泡菜（无文件IO）：TypeError：需要一个类似字节的对象，而不是“ str”

Question

I am getting this TypeError from object serialization alone, ie. no file IO involved (unlike other similar posts). One piece of the code is:

for itr in range(numiters):
    #Sample from proposal distribution
    d_star,Jratio,R_star,step = proposal(d_t,R_t,X,Y,alpha)
    #Compute the new posterior value, if necessary
    a_star = Pickle.dumps(d_star[:R_star+1])
    print(type(a_star))
    permsdic[a_star] = ...

and another is:

for perm in permsdic.keys():
    print(type(perm))
    print(perm)
    d_t = Pickle.loads(perm)

and yet I get the above error in: d_t = Pickle.loads(perm) The output of print statements in the first block is:

class 'bytes'

and yet in 2nd block it is somehow:

class 'str'

b'\\x80\\x03]q\\x00(K\\x01K\\x03K\\x05K\\x0cK\\x00e.'

This is in Python 3.6 (I am trying to adapt an older Python library)

Answer 1

I was able to make it work by using eval() function, ie the following works

for perm in permsdic.keys():
    print(type(perm))
    print(perm)
    d_t = Pickle.loads(eval(perm))

I still don't understand why this is necessary...