How to access and manipulate strings in Stata variable labels?

Question

I have a Stata dataset with variable labels describing the year of measurement. I need to access the year of measurement from the variables to later rename each variable using a suffix showing the year. For example V95 has a label GNP/CAPITA,75, and I want to rename it to gnp_capita_75. Some variables have labels like GNP/C:GROWTH RTS,60-75, and I want to add the midpoint of the interval after the comma to the variable name.

So far my code for accessing the variable labels looks like this:

local varlist V1 V2 V3 V28 V29 V30 V85 V86 V87 V88 V89 V90 V91 V92 V93 V94
foreach variable in `varlist' {
 //get the label of the variable
 local label : var label `variable'
 //find the position of the comma
 local commapos : strpos(`label', ",")
 //find the stub before the comma
 local namestub : substr(`label', 1, `commapos' - 1)
 //find year after the comma
 local year : substr(`label', `commapos' + 1, `commapos' + 2)
 //replace any illegal character (%,/,:," ") with underscores:
 local namestub : subinstr(`namestub', "%", "_")
 local namestub : subinstr(`namestub', "/", "_")
 local namestub : subinstr(`namestub', ":", "_")
 local namestub : subinstr(`namestub', " ", "_")
 //rename variable with the new stub and the year 
 rename `variable' `namestub'`year'
}

I get an error saying that strpos() is not allowed. Is it because I'm trying to make that a local macro? The examples I saw use it to generate a variable in the dataset, but I'm dealing with the variable labels. Am I in the right direction here? How do I fix this error?

• Roberto Ferrer: See help strtoname() for replacing characters not allowed in Stata names.

Answer 1

The syntax you need is

local commapos = strpos("`label'", ",")` 

and similarly for substr().

Note the extra bug that you need to enclose the label text in " ". That is needed elsewhere too.

The colon syntax is for extended macro functions, not functions in general.

(Learning how to automate this is a good ambition, but if the problem is this size, using varm interactively would have been quicker.)

• Kenji: Thank you for the replies and the comments. The problem is a little larger, as I have about ten times more variables than I put in the question (to save space). Therefore I don't want to use the variables manager.