pycloud.py

Go to the documentation of this file.

import numpy as np
from scipy.sparse import lil_matrix
from bisect import bisect_left

class PyCloud(lil_matrix):
        
        def __init__(self, arg1, row_step, point_step, shape=None, dtype=None, copy=False):
                lil_matrix.__init__(self, arg1, shape=shape, dtype=self.PyPoint, copy=copy)
                self.nan = self.PyPoint((np.nan, np.nan, np.nan))
                self.format = 'lil'
                self.dtype = self.PyPoint
                
                
        def __getitem__(self, index):
                item = super(PyCloud, self).__getitem__(index)
                assert type(item) == self.PyPoint or item == 0, 'Expecting %s, got: %s: %s' % (self.dtype.name, str(type(item)), item)
                
                return item.vals if type(item) == self.PyPoint else self.nan
                
        def __setitem__(self, index, item):
                # i, j = index
                if np.ma.extras.issequence(item):
                        super(PyCloud, self).__setitem__(index, self.PyPoint(item))
                        item = self.PyPoint(item)
                else:
                        super(PyCloud, self).__setitem__(index, item)
                
        class PyPoint(np.float64):
                name = "PyPoint"
                
                def __new__(cls, vals):
                        return np.float64.__new__(cls, np.nan)
                        
                def __init__(self, vals):
                        np.float64.__init__(self)
                        self.vals = vals
                
                @classmethod
                def type(cls, x):
                        if np.ma.extras.issequence(x): 
                                return cls(x)
                        elif(type(x) == cls):
                                return x
                        raise TypeError('Unable to convert value (%s) to dtype [%s]' % (x,cls.name))
                        
if __name__ == '__main__':
        l = PyCloud((100,100), 8, 8)
        l[5,5] = (1,2,3)
        assert (1,2,3) == l[5,5], 'Wrong value came out'
        assert np.isnan(l[5,7]), 'Unassigned index not nan'