Base-64 encoding--Just for the fun of it!

Ruby
1

Yes, there’s always:

require ‘base64’
Base64::encoe64(data)

But now and then it’s fun to write one’s own base64 encoder just for
the fun of it.

What I want to ask is this: How short can one get a function
definition that takes a string as input and returns a base-64
(MIME64-style, except for NO newlines/linefeeds–output is all one
giant line of encoded text in the character set ‘A’ to ‘Z’, ‘a’ to
‘z’, ‘0’ to ‘9’, ‘+’, and ‘/’)?

The shortest I’ve got it down to is 276 characters. WARNING: My email
client is gonna line-break this and mess it up. Just remove the
whitespace.)

def
e(s);c=(‘A’…‘Z’).to_a.join+(‘a’…‘z’).to_a.join+‘0123456789+/’;(t=s.unpack(‘C*’).inject([0,‘’,0]){|a,v|a[0]==0?[2,a[1]+c[v>>2,1],v16&48]:a[0]==2?[4,a[1]+c[v>>4|a[2],1],v4&60]:[0,a[1]+c[v>>6|a[2],1]+c[v&63,
1],0]})[1]+(t[0]==0?‘’:t[0]==2?c[t[2],1]+‘==’:c[t[2],1]+‘=’);end

If the above is too corrupted, I stuck it on my personal web site:
http://www.aarongifford.com/base64.rb.txt

The 276-character encoder works in Ruby 1.9.1 and Ruby 1.8.7 and
super-simplistic tests comparing results against the Base64 encoder
appear to work (as long as you strip out line breaks Base64 includes).

I’d love to hear suggestions to squeeze a few more characters out of it.

Ideas? Suggestions?

Having a bit too much fun fiddling with Ruby,
Aaron out.

2

This morning I realized I could shave off 3 characters by eliminating
one .join by adding () and converting + to | in the c= initialization.

Also, for Ruby 1.9 only, one can save 8 more characters since string
slices return 1-character strings in 1.9 instead of integers.

I updated http://www.aarongifford.com/base64.rb.txt so it now has the
original and the two improvements. And it adds a random data test on
Unix-based hosts supporting reading random data from /dev/urandom

Oh, I should add that performance is abysmal, but that obviously
wasn’t the goal. chuckle

Aaron out.

3

Aaron D. Gifford wrote:

(…)
The shortest I’ve got it down to is 276 characters. WARNING: My email
client is gonna line-break this and mess it up. Just remove the
whitespace.)

def
e(s);c=(‘A’…‘Z’).to_a.join+(‘a’…‘z’).to_a.join+‘0123456789+/’;(t=s.unpack(‘C*’).inject([0,’’,0]){|a,v|a[0]==0?[2,a[1]+c[v>>2,1],v16&48]:a[0]==2?[4,a[1]+c[v>>4|a[2],1],v4&60]:[0,a[1]+c[v>>6|a[2],1]+c[v&63,
1],0]})[1]+(t[0]==0?’’:t[0]==2?c[t[2],1]+’==’:c[t[2],1]+’=’);end
(…)
I’d love to hear suggestions to squeeze a few more characters out of it.

Ideas? Suggestions?

Having a bit too much fun fiddling with Ruby,
Aaron out.

Heh, in ruby 1.9 you can do:

c=(65…122).to_a.join(&:chr)+‘0123456789+/’

instead of
c=((‘A’…‘Z’).to_a|(‘a’…‘z’).to_a).join+‘0123456789+/’

hth,

Siep

4

On Mar 5, 1:11 am, “Aaron D. Gifford” [email protected] wrote:

The 276-character encoder works in Ruby 1.9.1 and Ruby 1.8.7 and
super-simplistic tests comparing results against the Base64 encoder
appear to work (as long as you strip out line breaks Base64 includes).

I’d love to hear suggestions to squeeze a few more characters out of it.

Ideas? Suggestions?

