spaCy/spacy/language.py

302 lines
9.8 KiB
Python
Raw Normal View History

from __future__ import absolute_import
from os import path
from warnings import warn
import io
try:
import ujson as json
except ImportError:
import json
from .tokenizer import Tokenizer
from .vocab import Vocab
from .syntax.parser import Parser
from .tagger import Tagger
from .matcher import Matcher
from .serialize.packer import Packer
from . import attrs
from . import orth
from .syntax.ner import BiluoPushDown
from .syntax.arc_eager import ArcEager
from . import util
2016-04-12 17:00:56 +03:00
from . import about
from .attrs import TAG, DEP, ENT_IOB, ENT_TYPE, HEAD
2015-08-25 16:37:17 +03:00
class Language(object):
2015-12-28 18:54:03 +03:00
lang = None
2015-08-25 16:37:17 +03:00
@staticmethod
def lower(string):
return string.lower()
@staticmethod
def norm(string):
return string
@staticmethod
def prefix(string):
return string[0]
@staticmethod
def suffix(string):
return string[-3:]
@staticmethod
def cluster(string):
return 0
@staticmethod
def is_digit(string):
return string.isdigit()
2015-08-25 16:37:17 +03:00
@staticmethod
def is_space(string):
return string.isspace()
2015-09-14 10:48:51 +03:00
@staticmethod
def is_stop(string):
return 0
@classmethod
2016-02-06 17:13:55 +03:00
def default_lex_attrs(cls, *args, **kwargs):
oov_prob = kwargs.get('oov_prob', -20)
2015-08-25 16:37:17 +03:00
return {
attrs.LOWER: cls.lower,
attrs.NORM: cls.norm,
attrs.SHAPE: orth.word_shape,
2015-08-25 16:37:17 +03:00
attrs.PREFIX: cls.prefix,
attrs.SUFFIX: cls.suffix,
attrs.CLUSTER: cls.cluster,
2016-02-06 17:13:55 +03:00
attrs.PROB: lambda string: oov_prob,
attrs.LANG: lambda string: cls.lang,
attrs.IS_ALPHA: orth.is_alpha,
attrs.IS_ASCII: orth.is_ascii,
2015-08-25 16:37:17 +03:00
attrs.IS_DIGIT: cls.is_digit,
attrs.IS_LOWER: orth.is_lower,
attrs.IS_PUNCT: orth.is_punct,
attrs.IS_SPACE: cls.is_space,
attrs.IS_TITLE: orth.is_title,
attrs.IS_UPPER: orth.is_upper,
attrs.IS_BRACKET: orth.is_bracket,
attrs.IS_QUOTE: orth.is_quote,
attrs.IS_LEFT_PUNCT: orth.is_left_punct,
attrs.IS_RIGHT_PUNCT: orth.is_right_punct,
attrs.LIKE_URL: orth.like_url,
attrs.LIKE_NUM: orth.like_number,
attrs.LIKE_EMAIL: orth.like_email,
2015-09-14 10:48:51 +03:00
attrs.IS_STOP: cls.is_stop,
2015-08-25 16:37:17 +03:00
attrs.IS_OOV: lambda string: True
}
@classmethod
def default_dep_labels(cls):
return {0: {'ROOT': True}}
@classmethod
def default_ner_labels(cls):
return {0: {'PER': True, 'LOC': True, 'ORG': True, 'MISC': True}}
2015-08-25 16:37:17 +03:00
@classmethod
def default_vocab(cls, package, get_lex_attr=None, vectors_package=None):
2015-08-25 16:37:17 +03:00
if get_lex_attr is None:
2016-02-06 17:13:55 +03:00
if package.has_file('vocab', 'oov_prob'):
with package.open(('vocab', 'oov_prob')) as file_:
oov_prob = float(file_.read().strip())
get_lex_attr = cls.default_lex_attrs(oov_prob=oov_prob)
else:
get_lex_attr = cls.default_lex_attrs()
if hasattr(package, 'dir_path'):
return Vocab.from_package(package, get_lex_attr=get_lex_attr,
vectors_package=vectors_package)
else:
return Vocab.load(package, get_lex_attr)
2015-08-25 16:37:17 +03:00
@classmethod
2015-12-07 08:01:28 +03:00
def default_parser(cls, package, vocab):
if hasattr(package, 'dir_path'):
data_dir = package.dir_path('deps')
else:
data_dir = package
2015-12-07 08:50:26 +03:00
if data_dir and path.exists(data_dir):
return Parser.from_dir(data_dir, vocab.strings, ArcEager)
else:
return None
2015-08-25 16:37:17 +03:00
@classmethod
2015-12-07 08:01:28 +03:00
def default_entity(cls, package, vocab):
if hasattr(package, 'dir_path'):
data_dir = package.dir_path('ner')
else:
data_dir = package
2015-12-07 08:50:26 +03:00
if data_dir and path.exists(data_dir):
return Parser.from_dir(data_dir, vocab.strings, BiluoPushDown)
else:
return None
2015-08-25 16:37:17 +03:00
def __init__(self,
data_dir=None,
vocab=None,
tokenizer=None,
tagger=None,
parser=None,
entity=None,
matcher=None,
serializer=None,
load_vectors=True,
package=None,
vectors_package=None):
"""
2016-04-14 11:36:57 +03:00
A model can be specified:
2016-04-14 11:36:57 +03:00
1) by calling a Language subclass
- spacy.en.English()
2016-04-14 11:36:57 +03:00
2) by calling a Language subclass with data_dir
- spacy.en.English('my/model/root')
- spacy.en.English(data_dir='my/model/root')
2016-04-14 11:36:57 +03:00
3) by package name
- spacy.load('en_default')
- spacy.load('en_default==1.0.0')
2016-04-14 11:36:57 +03:00
4) by package name with a relocated package base
