If you are skilled in languages such as python, R or C# and start coding in SAS, you might miss some of the functions and methods you’re are so familiar with. Or perhaps you just would like a function that isn’t there, or reshape what's already in the SAS toolbox.
But hey, no worries - Proc FCMP to the rescue! It allows you to create and mimic your favorite functions with a few lines of code.
One of the string manipulation functions I really like in .Net (C#, VB) and miss in SAS is the String.Format method, so this article describes how the custom-built function FormatString I created works and how you can implement it yourself. It mimics the String.Format method and my examples use the Numeric and the Named placeholders, so there are actually two functions, FmtStrNum and FmtStrName. I will sometimes refer to them both as FormatString in the article.
Calling the function FmtStrNum that I created with FCMP…
FmtStrNum('Of all the great Prog rock bands {1}, {2}, {3}, {4} and {5}, my current favorite is {2}.', b1, b2, b3, b4, b5);
…would return this string with variables b1-b5 set with the band names.
Of all the great Prog Rock bands Aristocrats, MÖRGlbl, Cosmosquad, Neal Morse and Dream Theater, my current favorite is MÖRGlbl.
The function FmtStrNum uses numeric placeholders from {1} to {5}, and that works well in the above example. Further down you will see the function FmtStrName that uses named placeholders which is more suitable to that text string. Obviously, you can call these functions whatever you want.
FormatString reduces the need for concatenations which often is time-consuming and tedious to construct, but more importantly – FormatString is much easier to read and understand for you and others maintaining your code.
It could look like this the traditional way and compared to FmtStrNum it’s rather messy. Of course, we have the excellent CAT functions but it still results in a not so easy to read statement. FormatString is concatenation and replacement in one go, and easy to read as well.
Band = 'Of all the great Prog Rock bands ' !! strip(b1) !! ', ' !! strip(b2) !! ', ' !! strip(b3) !! ', ' !!
strip(b4) !! ' and ' !! strip(b5) !! ', my current favorite is ' !! strip(b2) !! '.';
We all know SAS is a very rich language with buckets of powerful functions, and FormatString use one of them with the addition of a few lines of code, to achieve this powerful string formatting function.
Custom-built function adds great value to your SAS programs. Easier to read, maintain and debug. You can replace big chunks of Case-When-Else SQL with a one-liner, having the logic handled much nicer in a function where you can do whatever is needed, not limited to what SQL allows. We can also write functions to replace SAS Formats which offers greater flexibility than PROC FORMAT. The FCMP documentation have some great examples on what you can do, but this article is about doing some of what the String.Method can do in e.g. C#.
Example 1 – Numeric placeholders
This is the Proc FCMP code to build the function FmtStrNum.
- The first parameter (Text) is the string we are going to update using the other parameters V1-V5, and we store that in NewText (also $ 500) that finally will be returned.
- The function expects to find the string {1} in the Text parameter and replace it with the value in the parameter V1. Same with {2} and so on.
- If none of the placeholders {1} to {5} are found in parameter Text it will just be returned unchanged.
The replacement is done with the SAS function transtrn (translate string), but transtrn can only do one translation at a time, so you could say the FormatString is transtrn(transtrn(transtrn(transtrn(transtrn( .
Proc fcmp outlib=WORK.funcs.dev;
/* The placeholders in FmtStrNum are identified using numbered indexes e.g. {1}, {2} etc. */
Function FmtStrNum(Text $, V1 $, V2 $, V3 $, V4 $, V5 $) $ 500;
array Values [5] V1 V2 V3 V4 V5;
array PlaceHolders [5] $ 3 '{1}' '{2}' '{3}' '{4}' '{5}';
length NewText $ 500;
NewText = text;
do i = 1 to dim(Values);
if Values[i] > '' then
NewText = transtrn(NewText, PlaceHolders[i], strip(Values[i]));
end;
Return(NewText);
Endsub;
Run;
Now try it yourself with this small program.
