topic Re: How do I include all scenarios when writing Regular Expressions functions (prxchange) in SAS Programming

How do I include all scenarios when writing Regular Expressions functions (prxchange)

ciphercong — Wed, 10 Oct 2018 20:26:03 GMT

Hi guys,

I am really not good at using Perl Perl's regular expressions functions in data step... Below are several different expressions I wrote for the Prxchange function, and this is for searching for patterns and remove the <E T="XXX"><\E> tag but keep the content within this tag. I wonder what exactly is the difference between each expression and if any one can advise me on which one is the best (or even better -- propose a better one) that would be really awesome.

(XXX can be any integer)

data _null_;
A='<P><E T="12">Test </E>(a) <E T="03">Authority.</E> This part is issued pursuant to 12 U.S.C. 1 <E T="99">et seq.,</E> 12 U.S.C. 24 (Seventh), and 12 U.S.C. 93a.<E T="5">test</E></P>';
B=prxchange('s/(.+?)(<E.*?>(.+?)<\/E>)(.+?)/\1\3\4/',-1,A);
C=prxchange('s/(.+?)(<E T=.*?>(.+?)<\/E>)(.+?)/\1\3\4/',-1,A);
D=prxchange('s/(<E T=.*?>(.+?)<\/E>)(.+?)/\2\3/',-1,A);
put A=/B=C=D=;
run;

All of them generated desired results, but - none of them is completely satisfactory to me when I apply the function to the whole file( I have a large file to deal with, and this <E T="XXX"><\E> pattern can appear any where.

Thanks in advance!

Re: How do I include all scenarios when writing Regular Expressions functions (prxchange)

PGStats — Wed, 10 Oct 2018 21:23:17 GMT

Only specify the searched and replacement parts, the rest of the string will remain untouched

E=prxchange('s/<E T="\d+">(.*?)<\/E>/\1/', -1, A);

Re: How do I include all scenarios when writing Regular Expressions functions (prxchange)

ChrisNZ — Wed, 10 Oct 2018 22:22:51 GMT

All these expressions do what you want and give the same result for your string.

The difference between B and C is that B capture any <E tag while C also needs T= inside.

D is the same as C except that C needs something before <E, so won't capture <E if it's at the start of the string.

> propose a better one

What is 'better"?

What's wrong with these?

If you want restrict the capture to <E T="{digits"}> you could use this:

E=prxchange('s/(<E T="\d+">(.+?)<\/E>)(.+?)/\2\3/',-1,A);

[After reading @PGStats's reply: he is right about replacement of course, and his expression is a bit simpler.]

Re: How do I include all scenarios when writing Regular Expressions functions (prxchange)

Ksharp — Thu, 11 Oct 2018 13:03:47 GMT

This is what you are looking for ?

data _null_;
A='<P><E T="12">Test </E>(a) <E T="03">Authority.</E> This part is issued pursuant to 12 U.S.C. 1 <E T="99">et seq.,</E> 12 U.S.C. 24 (Seventh), and 12 U.S.C. 93a.<E T="5">test</E></P>';
B=prxchange('s/<[^<>]+>/ /',-1,A);
put A=/B=;
run;