What characters are more common in my MD2 hash?

Question

The challenge is simple

Write a script that, when given a string input, will hash the string using the MD2 hashing algorithm, and then return either a positive integer or negative integer output based on which character set below is more common in the resulting hash as a hexadecimal string:

01234567 - (positive)
89abcdef - (negative)

• The input will always be a string, but may be of any length up to 65535
• The entire input, whitespace and all, must be hashed
• For the purposes of this challenge, the integer 0 is considered neither positive nor negative (see tie output)
• The more common set is the one who's characters are more common within the 32 character hexadecimal hash string
• Your output may contain trailing whitespace of any kind, as long as the only non-whitespace characters are a valid truthy or falsey output
• In the event of a tie, where the hexadecimal string contains exactly 16 characters from each set, the program should output a 0

I/O Examples

Input: "" (Empty String)
Hash: 8350e5a3e24c153df2275c9f80692773
Output: 1

Input: "The quick brown fox jumps over the lazy cog" (Without quotes)
Hash: 6b890c9292668cdbbfda00a4ebf31f05
Output: -1

Input: "m" (Without quotes)
Hash: f720d455eab8b92f03ddc7868a934417
Output: 0

Winning Criterion

This is code-golf, fewest bytes wins!

Answer 1

Octave, 35 bytes

@(s)diff(hist(hash('md2',s),+'78'))

*Requires the latest version of Octave (at least 4.2).

Computes histcounts of the hash string with it's center of bins are 7 and 8 then computes the difference of counts.

Answer 2

Mathematica, 43 bytes

Tr@Sign[15-2#~Hash~"MD2"~IntegerDigits~16]&

Outputs the number of digits in 01234567 minus the number of digits in 89abcdef.

Answer 3

JavaScript (ES6), 731 bytes

This monster is implementing the MD2 algorithm, so it's embarrassingly long. Based on js-md2 by Chen Yi-Cyuan.

let f =

m=>{L=x=s=b=0,n=m.length,M=[],X=[],C=[],S=[...atob`KS5DyaLYfAE9NlSh7PAGE2KnBfPAx3OMmJMr2bxMgsoem1c8/dTgFmdCbxiKF+USvk7E1tqe3kmg+/WOuy/ueqloeZEVsgc/lMIQiQsiXyGAf12aWpAyJzU+zOe/95cD/xkws0iltdHXXpIqrFaqxk+4ONKWpH22dvxr4px0BPFFnXBZZHGHIIZbz2XmLagCG2Alra6wufYcRmFpNEB+D1VHoyPdUa86w1z5zrrF6iYsUw1uhSiECdPfzfRBgU1Satw3yGzBq/ok4XsIDL2xSniIlYvjY+ht6cvV/jsAHTny77cOZljQ5KZ3cvjrdUsKMURQtI/tHxrbmY0znxGDFA`].map(c=>c[O='charCodeAt']());for(l=1;l-2;){for(j=19;j--;)M[j]=M[16+j]||0;for(i=s;i<16;x++)L=(x-n||(b+=i-s,s=i-16,l=2),C[i]^=S[(M[i++]=x<n?m[O](x):16-(b&15))^L]);for(i=0;i<l;i++){for(j=16;j--;)X[32+j]=(X[16+j]=(i?C:M)[j])^X[j];for(t=j=0;j<18;t=t+j++&255)for(k=0;k<48;)t=X[k++]^=S[t]}}for(i=16,n=-i;i--;)n+=!(X[i]&8)+!(X[i]&128);return n}

console.log(f(''))
console.log(f('The quick brown fox jumps over the lazy cog'))
console.log(f('m'))

Answer 4

Python 2 + Crypto, 108 99 93 91 87 78 bytes

Python doesn't have a native builtin for MD2.

from Crypto.Hash import*
lambda s:sum(x<'8'for x in MD2.new(s).hexdigest())-16

_{Saved 12 bytes thanks to @ovs.
Saved 9 bytes thanks to @FelipeNardiBatista.}

Answer 5

Java 8, 173 bytes

-4 thanks to dzaima

-128 thanks to Oliver, this is basically his answer now.

a->{String h="";for(byte b:java.security.MessageDigest.getInstance("MD2").digest(a.ge‌​tBytes()))h+=h.forma‌​t("%02x",b);return h.codePoints().filter(c->c>47&&c<56).count()-16;}

Positive for truthy. Negative for falsy. 0 for 0.

Answer 6

PHP, 50 Bytes

prints 1 for truthy and -1 for false and 0 for a tie

<?=preg_match_all("#[0-7]#",hash(md2,$argn))<=>16;

PHP, 58 Bytes

prints 1 for truthy and -1 for false and 0 for a tie

<?=16<=>strlen(preg_filter("#[0-7]#","",hash(md2,$argn)));

Answer 7

Tcl + Trf package, 79

package require Trf
puts [expr [regexp -all \[0-7\] [hex -m e [md2 $argv]]]-16]

Try it online. (Thanks @Dennis for adding Tcl to TIO.)

Answer 8

PHP, 56 bytes

while($i<32)${hash(md2,$argn)[$i++]>'7'}++;echo$$_<=>16;

Answer 9

Java 137 130 124 123 bytes

a->{int c=32;for(int b:java.security.MessageDigest.getInstance("MD2").digest(a.getBytes()))c-=(b>>6&2)+(b>>2&2);return c;}

Test it online!

Basically, for each byte, we're asked to check against its 4th and 8th least significant bits. I don't go through the hex representation at all. So it seemed only natural to start playing with bits.

Values <0 are falsey, values >0 are truthy, value 0 is neither truthy or falsey. The usual truthy and falsey can't be applied to Java this time (because it can't be true or false or 0 with the rule if(<truthy>)), so I took the liberty to declare as such.

Saves

1. 137 -> 130 bytes: golfed by using bit operations, removing 2 everytime I get a "falsy" bit.
2. 130 -> 124 bytes: more bitwise operations
3. 124 -> 123 bytes: replaced byte by int in the for loop declaration.