Iterate vtkIdTypeArray in Python

Development
4

The iterator returned by NewIterator was designed for use from C++, it’s not usable from Python.

Probably the easiest way to deal with VTK arrays from Python is via numpy.
The numpy.asarray() function creates an array that is a “view” of the same memory buffer used by the VTK array:

from vtkmodules.vtkCommonCore import vtkIdTypeArray
import numpy as np

ids = vtkIdTypeArray()
ids.InsertNextValue(10)
ids.InsertNextValue(3)

# use numpy array as a "view" of a VTK array
np.asarray(ids)
# returns array([10,  3], dtype=int64)

# use numpy to modify the values in a VTK array
a = np.asarray(ids)
a *= 2
print([ids.GetValue(0), ids.GetValue(1)])
[20, 6]

There is a danger to this method, however: if the VTK array’s memory buffer is reallocated (for example, when the the array is resized), then the numpy array ends up with a dangling pointer. So the numpy array can only safely be used within a short scope where it is guaranteed that no resizing will occur.

You can also construct a new numpy array from a VTK array, which is safer if you plan to keep the numpy array arround for a while:

from vtkmodules.vtkCommonCore import vtkPoints
import numpy as np

points = vtkPoints()
points.InsertNextPoint(0,1,2)
points.InsertNextPoint(0,1,5)

a = np.array(points.GetData())
print(a)
# [[0. 1. 2.]
#  [0. 1. 5.]]

In the past there has been discussion of Python iterators for VTK arrays (I’m too lazy to search for a link). The idea was to have two iterators: one that would iterate over the tuples in the array, and one that would iterate over the values (i.e. iterate over the flattened array).

1 Like