spaCy/spacy/tokens/doc.pyx

900 lines
36 KiB
Cython
Raw Normal View History

2017-04-15 11:05:15 +00:00
# coding: utf8
2017-05-09 16:45:18 +00:00
# cython: infer_types=True
# cython: bounds_check=False
2017-04-15 11:05:15 +00:00
from __future__ import unicode_literals
2017-04-15 11:05:15 +00:00
cimport cython
cimport numpy as np
import numpy
import numpy.linalg
import struct
import dill
2017-04-15 11:05:15 +00:00
from libc.string cimport memcpy, memset
from libc.math cimport sqrt
from .span cimport Span
from .token cimport Token
2017-05-13 11:04:40 +00:00
from .span cimport Span
from .token cimport Token
from .printers import parse_tree
from ..lexeme cimport Lexeme, EMPTY_LEXEME
from ..typedefs cimport attr_t, flags_t
2017-05-28 12:06:40 +00:00
from ..attrs import intify_attrs
from ..attrs cimport attr_id_t
from ..attrs cimport ID, ORTH, NORM, LOWER, SHAPE, PREFIX, SUFFIX, LENGTH, CLUSTER
from ..attrs cimport LENGTH, POS, LEMMA, TAG, DEP, HEAD, SPACY, ENT_IOB, ENT_TYPE
from ..attrs cimport SENT_START
2017-05-13 11:04:40 +00:00
from ..parts_of_speech cimport CCONJ, PUNCT, NOUN, univ_pos_t
from ..util import normalize_slice
2017-04-15 11:05:15 +00:00
from ..compat import is_config
from .. import about
from .. import util
DEF PADDING = 5
cdef int bounds_check(int i, int length, int padding) except -1:
if (i + padding) < 0:
raise IndexError
if (i - padding) >= length:
raise IndexError
cdef attr_t get_token_attr(const TokenC* token, attr_id_t feat_name) nogil:
if feat_name == LEMMA:
return token.lemma
elif feat_name == POS:
return token.pos
elif feat_name == TAG:
return token.tag
elif feat_name == DEP:
return token.dep
elif feat_name == HEAD:
return token.head
elif feat_name == SENT_START:
return token.sent_start
elif feat_name == SPACY:
return token.spacy
elif feat_name == ENT_IOB:
return token.ent_iob
elif feat_name == ENT_TYPE:
return token.ent_type
else:
return Lexeme.get_struct_attr(token.lex, feat_name)
def _get_chunker(lang):
2017-06-04 20:53:05 +00:00
try:
cls = util.get_lang_class(lang)
except ImportError:
return None
except KeyError:
return None
return cls.Defaults.syntax_iterators.get(u'noun_chunks')
cdef class Doc:
"""A sequence of Token objects. Access sentences and named entities, export
annotations to numpy arrays, losslessly serialize to compressed binary strings.
The `Doc` object holds an array of `TokenC` structs. The Python-level
`Token` and `Span` objects are views of this array, i.e. they don't own
the data themselves.
EXAMPLE: Construction 1
>>> doc = nlp(u'Some text')
Construction 2
>>> from spacy.tokens import Doc
>>> doc = Doc(nlp.vocab, words=[u'hello', u'world', u'!'], spaces=[True, False, False])
"""
def __init__(self, Vocab vocab, words=None, spaces=None, orths_and_spaces=None):
"""Create a Doc object.
2016-09-28 09:15:13 +00:00
vocab (Vocab): A vocabulary object, which must match any models you want
to use (e.g. tokenizer, parser, entity recognizer).
words (list or None): A list of unicode strings to add to the document
as words. If `None`, defaults to empty list.
spaces (list or None): A list of boolean values, of the same length as
words. True means that the word is followed by a space, False means
it is not. If `None`, defaults to `[True]*len(words)`
RETURNS (Doc): The newly constructed object.
