views:

3077

answers:

33

It's Sunday, time for a round of code golf!

Challenge

Write the shortest source code by character count to determine if an input number is a "happy prime", "sad prime", "happy non-prime", or "sad non-prime."

Input

The input should be a integer that comes from a command line argument or stdin. Don't worry about handling big numbers, but do so if you can/want. Behavior would be undefined for input values less than 1, but 1 has a definite result.

Output

Output should print the type of number: "happy prime", "sad prime", "happy non-prime", or "sad non-prime." The trailing newline is optional.

Examples

$ happyprime 139
happy prime
$ happyprime 2
sad prime
$ happyprime 440
happy non-prime
$ happyprime 78
sad non-prime

Definitions

Just in case your brain needs a refresher.

Happy Number

From Wikipedia,

A happy number is defined by the following process. Starting with any positive integer, replace the number by the sum of the squares of its digits, and repeat the process until the number equals 1 (where it will stay), or it loops endlessly in a cycle which does not include 1. Those numbers for which this process ends in 1 are happy numbers, while those that do not end in 1 are unhappy numbers (or sad numbers).

For example,

  • 139
  • 1^2 + 3^2 + 9^2 = 91
  • 9^2 + 1^2 = 82
  • 8^2 + 2^2 = 68
  • 6^2 + 8^2 = 100
  • 1^2 + 0^2 + 0^2 = 1

Prime Number

A prime number is an integer greater than 1 and has precisely two divisors: 1 and itself.

Happy Prime

A happy prime, is therefore a number that is both happy and prime.

Answer Selection

Obviously the answer will be the shortest source code by character count that outputs the specified results in all cases that I test. I will mark the answer once the next (community decided) code golf challenge comes along, so we can focus all our energies on that one. :)

Decision

Well, it looks like the there is a new code golf in town and it has been about a week since this question was posted, so I've marked the shortest source code as the answer (gnibbler's 64 character Golfscript solution). That said, I enjoyed both the 99 character Mathematica solution by belisarius and the cryptic 107 character dc solution by Nabb.

To all others, great work! I've never had so many programming language environments on my computer. I hope everyone has learned some new, dirty tricks for their favorite language.

Reuse

I've re-published some of the code produced by this competition as an example for a script I wrote to test various programs against a reference implementation for auto-grading. The README in that directory explains where the source code comes from and states that all code is re-used under the CC BY-SA 2.5 license (as stated in SO's legal section). Each directory is labeled with your display name at the time of the submission.

If you have a problem with your code being re-used in this fashion or the attribution, let me know and I will correct the error.

+5  A: 

Ruby 1.9

169 168 146 chars

 h={1=>'happy'};s=->x{y=0;(y+=(x%10)**2;x/=10)while x>0;h[y]||(h[y]='sad';s[y])}
 $><<s[n=$*[0].to_i]+" #{'non-'if '1'*n=~/^1?$|^(11+?)\1+$/}prime"

If we use p instead of $><<, the code is shortened by 2 characters

Usage:

$ ruby happyprime.rb 139 happy prime $ ruby happyprime.rb 2 sad prime


Non golfed:

hash = {1->'happy'}
is_happy = lambda do |number|
  #sum = number.scan(/\d/).reduce(0){|acum, digit| acum + digit.to_i ** 2 }
  sum=0; 
  while (number > 0)
      sum+= (number%10)**2
      number/=10
  end
  return hash[sum] if hash[sum] # If 1, or if cycled and hash contains the number already
  h[sum] = 'sad'
  return is_happy.call(sum)
end
number = ARGV[0].to_i
string = ""
string += is_happy.call(number) # either 'happy' or 'sad'
string += is_prime(number) ? " non-prime" : "prime"
puts string

Where the is_prime method is left as an exercise to the reader ;)

Chubas
Nice use of regular expressions as primality check!
Gabe
You can save 1 character by getting rid of the () for that if statement (leaving a single space behind) as * is already higher precedence than =~.
shanna
+4  A: 

Python 2.6

happy.py: 280 314 333 chars, 14 lines.

import re
def q(z):
 while z!=1:z=sum((int(a)**2 for a in `z`));yield z
def h(g):
 l=[]
 while 1:
  try:z=g.next()
  except:return 'happy '
  if z in l:return 'sad '
  l.append(z)
p=lambda n:not re.match(r'^1$|^(11+?)\1+$','1'*n)
n=int(input())
print h(q(n))+'non-prime'[4*p(n):]

Usage:

$ echo 139 | python happy.py
happy prime
$ echo 2 | python happy.py
sad prime
$ echo 440 | python happy.py
happy non-prime
$ echo 1234567 | python happy.py
sad non-prime

--

Readable version:

import re, sys

def happy_generator(z):
    while z != 1:
        z = sum((int(a)**2 for a in str(z)))
        yield z

def is_happy(number):
    last = []
    hg = happy_generator(number)
    while True:
        try:
            z = hg.next()
        except StopIteration:
            return True

        if z in last:
            return False
        last.append(z)

def is_prime(number):
    """Prime test using regular expressions :)"""
    return re.match(r'^1?$|^(11+?)\1+$', '1'*number) is None

n = int(sys.argv[1])

print "%s %sprime" % (('sad','happy')[is_happy(n)], ('non-','')[is_prime(n)])
Seth
Couple of minor things I noticed at a brief glance:1: `while True` == `while 1`. 2: Don't need a space after the final print statement. 3: Rename the `ge` function to a single char.
sdolan
If you use the `match` function (instead of `search`), you don't need the `^` anchor in the regexp.
Roberto Bonvallet
This is a shorter `p`: `p=lambda n:not re.match(r'1?$|(11+?)\1+$','1'*n)`
Roberto Bonvallet
Since `h` is used only for indexing, it could return `0` and `1` instead of bools.
viraptor
+9  A: 

C, 188 187 185 184 180 172 171 165

h(c,C,r,p){for(;C>1&&C%++p;);for(;c;c/=10)r+=c%10*(c%10);r&~5?h(r,C,0,1):printf(
"%s %sprime",r-1?"sad":"happy",p>=C&C>1?"":"non-");}main(c){h(c,c,0,scanf("%d",&c));}

$ ./a.out
139
happy prime

$ ./a.out
2
sad prime

$ ./a.out
440
happy non-prime

$ ./a.out
78
sad non-prime

This is one recursive function that never issues a return but either calls itself or prints output when it's done. The recursive function sums squared digits and determines prime-ness in two for loops. The scanf returns 1 which is put as an argument to h(), saving one ; and one 1 (and at the cost of having to use prefix ++p instead of postfix p++ which would make p>C possible instead of p>=C)

r&~5 is 0 for 1 4 5, of which 1 indicates happiness and the others sadness.

Next attempt: drop h() and make main() recursive.

mvds
Hmmm, the golfed version seems to give me nothing but segfaults on two machines, the non-golf version works except 1 gives a floating point exception. Ideas? (compiled with `gcc -o hp happyprime.c`)
mjschultz
Golfed version does not compile at all for me (clang 1.0). It tells me that the second argument to main must be `char**`.
dreamlax
fixed. Although with people specifying language versions (or custom golfing-languages) I would rather keep it at 178 and just specify platform and compiler: 32 bit linux using gcc.
mvds
As long as I know that it requires a 32-bit compiler that is fine, are there any other `gcc` args I need? I still get an exception with the `1` case from the prime detection routine (initial call is `p(1,0)` leading to a modulo operation with 0 in the divisor).
mjschultz
No other args that I use (although if possible I would opt for a lot of `-D.=...` args). I'll have a look if I can blend in the non-primeness of 1.
mvds
+4  A: 

Java: 294 286 285 282 277 262 260 chars


  • Update 1: replaced BigInteger#isProbablePrime() by regex. Saved 8 chars.

  • Update 2: replaced && by & (oops). Saved 1 char.

  • Update 3: refactored s a bit. Saved 3 chars.

  • Update 4: the test on n!=1 was superfluous. Saved 5 chars.

  • Update 5: replaced regex by for loop and refactored happy for loops little bits. Saved 15 chars.

  • Update 6: replaced int/Integer by long/Long. Saved 2 chars.


import java.util.*;class H{public static void main(String[]a){long n=new Long(a[0]),p=n>1?1:0,s,d=1;while(++d<n)if(n%d<1)p=0;for(Set c=new HashSet();c.add(n);n=s)for(s=0;n>0;s+=d*d,n/=10)d=n%10;System.out.printf("%s %sprime",n>1?"sad":"happy",p>0?"":"non-");}}

With newlines:

import java.util.*;
class H{
 public static void main(String[]a){
  long n=new Long(a[0]),p=n>1?1:0,s,d=1;
  while(++d<n)if(n%d<1)p=0;
  for(Set c=new HashSet();c.add(n);n=s)for(s=0;n>0;s+=d*d,n/=10)d=n%10;
  System.out.printf("%s %sprime",n>1?"sad":"happy",p>0?"":"non-");
 }
}
BalusC
Shouldn't that be `new Char[p]`? Anyway, good job. You beat me
TheLQ
@BalusC, Character Count = 286. :-/
st0le
@BalusC, I can save 3 chars. :) I can't seem to `Save my Edit` some proxy error on my side, so i'll post it as a comment. Please Check it and if it's any good update the answer.
st0le
import java.util.*;class H{public static void main(String[]a){int n=new Integer(a[0]),p=n,s,d;Set c=new HashSet();for(;n!=1n=s)for(s=0;n>0;d=n%10,s+=d*d,n/=10);System.out.printf("%s %sprime",n==1?"happy":"sad",new String(new char[p]).matches(".?|(..+?)\\1+")?"non-":"");}}
st0le
@st0le: the trailing newline shouldn't count. It's really 285. As per your suggestion, yes that's better.
BalusC
@BalusC, 1 more byte. Move the `Set c=new HashSet()` into the first part of the `for(;c.add(n);)` loop. that saves a semicolon. :) `for(Set c=new HashSet();c.add(n);n=s)`
st0le
@BalusC,Haha, Ok, 7 bytes more. Instead of `new String(new char[p])` use `(""+new char[p])`. :D
st0le
@st0le: I saved after all 15 chars. Thanks you for the hints :)
BalusC
st0le
@st0le: 2 more chars off, those parentheses are superfluous. Regardless, that gives a false negative for `139`.
BalusC
Change `Integer` to `Long`, `int` to `long`, save couple more...
st0le
also `p=n>1?1:0` can be changed to `p=n`, i think. Since `Behavior would be undefined for input values less than 1` So it ought to do. :D
st0le
@BalusC,ignore this comment, just making sure you get notified....:D
st0le
@st0le: 1 is valid input. It should become 0 for the case that `while` loop doesn't enter (so that `p>0` in the print is still valid for 1). Changing int to long is a good one, thought of it as well in previous stages, but it didn't work well with new char[n] when the regex was still there, so I didn't use it.
BalusC
+28  A: 

Mathematica 115 108 107 102 100 99 91 Chars

Da monkey learnt a few tricks (91 chars)

 Print[
       If[Nest[Plus@@(IntegerDigits@ #^2) &, #, 9] > 1, Sad, Happy ],
       If[PrimeQ@#, " ", " non-"], prime
      ] &

Invoke with %[7]

Edit 5 - 99 Chars/

Nine iterations is enough. Thanks @Nabb, @mjschultz

h = Print[
    If[Nest[Plus @@ (IntegerDigits@#^2) &, #, 9] > 1, "Sad ", "Happy "]
   , If[PrimeQ@#, "", "non-"], "prime"] &

Edit 4 - 100 Chars/

Same as Edit 3, replacing 10^2 by 99 (allowing 84 digits for input values) ... Thanks, @Greg

h = Print[
    If[Nest[Plus @@ (IntegerDigits@#^2) &, #, 99] > 1, "Sad ", "Happy "]
   , If[PrimeQ@#, "", "non-"], "prime"] &

Edit 3 - 102 Chars/

Reworked the loop again.

