topic Re: regular expressions novice in SAS Programming

regular expressions novice

frisco — Wed, 29 Jun 2011 14:20:04 GMT

Hi, I need to parse a comment field that has TWO sets of quoted phrases - the kicker is that I only need the contents of the SECOND quoted phrase.

For example:

Blah blah Blah "Dont need this" yada yada yada "Definately need this" so forth and so on...

Is this possible using Perl regular expressions?

TIA.

regular expressions novice

polingjw — Wed, 29 Jun 2011 14:42:05 GMT

Here is an easy solution that does not use regular expressions.

data test;

comment = 'Blah blah Blah "Dont need this" yada yada yada "Definately need this" so forth and so on...';

quote=scan(comment, -2, '"');

run;

Re: regular expressions novice

DLing — Wed, 29 Jun 2011 16:17:18 GMT

Here's a code snippet that should work, typo notwithstanding. Modify to suite your needs.

data _null_;

if _n_ = 1 then do;

regid = prxparse( '/^.*"(.*)".*"(.*)"/' );

retain regid;

if regid = 0 then do;

put 'improper reg expression';

stop;

end;

length str s1 s2 $ 1000; /* or whatever length */

input str;

s1 = prxposn( regid, 1, str ); /* quick & short */

s2 = prxposn( regid, 2, str ); /* believed to do two independent scans of the str */

/* alternative form - believed to be more efficient as prxmatch is called only once */

if prxmatch( regid, str ) then do; /* if patterns matches then... */

call prxposn( regid, 1, pos, len );

s1 = substr( str, pos, len );

call prxposn( regid, 2, pos, len );

s2 = substr( str, pos, len );

end;

cards;

Blah blah Blah "Dont need this" yada yada yada "Definately need this" so forth and so on...

;;;;

The regular expression '/^.*"(.*)".*"(.*)"/' deconstructs as follows:

/ is the beginning of the regular expression. standard convention.

^ is the beginning of the string

.* matches any character 0 or more times (indeterminate)

" matches this character - beginning of 1st quoted string

(.*) is the 1st capture buffer, marked by (). the content is between the (), in this case .* says everything.

" finishes the 1st quotation

.* says anything can separate the two quoted strings

"(.*)" is the capture buffer for the second quoted string.

/ closes off this regular expression

Hope this helps. Enjoy.

Re: regular expressions novice

Ksharp — Thu, 30 Jun 2011 08:16:02 GMT

Agree with polingjw . Since you want the second part of quote.

data test;

comment = 'Blah blah Blah "Dont need this" yada yada yada "Definately need this" so forth and so on...';

quote=scan(comment, 4 , '"','m');

run;

Ksharp

Message was edited by: xia keshan

Re: regular expressions novice

manojinpec — Thu, 30 Jun 2011 10:22:11 GMT

Thanks Everybody.. Even i din't know about the modifier in scan function.