2017-04-15 11:05:15 +00:00
"""
self.vocab = vocab
size = 20
self.mem = Pool()
# Guarantee self.lex[i-x], for any i >= 0 and x < padding is in bounds
# However, we need to remember the true starting places, so that we can
# realloc.
data_start = <TokenC*>self.mem.alloc(size + (PADDING*2), sizeof(TokenC))
cdef int i
for i in range(size + (PADDING*2)):
data_start[i].lex = &EMPTY_LEXEME
data_start[i].l_edge = i
data_start[i].r_edge = i
2015-11-03 13:15:14 +00:00
self.c = data_start + PADDING
self.max_length = size
self.length = 0
self.is_tagged = False
self.is_parsed = False
self.sentiment = 0.0
2017-07-21 22:34:15 +00:00
self.cats = {}
2016-10-19 19:15:16 +00:00
self.user_hooks = {}
self.user_token_hooks = {}
self.user_span_hooks = {}
2016-10-17 13:23:47 +00:00
self.tensor = numpy.zeros((0,), dtype='float32')
2016-10-17 09:43:22 +00:00
self.user_data = {}
self._py_tokens = []
self._vector = None
self.noun_chunks_iterator = _get_chunker(self.vocab.lang)
2016-09-21 12:52:05 +00:00
cdef unicode orth
cdef bint has_space
if orths_and_spaces is None and words is not None:
if spaces is None:
spaces = [True] * len(words)
2016-10-16 16:16:42 +00:00
elif len(spaces) != len(words):
raise ValueError(
"Arguments 'words' and 'spaces' should be sequences of the "
"same length, or 'spaces' should be left default at None. "
"spaces should be a sequence of booleans, with True meaning "
"that the word owns a ' ' character following it.")
orths_and_spaces = zip(words, spaces)
2016-09-21 12:52:05 +00:00
if orths_and_spaces is not None:
for orth_space in orths_and_spaces:
if isinstance(orth_space, unicode):
orth = orth_space
has_space = True
elif isinstance(orth_space, bytes):
raise ValueError(
"orths_and_spaces expects either List(unicode) or "
"List((unicode, bool)). Got bytes instance: %s" % (str(orth_space)))
else:
orth, has_space = orth_space
# Note that we pass self.mem here --- we have ownership, if LexemeC
# must be created.
self.push_back(
<const LexemeC*>self.vocab.get(self.mem, orth), has_space)
# Tough to decide on policy for this. Is an empty doc tagged and parsed?
# There's no information we'd like to add to it, so I guess so?
if self.length == 0:
self.is_tagged = True
self.is_parsed = True
def __getitem__(self, object i):
"""Get a `Token` or `Span` object.
2017-05-18 22:30:51 +00:00
i (int or tuple) The index of the token, or the slice of the document to get.
RETURNS (Token or Span): The token at `doc[i]]`, or the span at
`doc[start : end]`.
EXAMPLE:
>>> doc[i]
Get the `Token` object at position `i`, where `i` is an integer.
Negative indexing is supported, and follows the usual Python
semantics, i.e. `doc[-2]` is `doc[len(doc) - 2]`.
>>> doc[start : end]]
Get a `Span` object, starting at position `start` and ending at
position `end`, where `start` and `end` are token indices. For
instance, `doc[2:5]` produces a span consisting of tokens 2, 3 and 4.
Stepped slices (e.g. `doc[start : end : step]`) are not supported,
as `Span` objects must be contiguous (cannot have gaps). You can use
negative indices and open-ended ranges, which have their normal
Python semantics.
2017-04-15 11:05:15 +00:00
"""
if isinstance(i, slice):
start, stop = normalize_slice(len(self), i.start, i.stop, i.step)
return Span(self, start, stop, label=0)
if i < 0:
i = self.length + i
bounds_check(i, self.length, PADDING)
if self._py_tokens[i] is not None:
return self._py_tokens[i]
else:
2015-11-03 13:15:14 +00:00
return Token.cinit(self.vocab, &self.c[i], i, self)
def __iter__(self):
"""Iterate over `Token` objects, from which the annotations can be
easily accessed. This is the main way of accessing `Token` objects,
which are the main way annotations are accessed from Python. If faster-
than-Python speeds are required, you can instead access the annotations
as a numpy array, or access the underlying C data directly from Cython.
EXAMPLE:
>>> for token in doc
2017-04-15 11:05:15 +00:00
"""
cdef int i
for i in range(self.length):
if self._py_tokens[i] is not None:
yield self._py_tokens[i]
else:
2015-11-03 13:15:14 +00:00
yield Token.cinit(self.vocab, &self.c[i], i, self)
def __len__(self):
"""The number of tokens in the document.
2017-05-19 16:47:39 +00:00
RETURNS (int): The number of tokens in the document.