It is interesting that the recursion depth until eventually reaching 1 is bounded by (15 + Number of digits of the argument). See here

So for numbers with less than 85 digits (I think this limit is pretty well into the OP's "Don't worry about handling big numbers" consideration) the following code works

h = Print[
    If[Nest[Plus @@ (IntegerDigits@#^2) &, #, 10^2] > 1, "Sad ", "Happy "]
   , If[PrimeQ@#, "", "non-"], "prime"] &

I changed the "NestWhile" for the shorter "Nest", and so, instead of specifying a stop condition for the recursion, is enough to hardcode the desired recursion depth (10^2).

It is not very efficient, but that's the golfer's life :D

Edit 2 - 107 Chars/

Reworked the Sad/Happy assignment

h = Print[
     If[NestWhile[Plus @@ (IntegerDigits@#^2) &, #, # > 4 &] > 1,"Sad ","Happy "]
    ,If[PrimeQ@#, "", "non-"]
    , "prime"] &

All spaces/newlines except on literals are optional and added for readability

Explanation:

The

    NestWhile[Plus @@ (IntegerDigits@#^2) &, #, # > 4 &]

Recurses applying "function" [Add up sum of digits squared] until the result is 4 or less. The function has the property that it stagnates at "1", or enters the cycle {4, 16, 37, 58, 89, 145, 42, 20, 4, ...}.

So, when the outcome is "1", the number is "Happy" and when the outcome is "4", it is "Sad".

If the result is "2", the number is also SAD, because it'll enter the SAD cycle in the next iteration (2^2 = 4).

If the result is 3, the cycle is 3->9->81->65->61->37->58->89->145-> .... (Enters the SAD loop).

So, we can stop the recursion when the result is 4 or less, knowing that only a result of "1" will lead to a Happy number.

Perhaps other solutions may take advantage of this fact.

In fact, the results 5 and 6 also lead to SAD numbers, but that gives us only an efficiency boost and not a golfing advantage (I guess).

Edit 1 - 108 Chars/

Reworked the Loop Control logic

    h = Print[
        NestWhile[Plus@@(IntegerDigits@#^2) &, #, #>4 &] /.{1 →"Happy ",_→"Sad "}
          , If[PrimeQ@#, "", "non-"]
          , "prime"] &

Original - 115 Chars/

h = Print[
    If[NestWhile[Plus @@ (IntegerDigits@#^2) &, #, Unequal, All] == 1
        ,"Happy ", "Sad "],      
    If[PrimeQ@#, "", "non-"], "prime"] &

The statement

NestWhile[Plus @@ (IntegerDigits@#^2) &, #, Unequal, All]

performs the recursive application of the sums of the digits squared, until some value repeats. The "Unequal,All" part takes care of the comparison across the preceding values list. Finally returns the repeated value, which is "1" for Happy Numbers.

Sample run

h[7]
Happy prime
h[2535301200456458802993406410753]
Sad non-prime

Looping (Slightly changing the Print statement)

1 Happy non-prime
2 Sad prime
3 Sad prime
4 Sad non-prime
5 Sad prime
6 Sad non-prime
7 Happy prime
8 Sad non-prime
9 Sad non-prime
10 Happy non-prime
11 Sad prime
12 Sad non-prime
13 Happy prime
belisarius
Oi, you're going to make me install Mathematica just so I can verify this! (Though I am fully convinced it is correct.)
mjschultz
Why do you use `10^2` instead of `100` or `99`, both of which are shorter?
Greg
@Greg yep, tnx ... I started using 10^99 for allowing greater input values and forgot to change that when came down to 85 digits ... Anyway the stack crashed with very big numbers :(
belisarius
belisarius: I'd bump it all the way down to `9` which puts you at 99 characters total. It still managed to evaluate `2535301200456458802993406410753` in no time on my system.
mjschultz
@mjschultz The required depth is not a monotone function. For example, calculating h[30] requires 11 iterations. The max depth value for numbers under 1000 is 15. So, 9 is way toooo low.
belisarius
+1, nice explanation!
Alex. S.
belisarius
belisarius: Thanks for doing that, but as a student I have access to Mathematica. However, I'm sure others do not, so it's still useful!
mjschultz
`9` iterations should be sufficient up to `10^974` digits.
Nabb
@Nabb I guess you've a proof there :). I just followed a few experiments and the Wikipedia statement. Without a proof I didn't want to assume that, and reaching the SAD cycle may take as much as 16 iterations for very small numbers (15999, for example)
belisarius
@belisarius Your test condition is `>1`, i.e. happy numbers will have reached 1, not unhappy numbers some arbitrary value. The smallest values for n steps: s(5) = 7, s(6) = 356, s(7) = 78999. log₁₀(s(8)) ~ 975 (=78999/81). log₁₀(log₁₀(s(9))) ~ 974. log₁₀(log₁₀(log₁₀(s(10)))) ~ 974. 9 iterations is therefore sufficient for everything up to `10^10^10^974`, or `10^10^974` digits. I suppose I missed a step earlier.
Nabb
belisarius
+7  A: 

