# Converting Bytes to a Negative Integer

Let’s say I have 4 bytes:

255 255 255 236

The binary representation of this is:

11111111
11111111
11111111
11101100

The decimal number should be -20. Does anyone know how to convert these
4 bytes into an Integer?

~harris

On Feb 14, 2007, at 2:27 PM, Harris R. wrote:

Let’s say I have 4 bytes:

255 255 255 236

bytes = [255, 255, 255, 236].map { |b| b.chr }.join
=> “\377\377\377\354”

The binary representation of this is:

11111111
11111111
11111111
11101100

bytes.unpack(“B*”).first.scan(/[01]{8}/)
=> [“11111111”, “11111111”, “11111111”, “11101100”]

The decimal number should be -20. Does anyone know how to convert
these 4 bytes into an Integer?

This is where I had trouble. I found the following:

bytes.reverse.unpack(“l*”).first
=> -20

This is tied to my processor though. I had to reverse the bytes
because they are not in the order my machine expects.

unpack() has directives for when the order is known, but all of those
seemed to be unsigned.

I’m anxious to see the cross-platform solution for this…

James Edward G. II

On 2/14/07, James Edward G. II [email protected] wrote:

I’m anxious to see the cross-platform solution for this…

LITTLE_ENDIAN = [42].pack(‘i’)[0] == 42

bytes = [255, 255, 255, 236]
str = bytes.pack(‘C*’)
str.reverse! if LITTLE_ENDIAN
str.unpack(‘i’)[0]

Blessings,
TwP

On 2/14/07, James Edward G. II [email protected] wrote:

11111111
This is where I had trouble. I found the following:
I’m anxious to see the cross-platform solution for this…
Joel VanDerWerf had something that I’ll be adapting if I need it
elsewhere.

x = -123
s = [x].pack(“N”)

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

puts to_signed[s.unpack(“N”).first]

I need it for unsigned->signed words, so:

x = -123
s = [x].pack(“n”)

bits = 16
max_unsigned = 2 ** bits
max_signed = 2 ** (bits - 1)
to_signed = proc { |n| (n >= max_signed) ? n - max_unsigned : n }

puts to_signed[s.unpack(“n”).first]

-austin

On Feb 14, 2007, at 3:27 PM, Harris R. wrote:

The decimal number should be -20. Does anyone know how to convert
these 4 bytes into an Integer?

Maybe this will help:

data = “”
[255, 255, 255, 236].each { |e| data << e }
data # => “\377\377\377\354”
data.unpack(‘l’).first # => -20

Regards, Morton

On Feb 14, 2007, at 6:37 PM, James Edward G. II wrote:

=> “”

[255, 255, 255, 236].each { |e| data << e }
=> [255, 255, 255, 236]

data
=> “\377\377\377\354”

data.unpack(‘l’).first
=> -318767105

I see your point. Many new computers have 64-bit CPUs and we should
remember that. Also, you want to be sure I, lowly 32-bit CPU owner,
envy you for having one of these .

Regards, Morton

On Feb 14, 2007, at 5:02 PM, Morton G. wrote:

data = “”
[255, 255, 255, 236].each { |e| data << e }
data # => “\377\377\377\354”
data.unpack(‘l’).first # => -20

Watch what happens when I run the same code:

data = “”
=> “”

[255, 255, 255, 236].each { |e| data << e }
=> [255, 255, 255, 236]

data
=> “\377\377\377\354”

data.unpack(‘l’).first
=> -318767105

James Edward G. II

On Feb 15, 2007, at 9:47 AM, Morton G. wrote:

remember that. Also, you want to be sure I, lowly 32-bit CPU owner,
envy you for having one of these .

I’ll prop you back up then, this isn’t a 32 to 64 bit (size) issue.
It’s an endian (order) issue:

James Edward G. II

On Feb 15, 2007, at 4:58 PM, Morton G. wrote:

which shows it’s big-endian because both unpack directives produce
the same result.

1. On your computer (which I think is also a Macintosh), you would get

Yes, it’s an Intel Core Duo Mac.

[-20].pack(‘i’) # => “\354\377\377\377”
[-318767105].pack(‘i’) # => “\377\377\377\354”

and

[-20].pack(‘N’) # => “\377\377\377\354”
[-318767105].pack(‘N’) # => “\354\377\377\377”

which would show it’s little-endian and, presumably, some kind of
Intel-based Mac.

Exactly.

Do I have it right now?

Sure do.

P.S. But I don’t see how this props me back up

I wasn’t lording 64 bit processing over you.

James Edward G. II

On Feb 15, 2007, at 11:04 AM, James Edward G. II wrote:

I see your point. Many new computers have 64-bit CPUs and we
should remember that. Also, you want to be sure I, lowly 32-bit
CPU owner, envy you for having one of these .

I’ll prop you back up then, this isn’t a 32 to 64 bit (size)
issue. It’s an endian (order) issue:

Mea culpa – I didn’t see your point at all. Let’s see if I’ve got it
now.

1. Your result couldn’t the result of conversion into a 64-bit
integer with big-endian byte order, because if that were the case we
would see either -20 or a large positive number (if sign extension
didn’t occur).

2. On my computer (iMac G5), I get

[-20].pack(‘i’) # => “\377\377\377\354”
[-318767105].pack(‘i’) # => “\354\377\377\377”

and

[-20].pack(‘N’) # => “\377\377\377\354”
[-318767105].pack(‘N’) # => “\354\377\377\377”

which shows it’s big-endian because both unpack directives produce
the same result.

1. On your computer (which I think is also a Macintosh), you would get

[-20].pack(‘i’) # => “\354\377\377\377”
[-318767105].pack(‘i’) # => “\377\377\377\354”

and

[-20].pack(‘N’) # => “\377\377\377\354”
[-318767105].pack(‘N’) # => “\354\377\377\377”

which would show it’s little-endian and, presumably, some kind of
Intel-based Mac.

Do I have it right now?

Regards, Morton

P.S. But I don’t see how this props me back up