How do you get the nth character of a string

[danielFHcode]

Looking through the standard library documentation there seems to be no method for getting the nth character of a string (Int -> String -> Maybe Char), and so if you want to get it you have to convert the string into an array (|> String.toList |> List.toArray) which is not optimal as that takes O(n), does anyone know if there is a way to get the nth character of a string in O(1)?

[lydell]

Hi! You can do myString |> String.dropLeft (n - 1) |> String.uncons |> Maybe.map Tuple.first, but I don’t know how efficient it is.

You could also mention what you are trying to do. Maybe you don’t even need to get a char at a certain index if you change something else.

[pit]

FWIW: I used exactly the same code as Simon (lydell) in my port of the Elm compiler’s parser, and there it was fast enough to even parse >1MB files, so depending on what you are trying to do it might still be sufficient for your use case.

But as Simon said: maybe there’s another way to solve you problem?

[danielFHcode]

Thank you both very much!
I originally stumbled across this problem while writing a parser in elm, which I was able to solve by using String.slice to get a 1-character long strings, so as you two suggested there was another solution to my problem, but regardless I thought it was odd there was no way to do this.
@lydle, regarding the efficiency of your solution it is certainly good enough for me as it is O(1) just like I wanted, so again thank you!
Finally, I just wanted to add that it is possible to make the method lydell suggested more generalized for getting the nth character of a string (as their method was only to get the last one):

get : Int -> String -> Maybe Char
get i str = String.slice i (i+1) str |> String.uncons |> Maybe.map Tuple.first

[perkee]

@lydle, regarding the efficiency of your solution it is certainly good enough for me as it is O(1)

Not sure if it’s O(1) or not. It ends up calling String.prototype.slice in JS. It seems that JS slices by the codepoint (of which there can be, in the extreme case, 30 in any extant character and for which there is no theoretical limit) which simplifies things, but even then—a codepoint can stretch across many bytes. So you can’t tell where any codepoint begins without starting at the start of the string and parsing the bytes, right? Only thing that could affect that is if the JS string is implemented as a JS array which is itself implemented as a hashmap and then, yeah, you’d have O(1) lookup.

[dta]

JS strings operate on UTF-16 codepoints, so they are always 2 bytes. It should be constant time to index into a string, but as you point out you may wind up extracting only part of a character if you slice a single code point.