Perl, 140 chars

sub h{$_==1&& happy||$s{$_}++&& sad
||do{$m=0;$m+=$_**2for split//;$_=$m;&h}}$n=$_=pop;
die h,$",(1x$n)=~/^1?$|^(11+?)\1+$/&&"non-","prime\n"

Linebreaks are optional.

hobbs
Ah, perl my first language. Interesting use of `die` for output, made me fix up my checker script.
mjschultz
Sorry, can be changed if necessary, but it's common practice here if nobody says "absolutely must go to stdout" :)
hobbs
if you have perl 5.10+, you can also -E and say. also, you can drop \n and just use newline.
sreservoir
+1  A: 

Python 2.6, 300 298 294 chars

Different from previous answer in that this doesn't use regex.

I'm sure there's some way of shortening my h(x) function, but I'm still learning Python so I've got no idea.

p(x) returns True if it's a non-prime. h(x) returns True if it's happy. I've done the t = True so that it shortens the character count when doing the truth check.

x=input()
def p(x):
 if x==1 or 1 in [1 for i in range(2,x) if x%i==0]: return True
def h(x):
 l=[]
 while x not in l:
  l.append(x)
  x=sum([int(i)**2 for i in str(x)])
 if 1 in l: return True
if h(x):print'happy',
elif not h(x):print'sad',
if p(x):print'non-prime'
elif not p(x):print'prime'
vlad003
Close, you'll need to fix it for the `1` case (1 being non-prime). Also you can strip the `[]` in the `sum()`. You could also probably just do `if h(x):` and the like.
mjschultz
Can someone please test this for me. I'm getting a weird problem: if I run this as a .py file from CLI, it says that 1 is prime. If I write the exact same function in an interpreter, it says 1 is not prime...
vlad003
So I think I fixed the `1` case this time. I'm still confused a bit about the whole logic of it though :P
vlad003
+64  A: 

dc - 107 chars

$ cat happyprimes
[happy][sad]?dsI[0rsc[lcI~rscd*+lc0<H]dsHxd4<h]dshx[r]sr1=rP[ ][ non-]_1lI[1-d2>rdlIr%0<p]dspx-2=rP[prime]p
$ echo 1  |dc happyprimes
happy non-prime
$ echo 139|dc happyprimes
happy prime
$ echo 2  |dc happyprimes
sad prime
$ echo 440|dc happyprimes
happy non-prime
$ echo 78 |dc happyprimes
sad non-prime
Nabb
Boy, that is nice a cryptic, but it works!
mjschultz
The hell? (15 chars)
TheLQ
What planet is this from?
Marko
About `dc`: [Wikipedia](http://en.wikipedia.org/wiki/Dc_%28computer_program%29), [manpage](http://linux.die.net/man/1/dc).
ShreevatsaR
Yeah I've already read it - still doesn't look human.
Marko
Code golfing is the third use for dc, the other two are doing modular exponentiation and implementing bc on top of dc (and using bc instead of dc ;P ).
ninjalj
this is hardcore
Jasconius
+12  A: 

C#, 380 378 374 372 364 363 315 280 275 274 Chars

By replacing the recursive function with nested loops, I was able to bring the stroke count to a respectable 280 (100 less than the original).

class P{static void Main(string[]a){var s=new System.Collections.Generic.HashSet<int>();int n=int.Parse(a[0]),p=n>1?4:0,c,d=1;for(;++d<n;)if(n%d<1)p=0;for(;n>1&s.Add(n);n=c)for(c=0;n>0;c+=d*d,n/=10)d=n%10;System.Console.Write((n>1?"sad":"happy")+" non-prime".Remove(1,p));}}

Here it is with whitespace:

class P
{
    static void Main(string[] a)
    {
        var s = new System.Collections.Generic.HashSet<int>();
        int n = int.Parse(a[0]),
            p = n > 1 ? 4 : 0,
            c,
            d = 1;
        // find out if the number is prime
        while (++d < n)
            if (n % d < 1)
                p = 0;
        // figure out happiness
        for (; n > 1 & s.Add(n); n = c)
            for (c = 0; n > 0; c += d * d, n /= 10)
                d = n % 10;

        System.Console.Write(
            (n > 1 ? "sad" : "happy")
            + " non-prime".Remove(1,p)
            );
    }
}
Gnafoo
I was able to cut out quite a bit by doing some obvious golfing, not changing the algorithm at all.
Gabe
+1 for "figure out happiness"
belisarius
Added some syntactical updates to save 5 more characters, have a bigger update which removes the need for `p` altogether, will post later.
NickLarsen
shaved a character by rewriting the output
NickLarsen
+2  A: 

