[Qt-interest] Converting first character of a char* from local charset

[Nikos Chantziaras]

On 05/27/2011 07:39 PM, Thiago Macieira wrote:
> On Friday, 27 de May de 2011 19:17:16 Nikos Chantziaras wrote:
>> On 05/27/2011 07:06 PM, Thiago Macieira wrote:
>>> On Friday, 27 de May de 2011 17:21:15 Nikos Chantziaras wrote:
>>>> I can imagine that the performance of this code ain't the best, since
>>>> what it does is call QTextCodec::toUnicode() repeatedly, increasing
>>>> the
>>>> amount of characters each time, until it hits an amount of bytes that
>>>> can be converted.
>>>>
>>>> Is there a better way to achieve this?
>>>
>>> Why do you want to achieve this first?
>>
>> I'm porting a piece of software from Windows to Unix, and the interfaces
>> (abstract classes) I need to implement work like that.  The pure virtual
>> I need to implement gets a char* and needs to return the first character
>> of it converted to Unicode, from whatever character set the user's
>> system is using.
>
> That didn't answer the question...
>
> Doing it because some requirement with no reason asks for it is not enough.
> Blindly accepting bad requirements is bad practice and leads to bad code.

It's for a charset-aware getch() implementation (same semantics as the 
one from curses.)  This version however, needs to read the first 
character (which can be multi-byte) from the input buffer, not just the 
first byte.


> Try converting the first 32 bytes to QString and then get the first one out of
> the string.

This won't work, because then I'd have no idea how many bytes the 
character used.  getch() needs to remove the bytes that were read from 
the buffer, so that the next call won't read them again.

Anyway, I guess my question was answered.  There's no routine in Qt that 
does this directly, so I guess I'll have to live with it.  Fortunately, 
the code is in a place that isn't performance-critical.  Although a 
getch() is needed, it's used rarely; a getstr()-like routine is used far 
more often, where I simply convert the whole buffer.