Get the character from mix of symbols using COMPRESS, INDEX, LEFT, TRIM?

[dupp99]

I have a data set named names with only one variable x.

x

* '''[[Nick]]'''

* [[KK Circle]]

* [[Spenser]]

||-

* Cool

||

Name

I want to create a new dataset with two variables, the first being the whole old line, and the second variable with only the names.

And each like should be only the line with character in it.

Like:

x                              newX

* '''[[Nick]]'''             Nick

* [[KK Circle]]           KK Circle

* [[Spenser]]             Spenser

* Cool                      Cool

Name                      Name

I tried

newX=compress(x ,";{|-*![=],");

but I couldn't get rid of the "" on the line Nick.

Might not need to use all of COMPRESS, INDEX, LEFT, TRIM, but these might be useful. 

Thank you.

[ballardw · Accepted solution]

The example data you show for

* '''[[Nick]]''' 

is 3 single quotes.

which would work with

data junk;
x="* '''[[Nick]]'''";
newX=compress(x ,";{|-*![=],'");
run;

If you need to find one of the quote character you can double in the search string such as

newX=compress(x ,";{|-*![=]"",'");

I placed the double quote to use in the search away from the quotes used to delimit the string so you can read it a bit more clearly. Note that in the editor a single " in the middle would show as an unbalance quote by having one or more characters appear differently than the quoted string. With the doubled character the entire string  appears and is treated as a single string.

[novinosrin]

data have;
input  x & $20.;
cards;
* '''[[Nick]]'''
* [[KK Circle]]
* [[Spenser]]
||-
* Cool
||
Name
;

data want;
set have;
want=compress(x,' ','ka');
if want>' ';
run;

[data_null__]

Use compress and tell it which class of characters to keep.

data x;
   input x $32.;
   xnew = compress(x,,'KADS');
   cards;
x
* '''[[Nick]]'''
* [[KK Circle]]
* [[Spenser]]
||-
* Cool
* Cool 9
||
Name
;;;;
   run;