Clojure, 353 318 298 261 230 chars

(defn h[x m](cond(= x 1)"happy "(m x)"sad ":else(recur(reduce +(for[n(map #(-(int %)48)(str x))](* n n)))(assoc m x 1))))(println(let [x (read)](str(h x{})(if(re-matches #"^1$|^(11+)?\1+"(apply str(repeat x\1)))"non-""")"prime")))

ptimac:clojure pti$ clj  happy.clj 139
CP=/Users/pti/playpen/clojure:/Users/pti/Library/Clojure/lib/clojure.jar:/Users/pti/Library/Clojure/lib/jline.jar:/Users/pti/Library/Clojure/lib/clojure-contrib.jar
happy prime
ptimac:clojure pti$ clj  happy.clj 440
CP=/Users/pti/playpen/clojure:/Users/pti/Library/Clojure/lib/clojure.jar:/Users/pti/Library/Clojure/lib/jline.jar:/Users/pti/Library/Clojure/lib/clojure-contrib.jar
happy non-prime
ptimac:clojure pti$ clj  happy.clj 2
CP=/Users/pti/playpen/clojure:/Users/pti/Library/Clojure/lib/clojure.jar:/Users/pti/Library/Clojure/lib/jline.jar:/Users/pti/Library/Clojure/lib/clojure-contrib.jar
sad prime
ptimac:clojure pti$ clj  happy.clj 78
CP=/Users/pti/playpen/clojure:/Users/pti/Library/Clojure/lib/clojure.jar:/Users/pti/Library/Clojure/lib/jline.jar:/Users/pti/Library/Clojure/lib/clojure-contrib.jar
sad non-prime

I am leaning on the clojure contrib for the primes sequence. I wonder if using for loops would be shorter than recursion?

I read up on the regex prime number check. It's awesome and removes 30 chars and my dependency on clojure.contrib. I also refactored the command line parsing somwhat and inlined a function.

Pre golf(somewhat outdated):

(defn h[x m]
  (cond
   (= x 1) "happy "
   (m x) "sad "
   :else (recur
               (reduce +
                 (for [n (map #(- (int %) 48) (str x))] (* n n))) 
               (assoc m x 1))))

    (println
      (let [x (read)]
        (str
           (h x{})
           (if (re-matches #"^1$|^(11+)?\1+"(apply str(repeat x \1)))
             "non-"
             "")
           "prime")))
Peter Tillemans
Nice, I've never worked with clojure before, but I had to change the `(nth *command-line-args* 1)` to `(nth *command-line-args* 2)`. (1 being your script's name).
mjschultz
Yes that seems to have to do with the bash scripts and the change from the Script class to clojure.main for script support.
Peter Tillemans
+3  A: 

Perl, 135C

sub h{my$s;$s+=$_**2for split//,pop;($s-4)?($s-1)?&h($s):1:0}$a=pop;
print h($a)?happy:sad,$",(1x$a)=~/^1?$|^(11+?)\1+$/&&"non-",prime

Combined C and Perl

sepl
It looks like you have "non-prime" and "prime" flip-flopped.
mjschultz
Whoops! Fixed and still 135C.
sepl
+2  A: 

F#, 249 chars

let n=stdin.ReadLine()|>int
let rec s x=seq{yield x;yield!string x|>Seq.sumBy(fun c->(int c-48)*(int c-48))|>s}
printfn"%s %sprime"(if s n|>Seq.take 99|>Seq.exists((=)1)then"happy"else"sad")(if[2..n/2]|>Seq.exists(fun d->n%d=0)then"non-"else"")
Brian
+1  A: 

Python (285 270 269 246 241 247 240 237 chars, 21 20 21 18 19 lines)

n=input()
s='prime'
for i in range(2,n):
    if n%i==0: 
        s='non'+s
        break
f=list(str(n))
g=set()
while n!=1:
    n=sum([int(z)**2 for z in f])
    if n in g:
        s='sad '+s
        break
    else:
        f=list(str(n))
        g.add(n)
else:
    s='happy '+s
print s

EDIT: Yes, the number went up, there was a bug :-P

Chinmay Kanchi
using `n=input()` will really cut down on your char count. And it seems to be accepted in the rules.And some things to fix: since you use a `for` loop it appends lots of `non`'s to the `s`. Also, `1` is not prime so you'd need to fix that.
vlad003
Yes, I realised that as soon as I posted. Edited now.
Chinmay Kanchi
You don't need the `int()`. `input()` already evaluates the input and makes it an int if it already is one.
vlad003
I fixed the loop problem and the unnecessary int(). You just have to fix the `1` case.
vlad003
Oh, I didn't know that. Also managed to cut down one more line by putting the if and statement inside a single line. Also, the lambda was unnecessary, so, one more line and a bunch of characters gone there...
Chinmay Kanchi
Did your changes get posted? I don't see them here. Oh, and the `int(n)` in your `for` loop is unnecessary because `n` is already an integer. **EDIT:** Nevermind; the changes are here now.
vlad003
Done :). I don't think it can be shortened any more now without serious mental effort. It's the shortest Python solution here by a long way.
Chinmay Kanchi
You need that `break` statement in your for loop or else you get this result for 440: `happy nonnonnonnonnonnonnonnonnonnonnonnonnonnonprime`
vlad003
Yes, just realised that.
Chinmay Kanchi
Made it even shorter using `while: else:`.
Chinmay Kanchi
Change `range(2,n/2+1)` to `range(2,n-1)` to save some characters? Who cares about efficiency? Don't you need a `-` after `non`?
trinithis
Actually, since `xrange(start, stop)` is non-inclusive of the last number, just `xrange(2,n)` will do :)
Chinmay Kanchi
Rather than `s='non'+s;break` you can use `s='non-prime'` and save a few chars.
Gabe
@Gabe: but he'd still need the `break` statement in there so it doesn't do the `nonononononononononononprime` problem.
vlad003
vlad003: The `break` was only necessary because he was concatenating with the previous value. Since `s='non-prime'` is idempotent, there's no need for the `break`.
Gabe
Sorry; I misread your comment.
vlad003
+9  A: 

Python 2.6: 194 180 chars, 4 lines

import re
s=lambda n,l:0if n==1 else n in l or s(sum(int(a)**2for a in str(n)),l+[n])
n=input()
print['happy','sad'][s(n,[])],'non-'*bool(re.match(r'1?$|(11+?)\1+$','1'*n))+'prime'

The lexer being able to split 0if and 2for into two tokens each was a nice surprise to me :) (it doesn't work with else though)

Function s (sad) is recursive, and receives the list of previous numbers in the cycle as its second parameter. Primality is tested inline using the regexp trick.

