Compare two strings based on ASCII dictionary order

I know the sort builtin function will sort cell arrays of strings in ascii dictionary order. But how may I simply compare two strings to determine which is first (by dictionary order) ? This must be a constituent part of the sort routine, but I cannot find a way to do it. (and sort is builtin so I cannot inspect it).

[Aravin]

Can you give one example ?

Answer 1

[Oleg Komarov]

[trash,idx] = sort({'abc';'a'})

Then just look at idx(1)

[Geoff]

Oh, that's nice and easy =) I would use [~, idx] though, rather than create a variable called 'trash'.

[Brandon Kuczenski]

This plus diff(idx) gives a usable result. it does not detect equal strings, however.

Answer 2

[Geoff]

Yeah this is a real failing of the strcmp function in my opinion. That function originated in C, and would tell you whether a string was less, equal or greater than another. But it was probably too non-intuitive for MatLab users because it returned zero if the strings were equal.

Basically, what the C function does is subtract the strings character by character. So you can do that here:

function cmp = cstrcmp( a, b )

    % Force the strings to equal length
    x = char({a;b});

    % Subtract one from the other
    d = x(1,:) - x(2,:);

    % Remove zero entries
    d(~d) = [];
    if isempty(d)
        cmp = 0;
    else
        cmp = d(1);
    end

end

The output is:

 a == b : 0
 a > b  : positive
 a < b  : negative

There are probably more efficient ways to do this in MatLab. I just stuck to the easy matrix operations.

[Brandon Kuczenski]

char({a;b})- now that is useful. so is ~ instead of Trash; and I like d(~d) = [] as well.. thank you!

[Geoff]

Haha yeah that logic negation of non-logic values is an old C habit. It's technically not very good programming practice (because it's not as readable as "d==0"), but MatLab does document the behaviour. Use at your own peril =)

Answer 3

[Junaid]

Can you give one example. As I understand, you can do it by compare operator.

a = 'abd';
b = 'abc';

a <= b

output is [1 1 0] where 0 indicates that some character in b comes before in by dictionary order.

[Brandon Kuczenski]

That only works if the strings are the same length. In order to do the comparison in general, I have to first test to see which string is longer and either truncate it or pad the other. I suppose that is not too much trouble, but I am surprised that it must be done manually.

[Brandon Kuczenski]

Also, the single comparison is insufficient to determine which string comes first.

a='hellob'

b='hellbo'

According to a dictionary test, a>b. In order for me to know that, I would need to do both comparisons:

a>=b

>> a>=b

ans =

1 1 1 1 1 0

>> b>=a

ans =

1 1 1 1 0 1

>>

and then see which one has the earlier zero.