ctypes And The WACAH Principle

[Lawrence D’Oliveiro]

On Wednesday, August 10, 2016 at 10:13:53 PM UTC+12, eryk sun wrote:

> ctypes classes have from_buffer and from_buffer_copy methods that use
> the buffer protocol. For example, for a read-only buffer you can use
> from_buffer_copy:
> 
>     >>> b = bytes(b'1234')
>     >>> a = (ctypes.c_char * 3).from_buffer_copy(b, 1)
>     >>> a[:]
>     b'234'

Can I avoid the copying? For example, in Qahirah <https://github.com/ldo/qahirah> the following method creates and fills in an ImageSurface from a stream of bytes in PNG format:

    @classmethod
    def create_from_png_bytes(celf, data) :
        "creates an ImageSurface from a PNG-format data sequence. This can be" \
        " of the bytes or bytearray types, or an array.array with \"B\" type code."

        offset = 0

        def read_data(_, data, length) :
            nonlocal offset
            if offset + length <= len(data) :
                ct.memmove(data, baseadr + offset, length)
                offset += length
                status = CAIRO.STATUS_SUCCESS
            else :
                status = CAIRO.STATUS_READ_ERROR
            #end if
            return \
                status
        #end read_data

    #begin create_from_png_bytes
        if isinstance(data, bytes) or isinstance(data, bytearray) :
            data = array.array("B", data)
        elif not isinstance(data, array.array) or data.typecode != "B" :
            raise TypeError("data is not bytes, bytearray or array of bytes")
        #end if
        baseadr = data.buffer_info()[0]
        return \
            celf.create_from_png_stream(read_data, None)
    #end create_from_png_bytes

But it has to copy the bytes into an array.array object, then decode that. Is there a way it could access the bytes memory directly?