- spacy.load('en_default', via='/my/package/root')
- spacy.load('en_default==1.0.0', via='/my/package/root')
"""
if package is None:
2016-01-16 14:23:45 +03:00
if data_dir is None:
2016-04-12 17:00:56 +03:00
package = util.get_package_by_name(about.__models__[self.lang])
else:
2016-01-16 14:23:45 +03:00
package = util.get_package(data_dir)
if load_vectors is not True:
warn("load_vectors is deprecated", DeprecationWarning)
if vocab in (None, True):
vocab = self.default_vocab(package, vectors_package=vectors_package)
2015-12-28 18:54:03 +03:00
self.vocab = vocab
if tokenizer in (None, True):
tokenizer = Tokenizer.from_package(package, self.vocab)
2015-12-28 18:54:03 +03:00
self.tokenizer = tokenizer
if tagger in (None, True):
tagger = Tagger.from_package(package, self.vocab)
2015-12-28 18:54:03 +03:00
self.tagger = tagger
if entity in (None, True):
2015-12-28 18:54:03 +03:00
entity = self.default_entity(package, self.vocab)
self.entity = entity
if parser in (None, True):
2015-12-28 18:54:03 +03:00
parser = self.default_parser(package, self.vocab)
self.parser = parser
if matcher in (None, True):
matcher = Matcher.from_package(package, self.vocab)
2015-12-28 18:54:03 +03:00
self.matcher = matcher
2015-08-25 16:37:17 +03:00
def __reduce__(self):
args = (
None, # data_dir
self.vocab,
self.tokenizer,
self.tagger,
self.parser,
self.entity,
2015-12-28 18:54:03 +03:00
self.matcher
)
return (self.__class__, args, None, None)
def __call__(self, text, tag=True, parse=True, entity=True):
2015-08-25 16:37:17 +03:00
"""Apply the pipeline to some text. The text can span multiple sentences,
and can contain arbtrary whitespace. Alignment into the original string
is preserved.
Args:
text (unicode): The text to be processed.
Returns:
tokens (spacy.tokens.Doc):
>>> from spacy.en import English
>>> nlp = English()
>>> tokens = nlp('An example sentence. Another example sentence.')
>>> tokens[0].orth_, tokens[0].head.tag_
('An', 'NN')
"""
tokens = self.tokenizer(text)
if self.tagger and tag:
self.tagger(tokens)
if self.matcher and entity:
self.matcher(tokens)
if self.parser and parse:
self.parser(tokens)
if self.entity and entity:
# Add any of the entity labels already set, in case we don't have them.
for tok in tokens:
if tok.ent_type != 0:
self.entity.add_label(tok.ent_type)
2015-08-25 16:37:17 +03:00
self.entity(tokens)
return tokens
def pipe(self, texts, tag=True, parse=True, entity=True, n_threads=2,
batch_size=1000):
2016-02-05 14:20:29 +03:00
stream = self.tokenizer.pipe(texts,
n_threads=n_threads, batch_size=batch_size)
if self.tagger and tag:
2016-02-05 14:20:29 +03:00
stream = self.tagger.pipe(stream,
n_threads=n_threads, batch_size=batch_size)
if self.matcher and entity:
2016-02-05 14:20:29 +03:00
stream = self.matcher.pipe(stream,
n_threads=n_threads, batch_size=batch_size)
if self.parser and parse:
2016-02-05 14:20:29 +03:00
stream = self.parser.pipe(stream,
n_threads=n_threads, batch_size=batch_size)
if self.entity and entity:
2016-02-05 14:20:29 +03:00
stream = self.entity.pipe(stream,
2016-02-05 14:47:57 +03:00
n_threads=1, batch_size=batch_size)
for doc in stream:
yield doc
2015-08-25 16:37:17 +03:00
def end_training(self, data_dir=None):
if data_dir is None:
data_dir = self.data_dir
if self.parser:
self.parser.model.end_training()
self.parser.model.dump(path.join(data_dir, 'deps', 'model'))
if self.entity:
self.entity.model.end_training()
self.entity.model.dump(path.join(data_dir, 'ner', 'model'))
if self.tagger:
self.tagger.model.end_training()
self.tagger.model.dump(path.join(data_dir, 'pos', 'model'))
strings_loc = path.join(data_dir, 'vocab', 'strings.json')
with io.open(strings_loc, 'w', encoding='utf8') as file_:
self.vocab.strings.dump(file_)
self.vocab.dump(path.join(data_dir, 'vocab', 'lexemes.bin'))
2015-08-25 16:37:17 +03:00
if self.tagger:
tagger_freqs = list(self.tagger.freqs[TAG].items())
else:
tagger_freqs = []
if self.parser:
dep_freqs = list(self.parser.moves.freqs[DEP].items())
head_freqs = list(self.parser.moves.freqs[HEAD].items())
else:
dep_freqs = []
head_freqs = []
if self.entity:
entity_iob_freqs = list(self.entity.moves.freqs[ENT_IOB].items())
entity_type_freqs = list(self.entity.moves.freqs[ENT_TYPE].items())
else:
entity_iob_freqs = []
entity_type_freqs = []
2015-08-25 16:37:17 +03:00
with open(path.join(data_dir, 'vocab', 'serializer.json'), 'w') as file_:
file_.write(
json.dumps([
(TAG, tagger_freqs),
(DEP, dep_freqs),
(ENT_IOB, entity_iob_freqs),
(ENT_TYPE, entity_type_freqs),
(HEAD, head_freqs)
]))