By using the deprecated `n` syntax instead of str(n), one can further reduce the character count by 4 characters, but I choose not to use it.

Roberto Bonvallet
That looks awesome :P. And if you replace your `int(sys.argv[1])` with `input()` it'll bring your char count down to 180.
vlad003
Thanks! I didn't notice stdin was allowed.
Roberto Bonvallet
Nice! You beat my solution by 60 chars and 15 lines!
Chinmay Kanchi
+1, the regex for prime numbers is VERY nice!
Yassin
@Yassin: Thanks for mentioning that's what it did! I'm just learning programming and I didn't even touch regex yet so I didn't know what it did :-). Now that I look, a lot of the answers use that regex.
vlad003
@Yassin @vlad003 I just added a link to an explanation of the trick in another SO question.
Roberto Bonvallet
I just submitted a python answer that uses 169 chars, but it's totally different than this one, so I didn't put it here (i really like the regex trick)
thepandaatemyface
+5  A: 

Haskell 172

h s n|n`notElem`s=h(n:s)$sum[read[k]^2|k<-show n]|1`elem`s="happy "|0<1="sad "
c n|n<2||any((0==).mod n)[2..n-1]="non-"|0<1=[]
y n=h[]n++c n++"prime"
main=readLn>>=putStr.y
trinithis
You must assume the input is greater than or equal to 1. (1 is a happy non-prime)
Gabe
Yeah, wasn't sure at first. Fixed now.
trinithis
The newline is not necessary after the output. so feel free to get take two characters.
mjschultz
+5  A: 

Javascript 244 250

function h(n){p=n=n<2?10:n;a=",";w="";s=[];while((y=a+s.join(a)+a).indexOf(a+n+a)<0){s.push(n);k=""+n;n=0;for(i=0;i<k.length;)c=k.charAt(i++),n+=c*c}w+=y.indexOf(",1,")<0?"sad ":"happy ";for(i=2;i<p;)p=p%i++?p:0;w+=p?"":"non-";return w+"prime"}

The above code should work in browsers without additional fancy functions and features (such as Array.prototype.indexOf and [] notation for strings), but I haven't tested it outside of Firefox.

Be aware that all but n are global variables (I'm just being cheap).

Usage

h(139) // returns "happy prime"
trinithis
Nice, I had to add a `}` at the end, but that is probably because I was using Firefox's Web Console. I'm impressed that Javascript and your code was able to handle `h(2535301200456458802993406410753)`!
mjschultz
Leaving the `}` out was a cut and paste mistake. Thanks.
trinithis
+3  A: 

C++, 258 231 230 227 chars

#include<iostream>
#define w while
int m,n,i,j,t=10;int main(){w(std::cin>>n){j=0,m=n;w(n>1){i=0;do i+=n%t*(n%t);w(n/=t);n=n*n+i;n=++j&0xFFFF?n:0;}i=1;w(m%++i&&j>1);std::cout<<(n?"happy":"sad")<<(i-m?" non-":" ")<<"prime\n";}}

not the best golfing language, gave it a good shot anyway. Most of this is straight C so would probably be shorter in C as well.

EDIT

Generally tidied it up, think it's pretty much at the limit now without a complete rerwrite.

Also forgot to add that this assumes that there are no numbers with a sequence with over 0xFFFF numbers which a pretty sensible assumption.

EDIT 2

fixed a bug. rearranged to remove the excessive calls to std::cout.

Bunnit
A few C golfing tricks which should shave off some: You could try to use the ternary conditional `cond?true:false` which also has useful low precedence properties, comparison for (non-)equality is best done using `a-b` rather than `a!=b`, and looping is shorter with `for` - although `while()` takes as much room as `for(;;)`, the latter gives space to put statements in without the cost of an extra `;`. And there are a few more chars to win still.
mvds
@mvds: thanks for the tips.
Bunnit
I know this will cost you a few character, but I get stuck in the last `while` loop when the input is `1`.
mjschultz
@mjschultz: thanks fixed it now. thanks for doing my debugging for me as well. ;)
Bunnit
+2  A: 

Scala, 253 247 246 chars

object H{def main(a:Array[String]){var s=Set(0)
val n=a(0)toInt
def r(m:Int):String={val k=""+m map(c=>c*(c-96)+2304)sum;if(k<2)"happy"else if(s(k))"sad"else{s+=k;r(k)}}
printf("%s %sprime",r(n),if(n<2|(2 to n-1 exists(n%_==0)))"non-"else"")}}

Probably there is some room for improvements. The damn test for 1 as non-prime costs 6 chars :-(

Landei
A: 

pseudocode 16 char:

Do the dc answer

it analyst
No input parms? :)
belisarius
This does not meet the "source code" requirement.
mjschultz
"Do the dc answer" that sounds like the worst dance move ever...
bta
It seems that every golf contest begets this answer. I wonder if some day somebody will write the REAL code for doing this.
belisarius
+1  A: 

Python, 169 168 158 157 166 164 162 chars, 4 lines

l=n=input()
while l>4:l=sum(int(i)**2for i in str(l))
print['sad','happy'][l==1and str(n)!=1],
print['non-',''][n!=1 and sum(n%i==0for i in range(1,n))<2]+"prime"

Takes a number from stdin and doesn't muck around with regexes like the other python answer, although I must admit that is pretty cool. I could also shave off 6 chars by using backticks instead of the str-function, but let's play nice.

