mirror of https://github.com/explosion/spaCy.git
Fixing encoding issue #4
This commit is contained in:
parent
fcbfff043f
commit
f07e4accd7
|
@ -19,10 +19,7 @@ class Lemmatizer(object):
|
|||
index[pos] = read_index(path.join(data_dir, 'wordnet', 'index.%s' % pos))
|
||||
exc[pos] = read_exc(path.join(data_dir, 'wordnet', '%s.exc' % pos))
|
||||
if path.exists(path.join(data_dir, 'vocab', 'lemma_rules.json')):
|
||||
if sys.version_info[0] == 3:
|
||||
rules = json.load(open(path.join(data_dir, 'vocab', 'lemma_rules.json'), encoding="utf_8"))
|
||||
else:
|
||||
rules = json.load(open(path.join(data_dir, 'vocab', 'lemma_rules.json')))
|
||||
rules = json.load(codecs.open(path.join(data_dir, 'vocab', 'lemma_rules.json'), encoding='utf_8'))
|
||||
else:
|
||||
rules = {}
|
||||
return cls(index, exc, rules)
|
||||
|
|
Loading…
Reference in New Issue