How to select a date from a string conditional on a part of that string

[ilikesas]

Hi,

suppose I have the following table:

Company
ABC (merged 1/10/2012)
DEF (bankrupt 2/5/2009)
ZZZ

If the company merged, I would like to get a new column for the date when it happened, and nothing for the others, in the following way:

Company	Merged
ABC (merged 1/10/2012)	1/10/2012
DEF (bankrupt 2/5/2009)
ZZZ

Thank you

[Reeza · Accepted solution]

You're previous questions should allow you to isolate which variables have the word merged.

Adding on to that, you can use the scan function to get the date, and in conjunction with the input function convert it to a valid SAS date which is highly recommended. I'm assuming the format of the dates is mmddyy though it's unclear based on your date format.

data have;

length var $30.;

var="ABC (merged 1/10/2012)";output;

var="DEF (bankrupt 2/5/2009)";output;

var="ZZZ";output;

run;

data want;

set have;

date=scan(var, -1, "() ");

sas_date=input(scan(var, -1, "() "), mmddyy10.);

format sas_date date9.;

run;

[ilikesas]

Hi Reeza,

thank you for replying and for reading my previous question!!!

I would like to ask a question, suppose I had ABC ( merged in 1/10/2012 because of reasons), when applying the scan function the "date" that I will get is "reasons" because its the last word, so is there a way of getting the date regardless of its position in the string?

Thank you 

[Reeza]

Any other variations? Will there be other numbers or slashes in the text?

You can look into the index/find/substr functions.

[OleSteen]

I would go for Perl regular expression functions: E.g. PRXCHANGE, PRXMATCH etc.