import numpy as np
from matplotlib.tri import Triangulation

def split_quads(elnode=None):  # modified by FY
    '''
    Split quad elements to triangles and append additional elements to element table
    This script can be made much faster by using vector operation instead of the for-loop;
    just append additional elements to the end.
    '''
    if elnode is None:
        raise Exception('elnode should be a numpy array of (np,4)')

    elnode = np.ma.masked_values(elnode, -1)  # modified by FY
    if elnode.shape[1] == 4:
        eid = np.nonzero(~((np.isnan(elnode[:,-1]))|(elnode[:,-1]<0)))[0]
        elnode = np.r_[elnode[:,:3], np.c_[elnode[eid, 0][:, None], elnode[eid, 2:]]]
    #elements = np.ma.masked_values(elements, -1)  # modified by FY
    #for ele in elements:
    #    ele = ele[~ele.mask] - 1
    #    if len(ele) == 3:
    #        tris.append([ele[0], ele[1], ele[2]])
    #    elif len(ele) == 4:
    #        tris.append([ele[0], ele[1], ele[3]])
    #        tris.append([ele[1], ele[2], ele[3]])
    return elnode.astype('int')

def triangulation(lon, lat, tris):
    if tris.max() >= len(lon): 
        tris -= 1
    return Triangulation(lon, lat, tris)