Question

PyCall文件说：重要：与Python最大的区别在于，使用o [：attribute]而不是o.attribute访问对象属性/成员，因此Python中的o.method（...）将被Julia中的o：方法替换。此外，您使用get（o，key）而不是o [key]。（但是，您可以通过o [i]访问整数索引，就像在Python中一样，虽然使用基于1的Julian索引而不是基于0的Python索引。）

但我不知道要导入哪个模块或对象

Answer 1

这是一个让你入门的简单例子

using PyCall

@pyimport numpy as np           # 'np' becomes a julia module

a = np.array([[1, 2], [3, 4]])  # access objects directly under a module
                                # (in this case the 'array' function)
                                # using a dot operator directly on the module
#> 2×2 Array{Int64,2}:
#> 1  2
#> 3  4

a = PyObject(a)                 # dear Julia, we appreciate the automatic
                                # convertion back to a julia native type, 
                                # but let's get 'a' back in PyObject form
                                # here so we can use one of its methods:
#> PyObject array([[1, 2],
#>                 [3, 4]])

b = a[:mean](axis=1)            # 'a' here is a python Object (not a python 
                                # module), so the way to access a method
                                # or object that belongs to it is via the
                                # pythonobject[:method] syntax.
                                # Here we're calling the 'mean' function, 
                                # with the appropriate keyword argument
#> 2-element Array{Float64,1}:
#>  1.5
#>  3.5

pybuiltin(:type)(b)             # Use 'pybuiltin' to use built-in python
                                # commands (i.e. commands that are not 
                                # under a module)
#> PyObject <type 'numpy.ndarray'>

pybuiltin(:isinstance)(b, np.ndarray)
#> true

如何使用PyCall

1 个答案: