Bitwise question

I’m working on a project that has a bitmap of permissions and I need to
find out if a user has access.

Given permissions bit mask of 0001C0200F02000000000 where each bit
represents a specific permission

And a request for permissions check on bits [37, 12, 48]

What’s the best way to find out if user has access to all requested
permissions?

Andrew B. [email protected] writes:

And a request for permissions check on bits [37, 12, 48]

What’s the best way to find out if user has access to all requested
permissions?

(def check(permissions,bits)
(bits . inject(true) { | r , b |
(r and (0 != (permissions & (1 << b))))
})
end)
nil
(check 0xf0,[1,4,5,6]) → false
(check 0xf0,[4,5,6]) → true

On Jan 28, 2009, at 4:32 PM, Pascal J. Bourguignon wrote:

(def check(permissions,bits)
(bits . inject(true) { | r , b |
(r and (0 != (permissions & (1 << b))))
})
end)
nil
(check 0xf0,[1,4,5,6]) → false
(check 0xf0,[4,5,6]) → true


Pascal B.

Or a simpler version:

class Integer
def as_bits
ary = []
(self.size*8).times {|i| ary.unshift(i) if self[i].nonzero?}
ary
end

def check_permissions(*bits)
bits.all? {|bit| self[bit].nonzero? }
end
end

irb> 0x0001C0200F02000000000.as_bits
=> [68, 67, 66, 57, 47, 46, 45, 44, 37]
irb> 0x0001C0200F02000000000.check_permissions 37, 12, 48
=> false
irb> 0x0001C0200F02000000000.check_permissions 37, 47, 57
=> true

And written in normal Ruby rather than LispRuby :wink:

-Rob

Rob B. http://agileconsultingllc.com
[email protected]

Hi –

On Thu, 29 Jan 2009, Pascal J. Bourguignon wrote:

(bits . inject(true) { | r , b |
(r and (0 != (permissions & (1 << b))))
})
end)
nil
(check 0xf0,[1,4,5,6]) → false
(check 0xf0,[4,5,6]) → true

Possibly more efficient since it short-circuits on the first failed
bit:

def check(permissions,bits)
bits.all? {|bit| (permissions & 1 << bit).nonzero? }
end

David


David A. Black / Ruby Power and Light, LLC
Ruby/Rails consulting & training: http://www.rubypal.com
Coming in 2009: The Well-Grounded Rubyist (The Well-Grounded Rubyist)

http://www.wishsight.com => Independent, social wishlist management!

Andrew B. wrote:

I’m working on a project that has a bitmap of permissions and I need to
find out if a user has access.

Given permissions bit mask of 0001C0200F02000000000 where each bit
represents a specific permission

And a request for permissions check on bits [37, 12, 48]

What’s the best way to find out if user has access to all requested
permissions?

Whatever ‘best’ is supposed to mean. Here is one way:

requested_bits.all? { |bit| permissions[bit] == 1 }

Regards
Stefan

Rob B. [email protected] writes:

            })

def check_permissions(*bits)
bits.all? {|bit| self[bit].nonzero? }
end
end

Simplier? A class, two methods, a lot of operations, a lot of memory!

Pascal J. Bourguignon wrote:

(def check(permissions,bits)
(bits . inject(true) { | r , b |
(r and (0 != (permissions & (1 << b))))
})
end)

That what happens when one spends his time typing parenthesis instead of
programming, my dear Pascal. I managed to count at least two bugs there.
One, you need to do (1 << (b-1)). Two, if you pass [] as the bits, the
result will be ‘true’.

Albert S. [email protected] writes:

One, you need to do (1 << (b-1)).
Bit 0 is the first bit. 2^0 = 1

Two, if you pass [] as the bits, the
result will be ‘true’.

Which is of course the correct answer.
All the bits in the empty set are in any bitfield.
There is no bit in the empty set hat are not in a given bitfield.

Pascal J. Bourguignon wrote:

Albert S. [email protected] writes:

One, you need to do (1 << (b-1)).
Bit 0 is the first bit. 2^0 = 1

I stand corrected.

Two, if you pass [] as the bits, the
result will be ‘true’.

Which is of course the correct answer.
All the bits in the empty set are in any bitfield.
There is no bit in the empty set hat are not in a given bitfield.

All right, all right, you win :wink: I now see that [].all? too returns
‘true’.

(So my attempt at dissuading you from using parenthesis failed
miserably. Say, won’t you consider dropping them for humanitarian
reasons?)

Hi –

On Thu, 29 Jan 2009, Rob B. wrote:

Given permissions bit mask of 0001C0200F02000000000 where each bit

def check_permissions(*bits)
bits.all? {|bit| self[bit].nonzero? }
end
end

Thanks for the reminder about Integer#[], one of those cool Ruby
things that come back and re-surprise me with their coolness
occasionally :slight_smile:

David


David A. Black / Ruby Power and Light, LLC
Ruby/Rails consulting & training: http://www.rubypal.com
Coming in 2009: The Well-Grounded Rubyist (The Well-Grounded Rubyist)

http://www.wishsight.com => Independent, social wishlist management!

On Thu, 29 Jan 2009, Pascal J. Bourguignon wrote:

(bits . inject(true) { | r , b |
ary
end

def check_permissions(*bits)
bits.all? {|bit| self[bit].nonzero? }
end
end

Simplier? A class, two methods, a lot of operations, a lot of memory!

as_bits is just there to provide information for the sake of
understanding the examples. check_permissions is Rob’s answer to the
original question. It looks about as simple as it could be: very
idiomatic, clear, and concise.

If you don’t like adding to Integer you can make it functional-style
by adding a second argument.

David


David A. Black / Ruby Power and Light, LLC
Ruby/Rails consulting & training: http://www.rubypal.com
Coming in 2009: The Well-Grounded Rubyist (The Well-Grounded Rubyist)

http://www.wishsight.com => Independent, social wishlist management!

Hi –

On Thu, 29 Jan 2009, Albert S. wrote:

Which is of course the correct answer.
All the bits in the empty set are in any bitfield.
There is no bit in the empty set hat are not in a given bitfield.

All right, all right, you win :wink: I now see that [].all? too returns
‘true’.

I could imagine wanting to handle the [] case conservatively. If the
idea is to screen for permissions, and you want the default to be
rejecting the person, then you’d probably want to reject an empty set
of bits. [].all? does indeed always return true, so you’d have to do a
separate empty? test somewhere.

David


David A. Black / Ruby Power and Light, LLC
Ruby/Rails consulting & training: http://www.rubypal.com
Coming in 2009: The Well-Grounded Rubyist (The Well-Grounded Rubyist)

http://www.wishsight.com => Independent, social wishlist management!

David A. Black wrote:

Thanks for the reminder about Integer#[], one of those cool Ruby
things that come back and re-surprise me with their coolness
occasionally :slight_smile:

Cool indeed… pity it doesn’t take a range however:

9547[0…-1]
TypeError: can’t convert Range into Integer

oops.

Thanks guys, the answers provided contained what I was looking for. I
had hacked together something that worked but seemed so un-ruby like.
This is nice looking very fast code.

Andy


From: Clifford H. [mailto:[email protected]]
Sent: Wed 1/28/2009 10:37 PM
To: ruby-talk ML
Subject: Re: Bitwise question

David A. Black wrote:

Thanks for the reminder about Integer#[], one of those cool Ruby
things that come back and re-surprise me with their coolness
occasionally :slight_smile:

Cool indeed… pity it doesn’t take a range however:

9547[0…-1]
TypeError: can’t convert Range into Integer

oops.