EXAMPLE:
>>> len(doc)
2017-04-15 11:05:15 +00:00
"""
return self.length
def __unicode__(self):
return u''.join([t.text_with_ws for t in self])
2015-11-02 18:22:18 +00:00
def __bytes__(self):
return u''.join([t.text_with_ws for t in self]).encode('utf-8')
2015-11-02 18:22:18 +00:00
2015-07-24 01:49:30 +00:00
def __str__(self):
2017-04-15 11:05:15 +00:00
if is_config(python3=True):
2015-11-02 18:22:18 +00:00
return self.__unicode__()
return self.__bytes__()
2015-07-24 01:49:30 +00:00
def __repr__(self):
2015-11-02 18:22:18 +00:00
return self.__str__()
2016-11-24 10:47:20 +00:00
@property
def doc(self):
return self
def char_span(self, int start_idx, int end_idx, label=0, vector=None):
"""Create a `Span` object from the slice `doc.text[start : end]`.
doc (Doc): The parent document.
start (int): The index of the first character of the span.
end (int): The index of the first character after the span.
label (uint64 or string): A label to attach to the Span, e.g. for named entities.
vector (ndarray[ndim=1, dtype='float32']): A meaning representation of the span.
RETURNS (Span): The newly constructed object.
"""
if not isinstance(label, int):
label = self.vocab.strings.add(label)
cdef int start = token_by_start(self.c, self.length, start_idx)
if start == -1:
return None
cdef int end = token_by_end(self.c, self.length, end_idx)
if end == -1:
return None
# Currently we have the token index, we want the range-end index
end += 1
cdef Span span = Span(self, start, end, label=label, vector=vector)
return span
def similarity(self, other):
"""Make a semantic similarity estimate. The default estimate is cosine
2016-11-01 11:25:36 +00:00
similarity using an average of word vectors.
other (object): The object to compare with. By default, accepts `Doc`,
`Span`, `Token` and `Lexeme` objects.
RETURNS (float): A scalar similarity score. Higher is more similar.
2017-04-15 11:05:15 +00:00
"""
if 'similarity' in self.user_hooks:
return self.user_hooks['similarity'](self, other)
2015-09-22 00:10:01 +00:00
if self.vector_norm == 0 or other.vector_norm == 0:
return 0.0
return numpy.dot(self.vector, other.vector) / (self.vector_norm * other.vector_norm)
property has_vector:
"""A boolean value indicating whether a word vector is associated with
the object.
RETURNS (bool): Whether a word vector is associated with the object.
2017-04-15 11:05:15 +00:00
"""
def __get__(self):
if 'has_vector' in self.user_hooks:
return self.user_hooks['has_vector'](self)
elif any(token.has_vector for token in self):
return True
2017-06-04 19:25:30 +00:00
elif self.tensor is not None:
return True
else:
return False
property vector:
"""A real-valued meaning representation. Defaults to an average of the
token vectors.
RETURNS (numpy.ndarray[ndim=1, dtype='float32']): A 1D numpy array
representing the document's semantics.
2017-04-15 11:05:15 +00:00
"""
def __get__(self):
if 'vector' in self.user_hooks:
return self.user_hooks['vector'](self)
if self._vector is not None:
return self._vector
elif self.has_vector and len(self):
2017-08-22 17:46:52 +00:00
vector = numpy.zeros((self.vocab.vectors_length,), dtype='f')
for token in self.c[:self.length]:
vector += self.vocab.get_vector(token.lex.orth)
self._vector = vector / len(self)
return self._vector
2017-06-04 19:25:30 +00:00
elif self.tensor is not None:
self._vector = self.tensor.mean(axis=0)
return self._vector
else:
return numpy.zeros((self.vocab.vectors_length,), dtype='float32')
def __set__(self, value):
self._vector = value
property vector_norm:
"""The L2 norm of the document's vector representation.
RETURNS (float): The L2 norm of the vector representation.
"""
def __get__(self):
if 'vector_norm' in self.user_hooks:
return self.user_hooks['vector_norm'](self)
cdef float value
cdef double norm = 0
if self._vector_norm is None:
norm = 0.0
for value in self.vector:
norm += value * value
self._vector_norm = sqrt(norm) if norm != 0 else 0
return self._vector_norm
def __set__(self, value):
self._vector_norm = value
property text:
"""A unicode representation of the document text.
RETURNS (unicode): The original verbatim text of the document.
2017-04-15 11:05:15 +00:00
"""
2016-11-01 11:25:36 +00:00
def __get__(self):
return u''.join(t.text_with_ws for t in self)
2016-11-01 11:25:36 +00:00
property text_with_ws:
"""An alias of `Doc.text`, provided for duck-type compatibility with
`Span` and `Token`.
RETURNS (unicode): The original verbatim text of the document.
