Re: Handling bytes

Hi,

Brad Schick wrote:
> On 5/8/2008 12:05 AM, Stefan Behnel wrote:
>>> It just seems
>>> so ugly in what is otherwise a fairly clean language. Particularly when
>>> the data sometimes starts life as a list. How do you go about stuffing
>>> 'bytes' into a string? With arrays as below? Or with cStringIO? Yuck and
>>> yuck ;)
>>>     
>> Sure, why not? You can also create an empty string and fill it yourself:
>>
>>    cdef int i
>>    cdef char* s
>>    py_s = PyString_FromStringAndSize(NULL, len(data))
>>    s = py_s
>>    for i in range(len(data)):
>>        s[i] = l[i]
>>   
> 
> I meant in python code. In other words, the code calling into the pyrex
> module that will need to cram everything into strings. That's fine if it
> the data starts in a string or file, but its pretty ugly and inefficient
> if it starts in some other sequence like a list.

But that's just an API design question. At the API leve, you can accept byte
strings and iterables that return byte numbers, and just write a Pyrex
function that detects what you actually got and return a normalised internal
representation (and maybe raise an exception for everything else). Then just
call it at the beginning of every API function and you're done thinking about
this issue.

Stefan