Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: RE: AW: RE: RE: RE: FW: Using regex to identify strings with capital letters

From   "Nick Cox" <>
To   <>
Subject   st: RE: AW: RE: RE: RE: FW: Using regex to identify strings with capital letters
Date   Thu, 27 May 2010 12:58:36 +0100

The sort order for strings is, to the best of my knowledge, the order of
-char()-. -asciiplot- from SSC is one of various ways to make it


Martin Weiss

So Erik can play around with the example code I created last night to
explain the -inrange()- thing to myself. Strangely, I did not find a
good source on the -sort- order for strings, in particular with regard
upper and lower case issues. The index points to [U], 13.2.3, which is a
little terse.


//from NJC`s
//follow special sequences

//create dataset
set obs 104

gen str2 myvar=""

token `c(ALPHA)'

forv i=1/26{
	replace myvar="``i''"+"``i''" in `i'
	replace myvar="``i''" in `=`i'+26'
	replace myvar="``i''" in `=`i'+52'

token `c(alpha)'

forv i=1/26{
	replace myvar=myvar+"``i''" in `=`i'+26'
	replace myvar="``i''"+myvar in `=`i'+52'
	replace myvar="``i''"+"``i''" in `=`i'+78'

//try -inrange()-
l if inrange(myvar,"aA","zZ")
l if inrange(myvar,"AA","ZZ")

//try -regexm()-
l if regexm(myvar, "^([A-Z][A-Z])")
l if regexm(myvar, "^([A-Z][a-z])")
l if regexm(myvar, "^([a-z][A-Z])")
l if regexm(myvar, "^([a-z][a-z])")

Nick Cox

Martin is correct that -inrange("Er", "AA", "ZZ")- is true. Possibly
this is Erik's specific problem, namely that having the first capital
letter in "A" ... "Z" is necessary but not sufficient. 

I offer as a stronger criterion 

inrange(substr(myvar,1,1), "A", "Z") & inrange(substr(myvar,2,1), "A",

I continue to like regex solutions when they are the simplest available!


Martin Weiss

Erik does have a point, though, in that Nick`s -inrange()- proposal
seems to
check for the first character only:

di inrange(substr("erik in lower case",1,2) , "AA", "ZZ")
di inrange(substr("Erik in lower case",1,2) , "AA", "ZZ")

BTW, why was -di inrange("erik in lower case", "AA", "ZZ")- a good
earlier, even though the -substr()- part was missing?

Nick Cox

Not true of my Stata: 

. di inrange("erik in lower case", "AA", "ZZ")

I think -- you've heard this before -- we need to see your code and some
your results, not your speculation about what might be happening. 


Beecroft, Erik (VDSS)

I tried Nick's suggestion, pasted below, but inrange does not seem to
distinguish between lower and upper case.  In other words, the statement
below keeps all observations that begin with two letters, whether
capital or lower case.

Nick Cox 

You don't need regex for this. 

... if inrange(substr(myvar,1,2), "AA", "ZZ") 

should be enough, or even "AK" to "WY" or whatever it is. (Remember this
is an international list!) 

From: Beecroft, Erik (VDSS) 

I need to extract certain observations from a series of text files.
Each file contains only one variable, which is string.  The
observations I want all begin with two capital letters. (They are state
abbreviations, such as VA or AK).  The other observations do not begin
with two capital letters.

Is there a way to tell Stata to keep only observations for which the
variable begins with two capital letters?

It seems like the regex function might work, but I have never worked
with regular expression syntax before.  

For example, a portion of a text file might look like:
	VA department of Social Services

I want to keep only the third observation above.

I am using Stata for Windows 10.1.

*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index