While lacking in the joy of code golf you may still be interested to
know about Radix

5

Thomas S. wrote:

On Mar 5, 1:11�am, “Aaron D. Gifford” [email protected] wrote:

The 276-character encoder works in Ruby 1.9.1 and Ruby 1.8.7 and
super-simplistic tests comparing results against the Base64 encoder
appear to work (as long as you strip out line breaks Base64 includes).

I’d love to hear suggestions to squeeze a few more characters out of it.

Ideas? �Suggestions?

While lacking in the joy of code golf you may still be interested to
know about Radix

Yes i am. And please ignore my previous post in this thread, it’s
rubbish.

Siep

6

On Thu, Mar 4, 2010 at 11:11 PM, Aaron D. Gifford [email protected]
wrote:

The 276-character encoder works in Ruby 1.9.1 and Ruby 1.8.7 and
super-simplistic tests comparing results against the Base64 encoder
appear to work (as long as you strip out line breaks Base64 includes).

def e(s);[s].pack(‘m’);end

Kirk H.

7

Yes, there’s always:

require ‘base64’
Base64::encoe64(data)

Actually that’s deprecated in 1.9 afaik. Somebody else already pointed
out that you can use Array#pack.

I’d love to hear suggestions to squeeze a few more characters out of it.

Yours:
c=(‘A’…‘Z’).to_a.join+(‘a’…‘z’).to_a.join+‘0123456789+/’

Mine (1.9 only):
c=[?A…?Z,?a…?z,?0…?9]’’+’+/’

That’s 23 characters.

Yours:
(t=s.unpack(‘C*’).inject([0,’’,0]){|a,v|a[0]==0?[2,a[1]+c[v>>2,1],v16&48]:a[0]==2?[4,a[1]+c[v>>4|a[2],1],v4&60]:[0,a[1]+c[v>>6|a[2],1]+c[v&63,1],0]})[1]+(t[0]==0?’’:t[0]==2?c[t[2],1]+’==’:c[t[2],1]+’=’)

Mine:
(s.unpack(“B*”)[0]+‘0’((6-s.size8%6)%6)).scan(/.{6}/).map{|e|c[e.to_i(2)]}’’+’=’((3-s.size%3)%3)

That’s another 104 characters off. So 127 characters less in total :slight_smile:

I also did only some very superficial tests, and mine doesn’t insert
newlines either. Though some clever gsub could do that easily.

Regards & thanks for the challenge :wink:
Stefan (aka apeiros)

8

On Fri, Mar 5, 2010 at 6:55 PM, Kirk H. [email protected] wrote:

def e(s);[s].pack(‘m’);end

Thanks for contributing something useful to me. I’ve always used the
‘base64’ module in the past, having overlooked that handy little
tidbit.

Aaron out.

9

On Sat, Mar 6, 2010 at 4:31 PM, [email protected] wrote:

Mine (1.9 only):
c=[?A…?Z,?a…?z,?0…?9]‘’+‘+/’

That’s 23 characters.

Gorgeous!

Mine:
(s.unpack(“B*”)[0]+‘0’((6-s.size8%6)%6)).scan(/.{6}/).map{|e|c[e.to_i(2)]}‘’+‘=’((3-s.size%3)%3)

Beautiful!

I love learning new things every day, and you’ve provided plenty for
me. Thanks for “golf” lessons. smile

Aaron out.

10

On Sat, Mar 6, 2010 at 4:31 PM, [email protected] wrote:

(s.unpack(“B*”)[0]+‘0’((6-s.size8%6)%6)).scan(/.{6}/).map{|e|c[e.to_i(2)]}‘’+‘=’((3-s.size%3)%3)

I notice you could shave off 2 more characters by using the same ?
character literal for the ‘0’ and the ‘=’ characters:
(s.unpack(“B*”)[0]+?0*((6-s.size8%6)%6)).scan(/.{6}/).map{|e|c[e.to_i(2)]}‘’+?=*((3-s.size%3)%3)

I hope you don’t mind, but I added your version to my file at:
http://www.aarongifford.com/base64.rb.txt

And I stole your character literal and range expansion (splat) ideas
to improve my longer algorithm a bit too.

Thanks again for teaching me about Range expansion inside method calls.

Aaron out.