diff --git a/day19/README.md b/day19/README.md
new file mode 100644
index 0000000..6b4d032
--- /dev/null
+++ b/day19/README.md
@@ -0,0 +1,72 @@
+question of the day: https://leetcode.com/problems/01-matrix/#/description
+
+Given a matrix consists of 0 and 1, find the distance of
+the nearest 0 for each cell.
+
+The distance between two adjacent cells is 1.
+
+*Example 1:*
+
+Input:
+
+```ruby
+0 0 0
+0 1 0
+0 0 0
+```
+
+Output:
+
+```ruby
+0 0 0
+0 1 0
+0 0 0
+```
+
+*Example 2:*
+
+Input:
+
+```ruby
+0 0 0
+0 1 0
+1 1 1
+```
+
+Output:
+
+```ruby
+0 0 0
+0 1 0
+1 2 1
+```
+
+Assumptions we can make:
+
+1. The number of elements of the given matrix will not exceed 10,000.  
+2. There are at least one 0 in the given matrix.  
+3. The cells are adjacent in only four directions: up, down, left and right.  
+
+## Ideas
+
+This is like drawing out a contour map. The 0's are the peaks of
+hills and mountains, while the parts of the matrix that are far away
+from any 0's are like the valleys.
+
+We can start from each peak and go outwards. It'd be a BFS-like
+approach. Do a search through the whole matrix once first and find
+where all the 0's are. Add each of those positions into a queue,
+and then use that queue to start off a BFS. During this BFS, we
+check the neighboring cells to see what the minimal value is that
+we can place in this cell.
+
+This solution is `O(n)` to find the 0s, and then `O(n)` to run the
+BFS. My implementation also requires `O(n)` space, although I think
+it's possible to modify the matrix in-place with some more clever
+checking.
+
+## Code
+
+[Ruby](./matrixCountours.rb)
+
+## Follow up
diff --git a/day19/matrixCountours.rb b/day19/matrixCountours.rb
new file mode 100644
index 0000000..fe8d923
--- /dev/null
+++ b/day19/matrixCountours.rb
@@ -0,0 +1,108 @@
+def contourDistances(matrix)
+  rows, cols = matrix.size, matrix[0].size
+
+  if rows == 0 || cols == 0
+    return matrix
+  end
+
+  contours = Array.new(rows).map { |x| [nil] * cols }
+  queue = []
+  # start by finding 0s
+  for row in 0..rows-1
+    for col in 0..cols-1
+      if matrix[row][col] == 0
+        queue << [row, col]
+        contours[row][col] = 0
+      end
+    end
+  end
+
+  # do bfs
+  while queue.size > 0
+    current = queue.shift
+    row, col = current
+    contours[row][col] = getMinNeighbor(contours, row, col) + 1
+    for i, j in [[row-1, col], [row, col-1], [row, col+1], [row+1, col]]
+      if withinBounds([i, j], rows, cols) && contours[i][j] == nil
+        queue << [i, j]
+      end
+    end
+  end
+
+  contours
+end
+
+def withinBounds(coordinate, rows, cols)
+  x, y = coordinate
+  x >= 0 && x < rows && y >= 0 && y < cols
+end
+
+def getMinNeighbor(matrix, row, col)
+  if matrix[row][col] == 0
+    return -1
+  end
+
+  vals = []
+  rows, cols = matrix.size, matrix[0].size
+  for i, j in [[row-1, col], [row, col-1], [row, col+1], [row+1, col]]
+    if withinBounds([i, j], rows, cols)
+      if matrix[i][j] != nil
+        vals << matrix[i][j]
+      end
+    end
+  end
+
+  vals.min
+end
+
+#########
+# Tests #
+#########
+
+class AssertionError < RuntimeError
+end
+
+def assert &block
+  raise AssertionError unless yield
+end
+
+def tests
+  # edge cases
+  empty = [[]]
+  emptyOutput = [[]]
+  assert { contourDistances(empty) == emptyOutput }
+
+  one = [[0]]
+  oneOutput = [[0]]
+  assert { contourDistances(one) == oneOutput }
+
+  # other cases
+  matrix = [[1,1,1],
+            [1,0,1],
+            [1,1,1]]
+
+  output = [[2,1,2],
+            [1,0,1],
+            [2,1,2]]
+  assert { contourDistances(matrix) == output }
+
+  matrix1 = [[0,0,0],
+             [0,1,0],
+             [0,0,0]]
+
+  output1 = [[0,0,0],
+             [0,1,0],
+             [0,0,0]]
+  assert { contourDistances(matrix1) == output1 }
+
+  matrix2 = [[0,0,0],
+             [0,1,0],
+             [1,1,1]]
+             
+  output2 = [[0,0,0],
+             [0,1,0],
+             [1,2,1]]
+  assert { contourDistances(matrix2) == output2 }
+end
+
+tests()