2017-04-15 11:05:15 +00:00
"""
2016-11-01 11:25:36 +00:00
def __get__(self):
return self.text
property ents:
"""Iterate over the entities in the document. Yields named-entity `Span`
objects, if the entity recognizer has been applied to the document.
YIELDS (Span): Entities in the document.
EXAMPLE: Iterate over the span to get individual Token objects, or access
the label:
>>> tokens = nlp(u'Mr. Best flew to New York on Saturday morning.')
>>> ents = list(tokens.ents)
>>> assert ents[0].label == 346
>>> assert ents[0].label_ == 'PERSON'
>>> assert ents[0].orth_ == 'Best'
>>> assert ents[0].text == 'Mr. Best'
2017-04-15 11:05:15 +00:00
"""
def __get__(self):
cdef int i
cdef const TokenC* token
cdef int start = -1
2017-05-28 16:09:27 +00:00
cdef attr_t label = 0
output = []
for i in range(self.length):
2015-11-03 13:15:14 +00:00
token = &self.c[i]
if token.ent_iob == 1:
assert start != -1
elif token.ent_iob == 2 or token.ent_iob == 0:
if start != -1:
output.append(Span(self, start, i, label=label))
start = -1
label = 0
elif token.ent_iob == 3:
if start != -1:
output.append(Span(self, start, i, label=label))
start = i
label = token.ent_type
if start != -1:
output.append(Span(self, start, self.length, label=label))
return tuple(output)
def __set__(self, ents):
# TODO:
# 1. Allow negative matches
# 2. Ensure pre-set NERs are not over-written during statistical prediction
# 3. Test basic data-driven ORTH gazetteer
# 4. Test more nuanced date and currency regex
cdef int i
for i in range(self.length):
2015-11-03 13:15:14 +00:00
self.c[i].ent_type = 0
# At this point we don't know whether the NER has run over the
# Doc. If the ent_iob is missing, leave it missing.
if self.c[i].ent_iob != 0:
self.c[i].ent_iob = 2 # Means O. Non-O are set from ents.
cdef attr_t ent_type
cdef int start, end
for ent_info in ents:
if isinstance(ent_info, Span):
ent_id = ent_info.ent_id
ent_type = ent_info.label
start = ent_info.start
end = ent_info.end
elif len(ent_info) == 3:
ent_type, start, end = ent_info
else:
ent_id, ent_type, start, end = ent_info
2015-08-06 15:28:43 +00:00
if ent_type is None or ent_type < 0:
# Mark as O
for i in range(start, end):
2015-11-03 13:15:14 +00:00
self.c[i].ent_type = 0
self.c[i].ent_iob = 2
else:
# Mark (inside) as I
for i in range(start, end):
2015-11-03 13:15:14 +00:00
self.c[i].ent_type = ent_type
self.c[i].ent_iob = 1
# Set start as B
2015-11-03 13:15:14 +00:00
self.c[start].ent_iob = 3
2016-09-28 09:39:49 +00:00
property noun_chunks:
"""Iterate over the base noun phrases in the document. Yields base
noun-phrase #[code Span] objects, if the document has been syntactically
parsed. A base noun phrase, or "NP chunk", is a noun phrase that does
not permit other NPs to be nested within it so no NP-level
coordination, no prepositional phrases, and no relative clauses.
YIELDS (Span): Noun chunks in the document.
2017-04-15 11:05:15 +00:00
"""
2016-09-28 09:15:13 +00:00
def __get__(self):
if not self.is_parsed:
raise ValueError(
"noun_chunks requires the dependency parse, which "
"requires data to be installed. For more info, see the "
"documentation: \n%s\n" % about.__docs_models__)
2016-09-28 09:15:13 +00:00
# Accumulate the result before beginning to iterate over it. This prevents
# the tokenisation from being changed out from under us during the iteration.
# The tricky thing here is that Span accepts its tokenisation changing,
# so it's okay once we have the Span objects. See Issue #375
spans = []
for start, end, label in self.noun_chunks_iterator(self):
spans.append(Span(self, start, end, label=label))
for span in spans:
yield span
property sents:
"""Iterate over the sentences in the document. Yields sentence `Span`
objects. Sentence spans have no label. To improve accuracy on informal
texts, spaCy calculates sentence boundaries from the syntactic
dependency parse. If the parser is disabled, the `sents` iterator will
be unavailable.
EXAMPLE:
>>> doc = nlp("This is a sentence. Here's another...")
