Problem with accented letters in the prx matching functions

[DonH]

I have PRX functions that I am using to validate strings. Here are a couple of examples:

/^[A-ZÑ&]{3}\d{6}[A-Z0-9]{3}$/

/^[A-ZÑ‘&]{3}\d{6}[A-Z0-9]{3}$/

/^[A-ZÃÑ‘&]{3}\d{6}[A-Z0-9]{3}$/

The difference is sometimes only à is allowed in addtion to the upper case letters, sometimes only Ñ, and sometimes both ÃÑ. I can easily handle those variables. It appears the prxmatch function is counting the accented characters as two characters. For example, the following string"

LÑL17010ZZZ

fails (i.e., returns a 0) when using prxmatch. However if I change the {3} to {3,4}, it returns a 1. So my speculation is that it sees LÑL as four characters. I have run other tests and combinations that seem to confirm this speculation.

So my question is whether there is a way to specify the à and the Ñ characters in the pattern so they are treated as a single character.

And note that it is only these two accented letters that are allowed.

TIA

[DonH]

I meant to add that I did try specifying these two accented characters in the patter in hex, e.g.,:

/^[\xD1\xC3A-Z&]{3}\d{6}[A-Z0-9]{3}$/

that also failed.

[Shmuel]

If ignoring accented letters is possible than see next solution:

https://communities.sas.com/t5/Base-SAS-Programming/How-to-ignore-accented-text/td-p/140883

[Patrick · Accepted solution]

So far the SAS implementation for Regular Expressions (the prx... functions) only support single byte character sets. 

Here the list which string function supports what (SBCS, DBCS, MBCS):

http://support.sas.com/documentation/cdl/en/nlsref/69741/HTML/default/viewer.htm#p1pca7vwjjwucin178l...

[DonH]

Thanks Patrick, that is what I was afraid the answer would be.

And to answer Schmuel's question, ignoring them in not an option.

It is only these characters, and accepting these characters is a key requirement for a large application.

So I am going to look into workaround for this (have a few ideas).

[Shmuel]

One more try:

I'm not famillar with prxmatch, but can you use hexadecimal in the expression ?

If YES then 

1) use tranwrd to replace the accented letter into a non printable hexa (like 'FA'x)

     to create a temporary variable

2) use prxmath with the hex expression to validate the temporary variable

[DonH]

Thanks Shmuel,

That is exactly the approach that I was considering. This is part of a large application and we have parameterized most of the checks, including the PRX checks.

At the point in the process where the prxmatch function is used, I have a data set that has the values to be validated along with the pattern. So I will add logic to detect if there are accented A or accented N characters in the string to be validated. If so, I will add code to convert them to lower case a and N in both the data value and the pattern - by using temp variables. The lower case letters work for this since another part of the requirements is that only upper case characters are allows for any text/string variable.

So thanks for suggesting this approach as it gives me a bit more confidence that I am going down the right path.