Kata

6 kyu? Hmmm... I spent 2 days to solve this, and still not elegant comparing to other solutions. I solved other 6 kyu katas less than an hour sometimes. Is it only me or do you also think that difficulties need some review?

JavaScript: random tests still contain errors in both neighbourhood types. @rekill has a potential fix below.

Python (at least): empty matrix [] is never tested, only [[]]

Perhaps updating the description is enough.

javascript:

tests are contradicting with each other... some want the middle included, some don't? also

von_neumann
[ [ 18, 71, 7, 72, 59, 89, 54, 40, 69, 90 ],
  [ 58, 64, 72, 55, 79, 28, 20, 91, 100, 28 ],
  [ 67, 82, 65, 79, 30, 78, 78, 95, 89, 12 ],
  [ 68, 89, 14, 70, 31, 72, 42, 38, 57, 24 ],
  [ 39, 51, 1, 80, 54, 24, 83, 58, 55, 53 ],
  [ 73, 14, 38, 70, 88, 21, 29, 65, 18, 93 ],
  [ 37, 58, 13, 79, 56, 50, 48, 73, 31, 29 ],
  [ 18, 64, 89, 16, 77, 55, 19, 47, 86, 27 ],
  [ 17, 22, 13, 53, 89, 3, 25, 77, 54, 74 ],
  [ 92, 19, 80, 81, 46, 71, 91, 78, 95, 15 ] ]
  
Expected: '[24, 78]', instead got: '[24, 31, 42, 78]'

surprised anyone has passed this in JS

Either I have misunderstood this kata entirely, or there is some error in the random tests (I pass all the basic tests, though). For example, for the von Neumann case with coordinates = [3, 7], distance = 1, and

matrix = 
[ [ 38, 87, 75, 37, 54, 11, 5, 2, 36, 57 ],
  [ 50, 60, 56, 26, 32, 22, 46, 60, 75, 65 ],
  [ 93, 58, 91, 90, 50, 45, 3, 35, 39, 23 ],
  [ 4, 0, 71, 33, 67, 19, 76, 44, 62, 57 ],
  [ 42, 37, 16, 39, 91, 60, 100, 55, 90, 76 ],
  [ 69, 17, 22, 49, 4, 87, 59, 57, 60, 64 ],
  [ 90, 62, 91, 16, 84, 76, 20, 9, 42, 32 ],
  [ 67, 97, 12, 23, 57, 83, 12, 34, 47, 48 ],
  [ 19, 25, 17, 4, 27, 61, 2, 53, 81, 97 ],
  [ 36, 31, 16, 66, 32, 14, 54, 48, 20, 67 ] ]

I get this:

Expected: '[35]', instead got: '[35, 55, 62, 76]'

Bug in tests?

matrix = [ [ 100, 25, 59, 35, 22, 11, 42, 75, 99, 7 ], [ 35, 61, 76, 74, 72, 45, 31, 7, 70, 82 ], [ 46, 37, 30, 55, 14, 51, 18, 44, 6, 35 ], [ 41, 2, 34, 34, 43, 56, 13, 71, 44, 17 ], [ 1, 8, 61, 35, 19, 16, 66, 25, 4, 39 ], [ 9, 71, 49, 11, 49, 53, 50, 81, 86, 62 ], [ 97, 19, 87, 16, 15, 37, 95, 38, 37, 21 ], [ 2, 55, 49, 96, 88, 21, 79, 76, 60, 54 ], [ 75, 86, 98, 92, 91, 73, 75, 77, 100, 20 ], [ 96, 82, 49, 5, 49, 81, 60, 60, 51, 40 ] ];

get_neighbourhood("von_neumann", matrix, [ 5, 8 ], 1);

Expected: '[4]', instead got: '[37, 4, 62, 81]'

Why 4?

This also goes to every kata in the series: neighbourhood type only makes sense for distance 1, for every other distance the only notation that makes sense is "metric norm", aka how distance is counted.

In the case of Moore and Von Neumann metric, the former evaluates to the maximum distance along one dimension, while the latter evaluates to the sum of the distances along each dimension.

(Alternatively, they are just L_infinity and L_1 metric respectively.)

(Also alternatively, metric can be extended to other forms like L_2 (Pythagorean metric).)

This goes to every kata in the series: You did not mention what are the order of the coordinates.

e.g in these two 2D katas, you only said m,n in the table, but what does the coordinate mean? (m,n) or (n,m)? Does the coordinate (a,b,c,...) mean mat[a][b][c]..., or mat...[c][b][a]?

It also doesn't help that all your examples have the same m and n.

Moore neighborhood != manhattan distance. Don't mix up different things.

And, of course, cells can't be called neighbors if the distance between them is > 1:

X | X | X | X | X         + - center
-----------------         O - neighbor
X | O | O | O | X         X - not a neighbor
-----------------
X | O | + | O | X
-----------------
X | O | O | O | X
-----------------
X | X | X | X | X

Sample tests expect function name to be get_neighborhood instead of get_neighbourhood