PRXMATCH, Regular expression id not changing

[lindseyj79]

Hi,

Base SAS 9.4 user; 

I'm a first time user of prxmatch but I'm having issues getting it to work in my datastep.

I have 2 variables set up in a dataset and I want to cross reference all the words in one variable with all the words in the other variable to see if there were any matches.  Here is an example of my code:

data xxx ;
   x="Drug: Pimavanserin tartrate (ACP-103)" ;
   y="Drug: pimavanserin tartrate" ;
    output ;
   x="Drug: masitinib" ;
   y="Drug: masitinib (AB1010)" ;
    output ;
run ;

What I want now it to see if any of the words in y appear in x, I decided to see if PRXMATCH would work, so I've created a pattern (which hopefully contains all the pearl stuff and the words to match 'm/word1|word2|word3/io')

data yyy ;
   set xxx ;

/*create pattern to search for*/
   pattern='m/'||lowcase(translate(compress(strip(scan(y,2,":")),,"adsk"),"|"," "))||'/io' ;
   parse=prxparse(pattern) ;

/*this is my checking criteria... so eventually if z>0 then there is a match with >= 1 of the keywords*/
   z=prxmatch(prxparse(pattern),x) ;
run ;

What I'm finding is that because the regular expression id (from the prxparse) isn't changing it will only find the first match and then nothing else.

Can anyone help?

thanks

Lindsey

[HB]

What would be the values of Z you would want for the following data?

x	y	z
pimavanserin tartrate (ACP-103)	pimavanserin tartrate
masitinib	masitinib (AB1010)
aspirin tablets 100mg	aspirin
calcium tartrate	calcium
calcium tartrate	tartrate
aluminum tartrate	tartrate
doxycycline	lexapro
hydrochlorothiazide oxide	oxide
Ibuprofen tablets 100mg red	tab
one two three four	two three

[lindseyj79]

I know this is very imperfect but a colleague is finding the minimum matches and I'm finding the maximum so we know the true answer will lie somewhere in between

x	y	z
pimavanserin tartrate (ACP-103)	pimavanserin tartrate	1
masitinib	masitinib (AB1010)	1
aspirin tablets 100mg	aspirin	1
calcium tartrate	calcium	1
calcium tartrate	tartrate	1
aluminum tartrate	tartrate	1
doxycycline	lexapro	0
hydrochlorothiazide oxide	oxide	1
Ibuprofen tablets 100mg red	tab	0
one two three four	two three	1

[s_lassen · Accepted solution]

The problem is probably that you use the "o" option on your expression, meaning that it will compile only once in the datastep. Try changing "/io" in the end of the expression to just "/i".

[lindseyj79]

BINGO!  thank you so much!