>>> assert [s.root.text for s in doc.sents] == ["is", "'s"]
"""
2016-09-28 09:15:13 +00:00
def __get__(self):
if 'sents' in self.user_hooks:
2017-06-03 09:31:11 +00:00
yield from self.user_hooks['sents'](self)
return
2016-09-28 09:15:13 +00:00
if not self.is_parsed:
raise ValueError(
"sentence boundary detection requires the dependency parse, which "
"requires data to be installed. For more info, see the "
"documentation: \n%s\n" % about.__docs_models__)
2016-09-28 09:15:13 +00:00
cdef int i
start = 0
for i in range(1, self.length):
if self.c[i].sent_start:
yield Span(self, start, i)
start = i
if start != self.length:
yield Span(self, start, self.length)
cdef int push_back(self, LexemeOrToken lex_or_tok, bint has_space) except -1:
if self.length == 0:
# Flip these to false when we see the first token.
self.is_tagged = False
self.is_parsed = False
if self.length == self.max_length:
self._realloc(self.length * 2)
2015-11-03 13:15:14 +00:00
cdef TokenC* t = &self.c[self.length]
if LexemeOrToken is const_TokenC_ptr:
t[0] = lex_or_tok[0]
else:
t.lex = lex_or_tok
if self.length == 0:
t.idx = 0
else:
t.idx = (t-1).idx + (t-1).lex.length + (t-1).spacy
t.l_edge = self.length
t.r_edge = self.length
assert t.lex.orth != 0
t.spacy = has_space
self.length += 1
self._py_tokens.append(None)
return t.idx + t.lex.length + t.spacy
@cython.boundscheck(False)
cpdef np.ndarray to_array(self, object py_attr_ids):
"""Given a list of M attribute IDs, export the tokens to a numpy
`ndarray` of shape `(N, M)`, where `N` is the length of the document.
The values will be 32-bit integers.
attr_ids (list[int]): A list of attribute ID ints.
RETURNS (numpy.ndarray[long, ndim=2]): A feature matrix, with one row
per word, and one column per attribute indicated in the input
`attr_ids`.
EXAMPLE:
>>> from spacy.attrs import LOWER, POS, ENT_TYPE, IS_ALPHA
>>> doc = nlp(text)
>>> # All strings mapped to integers, for easy export to numpy
>>> np_array = doc.to_array([LOWER, POS, ENT_TYPE, IS_ALPHA])
"""
cdef int i, j
cdef attr_id_t feature
cdef np.ndarray[attr_t, ndim=2] output
# Make an array from the attributes --- otherwise our inner loop is Python
# dict iteration.
2017-05-28 12:06:40 +00:00
cdef np.ndarray[attr_t, ndim=1] attr_ids = numpy.asarray(py_attr_ids, dtype=numpy.uint64)
output = numpy.ndarray(shape=(self.length, len(attr_ids)), dtype=numpy.uint64)
for i in range(self.length):
for j, feature in enumerate(attr_ids):
2015-11-03 13:15:14 +00:00
output[i, j] = get_token_attr(&self.c[i], feature)
return output
2015-07-14 01:20:09 +00:00
def count_by(self, attr_id_t attr_id, exclude=None, PreshCounter counts=None):
"""Count the frequencies of a given attribute. Produces a dict of
`{attribute (int): count (ints)}` frequencies, keyed by the values of
the given attribute ID.
attr_id (int): The attribute ID to key the counts.
RETURNS (dict): A dictionary mapping attributes to integer counts.
EXAMPLE:
>>> from spacy import attrs
>>> doc = nlp(u'apple apple orange banana')
>>> tokens.count_by(attrs.ORTH)
{12800L: 1, 11880L: 2, 7561L: 1}
>>> tokens.to_array([attrs.ORTH])
array([[11880], [11880], [7561], [12800]])
"""
cdef int i
cdef attr_t attr
cdef size_t count
2015-07-14 01:20:09 +00:00
if counts is None:
counts = PreshCounter()
2015-07-14 01:20:09 +00:00
output_dict = True
else:
output_dict = False
# Take this check out of the loop, for a bit of extra speed
if exclude is None:
for i in range(self.length):
2015-11-03 13:15:14 +00:00
counts.inc(get_token_attr(&self.c[i], attr_id), 1)
2015-07-14 01:20:09 +00:00
else:
for i in range(self.length):
if not exclude(self[i]):
2015-11-03 13:15:14 +00:00
attr = get_token_attr(&self.c[i], attr_id)
2015-07-14 01:20:09 +00:00
counts.inc(attr, 1)
if output_dict:
return dict(counts)
def _realloc(self, new_size):
self.max_length = new_size
n = new_size + (PADDING * 2)
