Dada uma matriz 2D de números inteiros, vamos classificar suas linhas e colunas em blocos. Isso significa que você só precisa classificar uma linha ou coluna, mas aplicar as transformações necessárias para classificá-la em todas as outras linhas ou colunas da matriz 2D.

Regras

• A entrada será uma matriz 2D de números inteiros e um inteiro indexado em 1. Esse número inteiro representará a linha a ser classificada se o número for positivo ou a coluna a ser classificada se o número for negativo (ou o contrário que você desejar). Exemplo: Dada uma 4x3matriz (linhas x colunas), você pode classificar a segunda coluna com um -2argumento ou a terceira linha com um 3argumento. Este segundo argumento nunca será zero e seu valor absoluto nunca será maior que a dimensão correspondente da matriz.
• A saída também será uma matriz 2D de números inteiros com as transformações necessárias aplicadas para classificar a linha ou coluna especificada. Como alternativa, você pode simplesmente escrever a matriz em STDOUT.
• A matriz de saída terá a linha ou coluna especificada classificada em ordem crescente. Observe que quando você precisar trocar dois números seguidos, as colunas inteiras onde estão os números serão trocadas. E quando você precisar trocar dois números em uma coluna, as linhas inteiras onde estão os números serão trocadas.
• No caso em que o mesmo número aparecer várias vezes na linha / coluna a ser classificada, haverá várias soluções possíveis de acordo com a maneira como você troca os valores, apenas faça o mesmo com o restante das linhas / colunas a serem trocadas.

Exemplos

Positive indices for rows and negative indices for columns

[5  8  7  6                                  [1  3  2  4
 1  3  2  4   order by -3 (3rd column)  -->   9  6  3  0
 9  6  3  0]                                  5  8  7  6]

[5  8  7  6                                  [9  6  3  0
 1  3  2  4   order by -4 (4th column)  -->   1  3  2  4
 9  6  3  0]                                  5  8  7  6]

[5  8  7  6                                  [5  7  8  6
 1  3  2  4     order by 2 (2nd row)  -->     1  2  3  4
 9  6  3  0]                                  9  3  6  0]

[5  8  7  6                                  [6  7  8  5
 1  3  2  4     order by 3 (3rd row)  -->     4  2  3  1
 9  6  3  0]                                  0  3  6  9]

[1  2                                    [1  2     [3  2
 3  2]   order by -2 (2nd column)  -->    3  2] or  1  2]  (both are valid)

[7  5  9  7                                  [5  7  7  9     [5  7  7  9
 1  3  2  4     order by 1 (1st row)  -->     3  1  4  2  or  3  4  1  2
 9  6  3  0]                                  6  9  0  3]     6  0  9  3]

Este é o código-golfe , portanto, pode ganhar o código mais curto para cada idioma!

R , 55 bytes

function(x,n)`if`(n>0,x[,+x[n,]],x[+x[,-n],])
`+`=order

Experimente online!

Reatribui o +operador (na verdade uma função em R) à orderfunção, que retorna os índices de um vetor do menor para o maior. Então é apenas manipulação de array.

R , 55 bytes

f=function(m,i)"if"(i<0,t(f(t(m),-i)),m[,order(m[i,])])

Experimente online!

Alternativa à resposta de ngm ; uma função recursiva que foi inspirada na resposta de DimChtz

Matlab, 73 62 47 bytes

@(m,i){sortrows(m,-i) sortrows(m',i)'}{(i>0)+1}

Experimente Online!

-11 bytes graças a @ Giuseppe.

-15 bytes graças a @LuisMendo.

Japonês , 18 17 bytes

negativo para linhas e positivo para colunas

>0?VñgUÉ:ßUa Vy)y

Experimente online!

05AB1E , 25 24 14 bytes

diø}Σ¹Ä<è}¹diø

Whopping -10 bytes graças a @Emigna .

Usa uma entrada inteira positiva para classificar as linhas, negativas para as colunas.

Experimente online ou verifique todos os casos de teste .

