Question about mb_metastrlen

Question about mb_metastrlen

[Sebastian Gniazdowski]

Hello,
the function counts bytes in last incomplete wide character:

        ret = mbrtowc(&wc, &inchar, 1, &mb_shiftstate);
        if (ret == MB_INCOMPLETE) {
            num_in_char++;
        } else {

When returning, it makes use of the count:

    /* If incomplete, treat remainder as trailing single bytes */
    return num + num_in_char;

Strings are stored in wchar_t arrays. The incomplete character will
occupy single index, correct? So maybe the return should be:

    return num + ( num_in_char > 0 ? 1 : 0 );

?

Best regards,
Sebastian Gniazdowski

Re: Question about mb_metastrlen

[Peter Stephenson]

On Tue, 27 Oct 2015 09:31:02 +0100
Sebastian Gniazdowski <sgniazdowski@gmail.com> wrote:
> Hello,
> the function counts bytes in last incomplete wide character:
> 
>         ret = mbrtowc(&wc, &inchar, 1, &mb_shiftstate);
>         if (ret == MB_INCOMPLETE) {
>             num_in_char++;
>         } else {
> 
> When returning, it makes use of the count:
> 
>     /* If incomplete, treat remainder as trailing single bytes */
>     return num + num_in_char;
> 
> Strings are stored in wchar_t arrays. The incomplete character will
> occupy single index, correct? So maybe the return should be:
> 
>     return num + ( num_in_char > 0 ? 1 : 0 );

The function you're talking about is for a string length, not a
character length.  num_in_char counts the number of trailing bytes that
didn't form a wide character.  Each will be treated as a single byte.
So each counts 1 for the length of the string.

I think your answer would be correct for a function that counts just the
next character (i.e. mb_metacharlenconv()), but I think that's already doing
what you'd expect in that case.

That's the intention of the function, anyway, but if you can see a use
that's inconsistent with it there could be a bug there.

pws

Re: Question about mb_metastrlen

[Sebastian Gniazdowski]

On 27 October 2015 at 10:10, Peter Stephenson <p.stephenson@samsung.com> wrote:
> On Tue, 27 Oct 2015 09:31:02 +0100
> The function you're talking about is for a string length, not a
> character length.  num_in_char counts the number of trailing bytes that
> didn't form a wide character.  Each will be treated as a single byte.
> So each counts 1 for the length of the string.

There is the condition:
            if (ret == MB_INVALID) {

Isn't it that if there are many trailing bytes that do not form a
character, they will be catched into MB_INVALID, and only last
"character" can stay as not yet complete?

Best regards,
Sebastian Gniazdowski

Re: Question about mb_metastrlen

[Peter Stephenson]

On Tue, 27 Oct 2015 11:34:35 +0100
Sebastian Gniazdowski <sgniazdowski@gmail.com> wrote:

> On 27 October 2015 at 10:10, Peter Stephenson <p.stephenson@samsung.com> wrote:
> > On Tue, 27 Oct 2015 09:31:02 +0100
> > The function you're talking about is for a string length, not a
> > character length.  num_in_char counts the number of trailing bytes that
> > didn't form a wide character.  Each will be treated as a single byte.
> > So each counts 1 for the length of the string.
> 
> There is the condition:
>             if (ret == MB_INVALID) {
> 
> Isn't it that if there are many trailing bytes that do not form a
> character, they will be catched into MB_INVALID, and only last
> "character" can stay as not yet complete?

Only the last multibyte character can consist of multiple individual
bytes that look like part of an incomplete character rather than simply
as invalid, that's correct.  Hence the note at the end of the function
about use of num_in_char, and hence we reset num_in_char to 0 any time
we get a full multibyte character or mark a byte as invalid rather than
incomplete.

pws