I need to rotate an array by 90 degrees, or have writefln figure that out.

I need, say:

0 4 5 6
0 0 0 0
0 0 0 0
0 0 0 0

But it's outputting:

0 0 0 0
4 0 0 0
5 0 0 0
6 0 0 0

int [4][4] data;
file.writeln(format("%(%-(%d %)\n%)", data));

On Wednesday, 18 April 2018 at 06:54:29 UTC, Chris Katko wrote:
> I need to rotate an array by 90 degrees, or have writefln figure that out.
>
> I need, say:
>
> 0 4 5 6
> 0 0 0 0
> 0 0 0 0
> 0 0 0 0
>
> But it's outputting:
>
> 0 0 0 0
> 4 0 0 0
> 5 0 0 0
> 6 0 0 0
>
> int [4][4] data;
> file.writeln(format("%(%-(%d %)\n%)", data));

You can transpose the matrix :)

On Wednesday, 18 April 2018 at 06:54:29 UTC, Chris Katko wrote:
> I need to rotate an array by 90 degrees, or have writefln figure that out.
>
> I need, say:
>
> 0 4 5 6
> 0 0 0 0
> 0 0 0 0
> 0 0 0 0
>
> But it's outputting:
>
> 0 0 0 0
> 4 0 0 0
> 5 0 0 0
> 6 0 0 0
>
> int [4][4] data;
> file.writeln(format("%(%-(%d %)\n%)", data));

Generally, the solution would be std.range.transposed. However, since you're using a int[4][4], that's not a range-of-ranges, and transposed don't work out of the box. This helper function should help:

T[][] ror(T, size_t N1, size_t N2)(ref T[N1][N2] arr)
{
    T[][] result = new T[][N2];
    foreach (i, e; arr) {
        result[i] = e.dup;
    }
    return result;
}

unittest
{
    import std.stdio;
    import std.range;

    int [4][4] data;
    data[2][3] = 4;
    writefln("%(%-(%d %)\n%)", data);
    writefln("%(%-(%d %)\n%)", data.ror.transposed);
}

--
  Simen

On Wednesday, 18 April 2018 at 07:15:47 UTC, Simen Kjærås wrote:
> On Wednesday, 18 April 2018 at 06:54:29 UTC, Chris Katko wrote:
>> I need to rotate an array by 90 degrees, or have writefln figure that out.
>>
>> I need, say:
>>
>> 0 4 5 6
>> 0 0 0 0
>> 0 0 0 0
>> 0 0 0 0
>>
>> But it's outputting:
>>
>> 0 0 0 0
>> 4 0 0 0
>> 5 0 0 0
>> 6 0 0 0
>>
>> int [4][4] data;
>> file.writeln(format("%(%-(%d %)\n%)", data));
>
> Generally, the solution would be std.range.transposed. However, since you're using a int[4][4], that's not a range-of-ranges, and transposed don't work out of the box. This helper function should help:
>
> T[][] ror(T, size_t N1, size_t N2)(ref T[N1][N2] arr)
> {
>     T[][] result = new T[][N2];
>     foreach (i, e; arr) {
>         result[i] = e.dup;
>     }
>     return result;
> }
>
> unittest
> {
>     import std.stdio;
>     import std.range;
>
>     int [4][4] data;
>     data[2][3] = 4;
>     writefln("%(%-(%d %)\n%)", data);
>     writefln("%(%-(%d %)\n%)", data.ror.transposed);
> }
>
> --
>   Simen

That makes sense why transpose wouldn't work for my arrays!

So you're saying if I used [][] (dynamic array) that's a range of ranges, and it would work?

Why is it you have to rework your templates for static vs dynamic ranges? Thanks!

On Thursday, 19 April 2018 at 10:10:41 UTC, Chris Katko wrote:
> That makes sense why transpose wouldn't work for my arrays!
>
> So you're saying if I used [][] (dynamic array) that's a range of ranges, and it would work?

Yup. Static arrays can't be ranges, since popFront must mutate the length, and the length is a part of the type for static arrays.


> Why is it you have to rework your templates for static vs dynamic ranges? Thanks!

Unless I answered this question above, I'm not entirely sure what you're asking.

Anyways, for matrix work (which seems to be what you're doing), I would suggest taking a look at https://github.com/libmir/mir-algorithm. I haven't actually used it myself, but seems to be a very good and comprehensive library for this purpose.

--
  Simen

On 4/18/18 3:15 AM, Simen Kjærås wrote:
> On Wednesday, 18 April 2018 at 06:54:29 UTC, Chris Katko wrote:
>> I need to rotate an array by 90 degrees, or have writefln figure that out.
>>
>> I need, say:
>>
>> 0 4 5 6
>> 0 0 0 0
>> 0 0 0 0
>> 0 0 0 0
>>
>> But it's outputting:
>>
>> 0 0 0 0
>> 4 0 0 0
>> 5 0 0 0
>> 6 0 0 0
>>
>> int [4][4] data;
>> file.writeln(format("%(%-(%d %)\n%)", data));
> 
> Generally, the solution would be std.range.transposed. However, since you're using a int[4][4], that's not a range-of-ranges, and transposed don't work out of the box. This helper function should help:
> 
> T[][] ror(T, size_t N1, size_t N2)(ref T[N1][N2] arr)
> {
>      T[][] result = new T[][N2];
>      foreach (i, e; arr) {
>          result[i] = e.dup;
>      }
>      return result;
> }
> 
> unittest
> {
>      import std.stdio;
>      import std.range;
> 
>      int [4][4] data;
>      data[2][3] = 4;
>      writefln("%(%-(%d %)\n%)", data);
>      writefln("%(%-(%d %)\n%)", data.ror.transposed);
> }


A version without allocating:

T[][N2] ror(T, size_t N1, size_t N2)(ref T[N1][N2] arr)
{
    T[][N2] result;
    foreach (i, ref e; arr) {
        result[i] = e[];
    }
    return result;
}

...

writefln("%(%-(%d %)\n%)" data.ror[].transposed); // need the slice operator here

Keep in mind, you can't simply assign a variable to data.ror[], as the backing goes away immediately (OK to use as an rvalue though). And you must keep data in scope as long as you are using the result of data.ror.

-Steve