Explicação:

di }      # If the (implicit) integer input is positive:
  ø       #  Swap the rows and columns of the (implicit) matrix input
          #   i.e. 3 and [[5,8,7,6],[1,3,2,4],[9,6,3,0]]
          #    → [[5,1,9],[8,3,6],[7,2,3],[6,4,0]]
Σ    }    # Sort the rows of this matrix by:
 ¹Ä       #  Take the absolute value of the input
          #   i.e. -3 → 3
   <      #  Decreased by 1 to make it 0-indexed
          #   i.e. 3 → 2
    è     #  And index it into the current row
          #   i.e. [5,8,7,6] and 2 → 7
          #   i.e. [5,1,9] and 2 → 9
          #  i.e. [[5,1,9],[8,3,6],[7,2,3],[6,4,0]] sorted by [9,6,3,0]
          #   → [[6,4,0],[7,2,3],[8,3,6],[5,1,9]]
          #  i.e. [[5,8,7,6],[1,3,2,4],[9,6,3,0]] sorted by [7,2,3]
          #   → [[1,3,2,4],[9,6,3,0],[5,8,7,6]]
¹di       # And if the integer input was positive:
   ø      #  Swap the rows and columns back again now that we've sorted them
          #   i.e. 3 and [[6,4,0],[7,2,3],[8,3,6],[5,1,9]]
          #    → [[6,7,8,5],[4,2,3,1],[0,3,6,9]]
          # (And implicitly output the now sorted matrix)

JavaScript (ES6), 90 bytes

t=m=>m[0].map((_,x)=>m.map(r=>r[x]))
f=(m,k)=>k<0?m.sort((a,b)=>a[~k]-b[~k]):t(f(t(m),-k))

Experimente online!

Quão?

JS não tem método de transposição nativo, portanto, precisamos definir um:

t = m =>              // given a matrix m[]
  m[0].map((_, x) =>  // for each column at position x in m[]:
    m.map(r =>        //   for each row r in m[]:
      r[x]            //     map this cell to r[x]
    )                 //   end of map() over rows
  )                   // end of map() over columns

Função principal:

f = (m, k) =>         // given a matrix m[] and an integer k
  k < 0 ?             // if k is negative:
    m.sort((a, b) =>  //   given a pair (a, b) of matrix rows, sort them:
      a[~k] - b[~k]   //     by comparing a[-k - 1] with b[-k - 1]
    )                 //   end of sort
  :                   // else:
    t(f(t(m), -k))    //   transpose m, call f() with -k and transpose the result

$k=2$

MATL, 17 bytes

y0>XH?!]w|2$XSH?!

Try it online!

Or verify all test cases

Explanation

y       % Implicit inputs: number n, matrix M. Duplicate from below: pushes n, M, n
0>      % Greater than 0?
XH      % Copy into clipboard H
?       % If true
  !     %   Transpose matrix. This way, when we sort the rows it will correspond
        %   to sorting the columns of the original M
]       % End
w       % Swap: moves n to top
|       % Absolute value
2$XS    % Two-input sortrows function: sorts rows by specified column
H       % Push contents from clipboard H
?       % If true
  !     %   Transpose again, to convert rows back to columns
        % Implicit end
        % Implicit display

APL (Dyalog Classic), 23 bytes

{⍺<0:⍉(-⍺)∇⍉⍵⋄⍵[;⍋⍺⌷⍵]}

Try it online!

Python 2, 71 70 bytes

f=lambda m,n:n<0and sorted(m,key=lambda l:l[~n])or zip(*f(zip(*m),-n))

Try it online!

If n is negative, the rows are sorted based on column n.

Otherwise the matrix is transposed, sorted the same way, and transposed back again.

Jelly, 12 bytes

ZṠ}¡
çị@ÞA}ç

Try it online!

C# (.NET Core), 186 bytes

(x,y)=>{Func<int[][],int[][]>shift=a=> a[0].Select((r,i)=>a.Select(c=>c[i]).ToArray()).ToArray();return y>0?shift(shift(x).OrderBy(e=>e[y-1]).ToArray()):x.OrderBy(e=>e[-y-1]).ToArray();}

Try it online!

Ungolfed:

    private static int[][] Blocksort0a(int[][] array, int sortingInstruction)
    {
        Func<int[][], int[][]> shift = a => a[0].Select((r, i) => a.Select(c => c[i]).ToArray()).ToArray();

        sortingInstruction++;

        array = sortingInstruction < 0 ? 
        shift(shift(array).OrderBy(e => e[-sortingInstruction]).ToArray()) 
             : 
        array.OrderBy(e => e[sortingInstruction]).ToArray();

        return null;
    }

The shift function we'll use twice, so a function variable will save space. The function iterates through the horizontal dimension of the array on index, and adds every item on that index in of each horizontal array to a new output array (horizontally) - much the same as in Arnoud's JS solution.

