def load_dictionary(dictionary_path):
"""
load dictionary
Parameters
----------
dictionary_path : str
a path of dictionary
"""
dictionary = DictionaryDataset(dictionary_path=dictionary_path)
return dictionary.data
def cache_or_load_dictionary():
dictionary_name = os.path.splitext(os.path.basename(
args.dictionary_path))[0]
cached_dictionary_path = os.path.join(
'./tmp', "cached_{}.pk".format(dictionary_name))
# If exist, load the cached dictionary
if os.path.exists(cached_dictionary_path):
with open(cached_dictionary_path, 'rb') as fin:
cached_dictionary = pickle.load(fin)
print("Loaded dictionary from cached file {}".format(
cached_dictionary_path))
dictionary, dict_sparse_embeds, dict_dense_embeds = (
cached_dictionary['dictionary'],
cached_dictionary['dict_sparse_embeds'],
cached_dictionary['dict_dense_embeds'],
)
else:
dictionary = DictionaryDataset(
dictionary_path=args.dictionary_path).data
dictionary_names = dictionary[:, 0]
dict_sparse_embeds = biosyn.embed_sparse(names=dictionary_names,
show_progress=True)
dict_dense_embeds = biosyn.embed_dense(names=dictionary_names,
show_progress=True)
cached_dictionary = {
'dictionary': dictionary,
'dict_sparse_embeds': dict_sparse_embeds,
'dict_dense_embeds': dict_dense_embeds
}
if not os.path.exists('./tmp'):
os.mkdir('./tmp')
with open(cached_dictionary_path, 'wb') as fin:
pickle.dump(cached_dictionary, fin)
print("Saving dictionary into cached file {}".format(
cached_dictionary_path))
return dictionary, dict_sparse_embeds, dict_dense_embeds
def load_dictionary(dictionary_path):
dictionary = DictionaryDataset(dictionary_path=dictionary_path)
return dictionary.data