Given two arrays of non-negative integers \$A = [A_1,A_2,\ldots,A_n]\$ and \$R = [R_1,R_2,\ldots,R_n]\$ which are equal in length, return an array which has the element \$A_1\$ repeated \$R_1\$ times, then element \$A_2\$ repeated \$R_2\$ times, all the way up to \$A_n\$.
Standard loopholes are forbidden. As this is code-golf, the shortest program wins.
| \$A\$ | \$R\$ | Output |
|---|---|---|
[1,2,3] |
[1,2,3] |
[1,2,2,3,3,3] |
[6,0,0,6] |
[5,1,1,0] |
[6,6,6,6,6,0,0] |
[100,100] |
[0,0] |
[] |
(concat.).zipWith replicate
These built-ins have long names, but I haven't found anything shorter.
(*Zxx*).flat
xx is Raku's replication operator. It produces a list of a number of copies of its left argument given by its right argument. (Very conveniently, though irrelevant for this problem, the left-hand side is re-evaluated the requested number of times, so for example rand xx 10 will produce a list of ten different random numbers.)Z is the zip "meta-operator." It can be prepended to any other operator to produce a new operator that zips two lists together using the original operator..flat flattens the list of replicated lists.
lambda*a:sum(map(lambda x,*y:x*y,*a),())
-1 thanks to Jonathan Allan!
-1 thanks to loopy walt!
lambda a,b:[0,*b]*1**b[:1]@[(),*zip(a)]
Of course, there is also the repeat builtin.
Beside the usual bending over backwards to avoid the explicit numpy import this uses the fact that if we write \$c_i = [b_i]\$ then, formally, \$c_1 a_1 + c_2 a_2 + ... + c_n a_n\$ is a dot product. It is mildly tricky to set up an array of equal length sequences because numpy will just create an additional dimension and create an array of scalars. Here we prepend a different length sequence, forcing a ragged array (after using zip to create a list of singleton tuples). To match we must also prepend one value to the length array. The 1**b[:1] factor, mathematically a nop, is there to trigger a coercion cascade to numpy arrays-
¨£ẋf
This two-byte solution should work, but it seems to bork when the length is an array only of zeros: Try it Online!
repelem and repelems, and their usage are different.
\$\endgroup\$
f(A,R)=[A[i<=∑_{n=1}^{[1...R.length]}R[n]][1]fori=[0...R.total][2...]]
Might post an explanation if I feel like it.
-3 bytes thanks to G B
->a,b{a.fl{[_1]*b.sh}}
-2 bytes thanks to G B
->a,b{a.flat_map{[_1]*b.shift}}
b.shift for repeating
\$\endgroup\$
=IFERROR(TEXTSPLIT(CONCAT(REPT(A1:C1&"|",A2:C2)),"|",,1),"")
Or, if you don't mind whole rows as input arrays, for 56 bytes:
=IFERROR(TEXTSPLIT(CONCAT(REPT(1:1&"|",2:2)),"|",,1),"")
Takes R as the first input.
A byte or 2 could be saved if output could be a 2D-array but I'm guessing a flat array is one of the points of the challenge.
cÈÇYgV
cÈÇYgV :Implicit input of arrays U=R & V=A
c :Flat map U
È :Pass each X at index Y through the following function
Ç : Map the range [0,X)
YgV : Index into V
z⟨kj₎t⟩ˢc
-2 bytes thanks to @DLosc.
-1 byte thanks to @Kroppeb
Given inputs [a1, …, an] and [r1, …, rn]:
z Zip: [[a1,r1], ..., [an, rn]]
⟨ ⟩ˢ For each sublist [ai, ri]
k Take [ai]
t Take ri
j₎ Juxtapose the list [ai] ri times to itself
c Concatenate into one list
=
\$\endgroup\$
z⟨kj₎t⟩ˢc with only 9 bytes Try it online!
\$\endgroup\$
k to get a list and not a simple integer.
\$\endgroup\$
f(*a,*b,n){for(;n;)~--*b?printf("%d ",*a):a++-b++-n--;}
0[b] unnecessarily ;-)
\$\endgroup\$
ＩΣＥθＥ§ηκι
Attempt This Online! Link is to verbose version of code. Explanation:
θ Array of values
Ｅ Map over values
η Array of counts
§ Indexed by
κ Current index
Ｅ Map over implicit range
ι Current value
Σ Concatenate arrays
Ｉ Cast to string
Implicitly print
The code defines an anonymous function that returns the expected result.
Nothing special here...
fn($A$R){$X=[]each$R as$k=>$v;if$v$X=&array_concat($X&array_fill($A[$k]$v))send$X;}
Just use it normally...
Example using the 2nd test case.
$fn = fn($A$R){$X=[]each$R as$k=>$v;if$v$X=&array_concat($X&array_fill($A[$k]$v))send$X;}
// should output: 6,6,6,6,6,0,0
echo &join(call $fn([6,0,0,6], [5,1,1,0]), ',');
Since it is an anonymous function, using call is required.
This code does exactly the same as the golfed version.
This is pretty close to pseudo-code.
Set $fn to an anonymous function($A, $R)
Begin.
Define the variable $result = [].
Loop through $R as value $value key $key.
Begin.
If $value then.
Begin.
Set $result to the result of calling &array_concat(
$result,
&array_fill($A[$key], $value)
).
End.
End.
Return the $result.
End.
A little further from pseudo-code...
$fn = fn($A, $R) => {
$result = [];
foreach $R as $key => $value {
if $value {
$result = &array_concat(
$result,
&array_fill($A[$key], $value)
);
}
}
return $result;
};
import sugar,sequtils
(x,y)=>concat zip(x,y).map x=>x[0].repeat x[1]
-3 bytes thanks to Michael Chatiskatzi
-p, 16 bytes{FA({aRLb}MZab)}
{FA({aRLb}MZab)} ; First argument = A = a
; Second argument = B = b
{ } ; Create a function that...
MZab ; maps a and b (as zipped pairs of elements) to...
{aRLb} ; a block that repeats the (first argument) by (second argument)...
FA( ) ; and flattens the resulting list
; (after which the list is implicitly printed out)
_RL B is the direct translation (_ is the same as {a} and B is the same as {b}), but you can save a byte if you flip the arguments: BRL_MZba. Then you can drop the outer curly braces, making it a full program rather than a function, using the -x flag to evaluate the inputs as lists: DSO
\$\endgroup\$