# What we're storing is a "padded" array. We've jumped forward PADDING
# places, and are storing the pointer to that. This way, we can access
# words out-of-bounds, and get out-of-bounds markers.
# Now that we want to realloc, we need the address of the true start,
# so we jump the pointer back PADDING places.
2015-11-03 13:15:14 +00:00
cdef TokenC* data_start = self.c - PADDING
data_start = <TokenC*>self.mem.realloc(data_start, n * sizeof(TokenC))
2015-11-03 13:15:14 +00:00
self.c = data_start + PADDING
cdef int i
for i in range(self.length, self.max_length + PADDING):
2015-11-03 13:15:14 +00:00
self.c[i].lex = &EMPTY_LEXEME
2016-01-30 19:27:52 +00:00
cdef void set_parse(self, const TokenC* parsed) nogil:
# TODO: This method is fairly misleading atm. It's used by Parser
# to actually apply the parse calculated. Need to rethink this.
# Probably we should use from_array?
self.is_parsed = True
for i in range(self.length):
2015-11-03 13:15:14 +00:00
self.c[i] = parsed[i]
def from_array(self, attrs, array):
if SENT_START in attrs and HEAD in attrs:
raise ValueError(
"Conflicting attributes specified in doc.from_array():\n"
"(HEAD, SENT_START)\n"
"The HEAD attribute currently sets sentence boundaries implicitly,\n"
"based on the tree structure. This means the HEAD attribute would "
"potentially override the sentence boundaries set by SENT_START.\n"
"See https://github.com/spacy-io/spaCy/issues/235 for details and "
"workarounds, and to propose solutions.")
cdef int i, col
cdef attr_id_t attr_id
2015-11-03 13:15:14 +00:00
cdef TokenC* tokens = self.c
cdef int length = len(array)
2017-05-09 16:45:18 +00:00
# Get set up for fast loading
cdef Pool mem = Pool()
cdef int n_attrs = len(attrs)
attr_ids = <attr_id_t*>mem.alloc(n_attrs, sizeof(attr_id_t))
for i, attr_id in enumerate(attrs):
attr_ids[i] = attr_id
# Now load the data
for i in range(self.length):
token = &self.c[i]
for j in range(n_attrs):
Token.set_struct_attr(token, attr_ids[j], array[i, j])
# Auxiliary loading logic
for col, attr_id in enumerate(attrs):
2017-05-09 16:45:18 +00:00
if attr_id == TAG:
for i in range(length):
2017-05-09 16:45:18 +00:00
if array[i, col] != 0:
self.vocab.morphology.assign_tag(&tokens[i], array[i, col])
2015-11-03 13:15:14 +00:00
set_children_from_heads(self.c, self.length)
self.is_parsed = bool(HEAD in attrs or DEP in attrs)
self.is_tagged = bool(TAG in attrs or POS in attrs)
return self
def to_disk(self, path, **exclude):
"""Save the current state to a directory.
path (unicode or Path): A path to a directory, which will be created if
it doesn't exist. Paths may be either strings or `Path`-like objects.
"""
with path.open('wb') as file_:
file_.write(self.to_bytes(**exclude))
def from_disk(self, path, **exclude):
"""Loads state from a directory. Modifies the object in place and
returns it.
path (unicode or Path): A path to a directory. Paths may be either
strings or `Path`-like objects.
RETURNS (Doc): The modified `Doc` object.
"""
with path.open('rb') as file_:
bytes_data = file_.read()
self.from_bytes(bytes_data, **exclude)
def to_bytes(self, **exclude):
"""Serialize, i.e. export the document contents to a binary string.
RETURNS (bytes): A losslessly serialized copy of the `Doc`, including
all annotations.
2017-04-15 11:05:15 +00:00
"""
array_head = [LENGTH,SPACY,TAG,LEMMA,HEAD,DEP,ENT_IOB,ENT_TYPE]
serializers = {
'text': lambda: self.text,
'array_head': lambda: array_head,
'array_body': lambda: self.to_array(array_head),
'sentiment': lambda: self.sentiment,
'tensor': lambda: self.tensor,
'user_data': lambda: self.user_data
}
return util.to_bytes(serializers, exclude)
def from_bytes(self, bytes_data, **exclude):
"""Deserialize, i.e. import the document contents from a binary string.
data (bytes): The string to load from.
