Slice syntax to object [duplicate]

问题

I have a class holding data (a numpy ndarray) that includes a method storing the data to a mat-file (using scipy.io.savemat). The data can be very large, so I may only want to store a segment of the data. Therefore I pass a slice-object, like this:

def write_mat(self, fn, fields=None, sel=None):
    # (set fields and sel to sensible values if None)
    scipy.io.savemat(fn, dict(data=self.data[fields][sel]))

Here, fields may be a list of strings (for self.data is a structured array), and sel is a slice-object. Of course, I cannot directly pass the slice-syntax into write_mat: obj.write_mat(fn, fields, [::10]) is a SyntaxError. Of course, I can pass in slice(None, None, 10) instead, but I don't like this solution very much.

Is there any builtin convenience object that will allow me to create a slice-object from the slice-syntax? Of course, it's easy to implement:

In [574]: class Foo:
     ...:     def __getitem__(self, item):
     ...:         return item
     ...:     

In [578]: slicer = Foo()

In [579]: slicer[::100]
Out[579]: slice(None, None, 100)

but even for something easy to implement there may already be a more standard solution. Is there? By standard I mean existing inside Python, numpy, or scipy.

回答1:

The answer on Passing Python slice syntax around to functions is correct, but as you're already using NumPy you can use np.s_:

import numpy as np
np.s_[1:2:3]
Out[1]: slice(1, 2, 3)

来源：https://stackoverflow.com/questions/15161856/slice-syntax-to-object

标签

python

numpy

slice