How can I achieve:
>>> foo = np.array([1,2,3])
array([1,2,3])
>>> foo.append(4)
array([1,2,3,4])
Instead of numpy's:
np.append(foo, 4)
I've tried stuff along the lines of:
import numpy as np
class myarrayclass(np.array):
def append(self, value):
self.object = np.append(self.object, value)
Also, is it possible to overwrite the numpy class instead of creating my own? I don't need this to work, just wondering if it is possible, thanks in advance!
The code in dubbbdan's answer definitely has its value but I would suggest the implementation below:
import numpy as np
class MyArrayClass(np.ndarray):
def __new__(cls, *args, **kwargs):
return np.ndarray.__new__(cls, *args, **kwargs)
def __init__(self, *args, **kwargs):
super()
def append(self, item):
return np.append(self, item)
The devil is in the details and that answer is not actually extending the numpy.ndarray class but creating a class wrapper for the append function.
I am not sure if this is the best way, but here is one approach.
import numpy as np
class myarrayclass(object):
def __init__(self, in_array):
self.array = in_array
def append(self, val):
self.array = np.append(self.array, val)
def show(self):
return self.array
foo = np.array([1,2,3])
myarray = myarrayclass(foo)
myarray.append(5)
myarray.show()
#Out[7]: array([1, 2, 3, 5])
When you make an object, you need to initialize it with def __init__(self, in_array). Here I have included a simple append and show method. I included the show method to demonstrate that the append worked.
np.appendbefore you try this. It is not a listappendclone, and shouldn't be treated as one.resizemethod. That can add zeros in-place. It may be instructive to look at the code for thenp.resizefunction, which does not operate in place. Rather it usesconcatenateandreshape, making a new array.