import numpy.linalg as npl
import numpy as np
import scipy.sparse as sp
from sklearn.preprocessing import normalize
def bigArray_to_playerArray(matrix):
  player_matrices = []
  for row in range(len(matrix)):
    diag_u = matrix[row]
    diag_l = [matrix[i][row] for i in range(len(diag_u))]

    diag_u = np.delete(diag_u,row)

    diag_l = np.delete(diag_l,row)

    for i in range(len(diag_u)):
      tot = diag_l[i] + diag_u[i]
      if tot != 0:
        diag_u[i] = diag_u[i] /tot
        diag_l[i] /= tot
      else:
        diag_l[i] = 0
        diag_u[i] = 0
    player = sp.diags([diag_u,diag_l],[1,-1]).todense()
    player_matrices.append(player)
  return player_matrices

def clean_matrices(players):
  i,j,k = [0,0,0]
  diags = []
  for player in players:
    diag_u = np.diag(player,1)
    u_sum = sum(diag_u)
    diag_l = np.diag(player,-1)
    l_sum = sum(diag_l)
    diags.append(u_sum/(u_sum+l_sum))
  for player in players:
    diag_u = np.diag(player,1)
    diag_l = np.diag(player,-1)
    for index in range(len(diag_u)):
      if diag_u[index] == 0 and diag_l[index] == 0:
        k = index
        if index >= i:
          j = index + 1

        else:
          j = index
        print(i,j,k)
        p1win = (diags[i]+1-diags[j])/2
        player[0+k,1+k] = p1win
        player[1+k,0+k] = 1-p1win
        players[j][0+i,1+i] = 1-p1win
        players[j][1+i,0+i] = p1win
    #print(player)
    i += 1

def ratiosToEigenvector(matrix):
  return npl.eig(matrix)
def fill_matrix(matrix):
  ratios = []
  for i in range(len(matrix)):
    col = sum(matrix[:,i])
    row = sum(matrix[i,:])
    ratios.append(row/(col+row))
  for p1 in range(len(matrix)):
    for p2 in range(len(matrix)):
      if p1 != p2:
        if matrix[p1,p2] == 0 and matrix[p1,p2] == 0:
          p1win = (ratios[p1] + 1 - ratios[p2])/2
          matrix[p1,p2] = p1win
          matrix[p2,p1] = 1- p1win
        else:
          tot =matrix[p1,p2] +  matrix[p2,p1]
          matrix[p1,p2] = matrix[p1,p2]/tot
          matrix[p2,p1] = matrix[p2,p1]/tot
  return matrix

matrix =np.array([ [0,3,2,4],[1,0,1,1],[1,0,0,0],[3,1,0,0]],float)
matrix2 = [[0,1.0,1,1],[0,0,1,1],[0,0,0,1],[0,0,0,0]]

arr = fill_matrix(matrix)
print(arr)
print(ratiosToEigenvector(matrix)[1][:,0])
"""
arr = bigArray_to_playerArray(matrix2)
clean_matrices(arr)
print(arr)
for i in range(len(arr)):
  print("norm", npl.eigvals(normalize(arr[i])))
  print(npl.eigvals(arr[i]))
"""