N
;;;;


From: nina on 3 May 2010 13:47
any danger in using something like

proc sql;
create table match as select * from testvalue
from testvalue x left join formatlookup b
on x.testvalue LIKE b.value
;quit;



From: Ya on 3 May 2010 16:36
On May 3, 10:33 am, "nina" <s...(a)mailinator.com> wrote:
> a data dictionary defines a format value based on a wildcard
>
> example:
>
> name, value
> A , M?H
> A,  GHH
> B , ANN
> C , K?N
>
> since the values that are defined with a "?" can be any character, what is a
> good method for parsing out the matches in order to come up with the Name
> value ?
>
> data test;
> input testvalue $4.;
> cards;
> MAH
> MOJ
> MOH
> ANU
> ANN
> KIN
> K*N
> ;;;;

You can construct your format based on the lookup dataset:

data xx;
input name $ value $;
cards;
A M*H
A GHH
B ANN
C K*N
;

data xx1;
set xx;
retain fmtname '$mycode' type 'C';
if index(value,'*') then do;
do s=65 to 65+25;
start=translate(value,byte(s),'*');
label=name;
output;
end;
start=value;
label=name;
output;
end;
else do;
start=value;
label=name;
output;
end;
run;

proc format cntlin=xx1;
run;

data test;
input testvalue $;
name=put(testvalue,$mycode.);
cards;
MAH
MOH
ANN
KIN
K*N
;

proc print;
run;

Obs testvalue name

1 MAH A
2 MOH A
3 ANN B
4 KIN C
5 K*N C
From: nina on 3 May 2010 21:46
did not work correctly, see below, three(3) errors

data xx;
input name $ value $;
cards;
A M*H
A GHH
B A?N
D J*N
D K?N
;

data xx1;
set xx;
retain fmtname '$mycode' type 'C';
if index(value,'*') then do;
do s=65 to 65+25;
start=translate(value,byte(s),'*');
label=name;
output;
end;
start=value;
label=name;
output;
end;
else do;
start=value;
label=name;
output;
end;
run;

proc format cntlin=xx1;
run;

data test;
input testvalue $;
name=put(testvalue,$mycode.);
cards;
MAH
MOH
ANN
KIN
K_N
;

data _null_;set ;put(_all_)(=);run;

testvalue=MAH name=A
testvalue=MOH name=A
testvalue=ANN name=A (should be B)
testvalue=KIN name=K (should be D)
testvalue=K_N name=K (should be D)

From: Arthur Tabachneck on 4 May 2010 00:34
Nina,

The following includes some lookups you may not want to include, but
it does work for your example:

data xx;
input name $ value $;
cards;
A M*H
A GHH
B A?N
D J*N
D K?N
;
data xx1 (drop=pos lookup);
set xx;
retain fmtname '$mycode' type 'C';
pos= index(value,'*') + index(value,'?');
if pos then do;
lookup=substr(value,pos,1);
do s=65 to 65+30;
start=translate(value,byte(s),lookup);
label=name;
output;
end;
start=value;
label=name;
output;
end;
else do;
start=value;
label=name;
output;
end;
run;


proc format cntlin=xx1;
run;


data test;
input testvalue $;
name=put(testvalue,$mycode.);
cards;
MAH
MOH
ANN
KIN
K_N
;


data _null_;set ;put(_all_)(=);run;

HTH,
Art
-------------
On May 3, 9:46 pm, "nina" <s...(a)mailinator.com> wrote:
> did not work correctly, see below, three(3) errors
>
> data xx;
> input name $ value $;
> cards;
> A  M*H
> A  GHH
> B  A?N
> D  J*N
> D  K?N
> ;
>
> data xx1;
>  set xx;
> retain fmtname '$mycode' type 'C';
> if index(value,'*') then do;
>  do s=65 to 65+25;
>  start=translate(value,byte(s),'*');
>  label=name;
>  output;
>  end;
>  start=value;
>  label=name;
>  output;
> end;
> else do;
>  start=value;
>  label=name;
>  output;
> end;
> run;
>
> proc format cntlin=xx1;
> run;
>
> data test;
> input testvalue $;
> name=put(testvalue,$mycode.);
> cards;
> MAH
> MOH
> ANN
> KIN
> K_N
> ;
>
> data _null_;set ;put(_all_)(=);run;
>
> testvalue=MAH name=A
> testvalue=MOH name=A
> testvalue=ANN name=A (should be B)
> testvalue=KIN name=K (should be D)
> testvalue=K_N name=K (should be D)

 |  Next  |  Last
Pages: 1 2
Prev: Selecting cases(data step) for non sequential IDs
Next: WHERE statement issue in data step