Page 1 of 1
PeekS byte length?
Posted: Sat Dec 31, 2016 3:22 am
by IdeasVacuum
PB5.51 x86
Code: Select all
sStr.s = "1ABCDEFGHJ2abcdefghj3ABCDEFGHJ4abcdefghj5ABCDEFGHJ"
*Buf = AllocateMemory(128)
PokeS(*Buf, sStr, -1, #PB_UTF8)
sVal.s = PeekS(*Buf, 32, #PB_UTF8 | #PB_ByteLength)
FreeMemory(*Buf)
Debug sVal
Debug StringByteLength(sVal)
I was expecting PeekS to return a string 32bytes long, but it returns a string 32 characters long?
Re: PeekS byte length?
Posted: Sat Dec 31, 2016 3:38 am
by Lunasole
IdeasVacuum wrote:PB5.51 x86
Code: Select all
sStr.s = "1ABCDEFGHJ2abcdefghj3ABCDEFGHJ4abcdefghj5ABCDEFGHJ"
*Buf = AllocateMemory(128)
PokeS(*Buf, sStr, -1, #PB_UTF8)
sVal.s = PeekS(*Buf, 32, #PB_UTF8 | #PB_ByteLength)
FreeMemory(*Buf)
Debug sVal
Debug StringByteLength(sVal)
I was expecting PeekS to return a string 32bytes long, but it returns a string 32 characters long?
As for me It is OK, because UTF8 uses 1 byte per character when you writing it into some custom buffer or file, etc.
It takes 2 bytes per char only if stored inside PB strings (which are UNICODE).
Just also specify format to StringByteLength(sVal, #PB_UTF8) for to get correct 32 from it.
Code: Select all
sStr.s = "1ABCDEFGHJ2abcdefghj3ABCDEFGHJ4abcdefghj5ABCDEFGHJ"
*Buf = AllocateMemory(128)
PokeS(*Buf, sStr, -1, #PB_UTF8)
ShowMemoryViewer(*Buf, 128)
CallDebugger
sVal.s = PeekS(*Buf, 32, #PB_UTF8 | #PB_ByteLength)
FreeMemory(*Buf)
Debug sVal
Debug StringByteLength(sVal)
Re: PeekS byte length?
Posted: Sat Dec 31, 2016 5:36 am
by IdeasVacuum
You are right, I should have qualified the type:
...does return 32
However, other functions such as AESEncode will 'see' that string as Unicode? In which case it's size needs to be 32bytes in Unicode if for example used as a Key.
It would be useful if PeekS could optionally return a Unicode string of n bytes.
Re: PeekS byte length?
Posted: Sat Dec 31, 2016 6:15 am
by TI-994A
IdeasVacuum wrote:It would be useful if PeekS could optionally return a Unicode string of n bytes.
I believe that the
PokeS()/PeekS() functions work on the
string length, Len(), and not the
byte length.
Re: PeekS byte length?
Posted: Sat Dec 31, 2016 7:35 am
by netmaestro
However, other functions such as AESEncode will 'see' that string as Unicode?
The AES commands don't take strings as input and they don't output any strings. All inputs and outputs are pointers to binary data (which may well contain strings) except keysize and mode, which are integers. If the string library ran up to the AES library and bit it on the ass it wouldn't recognize it.
Re: PeekS byte length?
Posted: Sat Dec 31, 2016 7:39 am
by Lunasole
IdeasVacuum wrote:
However, other functions such as AESEncode will 'see' that string as Unicode?
Looks like yes, if use *buffer_with_utf8string they will read it as UTF8 1 byte per char. If @StringVariable$, they will read unicode, which used with ciphers might damage security due to a lot of repeating zero bytes with known positions.
Re: PeekS byte length?
Posted: Sat Dec 31, 2016 7:56 am
by netmaestro
Seriously, it looks like yes? It's no. No, pointers to binary data only, size determined by keysize for the key and MemorySize for the buffers. Where you're getting this utf8 stuff is a mystery. Do you see string types somewhere in the AES command prototypes? If you have difficulty determining buffer size for stringbytelength reasons or for any other reason, trust me the AES library doesn't know a thing about it. Such considerations are for the coder to solve in advance, not the cipher library. It knows pointers and lengths and nothing more. Again, no-strings-in-or-out. Nostrings. InorOut. Give me a tune and I'll sing it to you. But I won't dance. No way. You have to draw the line somewhere.
Re: PeekS byte length?
Posted: Sat Dec 31, 2016 7:51 pm
by Lunasole
netmaestro wrote:Seriously, it looks like yes? It's no. No, pointers to binary data only, size determined by keysize for the key and MemorySize for the buffers. Where you're getting this utf8 stuff is a mystery. Do you see string types somewhere in the AES command prototypes? If you have difficulty determining buffer size for stringbytelength reasons or for any other reason, trust me the AES library doesn't know a thing about it. Such considerations are for the coder to solve in advance, not the cipher library. It knows pointers and lengths and nothing more. Again, no-strings-in-or-out. Nostrings. InorOut. Give me a tune and I'll sing it to you. But I won't dance. No way. You have to draw the line somewhere.
I know what you talking about, but you should better read before writing so much.
He asked about how AESEncoder() and other functions will read regular PB string, and they are reading it's content as Unicode, because built-in strings using Unicode format. If don't know about that, easily can think than "S$ = PeekS (#PB_UTF8)" will lead S$ to be UTF8 internally.