Recoding string variable - leaving capital letters only

[Konrad Zdeb]

Hi All,

I have a string variable with a few words per observation. I would like to generate an acronym variable contained only capital letters from the string variable, as in the example below:

Observation	StringVariable	AcronymVariable
OBS001	Some Phrase as a String	SPS

Needles to say, as always I will be grateful for any help.

[Robert Picard]

I can't think of a solution that does not require looping but here's a way to do it using regular expressions

clear
input str50 s
"Some Phrase as a String"
"Some Other String"
end

gen abbr = s
local more 1
while `more' {
    replace abbr = regexr(abbr,"[^A-Z]+","")
    count if regexm(abbr,"[^A-Z]+")
    local more = r(N)
}

[Joe Canner]

Konrad,

There may be a more elegant and efficient way to do this, but this brute force method is the first thing that came to mind:

gen Acronym=""
forvalues i=1/20 {
  replace Acronym=Acronym+substr(StringVariable,`i',1) if inrange(substr(StringVariable,`i',1),"A","Z")
}

Change the 20 to the length of the string. If you need this to be more general, there is a way to compute this automatically.

Regards,
Joe

[Konrad Zdeb]

Robert,

Thank you very much for the prompt reply, it worked perfectly.

Change the 20 to the length of the string. If you need this to be more general, there is a way to compute this automatically.

Thanks very much. Strings are of variable lengths. My initial thinking is that I could get string length and pass it as local macro instead 20 but then I would end up writing a loop, as I don't know alternative solution.

[Nick Cox]

Variable string lengths won't bite. It's just the longest string value you need to measure (i.e. count characters).

[Robert Picard]

Here's another technique that does not use regular expressions.

local chars `c(alpha)' 0 1 2 3 4 5 6 7 8 9
gen acronym = subinstr(s," ", "", .)
foreach c in `chars' {
    replace acronym = subinstr(acronym, "`c'", "", .)
}