Programming Puzzles & Code Golf Stack Exchange is a question and answer site for programming puzzle enthusiasts and code golfers. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top


Given a set of integers, find the sum of all positive integers in it.


  • t – number of test cases [t < 1000]
  • On each of next t lines, an integer N [-1000 ≤ N ≤ 1000]


The program should output the sum of all the positive integers.

Check your code in online judge


Score is equal to size of source code of your program except symbols with ASCII code ≤ 32.

Here is the best score list: Python Best Scores (Best score is 29)

share|improve this question

closed as unclear what you're asking by cat, rink.attendant.6, Dr Green Eggs and Iron Man, Downgoat, muddyfish Apr 23 at 22:47

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question.If this question can be reworded to fit the rules in the help center, please edit the question.

There are a lot more codegolf challenges at Don't see the point of duplicating them here though. – hallvabo Feb 24 '11 at 12:22
why is this question tagged python.. Are we interested in python solution only? – Aman ZeeK Verma Feb 24 '11 at 13:56
I don't think questions from contest sites should be posted here. – fR0DDY Feb 24 '11 at 14:01
I already did this one on SPOJ. A while ago they promoted all the Python2.6 answers to Python3 even though some of them wont run under Python3 and would be longer in Python3 - eg have to use int(input()) instead of input() and print(x) instead of print x. So I don't take SPOJ very seriously anymore. I'm tied with Tim Peters and that's good enough for me :) – gnibbler Feb 24 '11 at 21:22
I just want to point out, that skipping the T (number of ...errr...numbers (?) ) is not an option...since the testcases involve extra data after T numbers...your code will fail at SPOJ. Everyone (3 answers below) seemed to have cleverly skipped the first integer. – st0le Feb 25 '11 at 9:36

49 Answers 49

Whitespace, 0

I couldn't resist. S= space, T= tab, N = newline, all have ASCII codes <= 32.


Base64 encoded for easy copy & paste.

share|improve this answer
(+1) Nice program! A tiny "FWIW": 9 characters can be removed due to 9 instances of an unneeded S in the binary coding of a number. These are all in push-number-to-stack instructions of the form SSSS...N, where the 4th S codes a superfluous leading 0. (Of course this has no effect on the score.) – r.e.s. Mar 29 '12 at 12:18

Element, 17 characters plus 1 space

_'[_ 2:n;0>[n~+]]`

This is my first constructed language. It is designed to be very compact and human-readable. All of the instructions are one character long and perform a single function.

Element has two stacks and a hash as memory structures. The two stacks are called the main stack and the control stack. The main stack is where arithmetic, I/O, and hash manipulation occurs. The control stack is where logic operations occur, and this stack controls the while and for loops.

The basic idea behind Element is that there is a hash that stores numbers/strings, while the stack is used to perform calculations on these numbers. The results to these calculation can then assigned a certain place in the hash for future use. The different contents of the hash are called elements, so it is similar to an array but can have non-numerical names.

EDIT: You can find an interpreter for Element (written in Perl) here.

Here is the list of operators: In some of these examples, m and n represent numbers already on the stack.

text  --pushes the string "text" onto the main stack
'     --pops from main stack and pushes onto control stack
"     --pops from control stack and pushes onto main stack
#     --pops from main stack and destroys
[]    --FOR statement (view the top number number from control stack and eval those many times)
{}    --WHILE (loop until top number on control stack is 0)
(     --pops from main stack, removes first character, pushes the remaining string onto stack, and pushes the removed character onto stack
)     --pops from main stack, removes last character, pushes the remaining string onto stack, and pushes the removed character onto stack
~     --pops from main stack, pushes contents of the element with that name
+-*/%^ --pops two most recently named elements, adds/negates/multiplies/divides/modulates/exponentiates them, and places the result on the stack
mn;   --pops m and n and assigns element n the value of m
mn@   --pops m and n and moves mth thing in stack to move to place n in stack
m$    --pops m and pushs size of m onto the stack
mn:   --pops m and n and pushes m onto the stack n times
mn.   --pops m and n and pushes m concatonated with n
m?    --pops m and pushes 0 onto control stack if m is '0' or and empty string, else pushes 1 
\     --escapes out of next character, so it isn't an operator and con be pushed onto the stack
><=   --pops two numbers off of stack and tests, pushes 1 onto control stack if true and 0 if false
`     --pops from main stack and prints
&|    --pops two items from control stack, performs and/or respectively, and pushes result back onto control stack
!     --pops a number off of control stack, pushes 1 if 0 or empty string, 0 otherwise
_     --inputs a word and pushes onto main stack
m,    --pops m from main stack, coverts it to char and pushes, converts to num and pushes
Newlines and spaces separate different elements to be pushed onto the stack individually, but can pushed onto the stack using \

Here is a walkthrough of how the program works:

_'[    --take the first line of input, transfer it to the control stack, and start a for loop
_ 2:   --take one more line of input, and duplicate it so that there are two copies
n;     --take one copy and put into element n
0>     --push a zero onto the stack, remove the zero and the other copy of the input, and compare. A 1 will be placed on the control stack if the input was greater than zero, a 0 otherwise.
[      --starts another for loop if the comparison was true. This loop will be repeated once if the comparison was true and no times if it was false, so it is the same as an IF statement.
n~     --pushes n onto the main stack, then pops it ans replaces it with the contents of n, which is the number stored earlier
+      --takes the number and adds it to the running total, which is contained as the last item on the stack
]      --ends the inner for loop
]      --ends the outer for loop
`      --print the top item (also the only item) on the stack to output
share|improve this answer
An entry like this would be much improved with a pointer to a working environment. – dmckee Mar 26 '12 at 4:09
I don't think you understand what "human-readable" means. – wchargin Feb 25 '14 at 2:44
@WChargin he is used to Perl... – Caridorc Feb 10 '15 at 15:20
@WChargin Every language is unreadable until you learn it. ;) – Martin Ender Feb 10 '15 at 23:53

Perl, 31

share|improve this answer
Wouldn't using say make this just a bit shorter? It'd tie with the 29 character best. – Mr. Llama Apr 23 '12 at 15:52
No, because say isn’t built-in and (at least) requires a command-line switch which would count towards the character count. – Timwi Apr 23 '12 at 18:19
It can be shortened to 29 bytes by using $\ instead of $i: <>;$\+=$_*($_>0)while<>;print – Heiko Oberdiek Mar 5 '14 at 9:59

Ruby 1.9.2, 37

p eval [*$<].join.gsub(/\A\d+|-\d+|\n/, '+0')

Call like ruby scriptname file_with_ints .

share|improve this answer
I can't read much Ruby, but does that even read the number of test cases? – Joey Feb 25 '11 at 9:15
No it doesn't... – st0le Feb 25 '11 at 9:31
@st0le: Just noticed that apparently no solution currently solves the task. – Joey Feb 25 '11 at 23:49

Ruby, 52

t=gets.to_i;s=0;t.times{i=gets.to_i;s+=i if i>0};p s
share|improve this answer
I don't see where you are printing the value of s. – Dogbert Feb 27 '11 at 18:27
yeah I forgot – david4dev Feb 27 '11 at 19:21

Haskell, 58

Properly operates on only t integers. Haven't run it against Spoj because I just don't care to register there.

f (x:l) = take x l
main = interact $ show . sum . f . map (max . lines
share|improve this answer
What are "t integers"? – wchargin Feb 10 '15 at 20:25

code in C 89 characters

x="%d";  main(b,a,t)  {  
    {  scanf(x,&a); a>0?b+=a:a; }  printf(x,b-1);
       return 0; }

I tried a lot to reduce my code less than 63 bytes, but i can reduce it only to 89 bytes. Please help me to reduce it to 63 bytes or even less.

share|improve this answer
1) I have counted 90 characters. 2) return 0; is not necessary, the for cycle may be contracted to for(scanf(x,&t);t--;scanf(x,&a),a>0?b+=a:a); == which make that 78 characters... – V-X Mar 5 '14 at 11:34
Doesn't compiles with gcc 4.8.1 error: initializer element is not computable at load time x="%d" – manav m-n Jul 21 '14 at 8:49

Perl, 33

<>;while(<>){$i+=$_ if$_>0}print$i

Although the space is necessary, so it seems odd not to count it. Oh well, the rules is the rules.

Hmm. I could probably get away with using a variable name which doesn't count towards the total either. The thing is, I'm not sure how I'd paste the code then.

share|improve this answer
Just show them as $^A-$^Z, but beware that many of that variables have special meanings. – ninjalj Feb 24 '11 at 20:58

Clojure, 71

(reduce + (filter pos? (map #(Integer/parseInt %) (next (line-seq *in*)))))
share|improve this answer
This doesn't produce any output, and fails because *in* is not a, as required by line-seq. – John Cromartie Jul 11 '13 at 15:24
Also it ignores the number of lines input t. – John Cromartie Jul 11 '13 at 15:39

In memoriam Dennis M. Ritchie

unix 57¹72:

n=$(head -n1 i); echo $(($(head -n $((n+1)) i | tail -n $n | grep -v "-" | tr '\n' '+')0))

assuming i is the file, containing the ints.

¹) was wrong, included the number of lines, and added 1 line too less.

echo $(($(cat i | head -n $(head -n1 i) | grep -v "-" | tr '\n' '+')0))

share|improve this answer

Haskell, 51

main = interact $ show . f . lines
f (x:l) = foldl (+) 0 $ map read l

(extra spaces for clarity, since they don't count)

Haskell is ... interesting, since you tend to get programs with a significant number of necessary spaces.

share|improve this answer
You forgot a filter (>0). – FUZxxl Apr 10 '11 at 14:58


x="%d";  main(b,a,t)  {  
{  scanf(x,&a); a>0?b+=a:0; }  printf(x,b-1);
return 0; }

After another big effort, code is one character less, please help me to reduce it more.

share|improve this answer
just edit the original answer next time – ratchet freak Jul 3 '11 at 19:18
remove the (return 0;) and ({} for for) – l0n3sh4rk Apr 21 '12 at 19:39
b,x="%d";main(a,t){for(scanf(x,&t);t--&&scanf(x,&a);)b+=(a>0)*a;printf(x,b);} <- 77 bytes – walpen Jun 9 '12 at 19:59
@walpen: they used the fact, that their "argc" like parameter was set to 1, your b is uninitialized... – V-X Mar 5 '14 at 12:18

Befunge-98 (24)

(Make sure you use an interpreter that can read negative numbers (seems to be a somewhat common bug, but RcFunge works))


Perl (25)

(Perl allows control characters in variable names, I named my variable ^B (ASCII 2) so that it does not count towards the goal.)