RETURNS (Doc): Itself.
2017-04-15 11:05:15 +00:00
"""
if self.length != 0:
raise ValueError("Cannot load into non-empty Doc")
deserializers = {
'text': lambda b: None,
'array_head': lambda b: None,
'array_body': lambda b: None,
'sentiment': lambda b: None,
'tensor': lambda b: None,
'user_data': lambda user_data: self.user_data.update(user_data)
}
msg = util.from_bytes(bytes_data, deserializers, exclude)
2017-05-28 12:06:40 +00:00
cdef attr_t[:, :] attrs
cdef int i, start, end, has_space
self.sentiment = msg['sentiment']
self.tensor = msg['tensor']
start = 0
cdef const LexemeC* lex
cdef unicode orth_
text = msg['text']
attrs = msg['array_body']
for i in range(attrs.shape[0]):
end = start + attrs[i, 0]
has_space = attrs[i, 1]
orth_ = text[start:end]
lex = self.vocab.get(self.mem, orth_)
self.push_back(lex, has_space)
start = end + has_space
self.from_array(msg['array_head'][2:],
attrs[:, 2:])
2017-05-09 16:45:18 +00:00
return self
def merge(self, int start_idx, int end_idx, *args, **attributes):
"""Retokenize the document, such that the span at `doc.text[start_idx : end_idx]`
is merged into a single token. If `start_idx` and `end_idx `do not mark
start and end token boundaries, the document remains unchanged.
start_idx (int): The character index of the start of the slice to merge.
end_idx (int): The character index after the end of the slice to merge.
**attributes: Attributes to assign to the merged token. By default,
attributes are inherited from the syntactic root token of the span.
RETURNS (Token): The newly merged token, or `None` if the start and end
indices did not fall at token boundaries.
2016-11-01 11:25:36 +00:00
"""
cdef unicode tag, lemma, ent_type
if len(args) == 3:
# TODO: Warn deprecation
tag, lemma, ent_type = args
2017-05-28 12:06:40 +00:00
attributes[TAG] = tag
attributes[LEMMA] = lemma
attributes[ENT_TYPE] = ent_type
elif not args:
if "label" in attributes and 'ent_type' not in attributes:
2017-05-28 21:22:45 +00:00
if isinstance(attributes["label"], int):
attributes[ENT_TYPE] = attributes["label"]
else:
2017-05-28 21:30:40 +00:00
attributes[ENT_TYPE] = self.vocab.strings[attributes["label"]]
if 'ent_type' in attributes:
attributes[ENT_TYPE] = attributes['ent_type']
elif args:
raise ValueError(
"Doc.merge received %d non-keyword arguments. "
"Expected either 3 arguments (deprecated), or 0 (use keyword arguments). "
"Arguments supplied:\n%s\n"
"Keyword arguments:%s\n" % (len(args), repr(args), repr(attributes)))
# More deprecated attribute handling =/
if 'label' in attributes:
attributes['ent_type'] = attributes.pop('label')
2017-05-28 12:06:40 +00:00
attributes = intify_attrs(attributes, strings_map=self.vocab.strings)
cdef int start = token_by_start(self.c, self.length, start_idx)
if start == -1:
return None
cdef int end = token_by_end(self.c, self.length, end_idx)
if end == -1:
return None
# Currently we have the token index, we want the range-end index
end += 1
cdef Span span = self[start:end]
# Get LexemeC for newly merged token
new_orth = ''.join([t.text_with_ws for t in span])
if span[-1].whitespace_:
new_orth = new_orth[:-len(span[-1].whitespace_)]
cdef const LexemeC* lex = self.vocab.get(self.mem, new_orth)
# House the new merged token where it starts
2015-11-03 13:15:14 +00:00
cdef TokenC* token = &self.c[start]
token.spacy = self.c[end-1].spacy
2017-05-28 12:06:40 +00:00
for attr_name, attr_value in attributes.items():
if attr_name == TAG:
2017-06-03 09:31:18 +00:00
self.vocab.morphology.assign_tag(token, attr_value)
2017-05-28 12:06:40 +00:00
else:
Token.set_struct_attr(token, attr_name, attr_value)
