explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2024-12-24 17:06:29 +03:00
Code Issues Packages Projects Releases Wiki Activity

* draft de/serialization functions in doc.pyx

Browse Source
This commit is contained in:
Matthew Honnibal 2015-07-16 01:16:33 +02:00
parent 9d956b07e9
commit 0e07c1ed2a
1 changed files with 29 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

58
spacy/tokens/doc.pyx
View File

@ -278,7 +278,7 @@ cdef class Doc:
self.data[i].lex = &EMPTY_LEXEME self.data[i].lex = &EMPTY_LEXEME
cdef int set_parse(self, const TokenC* parsed) except -1: cdef int set_parse(self, const TokenC* parsed) except -1:
# TODO: This method is fairly misleading atm. It's used by GreedyParser # TODO: This method is fairly misleading atm. It's used by Parser
# to actually apply the parse calculated. Need to rethink this. # to actually apply the parse calculated. Need to rethink this.
self.is_parsed = True self.is_parsed = True
for i in range(self.length): for i in range(self.length):
@ -369,40 +369,40 @@ cdef class Doc:
# Return the merged Python object # Return the merged Python object
return self[start] return self[start]
def serialize(self, bits=None): def serialize(self, codecs, bits=None):
if bits is None: if bits is None:
bits = BitArray() bits = BitArray()
codec = self.vocab.codec array = self.to_array([codec.attr_id for codec in codecs])
ids = numpy.zeros(shape=(len(self),), dtype=numpy.uint32) for i, codec in enumerate(codecs):
cdef int i codec.encode(array[i,], bits)
for i in range(self.length):
ids[i] = self.data[i].lex.id
bits = codec.encode(ids, bits=bits)
for i in range(self.length):
bits.append(self.data[i].spacy)
return bits return bits
@staticmethod @staticmethod
def deserialize(Vocab vocab, bits): def deserialize(Vocab vocab, bits):
biterator = iter(bits) biterator = iter(bits)
ids = vocab.codec.decode(biterator) ids = vocab.lex_codec.decode(bits)
spaces = []
for bit in biterator:
spaces.append(bit)
if len(spaces) == len(ids):
break
string = u''
cdef const LexemeC* lex
for id_, space in zip(ids, spaces):
lex = vocab.lexemes[id_]
string += vocab.strings[lex.orth]
if space:
string += u' '
cdef Doc doc = Doc(vocab) cdef Doc doc = Doc(vocab)
cdef bint has_space = False cdef int id_
cdef int idx = 0 for id_ in ids:
for i, id_ in enumerate(ids): is_spacy = biterator.next()
lex = vocab.lexemes[id_] doc.push_back(vocab.lexemes.at(id_), is_spacy)
has_space = spaces[i]
doc.push_back(lex, has_space) cdef int i
for codec in vocab.annotation_codecs:
values = codec.decode(biterator)
if codec.attr_id == HEAD:
for i, head in enumerate(values):
doc.data[i].head = head
elif codec.attr_id == TAG:
for i, tag in enumerate(values):
doc.data[i].tag = tag
elif codec.attr_id == DEP:
for i, dep in enumerate(values):
doc.data[i].dep = dep
elif codec.attr_id == ENT_IOB:
for i, ent_iob in enumerate(values):
doc.data[i].ent_iob = ent_iob
elif codec.attr_id == ENT_TYPE:
for i, ent_type in enumerate(values):
doc.data[i].ent_type = ent_type
return doc return doc