(Normal variant (27 chars)):

share|improve this answer
i first ignored your perl answer when i saw the variable naming and completely missed the excellent one below it – ardnew Jul 9 '13 at 22:55

APL (10)



  • ⍳⎕: read a line, gives a list [1..N] for user's input N
  • ¨: for each element in this list... (i.e. do N times)
  • 0⌈⎕: read a line, return the maximum of 0 and the entered N
  • We now have a list with all positive Ns the user entered, and 0s where the user entered something negative.
  • +/ gives the sum of this list.
  • The result is output by default (because we're not doing anything else with it).
share|improve this answer

Mathematica: 18 16

share|improve this answer
Nice function, but how does this handle the specified newline-separated input? How does it not incorporate the number of test cases parameter t as part of the sum? How does it sum only up to the number of test cases specified, even if more are given? – Jonathan Van Matre Mar 5 '14 at 23:08

PowerShell, 44

share|improve this answer



sample usage

q){0+/x(&)x>0} 1 -1 2 3 -1
share|improve this answer

befunge, 3524


with a little inspiration by seeing marinus answer, I've also managed 24 characters. but I've got a completely different approach.

share|improve this answer

PYTHON 2.x, 50 chars

print sum(i for i in (r() for j in range(r())) if i>0)
share|improve this answer

C, 70 72 characters


The results on the SPOJ site definitely seem unreal - I have no idea how to get this down to 63.

However, 68 characters is reachable on some compilers by abusing undefined behaviour. The following works on x86 Linux with 32-bit gcc, on which all arguments are passed on the stack.

share|improve this answer

Clojure, 108

(let [[n & m] (->> *in* line-seq (map read-string))]
  (->> m (take n) (filter pos?) (apply +) println))

I really wish I could avoid the part, since it costs 24 chars itself. But AFAIK there's no facility to read lines from STDIN without it.

share|improve this answer

Perl, 20

I know it is old and trivial, but the Perl answer can be still improved:

#!perl -p
share|improve this answer
This is awesome! But what does }{ mean/do? – daniero Jul 19 '15 at 13:40


using namespace std;
int main()
    int c,n,s=0;cin>>c;
cout<<s;return 0;

115 characters long. Need to optimize it to 90. Any suggestions ?

share|improve this answer
Just the standard tricks: The return is unnecessary in standard C++ or C99, there is an implicit return 0 in main. By making the variables global you can drop the =0 initialization. Finally, for(;;) is the same number of characters as while() but you get two extra places to put an expression in. – han Mar 25 '12 at 11:10
This is already old, but also, writing std:: before cin and cout and getting rid of the using namespace std; can save 5 more characters. – Morwenn Jul 3 '13 at 11:35

PHP, 71

share|improve this answer

Python: (92 characters)

t = int(raw_input())
n = [int(raw_input()) for i in range(t)]
print(sum([n[i] for i in range(t) if n[i]>0]))
share|improve this answer
Using a=raw_input and r=range and using a() and r() later can save quite a few characters. – Morwenn Jul 3 '13 at 11:37

scala 55 54:

println ((for (r <- 1 to readInt;
    if i>0) yield i)sum)
share|improve this answer


void main()
int n;
for(int i=0;i<n;i++)
printf("sum of positive numbers is %d",sum);
share|improve this answer
Welcome to CodeGolf.SE! If you look at the other answer, you'll see that they have formatted code and a minimal header noting the implementation language; on more complicated challenges many also have notes on the implementation and any limits or surprises in the code. Without some of this, you answer is unlikely to be well received. – dmckee Oct 12 '11 at 15:28
I counted the chars, added the indentation to make code-layout work, and removed the decoration of the output. Oh - now I have to count again. :) – user unknown Oct 13 '11 at 14:35
Added language name. There's a lot of room for reductions here - sum can be reduced to s, the output string can just be "%d", etc. – Gareth Oct 31 '12 at 9:17

excel, 27


count t in A1, rest of data a2 and down

share|improve this answer

Ruby, 42

gets.to_i.times { 
    s+=i if i>0
p s

(based on david4dev's answer)

Best score for Ruby on spoj:

share|improve this answer

45 chars in python

for i in j()*[0]:
print c
share|improve this answer
How you counted that? It gives me 54 characters. – manatwork Jul 2 '13 at 7:02
@manatwork, this question has non-standard scoring rules which don't count whitespace. – Peter Taylor Jul 2 '13 at 11:00
Oops, sorry. I missed that. Thanks, @PeterTaylor. – manatwork Jul 2 '13 at 11:39

Not the answer you're looking for? Browse other questions tagged or ask your own question.