prxparse using a column values

[Rohit_1990]

Hi Experts,

I am trying to use prxparse to match a pattern

data c;

pattern=prxparse('/abc|bcd|esd/');

position =prxmatch(pattern,' abc fgg rrt);

run;

 now i have another table with column c1 as follows:

C1

abc

bcd

esd

so my question is here can we use column values directly into prxparse to define pattern

rather then manually typing each value.

My other concern is how to remove consecutive duplicate string in 

 lets say I have values as follows:

column1

steve steve is walking is walking

steve john steve john is walking

steve john steve john steve john

So i want mu output as

column1

Steve is walking

Steve John is walking

steve John

Please help me on this.

In case of more clarifications required please revert.

Regards,

Rohit

[PGStats · Accepted solution]

First question: build the pattern by concatenating words (hint: use string = catx("|", string, word) )

Second question:

data have;
input str :&$64.;
datalines;
steve steve is walking is walking
steve john steve john is walking
steve john steve john steve john
steve is is john steve
;

data want;
set have;
length oldStr newStr $64;
oldStr = str;
do i = 1 to 100;
    newStr = prxchange("s/(\b\w+\b)(.*)(\1\s*)/\1\2/io", -1, oldStr);
    if newStr = oldStr then leave;
    oldStr = newStr;
    end;
drop i oldStr;
run;

proc print; run;

[tomrvincent]

use the scan function to go word by word, skipping duplicates.