Cannot load pickled object

问题

The problem I am having is when I try to load the pickled object. I have tried using both pickle.loads and pickle.load Here are the results:

pickle.loads:

TypeError: 'str' does not support the buffer interface

pickle.load:

TypeError: file must have 'read' and 'readline' attributes

Can someone please tell me what I am doing wrong in this process?

elif str(parser) == 'SwissWithdrawn_Parser':
       # swissprot name changes
       print('Gathering SwissProt update info...')
       cache_hits = 0
       cache_misses = 0
       files = set()

       for f in os.listdir('out/cache/'):
           if os.path.isfile('out/cache/'+f):
               files.add(f)

       for name in sp_lost_names:

           cached = False
           url = 'http://www.uniprot.org/uniprot/?query=mnemonic%3a'+name+ \
               '+active%3ayes&format=tab&columns=entry%20name'
           hashed_url = str(hash(url))

           ################### For Testing Only - use cache ##################
           if hashed_url in files:
               cached = True
               cache_hits += 1
               content = pickle.loads('out/cache/' +hashed_url)  # <-- problematic line
           else:
               cache_misses += 1
               content = urllib.request.urlopen(url)

           # get the contents returned from the HTTPResponse object
           content_list = [x.decode().strip() for x in content.readlines()]
           if not cached:
               with open('out/cache/'+hashed_url, 'wb') as fp:
                   pickle.dump(content_list, fp)
           ####################################################################

           # no replacement
           if len(content_list) is 0:
               change_log['swiss-names'] = 
                  { name : 'withdrawn' }
           # get the new name
           else:
               new_name = content_list[1]
               change_log['swiss-names'] =
                  { name : new_name }

回答1:

You need to either read the file first (as binary bytes) and use pickle.loads(), or pass an open file object to the pickle.load() command. The latter is preferable:

with open('out/cache/' +hashed_url, 'rb') as pickle_file:
    content = pickle.load(pickle_file)

Neither method supports loading a pickle from a filename.

来源：https://stackoverflow.com/questions/18261898/cannot-load-pickled-object