topic Re: extracting a text string from a string in SAS Programming

extracting a text string from a string

Bluebonnet16 — Tue, 16 Oct 2018 21:25:33 GMT

Hi, I'm trying to extract a text string (name) from a longer string, which may occur either once within the string or multiple times (several names). I'm breaking the longer string into groups and using prxparse and prxmatch with grouping to extract only the group I need, but something is not working. Below is an example of my text string, my code and what I'm looking for at the end.

data have;

Prov_Info ="ProviderName: Spine and Pain Center of Whatchamikola; IDN: 2345678901; IsGroup: No;, ProviderName: Happy Toes; IDN: 3456789012; IsGroup: No;, ProviderName: IDN: 3456789012; IsGroup: Yes;, ProviderName: Bright Smiles of AZ; IDN: 1234567890 IsGroup: Yes;, ";

patternID = prxparse('/^(ProviderName:)( |.*; )(NPI: )/');

if prxmatch(patternID, strip(Prov_Info)) then do;

newname=prxposn(patternID, 2, Prov_Info);

end;

run;

Result needed:

newname=Spine and Pain Center of Whatchamikola; Happy Toes; Bright Smiles of AZ;

Any suggestions would be attreciated.

Thank you!

Re: extracting a text string from a string

kiranv_ — Wed, 17 Oct 2018 00:19:24 GMT

you may need to use prxnext. prxparse('/(ProviderName:\s+[a-zA-z ]+?;)/') indicates providers name folowed by space and words and till the ; prxnext capture position and length wherever you have this pattern. by doing substr(prov_info, position+13, length-13, we can remove ProviderName:

data have; 
Prov_Info ="ProviderName: Spine and Pain Center of Whatchamikola; 
IDN: 2345678901; IsGroup: No;,
ProviderName: Happy Toes; IDN: 3456789012; IsGroup: No;,
 ProviderName:  IDN: 3456789012; IsGroup: Yes;,
 ProviderName: Bright Smiles of AZ; IDN: 1234567890 IsGroup: Yes;, "; 
  run;
  
 


data want;
length val patternid $200.;
set have;
 start = 1;
   stop = length(prov_info);
 
   re = prxparse('/(ProviderName:\s+[a-zA-z ]+?;)/');
   set have;
   call prxnext(re, start, stop, trim(prov_info), position, length);
      do while (position > 0);
         val = substr(prov_info, position+13, length-13);
          patternID = catx(" ",  patternid, val);
         call prxnext(re, start, stop, trim(prov_info), position, length);
      end;
drop re start stop position length val;
run;

proc print data=want;
run;

Re: extracting a text string from a string

Bluebonnet16 — Wed, 17 Oct 2018 13:29:19 GMT

That works! Thank you so very much, kiranv_!!!!

Re: extracting a text string from a string

Bluebonnet16 — Wed, 17 Oct 2018 13:59:17 GMT

Hi,

The code works and does exactly what I need. However, I just realized, that if there is a comma in the provider name, a dot, or brackets, then it doesn't. 😞

Some of the examples, on which the code didn't work:

data have2;

Prov_Info ="ProviderName: Shapa Inc.;

IDN: 2345678901; IsGroup: No;, ProviderName: Smith, John; IDN: 3456789012; IsGroup: No;,

ProviderName: IDN: 3456789012; IsGroup: Yes;, ProviderName: Pharmacy TX-115; IDN: 1234567890 IsGroup: Yes;, ";

ProviderName: Star Med EMS (Medicare); IDN: 1234567890 IsGroup: Yes;, ";

ProviderName: CVS Pharmacy 50; IDN: 1234567890 IsGroup: Yes;, ";

ProviderName: CASE MANAGEMENT & IDN: 1234567890 IsGroup: Yes;, ";

run;

Please let me know if there is a solution to account for those special characters in the provider name? Thank you!

Re: extracting a text string from a string

kiranv_ — Wed, 17 Oct 2018 14:36:12 GMT

I guess ProviderName: CASE MANAGEMENT & IDN: 1234567890 IsGroup: Yes;, ";

should be like ProviderName: CASE MANAGEMENT & something; IDN: 1234567890 IsGroup: Yes;, ";

Re: extracting a text string from a string

Bluebonnet16 — Wed, 17 Oct 2018 14:43:17 GMT

Yes, very unusually spelled provider name: ProviderName: MEDICAL CASE MANAGEMENT & SOCIAL SERVICES; IDN: 1234567890; IsGroup: Yes;,

Thank you!

Re: extracting a text string from a string

kiranv_ — Wed, 17 Oct 2018 14:46:48 GMT

one last question. Do you have numbers in providers name

ProviderName: CVS Pharmacy 50

if it does not then you can use.

 re = prxparse('/(ProviderName:\s+\D+?;)/');

Re: extracting a text string from a string

Bluebonnet16 — Wed, 17 Oct 2018 14:49:17 GMT

Yes, unfortunately some provider names, like pharmacies for example, do contain digits in the name

Re: extracting a text string from a string

kiranv_ — Wed, 17 Oct 2018 14:53:56 GMT

this should work. i have used ?!IDN: ?! is negative lookahead, it will take everything till ; in providername unless IDN comes after providername and I guess this happens when you have no providername

re = prxparse('/(ProviderName:\s+(?!IDN:).+?;)/');

Re: extracting a text string from a string

Bluebonnet16 — Wed, 17 Oct 2018 14:58:17 GMT

Yeap, it worked!! No missing values for the patternID 🙂

Thank you So SO SO very much!!!!

Re: extracting a text string from a string

kiranv_ — Wed, 17 Oct 2018 14:59:42 GMT

You are welcome and I am glad it worked