2017-03-12 15:07:28 +03:00
|
|
|
# coding: utf8
|
2015-08-25 16:37:30 +03:00
|
|
|
from __future__ import unicode_literals, print_function
|
2015-07-07 15:00:07 +03:00
|
|
|
|
2015-08-25 16:37:30 +03:00
|
|
|
from os import path
|
2017-03-08 03:37:52 +03:00
|
|
|
from pathlib import Path
|
2014-12-24 09:42:00 +03:00
|
|
|
|
2016-12-19 00:29:31 +03:00
|
|
|
from ..util import match_best_version
|
2016-12-19 00:46:31 +03:00
|
|
|
from ..util import get_data_path
|
2015-08-25 16:37:30 +03:00
|
|
|
from ..language import Language
|
2016-09-25 15:49:53 +03:00
|
|
|
from ..lemmatizer import Lemmatizer
|
|
|
|
from ..vocab import Vocab
|
|
|
|
from ..tokenizer import Tokenizer
|
2016-10-18 19:52:48 +03:00
|
|
|
from ..attrs import LANG
|
2016-12-08 15:58:32 +03:00
|
|
|
|
2016-12-18 18:54:19 +03:00
|
|
|
from .language_data import *
|
2016-01-19 04:54:56 +03:00
|
|
|
|
2017-03-08 03:37:52 +03:00
|
|
|
try:
|
|
|
|
basestring
|
|
|
|
except NameError:
|
|
|
|
basestring = str
|
|
|
|
|
2016-12-18 18:58:28 +03:00
|
|
|
|
2015-08-25 16:37:30 +03:00
|
|
|
class English(Language):
|
2015-12-28 18:54:03 +03:00
|
|
|
lang = 'en'
|
2015-12-18 11:52:55 +03:00
|
|
|
|
2016-09-24 21:26:17 +03:00
|
|
|
class Defaults(Language.Defaults):
|
|
|
|
lex_attr_getters = dict(Language.Defaults.lex_attr_getters)
|
2016-10-18 19:52:48 +03:00
|
|
|
lex_attr_getters[LANG] = lambda text: 'en'
|
2016-09-24 21:26:17 +03:00
|
|
|
|
2016-12-08 15:58:32 +03:00
|
|
|
tokenizer_exceptions = TOKENIZER_EXCEPTIONS
|
|
|
|
tag_map = TAG_MAP
|
|
|
|
stop_words = STOP_WORDS
|
2016-12-19 00:29:31 +03:00
|
|
|
|
|
|
|
|
|
|
|
def __init__(self, **overrides):
|
|
|
|
# Make a special-case hack for loading the GloVe vectors, to support
|
|
|
|
# deprecated <1.0 stuff. Phase this out once the data is fixed.
|
|
|
|
overrides = _fix_deprecated_glove_vectors_loading(overrides)
|
|
|
|
Language.__init__(self, **overrides)
|
|
|
|
|
|
|
|
|
|
|
|
def _fix_deprecated_glove_vectors_loading(overrides):
|
|
|
|
if 'data_dir' in overrides and 'path' not in overrides:
|
|
|
|
raise ValueError("The argument 'data_dir' has been renamed to 'path'")
|
2016-12-19 01:12:49 +03:00
|
|
|
if overrides.get('path') is False:
|
|
|
|
return overrides
|
|
|
|
if overrides.get('path') in (None, True):
|
2016-12-19 00:46:31 +03:00
|
|
|
data_path = get_data_path()
|
|
|
|
else:
|
|
|
|
path = overrides['path']
|
2017-03-08 03:37:52 +03:00
|
|
|
if isinstance(path, basestring):
|
|
|
|
path = Path(path)
|
2016-12-19 00:29:31 +03:00
|
|
|
data_path = path.parent
|
2016-12-20 20:21:05 +03:00
|
|
|
vec_path = None
|
2016-12-19 00:46:31 +03:00
|
|
|
if 'add_vectors' not in overrides:
|
2016-12-19 00:29:31 +03:00
|
|
|
if 'vectors' in overrides:
|
|
|
|
vec_path = match_best_version(overrides['vectors'], None, data_path)
|
|
|
|
if vec_path is None:
|
2017-03-08 03:37:52 +03:00
|
|
|
return overrides
|
2016-12-19 00:29:31 +03:00
|
|
|
else:
|
|
|
|
vec_path = match_best_version('en_glove_cc_300_1m_vectors', None, data_path)
|
|
|
|
if vec_path is not None:
|
|
|
|
vec_path = vec_path / 'vocab' / 'vec.bin'
|
2016-12-19 01:06:44 +03:00
|
|
|
if vec_path is not None:
|
|
|
|
overrides['add_vectors'] = lambda vocab: vocab.load_vectors_from_bin_loc(vec_path)
|
2016-12-19 00:29:31 +03:00
|
|
|
return overrides
|