-1

Given a string, I'm trying to determine which letter of the alphabet it belongs to. For example, "apple" goes into the "A" section. "Banana" goes into the "B" section. I'm using this to identify the section:

NSRange range = [string rangeOfString:letter 
                              options:NSAnchoredSearch |
                                      NSCaseInsensitiveSearch |
                                      NSDiacriticInsensitiveSearch |
                                      NSWidthInsensitiveSearch
                                range:NSMakeRange(0, string.length)
                               locale:locale];

Where string is the string I'm trying to bucket and letter is a letter of the alphabet. I do this in a loop for each letter of the alphabet.

It works great, except for words like "æquo", which should be bucketed into the letter "A", but aren't. What to do?

Edit The plot thickens. I'm looking at Korean now. The word "것" should be bucketed into the letter "ㄱ". There's got to be some way to do this other than maintaining a huge mapping table.

Hilton Campbell
  • 6,065
  • 3
  • 47
  • 79
  • Why not just use `characterAtIndex:` to find the first character of the word? – Hot Licks Sep 07 '14 at 22:21
  • Suppose I use `characterAtIndex:`, which gives me "æ" for the word "æquo". How do I know that goes into the "A" section? – Hilton Campbell Sep 07 '14 at 22:27
  • 1
    Presumably because you have a table that tells you how you want the "odd" characters to map. – Hot Licks Sep 07 '14 at 22:28
  • I'm hoping there's a better solution than that, given the extensive string manipulation API's that Apple provides. – Hilton Campbell Sep 07 '14 at 23:01
  • 1
    @HiltonCampbell In my language the letter "æ" is at the end of the alphabet. This letter has nothing to do with "a", except it written as an "a" and "e" joined. Originally a ligature representing a Latin diphthong, it has been promoted to the full status of a letter in the alphabets of some languages, including Danish, Norwegian, Icelandic and Faroese. As a letter of the Old English Latin alphabet, it was called æsc ("ash tree") after the Anglo-Saxon futhorc rune ᚫ (Runic letter ansuz.svg), which it transliterated; its traditional name in English is still ash – Sverrisson Sep 08 '14 at 00:01
  • 1
    If you have some conception that "æ" should be treated as "a", it seems to me that you need to somehow encode that information into a table or some such. Note that it may be that, in some character set/locale, "æ" naturally sorts the way you want, but if that's the case you need to first identify that character set/locale. – Hot Licks Sep 08 '14 at 00:05
  • @HotLicks I've identified the locale (it's `locale` in the code snippet). – Hilton Campbell Sep 08 '14 at 00:36
  • @HannesSverrisson Yes, I did some reading about it and I understand that the character "æ" should be treated differently based on locale. As far as I can tell, in French it is treated as "ae". – Hilton Campbell Sep 08 '14 at 00:37
  • So, do a sortUsingWhatever with NSString's `compare:options:range:locale:` as the compare function. – Hot Licks Sep 08 '14 at 02:14

1 Answers1

1

I think I've figured it out: I was thinking about it wrong. The question isn't, does a given word begin with a certain letter of the alphabet. Rather, the question is, does a given word fall within the sorting range of a certain letter of the alphabet.

For example, in the case of "æquo", I can check if it falls within the sorting range of the "A" section by checking if it is or comes after "A", and comes before "B".

Apple's compare:options:range:locale: method knows the answer to those two questions for any given locale. In this particular example, for French it would say yes. For some other language, like Danish, it should say no.

I've tested this on English, Spanish, Portuguese, French, German, and Korean, and it appears to be giving the expected results.

Hilton Campbell
  • 6,065
  • 3
  • 47
  • 79
  • If you're still doing the check for "Where does the letter A come in this word" then you're still doing it wrong. – Fogmeister Sep 09 '14 at 06:59
  • 1
    Agreed. Doing a check for "What is the first letter?" also appears to be doing it wrong. The right thing to do is identify the range the word falls in, which requires a number of `compare:options:range:locale:` checks. – Hilton Campbell Sep 10 '14 at 13:08