Unpack 4 bytes to a signed integer

Hi
I' m using unpack to convert 4 bytes to local integer, but ruby just
supply the "N" modifer which means unsigned long integer.

My questions is :
How to unpack 4 bytes to a signed integer ?

Thank you.

···

--
wwg

There are "i" and "I" for signed and unsigned integer respectively, which deals in the local size of integer.

Hope this helps,

Mike

···

On 15-Jan-07, at 8:29 AM, w wg wrote:

Hi
I' m using unpack to convert 4 bytes to local integer, but ruby just
supply the "N" modifer which means unsigned long integer.

My questions is :
How to unpack 4 bytes to a signed integer ?

--

Mike Stok <mike@stok.ca>
http://www.stok.ca/~mike/

The "`Stok' disclaimers" apply.

s = 0xFF.chr * 4
i = s.unpack("L").shift

p i
p 2**32-1

s = 0xFF.chr * 4
i = s.unpack("L").shift

p i
p 2**32-1

w wg wrote:

Hi
I' m using unpack to convert 4 bytes to local integer, but ruby just
supply the "N" modifer which means unsigned long integer.

My questions is :
How to unpack 4 bytes to a signed integer ?

The best way I've found is to unpack with N (to get the swapping right) and then do some arithmetic to interpret the unsigned value as signed:

x = -123
s = .pack("N")
   # Note that for _pack_ there is no need for a
   # special signed version of N

p s # ==> "\377\377\377\205"

length = 32
mid = 2**(length-1)
max_unsigned = 2**length
to_signed = proc {|n| (n>=mid) ? n - max_unsigned : n}

p to_signed[s.unpack("N").first] # ==> -123

This is all very hard for me to remember, so I've written a library to do it, bit-struct (http://redshift.sourceforge.net/bit-struct\). This makes life easier:

require 'bit-struct'

class Packet < BitStruct
   signed :x, 32
end

pkt = Packet.new
pkt.x = -123

p pkt.to_s # ==> "\377\377\377\205"
p pkt.x # ==> -123

# given string data from a network:
pkt2 = Packet.new("\377\377\377\205")
p pkt2.x # ==> -123

···

--
       vjoel : Joel VanderWerf : path berkeley edu : 510 665 3407

But the "i" and "I" modifer don't care bytes order.
I'm reading data from network using IO#sysread, I need receive integer
, not unsigned integer.

Thank you for your reply.

···

2007/1/15, Mike Stok <mike@stok.ca>:

On 15-Jan-07, at 8:29 AM, w wg wrote:

> Hi
> I' m using unpack to convert 4 bytes to local integer, but ruby just
> supply the "N" modifer which means unsigned long integer.
>
> My questions is :
> How to unpack 4 bytes to a signed integer ?

There are "i" and "I" for signed and unsigned integer respectively,
which deals in the local size of integer.

Hope this helps,

Mike

--

Mike Stok <mike@stok.ca>
Mike Stok

The "`Stok' disclaimers" apply.

--
--
WenGe Wang

I tried your codes, it works exactly as what I expected. I havn't
tried your library, but I think it should get the correct result too.

Thank you.

···

The best way I've found is to unpack with N (to get the swapping right)
and then do some arithmetic to interpret the unsigned value as signed:

x = -123
s = .pack("N")
   # Note that for _pack_ there is no need for a
   # special signed version of N

p s # ==> "\377\377\377\205"

length = 32
mid = 2**(length-1)
max_unsigned = 2**length
to_signed = proc {|n| (n>=mid) ? n - max_unsigned : n}

p to_signed[s.unpack("N").first] # ==> -123

But the "i" and "I" modifer don't care bytes order.
I'm reading data from network using IO#sysread, I need receive integer
, not unsigned integer.

Thank you for your reply.

OK then, what about "V"? "treat 4 characters as an unsigned long in little-endian byte order"

The N uses network byte order "big-endian"

Mike

···

On 15-Jan-07, at 10:10 AM, w wg wrote:

2007/1/15, Mike Stok <mike@stok.ca>:

On 15-Jan-07, at 8:29 AM, w wg wrote:

> Hi
> I' m using unpack to convert 4 bytes to local integer, but ruby just
> supply the "N" modifer which means unsigned long integer.
>
> My questions is :
> How to unpack 4 bytes to a signed integer ?

There are "i" and "I" for signed and unsigned integer respectively,
which deals in the local size of integer.

Hope this helps,

Mike

--

Mike Stok <mike@stok.ca>
Mike Stok

The "`Stok' disclaimers" apply.

--
--
WenGe Wang

--

Mike Stok <mike@stok.ca>
http://www.stok.ca/~mike/

The "`Stok' disclaimers" apply.