EDIT: Fixed a bug with 1 being a prime, which bumped up the charcount by 10. I figure there must be a more concise way than mine for doing this.

EDIT 2: Apparently, python 2.6 allows print[1, 2] without a space between the two.

EDIT 3: Used another calculation for the happy numbers

thepandaatemyface
I do appreciate the non-use of regex, but yours seems to detect `1` as prime and `7` as sad. (Also, strictly speaking the last comma should be a `+`.)
mjschultz
Fixed 'em both!
thepandaatemyface
Hmmm, it still claims that `7` is a "sad prime", instead of "happy prime". Also, you could remove the space before both the `for`s as well.
mjschultz
+3  A: 

VBA 245 characters

Good starter, will trim if time allows. Its only my 2nd go at code golf!

Public Sub G(N)
Dim Z, D, X, O
X = N
Z = N
Do Until Z = 1 Or X > N Or X = 0
    X = 0
    For D = 1 To Len(CStr(Z))
        X = X + CLng(Mid(CStr(Z), D, 1) ^ 2)
    Next D
    Z = X
Loop
If Z = 1 Then O = "Happy" Else O = "Sad"
D = 2
Do
    If N / D = Int(N / D) Then O = O & " Not Prime": Debug.Print O: Exit Sub
    D = D + 1
Loop While D < N
O = O & " Prime"
Debug.Print O
End Sub
Kevin Ross
Also, VisualBasic is not a great language for golfing. (English keywords, not free form, etc.)
Andreas Rejbrand
I do golf challenges without regard to char count just to practice :P
Radu
+2  A: 

MATLAB - 166 chars

function happyprime(a)
h={'non-prime','prime'};
h=h{isprime(str2num(a))+1};
for i=1:99
   a=num2str(sum(str2num((a)').^2));
end
s={'Sad ','Happy '};
[s{(str2num(a)==1)+1},h]

Usage

happyprime 139
ans =
Happy prime
Jonas
+12  A: 

Python - 127 chars

Beating both perl answers at this moment!

l=n=input()
while l>4:l=sum(int(i)**2for i in`l`)
print['sad','happy'][l<2],'non-prime'[4*all(n%i for i in range(2,n))*(n>1):]

I also ported this answer to GolfScript and it is just over 1/2 the size!

gnibbler
That is pretty sweet. I wouldn't expect a python answer to be shorter than perl.
mjschultz
The slice trick is cool!
Roberto Bonvallet
You can use the backticks instead of `str`, as I explain in my answer above.
Roberto Bonvallet
@Roberto, good catch
gnibbler
I never knew a space isn't required before that first `for`.
vlad003
+1 Never thought I'd see python beat perl in a terse contest. Incidentally, not gzip, bzip2, or even xz can not compress this smaller than it already is.
Seth
@Seth: Most compression algorithms have some small amount of overhead and thus are miserable at compressing tiny files.
Brian
+7  A: 

MATLAB 7.8.0 (R2009a) - 120 characters

Whitespace, newlines, and comments added for readability

n=input('');
s=n;
c={'happy ','sad ','non-'};
while s>6,
  s=int2str(s)-48;
  s=s*s';                    %'# Comment to fix code highlighting
end;
disp([c{[s<2 s>1 ~isprime(n)]} 'prime'])
gnovice
I'm curious about the last line ("disp") ... Could you explain it, please?
belisarius
@belisarius: `c` is a cell array of strings, and its contents are being indexed by a 1-by-3 logical array with elements `s<2`, `s>1`, and `~isprime(n)`. For every true entry in the index array, the contents of the corresponding cell are dumped into a set of square brackets, which concatenates them into one string along with the word `'prime'`. It is then displayed using `disp`.
gnovice
Any reason for the `while s>9,` line? It seems to cause invalid output when the input is `7`. Would `while s>6,` work?
mjschultz
@mjschultz: Oops! That was an old typo I had in an earlier version. Fixing...
gnovice
@mjschultz @gnovice That's why I asked. I thought that the "disp" line made the magic to fix that. Tnx for the explanation!
belisarius
Clever use of int2str!
Jonas
+4  A: 

J: 113 characters

h=.1=$:@([:+/[:*:@"."0":)`]@.(e.&1 4)
1!:2&2;(({&('sad ';'happy '))@h,({&('non-prime';'prime'))@(1&p:))".(1!:1]3)

 

$ echo -n 7 | jc happy.ijs
happy prime
$ echo -n 139 | jc happy.ijs
happy prime
$ echo -n 2 | jc happy.ijs
sad prime
$ echo -n 440 | jc happy.ijs
happy non-prime
$ echo -n 78 | jc happy.ijs
sad non-prime
cobbal
+2  A: 

PHP 217 Chars

$t=$argv[1];for($z=$t-1,$p=1;$z&&++$p<$t;)$z=$t%$p;$f=array(1);while(!in_array($t,$f,1)){$f[]=$t;$t=array_reduce(str_split($t),function($v,$c){return $v+=$c*$c;});}print($t<2?"happy ":"sad ").(!$z?"non-":"")."prime";

Usage:

$ php -r '$t=$argv[1];for($z=$t-1,$p=1;$z&&++$p<$t;)$z=$t%$p;$f=array(1);while(!in_array($t,$f,1)){$f[]=$t;$t=array_reduce(str_split($t),function($v,$c){return $v+=$c*$c;});}print($t<2?"happy ":"sad ").(!$z?"non-":"")."prime";' 139
happy prime
Kevin
This does extremely well for everything except for the `1` case.
mjschultz
Oops! Fixed, added 3 characters (total 217 now) to handle the proper prime check for 1.
Kevin
+2  A: 

Javascript, 192 190 185 182 165 158 chars

The prime checking runs from 2 to square root of N. I wasted few chars there...

