topic Re: Perl regular expressions in SAS Programming

Perl regular expressions

RW9 — Wed, 30 Jul 2014 14:46:15 GMT

Hi,

Am struggling to get something basic working and I can't see why. Given the string:

{\rtf1\ansi\ansicpg1252\uc1\deff0\deflang1033\deflangfe1033

Why does the following code not give position at all (raw_base contains the above text):

data want;

set have;

tmp=prxparse("/rtf/");

tmp2=prxposn(tmp,1,raw_base);

run;

Had also tried ("/\\rtf/") and ("/ansi/"), none seem to work?

Re: Perl regular expressions

Ksharp — Wed, 30 Jul 2014 15:29:44 GMT

You missed prxmatch() . Check the documentation to find an example in it .

Re: Perl regular expressions

jakarman — Wed, 30 Jul 2014 15:49:12 GMT

As ksharp already made the note, prxmatch must have been executed before prxposn.

The prxparse call is only needed once it seems to be a compilation where after you can use that.

prxposn is using grouping results, fort that the ( ) usage. See:

SAS(R) 9.4 Functions and CALL Routines: Reference, Second Edition (prx) and SAS(R) 9.4 Functions and CALL Routines: Reference, Second Edition (prxposn function not the call)

try:

48 data want;

49 set have;

50 if ( _n_ =1 ) then do; tmp=prxparse('/(rtf)/'); end;

51 tmp1=prxmatch(tmp,raw_base);

52 tmp2=prxposn(tmp,1,raw_base);

53 put "output: " tmp1 tmp2 $8. ;

54 run;

output: 2 rtf

Re: Perl regular expressions

RW9 — Thu, 31 Jul 2014 08:48:25 GMT

Thanks Jaap, I couldn't get your example working however. I am using version 9.3 (not sure if that makes a difference), I copied and pasted and had the first set of errors regarding tmp not being on each row. So I modified per:

data want;

if (_n_=1) then do;tmp=prxparse('/rtf/');end;

set raw_base;

retain tmp;

if prxmatch(tmp,raw_base) then do;

tmp1=prxmatch(tmp,raw_base);

tmp2=prxposn(tmp,1,raw_base);

end;

run;

With the above, tmp1 is correct, only the first record has tmp1 not null and it is 3, where my string starts. However tmp2 is always empty.

Re: Perl regular expressions

Ksharp — Thu, 31 Jul 2014 12:39:43 GMT

Try this one :

data raw_base;

raw_base='{\rtf1\ansi\ansicpg1252\uc1\deff0\deflang1033\deflangfe1033';

run;

data want;

set raw_base;

if _n_=1 then do;tmp=prxparse('/(rtf)/');end;

if prxmatch(tmp,raw_base) then do;

tmp2=prxposn(tmp,1,raw_base);

end;

run;

Xia Keshan

Re: Perl regular expressions

haikuobian — Thu, 31 Jul 2014 12:43:22 GMT

prxposn() is to capture buffer which has to be defined using parenthesis (), try this:

data want;

if (_n_=1) then do;tmp=prxparse('/(rtf)/');end;

var='{\rtf1\ansi\ansicpg1252\uc1\deff0\deflang1033\deflangfe1033';

retain tmp;

if prxmatch(tmp,var) then do;

tmp1=prxmatch(tmp,var);

tmp2=prxposn(tmp,1,var);

end;

run;

Haikuo

Re: Perl regular expressions

gergely_batho — Thu, 31 Jul 2014 13:09:47 GMT

Jaap's code is correct. Of course you need the retain statement.

On the other hand you forgot the parenthesis: prxparse('/rtf/') prxparse('/(rtf)/')

Without the parenthesis there is no capture buffer, prxposn() returns the content of the capure buffer(s).

But I don't think you need it, since you already know what will be the the capture buffer: rtf (or missing).

So I suppose this is just an example. Your search string is more complex in your real problem.

Probably: prxparse('/\\(rtf.*?)\\/')

This will match this string: \rtf1\

Position will be: 2

Capture buffer content will be: rtf1

Re: Perl regular expressions

RW9 — Thu, 31 Jul 2014 13:14:05 GMT

Thank you HaiKuo and Gergely Batho, unfortunately I have run out of helpful answers, but your input was very helpful.

Re: Perl regular expressions

jakarman — Thu, 31 Jul 2014 13:44:39 GMT

Yes, I forgot the retain statement for running all observations. thx rw9/ batho..

Telling it is compilation reference and not taken action .... (sorry).
It was tested using UE (9.4) but the same approach as in 9.1.3. The grouping ( ) got my attention in the prx part.

I did mention the difference the difference between prxposn as function and as call.

The first one is returning the string the latter the position (number).