options cmplib=WORK.funcs;
data favorite;
length Band $ 300;
keep Band;
b1='Aristocrats';
b2='MÖRGlbl';
b3='Cosmosquad';
b4='Neal Morse';
b5='Dream Theater';
/* With FmtStrNum */
Band = FmtStrNum('Of all the great Prog Rock bands {1}, {2}, {3}, {4} and {5}, my current favorite is {2}.', b1, b2, b3, b4, b5);
output;
/* The Traditional way */
Band = 'Of all the great Prog Rock bands ' !! strip(b1) !! ', ' !! strip(b2) !! ', ' !! strip(b3) !! ', ' !! strip(b4) !! ' and ' !!
strip(b5) !! ', my current favorite is ' !! strip(b2) !! '.';
output;
run;
The content of the variable Band should now be:
Of all the great Prog Rock bands Aristocrats, MÖRGlbl, Cosmosquad, Neal Morse and Dream Theater, my current favorite is MÖRGlbl.
Example 2 – Named placeholders
The next example use named placeholders which works better when we’re going to build a URL string; the FmtStrName function.
We just need to do small changes to the FCMP code from FmtStrNum:
- Insert the PlaceHolder parameters PH1 to PH5 in the function statement, giving us value-pairs PH1/V1 PH2/V2 et cetera.
- Change the PlaceHolder array statement to use PH1-PH5 instead of the literals we had in FmtStrNum, and remove $ 3.
This is the FCMP code to create the FmtStrName function.
Proc fcmp outlib=WORK.funcs.dev;
/* The placeholders in FmtStrName are identified using named indexes e.g. {band} or {guitar}. */
Function FmtStrName(Text $, PH1 $, V1 $, PH2 $, V2 $, PH3 $, V3 $, PH4 $, V4 $, PH5 $, V5 $) $ 500;
array Values [5] v1 v2 v3 v4 v5;
array PlaceHolders [5] ph1 ph2 ph3 ph4 ph5;
length NewText $ 500;
NewText = Text;
do i = 1 to dim(Values);
if Values[i] > '' then
NewText = transtrn(NewText, PlaceHolders[i], strip(Values[i]));
end;
Return(NewText);
Endsub;
Run;
Try it yourself with this program building a typical URL string we often see in Stored Processes or Viya Job Execution code.
options cmplib=WORK.funcs;
data url;
Length url $ 300;
keep url;
Artist = 'Guthrie Govan';
Band = 'The Aristocrats';
Guitar = '/images/Charvel Cooked Ash.png';
Rig = 666;
/* With FmtStrName */
url = FmtStrName('<a href="javascript: f.Rig.value=''{rig}''; f.submit();"><img title="Full rig rundown - {artist} with {band}." src="{guitar}"></a>',
'{rig}', put(Rig, best.),
'{artist}', Artist,
'{band}', Band,
'{guitar}', Guitar,
'', '');
output;
/* The Traditional way */
url = '<a href="javascript: f.Rig.value=' !! "'" !! strip(put(Rig, best.)) !! "'" !!
'; f.submit();"><img title="Full rig rundown - ' !! Strip(Artist) !! ' with ' !!
trim(Band) !! '." src="' !! Guitar !! '"></a>';
output;
run;
The content of the variable Url should now be:
<a href="javascript: f.Rig.value='666'; f.submit();"><img title="Full rig rundown - Guthrie Govan with The Aristocrats." src="/images/Charvel Cooked Ash.png"></a>
As you perhaps have noticed above, the last value pair is blank. This is due to that I have set up the FmtStrName to take 5 value pairs, and as Proc FCMP does not support variable arguments I have to pass on ‘ ’,’ ’ as the last pair. The same goes for the first example FmtStrNum; if I didn’t need the fifth parameter I would just leave ‘ ’ as last value. An alternative to this is to copy the FmtStrNum FCMP code and create an FmtStrNum2 taking just 2 parameters, an FmtStrNum3 taking 3 parameters and so on. This is how I have set it up for myself.
These both examples return a string 500 characters long, and you probably need to adjust this according to your needs. Setting local strings to 32k has a performance cost to it, so be sensible. For more details on FCMP and performance, read the blog post Removing repeated characters in SAS strings by Leonid Batkhan.
If you have any questions or suggestions please reply to this post or reach out via email.