In one line:

for(x=2,y=m=n=prompt();x*x<y&&n%x++;);for(s={};!s[m];m=p)for(s[m]=1,p=0;m;m=(m-=k=m%10)/10,p+=k*k);alert((m-1?'sad':'happy')+(n-1&&x*x>y?' ':' non-')+'prime')

Formatted:

// Getting the number from the input and checking for primeness
// (ie. if after the loop x>y => n is prime)
for (x=2, y=m=n=prompt(); x*x<y && n%x++;)

// Checking for happiness
// the loop is broken out of if m is already encountered
// the m==1 after the loop indicates happy number
for(s={}; !s[m]; m=p)
    for (s[m]=1, p=0; m; m=(m -= k=m%10)/10, p+=k * k);

alert((m-1 ? 'sad' : 'happy') + (n-1 && x*x>y ? ' ' : ' non-') + 'prime')

Check: http://jsfiddle.net/TwxAW/6/

Marko Dumic
This fails in the case where `n=1` where it should say "happy non-prime" instead of "happy prime".
mjschultz
Thanks. I fixed that, making the code even smaller.
Marko Dumic
+16  A: 

GolfScript - 64 chars (works for 1)

~:@.{0\`{15&.*+}/}*1=!"happy sad "6/=@,{@\)%!},,2=4*"non-prime">

This program does n iterations to determine the happiness of the number, which is very wasteful for large numbers, but code-golf is not about conserving resources other than characters. The prime test is similarly inefficient - dividing n by all the values from 1 to n inclusive and checking that there are exactly two values with zero remainder. So while it is theoretically correct, running with really large numbers is not practical on real computers

GolfScript - 63 chars (fails for 1)

~:@9{0\`{15&.*+}/}*1=!"happy sad "6/=@,2>{@\%!},!4*"non-prime">
gnibbler
I get...odd behavior when the input is `1`. Is that from a limitation of golfscript or something else? (It says `../lib/golfscript.rb:353:in `select': private method `select' called for nil:NilClass (NoMethodError)`.)
mjschultz
@mjschultz, darn - `1` is a special case in that solution. One char extra for a better primality test
gnibbler
+2  A: 

Perl, 113 109 105 chars

Beating all Python answers at this moment! SCNR.

$n=$s=<>;$s=0,s/\d/$s+=$&*$&/ge while($_=$s)>4;die$s>1?sad:happy,$","non-"x(1x$n)=~/^1$|(^11+)\1+$/,prime
ninjalj
About time! Now see if you can beat the GolfScript :)
gnibbler
I'll pass on beating your Golfscript. Instead, I'll beat your python-fu with my sed-fu :P
ninjalj
+1  A: 

Python - 142 chars

I was playing around with this idea, but it turned out too long. Perhaps someone can find a way to make it shorter. Maybe it'll turn out better in Ruby. Should be fun to understand how it works anyway :)

n=input();L=[n];print"%s non-prime"[4*([1for f in range(1,n)if L.append(sum(int(x)**2for x in`L[-1]`))or n%f<1]==[1]):]%['sad','happy'][1in L]
gnibbler
+1  A: 

GNU sed, 146 125 chars

Run with sed -rf file. Using -r saves 5 backslashes.

Needs bc, printf, and a shell with support for here-strings.

h
s/^/printf %*s /e
s/^ $|^(  +)\1+$/non-/
s/ *$/prime/
x
:a
s/./+&*&/g
s//bc<<</e
tb
:b
s/^1$/happy/
s/^4$/sad/
Ta
G
s/\n/ /

GNU sed, 155 141 chars (needs neither printf nor here-strings)

Uses the more standard traditional yes and head instead of printf.

h
:a
s/./+&*&/g
s/.*/echo 0&|bc/e
tb
:b
s/^1$/happy/
s/^4$/sad/
Ta
x
s/^/yes|head -/e
s/\n//g
s/^y$|^(yy+)\1+$/non-/
s/y*$/prime/
x
G
s/\n/ /

GNU sed, 134 115 chars (slightly bad formatted output)

Slightly shorter version, doesn't respect output formatting (has extra spaces and a newline between happy/sad and (non-)prime).

h
:a
s/./+&*&/g
s//bc<<</e
tb
:b
s/^1$/happy/
s/^4$/sad/
Ta
p
g
s/^/printf %*s /e
s/^ $|^(  +)\1+$/non-/
s/$/prime/
ninjalj
A: 

Racket, 286 characters

Only issue is that 1 is prime.

(define d
  (lambda (n)
    (cond
      ((= n 0) 0)
      ((+
       (expt (modulo n 10) 2)
       (d (floor (/ n 10)))
       ))
      )
    )
  )

(define dd
  (lambda (n l)
    (cond
      ((= n 1) "happy ")
      ((for/or ([v l]) (= n v)) "sad ")
      ((dd (d n) (cons n l)))
      )
    )
  )

(define cg
  (lambda (n)
    (display (dd n '()))
    (display
    (if
     (for/and
       ([i (in-range 2 (floor (+ (sqrt n) 1)))]) (> (modulo n i) 0))
       "prime" "non-prime"
     ))
    )
  )
mangoDrunk
+1  A: 

PARI/GP

184 bytes

A Bit late to participate, but here's a short one.

a(x,v=[])=s=0;while(x,s+=(x%10)^2;x\=10);for(k=1,length(v),if(v[k]==s,return("sad ")));return(if(s==1,"happy ",v=concat(v,s);a(s,v)));f(x)=print(a(x),if(!isprime(x),"non-",""),"prime")

To use it. Write f(139)

st0le