Now the ordering is simple, order horizontal array by number-at-index (argument -1), optionally shifting the array before and after sorting.

Seen how the question talks about arrays specifically, we convert to array a few times (very, very wasteful). Feeling a bit silly to use such a verbose language in code golf hehe.

C# (.NET Core), 142/139 138/135 bytes (and yet another -1 by Kevin)

(a,s)=>s<0?a.OrderBy(e=>e[~s]).ToArray():a.Select(f=>a[s-1].Select((v,j)=>new{v,j}).OrderBy(e=>e.v).Select(e=>f[e.j]).ToArray()).ToArray()

Try it online!

Ungolfed:

    private static int[][] Blocksort0b(int[][] array, int sortingInstruction)
    {
        if (sortingInstruction < 0) { return array.OrderBy(e => e[-sortingInstruction - 1]).ToArray(); }
        var rowIndices = array[sortingInstruction - 1].Select((value, index) => (value, index)).OrderBy(e => e.value);
        var newRow = new int[array[0].Length];
        for (var i = 0; i < array.Length; i++)
        {
            int horizontalIndexer = 0;
            foreach (var e in rowIndices)
            {
                newRow[horizontalIndexer++] = array[i][e.index];
            }
            array[i] = newRow.ToArray();
        }
        return array;
    }

New all-inline approach; negative answer still orders arrays by element-at-index. Otherwise, a collection of value-index-pair is created of the array-at-index and sorted by value. This effectively creates a collection of indices in order of having-to-be-added. Then for each array, the elements in the predetermined positions are selected. Quite some trimming of code and ugly, ugly, ugly **silently sobs** reuse of input parameters is involved, and there you go ... 142 bytes.

Again, the arrays argument is strictly enforced, adding quite some overhead for .ToArray() calls.

135 bytes claim, eh?! C# 7.2 inferred value-tuples would trim an additional three bytes, but tio.run doesn't allow. Therefor, this is the answer i decided to post for easy verification.

Java (OpenJDK 8), 326 bytes

(a,b)->{int l=a.length,w=a[0].length,k,m,t,i;if(b>0){for(i=0;i<w;i++){for(k=1;k<(w-i);k++){if(a[b-1][k-1]>a[b-1][k]){for(m=0;m<l;m++){t=a[m][k];a[m][k]=a[m][k-1];a[m][k-1]=t;}}}}}else{b*=-1;for(i=0;i<l;i++){for(k=1;k<(l-i);k++){if(a[k-1][b-1]>a[k][b-1]){for(m=0;m<w;m++){t=a[k][m];a[k][m]=a[k-1][m];a[k-1][m]=t;}}}}}return a;}

Try it online!

Well guys, this question was very frustrating for me, and I posted my answer KNOWING I was forgetting something, luckily we have legends like Kevin Cruijssen out here to help us out :)

Java (OpenJDK 8), 281 bytes

a->b->{int l=a.length,w=a[0].length,k,m,t,i;if(b>0)for(i=0;i<w;i++)for(k=0;++k<w-i;)for(m=0;a[b-1][k-1]>a[b-1][k]&m<l;a[m][k]=a[m][k-1],a[m++][k-1]=t)t=a[m][k];else for(b*=-1,i=0;i<l;i++)for(k=0;++k<l-i;)for(m=0;a[k-1][b-1]>a[k][b-1]&m<w;a[k][m]=a[k-1][m],a[k-1][m++]=t)t=a[k][m];}

Try it online!

Clean, 95 bytes

import StdEnv,Data.List,Data.Func
$n#f=if(n>0)transpose id
=f o sortBy(on(<)\u=u!!(abs n-1))o f

Try it online!

Kotlin, 192 bytes

{m:Array<Array<Int>>,s:Int->if(s<0){m.sortBy{it[-s-1]}}else{val a=Array(m[0].size){c->Array(m.size){m[it][c]}}
a.sortBy{it[s-1]}
(0..m.size-1).map{r->(0..m[0].size-1).map{m[r][it]=a[it][r]}}}}

Try it online!

Ruby, 69 bytes

->a,n{f=->x{[0,x.transpose,x][n<=>0]};f[f[a].sort_by{|x|x[n.abs-1]}]}

Try it online!

Red, 190 185 bytes

func[b n][t: func[a][c: length? a/1 a: to[]form a
d: copy[]loop c[append/only d extract a c take a]d]d: does[if n > 0[b: t b]]d
m: absolute n sort/compare b func[x y][x/(m) < y/(m)]d b]

Try it online!

