结构化的numpy ndarray，如何获取值

Question

I have a structured numpy ndarray la = {'val1':0,'val2':1} and I would like to return the vals using the 0 and 1 as keys, so I wish to return val1 when I have 0 and val2 when I have 1 which should have been straightforward however my attempts have failed, as I am not familiar with this structure.

How do I return only the corresponding val, or an array of all vals so that I can read in order?

Answer 1

When you save a Python object (non-array), numpy wraps it in an array. The object is pickled :

In [112]: np.save('test.npy', {'foo':34})                                              

In newer numpy versions, you have to explicitly allow it to load pickled items:

In [113]: data = np.load('test.npy')                                                   
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-113-c5835e6fb31e> in <module>
----> 1 data = np.load('test.npy')

/usr/local/lib/python3.6/dist-packages/numpy/lib/npyio.py in load(file, mmap_mode, allow_pickle, fix_imports, encoding)
    451             else:
    452                 return format.read_array(fid, allow_pickle=allow_pickle,
--> 453                                          pickle_kwargs=pickle_kwargs)
    454         else:
    455             # Try a pickle

/usr/local/lib/python3.6/dist-packages/numpy/lib/format.py in read_array(fp, allow_pickle, pickle_kwargs)
    720         # The array contained Python objects. We need to unpickle the data.
    721         if not allow_pickle:
--> 722             raise ValueError("Object arrays cannot be loaded when "
    723                              "allow_pickle=False")
    724         if pickle_kwargs is None:

ValueError: Object arrays cannot be loaded when allow_pickle=False


In [115]: data = np.load('test.npy',allow_pickle=True)                                 
In [116]: data                                                                         

The result is an object dtype array with 1 element (0d)

Out[116]: array({'foo': 34}, dtype=object)

tolist can extract that object, as will item() .

In [117]: data.tolist()                                                                
Out[117]: {'foo': 34}
In [118]: data.tolist()['foo']                                                         
Out[118]: 34
In [119]: data.item()                                                                  
Out[119]: {'foo': 34}
In [120]: data[()]                                                                     
Out[120]: {'foo': 34}

Answer 2

That looks like a dictionary, rather than an ndarray. Assuming you don't have multiple keys associated with the same value, you reverse the dictionary with

la_reversed = {v: k for k, v in la.items()}

so la_reversed[0] would be 'val1' and la_reversed[1] would be 'val2' .

You can get a list of values in the dictionary with list(la.values()) .

Answer 3

Just found out that I can use la.tolist() and it returns a dictionary, somehow? when I wanted a list, alas from there on I was able to solve my problem.