Q: how do I alphabetize strings with diacritics?


I have a long list of author names, some of whom have letters with diacritics in their names (e.g., "á" or "é"). I want to sort this list alphabetically.

The problem: sorting the list with string-lessp does not sort them alphabetically.

According to the accepted answer in this post from a sister site, English ignores the diacritics in sorting except to break ties. (Other languages do it differently.)

Toy Example

Here's a toy example. The list of letters that I start with is already ordered alphabetically. When I sort this list with string-lessp, however, it sorts them in what I presume is unicode point order rather than alphabetical order:

(let ((letters '("a" "à" "á" "â" "b" "c" "e" "é" "ê")))
  (sort letters #'string-lessp))
;; => ("a" "b" "c" "e" "à" "á" "â" "é" "ê")

What Do I Do?

How can I alphabetize strings with diacritics in them?

At a minimum, I'd like to respect the "ignore diacritics except to break ties" rule described above. Ideally, I'd like to be able to alphabetize according to an arbitrarily defined language, but I'll happily settle for English.

  • Good question. I wonder whether it depends on - or should depend on - your OS language setting. Alphabetization is language-dependent, in general. – Drew Mar 22 at 20:42

If your system locale is set to something that will properly collate diacritics (not POSIX), this should work for you:

(let ((letters '("é" "a" "à" "c" "â" "b" "á" "e" "ê")))
  (sort letters #'string-collate-lessp))
;; => ("a" "á" "à" "â" "b" "c" "e" "é" "ê")

If that doesn't work, you can supply a locale string as the third argument to string-collate-lessp to get what you want. With American English on a POSIX system, for example:

(let ((letters '("é" "a" "à" "c" "â" "b" "á" "e" "ê")))
  (sort letters (lambda (a b) (string-collate-lessp a b "en_US.UTF-8"))))

(For American English on MS-Windows, replace "en_US.UTF-8" with "enu_USA.1252".)

If you want to dig into this a little more to see what happens under the hood, I recommend taking a look at the function definition of str_collate in src/sysdep.c.

  • Exactly what I was looking for, thank you! As an aside, could you mention where one can look up the locale strings for non-American English languages? – Dan Mar 22 at 22:58
  • @Dan That is system-dependent information AFAIK. Although not a listing of locale identifiers, this Q&A from unix.stackexchange should help you formulate good values on most unix-like systems. – Sam Mar 25 at 12:52
  • thanks for the follow-up! – Dan Mar 25 at 15:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.