Extract number adjacent to String

[#1]

Hi, can anybody help me to extract numbers adjacent to strings?

I have column1 shown below, and expect to extract numbers adjacent to "HFA" (column2).

I tried the following JSL, but could not get column2 that I expect

Num( Regex( :Column 1, "(\d\d\d)HFA | HFA(\d\d\d)", "\1" ) )

column1	column2
400HFA isa300	400
isa300 HFA400	400

Thank you for considering the question.

[#2 · Accepted solution]

Given the small example set, the following formula works:

Num( Regex( :column1, "(\d*)HFA(\d*)", "\1\2" ) )

This solution will run into problems if you have instances like "400HFA300".  It will return 400300.

[#3]

Hi Ngambels,

Thank you very much for your help. It resolved the issue.

I did try 

"(\d*)HFA(\d*)"

but apparently I missed (I did not know I can and must use both \1 and \2)

"\1\2"

I learned new things from you and through this discussion.

Once again, thank you very much!

[#4]

If for some reason there is a need to tackle the issue @ngambles mentioned, one fairly simple way would be to use Regex Match instead of Regex

Names Default To Here(1);
a = "400HFA isa300";
b = "isa300 HFA400";
c = "400HFA300";
d = "400HF300";

rgx_pattern = "(\d*)(?:HFA)(\d*)";
show(Regex Match(a, rgx_pattern));
show(Regex Match(b, rgx_pattern));
show(Regex Match(c, rgx_pattern));
show(Regex Match(d, rgx_pattern));

Regex Match will though require some extra checks as it returns a list of groups captured.

[#5]

Thanks @AprioriFish3134  for pointing out the original issue (no \2), and @ngambles  for a good solution.

If you need even more levels of nested parens, or the sequential parens in your example, just count the open-parens from left-to-right to discover the back reference number. The back reference for unused alternatives is an empty string.

Regex( "abac", "(((a)|(b)|(c))(c))+", "1=\1 2=\2 3=\3 4=\4 5=\5 6=\6" )

"1=ac 2=a 3=a 4= 5= 6=c"

One thing that might not be obvious: the back reference can also be used in the pattern. This could be useful for making sure a leading and trailing quotation mark or apostrophe are used consistently:

Write( Regex( 
	"before 'a \!"b\!" c' after", // before 'a "b" c' after
	"(\!"|')(.*?)\1", // ("|')(.*?)\1
	"\2" 
) )

a "b" c

(The \!" escape puts a quotation mark inside the strings; the comment shows the text of the string.) In this example, \1 group can match either " or ' and \2 matches all of the text up to another occurrence of \1.

Use some comments if you write regex like that! The next maintainer will appreciate it.

[#6]

Good place to practice and test regex is https://regex101.com/ .

Those comments with JMP escape characters removed were very helpful, I have to start adding those when I write more complicated regex patterns (Regex issue with escaped characters )

[#7]

HI @jthi 

Thank you very much for your alternative solution to the issue, and also for the link on regex.

[#8]

Hi @Craige_Hales ,

Thank you very much for enriching the discussion.

In fact I first learned about JMP's regex from your old article !
https://community.jmp.com/t5/Uncharted/Regex/ba-p/21008