Own strcmp function - non standard chars - c

Question

I am currently writing a little sort function. I can only use stdio libary, so I wrote my 'own strcmp' function.
int ownstrcmp(char a[], char b[])
{
int i = 0;
while( a[i] == b[i] )
{
if( a[i] == '\0' )
return 0;
++i;
}
return ( a[i] < b[i]) ? 1 : -1;
}
This works great for me. But there is one little problem: What can I do for 'non-Standard-Chars'? Like "ä,ü,ß Their decimal ASCII value is greater than the normal chars, so it sort the string 'example' behind 'ääää'.
I have already read about locale, but the only library that i can use is stdio.h. Is there a 'simple' solution for this problem?

Answer 1

Your question is somewhat vague. First of all, how characters with umlaut are represented depends on your encoding. For example, my computer's locale is set to Greek, meaning that in place of those special Latin characters I have Greek characters. You can't assume anything like that, as far as I can tell.
Second, the answer to your question depends on your representation. Are you still using a "one char per character" representation? If that's so, the above code might still work.
If you're using multi char representation, for example two chars per character, you should change your code so that it exits when two consecutive chars are \0.
Generally, you may want to look into how wchar_t and its family of functions (specifically wcscmp) are implemented.

Answer 2

For german the umlauts ä,ö,ü and ß will be sorted as if they occur in their 'expanded' form:
ä -> ae
ö -> oe
ü -> ue
ß -> ss
In order to get the collation according to the standard you could expand the strings before comparing.

Answer 3

You need to know the encoding the characters are in, and make sure you treat the strings properly. If the encoding is multi-byte, you must start reading (and comparing) individual characters, not bytes.
Also, the way to compare characters internationally varies with the locale, there's no single solution. In some languages, 'ä' sorts after 'z', in some it sorts right next to 'a'.
One simple way of implementing this is of course to create a table which holds the relative order for each character, like so:
unsigned char character_order[256];
character_order[(unsigned char) 'a'] = 1;
character_order[(unsigned char) 'ä'] = character_order[(unsigned char) 'a'];
/* ... and so on ... */
Then instead of subtracting the character's encoded value (which no longer can be used as a "proxy" for the sorting order of the character), you compare the character_order values.
The above assumes single-byte encoding, i.e. Latin-1 or something, since the array size is only 256.
Also note casts to unsigned char when indexing with character literals.

Answer 4

If you are using ISO/IEC_8859-16 encoding, which is the normal enconding for German Language, it's enough to transform your char to unsigned char.
In this way chars can be represented in interval 0-255, suitable for this standard.

Answer 5

Under UTF8 this can help, following your code
if ((a[i] > 0) ^ (b[i] > 0))
return a[i] > 0 ? 1 : -1;
else
return a[i] < b[i] ? 1 : -1;
But you have to check cases like ownstrcmp("ab", "abc");
Furthermore your code doesn't work like strcmp() in <string.h>
A value greater than zero indicates that the first character that does not match has a greater value in str1 than in str2; And a value less than zero indicates the opposite.
I would do it like this:
int ownstrcmp(char a[], char b[])
{
int i = 0;
while(a[i] == b[i]) {
if (a[i] == 0) return 0;
++i;
}
if ((a[i] == 0) || (b[i] == 0))
return a[i] != 0 ? 1 : -1;
if ((a[i] > 0) ^ (b[i] > 0))
return a[i] < 0 ? 1 : -1;
else
return a[i] > b[i] ? 1 : -1;
}