Explanation:

f: func [ b n ] [
    t: func [ a ] [                            ; helper transpose function 
        c: length? a/1                         ; c is the length of the rows
        a: to-block form a                     ; flatten the list
        d: copy []                             ; an empty block (list)
        loop c [                               ; do as many times as the number of columns  
            append/only d extract a c          ; extract each c-th element (an entire column)
                                               ; and append it as a sublist to d
            take a                             ; drop the first element
        ] 
        d                                      ; return the transposed block (list of lists)
    ]
   d: does [ if n > 0 [ b: t b ] ]             ; a helper function (parameterless) to transpose 
                                               ; the array if positive n
   d                                           ; call the function  
   m: absolute n                               ; absolute n
   sort/compare b func[ x y ] [ x/(m) < y/(m) ]; sort the array according to the chosen column 
   d                                           ; transpose if positive n
   b                                           ; return the array  
]

My actual solution is 175 bytes long, but it doesn't work in TIO. Here it is, working normalyl in the Red console:

Red, 175 bytes

func[b n][d: does[if n > 0[c: length? b/1 a: to-block form b
t: copy[]loop c[append/only t extract a c take a]b: t]]d
m: absolute n sort/compare b func[x y][x/(m) < y/(m)]d b]

VBA (Excel), 205 bytes

Yay! 2nd longest byte count! I didn't completely lose :D

Golfed:

Sub d(a)
With ActiveSheet.Sort
  .SortFields.Clear
  .SortFields.Add Key:=IIf(a<0,ActiveSheet.Columns(Abs(a)),ActiveSheet.Rows(Abs(a)))
  .SetRange ActiveSheet.UsedRange
  .Orientation=IIf(a<0,1,2)
  .Apply
End With
End Sub

This sorts all the data on the open (active) worksheet using UsedRange... which can be buggy, but should only contain cells that have been edited.

UnGolfed:

Sub d(a)
  'Clear any Sort preferences that already exists
  ActiveSheet.Sort.SortFields.Clear
  'Use the column if A is negative, the row if A is positive
  ActiveSheet.Sort.SortFields.Add Key:=IIf(a < 0, ActiveSheet.Columns(Abs(a)), ActiveSheet.Rows(Abs(a)))
  'Set the area to sort
  ActiveSheet.Sort.SetRange ActiveSheet.UsedRange
  'Orient sideways if sorting by row, vertical if by column
  ActiveSheet.Sort.Orientation = IIf(a < 0, xlTopToBottom, xlLeftToRight)
  'Actually sort it now
  ActiveSheet.Sort.Apply
End Sub

Perl 6, 43 bytes

{($!=$_>0??&[Z]!!*[])o*.sort(*[.abs-1])o$!}

Try it online!

Curried function.

Explanation

{                                         } # Block returning function composed of
                                       o$!  # 1. Apply $! (transpose or not)
                     o*.sort(*[.abs-1])     # 2. Sort rows by column abs(i)-1
     $_>0??&[Z]                             # 3. If i > 0 transpose matrix
               !!*[]                        #    Else identity function
 ($!=               )                       #    Store in $!

Physica, 45 bytes

Very similar to Arnauld's JS answer.

F=>n;m:n<0&&Sort[->u:u{~n};m]||Zip@F#Zip@m#-n

Try it online!

How it works?

A more elaborate and visual explanation can be found in the linked answer.

F=>n;m:           // Create a function F that takes two arguments, n and m.
       n<0&&      // If n < 0 (i.e. is negative)
Sort[->u{~n};m]   // Sort the rows u of m by the result of the function u[~n].
                  // In short, sort by indexing from the end with n.
||    F#Zip@m#-n  // Else, apply F to Zip[m] and -n. Uses a new feature, binding.
  Zip@            // And transpose the result.

J, 32 bytes

f=.[/:({"1~<:)
g=.(f&.|:|)`f@.(0<])

Try it online!

Note: The g=. of the main verb doesn't count.

An explicit version for the same bytes

J, 32 bytes

4 :'y(]/:{"1)&.(|:^:(x<0))~<:|x'

Try it online!

Clojure, 91 bytes

(fn f[A i](if(< i 0)(sort-by #(nth %(- -1 i))A)(apply map list(f(apply map list A)(- i)))))

Argh, apply map list * 2.