# Begin by setting all the head indices to absolute token positions
# This is easier to work with for now than the offsets
# Before thinking of something simpler, beware the case where a dependency
# bridges over the entity. Here the alignment of the tokens changes.
span_root = span.root.i
token.dep = span.root.dep
# We update token.lex after keeping span root and dep, since
# setting token.lex will change span.start and span.end properties
# as it modifies the character offsets in the doc
token.lex = lex
for i in range(self.length):
2015-11-03 13:15:14 +00:00
self.c[i].head += i
# Set the head of the merged token, and its dep relation, from the Span
2015-11-03 13:15:14 +00:00
token.head = self.c[span_root].head
# Adjust deps before shrinking tokens
# Tokens which point into the merged token should now point to it
# Subtract the offset from all tokens which point to >= end
offset = (end - start) - 1
for i in range(self.length):
2015-11-03 13:15:14 +00:00
head_idx = self.c[i].head
if start <= head_idx < end:
2015-11-03 13:15:14 +00:00
self.c[i].head = start
elif head_idx >= end:
2015-11-03 13:15:14 +00:00
self.c[i].head -= offset
# Now compress the token array
for i in range(end, self.length):
2015-11-03 13:15:14 +00:00
self.c[i - offset] = self.c[i]
for i in range(self.length - offset, self.length):
2015-11-03 13:15:14 +00:00
memset(&self.c[i], 0, sizeof(TokenC))
self.c[i].lex = &EMPTY_LEXEME
self.length -= offset
for i in range(self.length):
# ...And, set heads back to a relative position
2015-11-03 13:15:14 +00:00
self.c[i].head -= i
# Set the left/right children, left/right edges
2015-11-03 13:15:14 +00:00
set_children_from_heads(self.c, self.length)
# Clear the cached Python objects
self._py_tokens = [None] * self.length
# Return the merged Python object
return self[start]
def print_tree(self, light=False, flat=False):
"""Returns the parse trees in JSON (dict) format.
light (bool): Don't include lemmas or entities.
flat (bool): Don't include arcs or modifiers.
RETURNS (dict): Parse tree as dict.
EXAMPLE:
>>> doc = nlp('Bob brought Alice the pizza. Alice ate the pizza.')
>>> trees = doc.print_tree()
>>> trees[1]
{'modifiers': [
{'modifiers': [], 'NE': 'PERSON', 'word': 'Alice', 'arc': 'nsubj',
'POS_coarse': 'PROPN', 'POS_fine': 'NNP', 'lemma': 'Alice'},
{'modifiers': [
{'modifiers': [], 'NE': '', 'word': 'the', 'arc': 'det',
'POS_coarse': 'DET', 'POS_fine': 'DT', 'lemma': 'the'}],
'NE': '', 'word': 'pizza', 'arc': 'dobj', 'POS_coarse': 'NOUN',
'POS_fine': 'NN', 'lemma': 'pizza'},
{'modifiers': [], 'NE': '', 'word': '.', 'arc': 'punct',
'POS_coarse': 'PUNCT', 'POS_fine': '.', 'lemma': '.'}],
'NE': '', 'word': 'ate', 'arc': 'ROOT', 'POS_coarse': 'VERB',
'POS_fine': 'VBD', 'lemma': 'eat'}
"""
return parse_tree(self, light=light, flat=flat)
cdef int token_by_start(const TokenC* tokens, int length, int start_char) except -2:
cdef int i
for i in range(length):
if tokens[i].idx == start_char:
return i
else:
return -1
cdef int token_by_end(const TokenC* tokens, int length, int end_char) except -2:
cdef int i
for i in range(length):
if tokens[i].idx + tokens[i].lex.length == end_char:
return i
else:
return -1
cdef int set_children_from_heads(TokenC* tokens, int length) except -1:
cdef TokenC* head
cdef TokenC* child
cdef int i
# Set number of left/right children to 0. We'll increment it in the loops.
for i in range(length):
tokens[i].l_kids = 0
tokens[i].r_kids = 0
tokens[i].l_edge = i
tokens[i].r_edge = i
# Set left edges
for i in range(length):
child = &tokens[i]
head = &tokens[i + child.head]
if child < head:
if child.l_edge < head.l_edge:
head.l_edge = child.l_edge
head.l_kids += 1
# Set right edges --- same as above, but iterate in reverse
for i in range(length-1, -1, -1):
child = &tokens[i]
head = &tokens[i + child.head]
if child > head:
if child.r_edge > head.r_edge:
head.r_edge = child.r_edge
head.r_kids += 1
# Set sentence starts
for i in range(length):
if tokens[i].head == 0 and tokens[i].dep != 0:
tokens[tokens[i].l_edge].sent_start = True