topic Re: Finding words wrapped with single/double quotes in SAS Programming

Finding words wrapped with single/double quotes

gandikk — Tue, 19 Jul 2022 15:31:29 GMT

Hi,

I need to find the records that have words wrapped with single/double quotes. For example, "Tech Data" Company, "WOW" Company etc. But my code shouldn't count the record like this - John's & Gary's world.

Thanks in advance!!

Re: Finding words wrapped with single/double quotes

ballardw — Tue, 19 Jul 2022 16:24:45 GMT

Think very carefully and provide a rule that describes exactly why the second record is not counted. An example is not a rule that can be programmed.

Re: Finding words wrapped with single/double quotes

gandikk — Tue, 19 Jul 2022 16:42:01 GMT

Thanks for your reply!! I didn't get the rule from customer. I got examples only. I believe that the customer doesn't want to consider a single quote if it is an apostrophe. So, in the 2nd example, though there are 2 single quotes, both are apostrophes. My rule is that if a single with 's' as an immediate character, we shouldn't count it.

Re: Finding words wrapped with single/double quotes

ballardw — Tue, 19 Jul 2022 21:46:20 GMT

Problem with examples.

What if you have measurements in feet and inches? 5' 6"

Angular measurements, such as latitude and longitude where the ' and " can be used for minutes and seconds (fractions of a degree)?

Contractions: Don't

You might be considering exclusion when there are any 2 characters on either side of the ' (more common) or ". But the hanging possessives like Jones' are a headache.

There might be some slick approaches with Regular expressions but I'm frankly too lazy at the moment to tackle a potentially very obnoxious set of search and refine search codes for such.

I spent some time a long time ago dealing with cleaning up text with measurements such as 2" by 36" in the middle of other text and remember it was a headache to differentiate from the other quoted text.

Re: Finding words wrapped with single/double quotes

BrunoMueller — Wed, 20 Jul 2022 09:11:29 GMT

I guess this is a case for using regular expressions. Have a look at the code below. I am not regular expression expert and there might be easier ways of doing this.

data have;
  infile cards truncover;
  input
    text $char256.
  ;

  re1 = prxparse('/([\s"|"\A][^"]*"\s)/');
  qt1 = prxmatch(re1, text);
  re2 = prxparse("/([\s'|'\A][^']*'\s)/");
  qt2 = prxmatch(re2, text);

  if qt1 then do;
    value = prxposn(re1, 1, text);
  end;

  if qt2 then do;
    value = prxposn(re2, 1, text);
  end;

  value2 = dequote(value);
  cards;
"Tech Data1"
'Tech Data2' Company
abc "Tech Data" Company
def 'WOW' Company
John's & Gary's world
abc
123 "O'Reilly" here
456 'O"Reilly' here
;