Re: [v2 PATCH 1/4] shcore: Implement a bunch of string conversion functions.

[Nikolay Sivov]

On 11/29/18 9:10 AM, Dmitry Timoshkov wrote:
> Nikolay Sivov  wrote:
> > +DWORD WINAPI SHAnsiToAnsi(const char *src, char *dest, int dest_len)
> > +{
> > - DWORD ret = 0;
> > -
> > - TRACE("(%s, %p, %d)\n", debugstr_a(src), dest, dest_len);
> > -
> > - if (!src || !dest || dest_len <= 0)
> > -
> >    return 0;
> > -
> > - while (dest_len > 1 && *src)
> > - {
> > -
> >    dest_len--;
> > -
> >    *dest++ = *src++;
> > -
> >    ret++;
> > - }
> > -
> > - if (dest_len)
> > - {
> > -
> >    *dest = 0;
> > -
> >    ret++;
> > - }
> > -
> > - return ret;
> >
> > +}
>
> Still the same duplication without a reasonable explanation.

It's used to get correct return value.
> > +DWORD WINAPI SHUnicodeToAnsi(const WCHAR *src, char *dest, int dest_len)
> > +{
> > - int required = 1;
> > -
> > - TRACE("(%s, %p, %d)\n", debugstr_w(src), dest, dest_len);
> > -
> > - if (!dest || !dest_len)
> > -
> >    return 0;
> > -
> > - if (src)
> > - {
> > -
> >    required = WideCharToMultiByte(CP_ACP, 0, src, -1, NULL, 0, NULL, NULL);
> > -
> >    WideCharToMultiByte(CP_ACP, 0, src, -1, dest, dest_len, NULL, NULL);
>
> It should be possible to avoid doing the conversion twice.

Second call returns 0 for shorter buffer. If you have actual suggestion 
I'm sure nobody would mind if you posted it.
> Same comments apply for remaining implementations.