Rooba
/
spaCy
mirror of https://github.com/explosion/spaCy.git
1
1
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Revert "Changes to vocab for new stringstore scheme" 

This reverts commit a51149a717.
This commit is contained in:
Matthew Honnibal 2016-09-30 20:10:30 +02:00
parent 1e7d0af127
commit 1f1cd5013f
1 changed files with 13 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

37
spacy/vocab.pyx
View File

@ -104,7 +104,7 @@ cdef class Vocab:
# Need to rethink this.
for name in symbols.NAMES + list(sorted(tag_map.keys())):
if name:
self.strings.intern(name)
_ = self.strings[name]
self.lex_attr_getters = lex_attr_getters
self.morphology = Morphology(self.strings, tag_map, lemmatizer)
self.serializer_freqs = serializer_freqs
@ -159,38 +159,28 @@ cdef class Vocab:
if lex != NULL:
return lex
else:
return self._new_lexeme(mem, self.strings.decode_int(orth))
return self._new_lexeme(mem, self.strings[orth])
cdef const LexemeC* _new_lexeme(self, Pool mem, unicode string) except NULL:
# TODO: This is all quite a mess, and doesn't work currently.
# The current state is:
# - mem is None: Denotes that the string should be interned within the StringStore
# - mem is self.mem: Denotes that the string wll be permanently available,
# but that it will be stored within the *Vocab*
# - mem is Doc.mem: Denotes that the string will 'belong' to the Doc
cdef hash_t key
if mem is None:
lex = <LexemeC*>self.mem.alloc(sizeof(LexemeC), 1)
else:
lex = <LexemeC*>mem.alloc(sizeof(LexemeC), 1)
print("Add new lexeme", "id(mem)", id(mem), "id(self.mem)", id(self.mem))
cdef bint is_oov = mem is not self.mem
if len(string) < 3 or not is_oov:
mem = self.mem
lex = <LexemeC*>mem.alloc(sizeof(LexemeC), 1)
lex.orth = self.strings.intern(string, mem=mem)
lex.length = len(string)
lex.id = self.length
if mem is None:
lex.vector = <float*>self.mem.alloc(self.vectors_length, sizeof(float))
else:
lex.vector = <float*>mem.alloc(self.vectors_length, sizeof(float))
lex.vector = <float*>mem.alloc(self.vectors_length, sizeof(float))
if self.lex_attr_getters is not None:
for attr, func in self.lex_attr_getters.items():
value = func(string)
if isinstance(value, unicode):
value = self.strings.intern(value, mem=mem)
value = self.strings.intern(value)
if attr == PROB:
lex.prob = value
else:
Lexeme.set_struct_attr(lex, attr, value)
if mem not in (None, self.mem):
if is_oov:
lex.id = 0
else:
key = hash_string(string)
@ -231,9 +221,8 @@ cdef class Vocab:
instantiation.
'''
cdef attr_t orth
if isinstance(id_or_string, basestring):
print("Intern", id_or_string, id(self.mem))
orth = self.strings.intern(id_or_string, mem=self.mem)
if type(id_or_string) == unicode:
orth = self.strings.intern(id_or_string)
else:
orth = id_or_string
return Lexeme(self, orth)
@ -245,7 +234,7 @@ cdef class Vocab:
token = &tokens[i]
# Set the special tokens up to have morphology and lemmas if
# specified, otherwise use the part-of-speech tag (if specified)
token.lex = <LexemeC*>self.get(None, props['F'])
token.lex = <LexemeC*>self.get(self.mem, props['F'])
if 'pos' in props:
self.morphology.assign_tag(token, props['pos'])
if 'L' in props:
@ -356,7 +345,7 @@ cdef class Vocab:
raise VectorReadError.mismatched_sizes(file_, line_num,
vec_len, len(pieces))
orth = self.strings[word_str]
lexeme = <LexemeC*><void*>self.get_by_orth(None, orth)
lexeme = <LexemeC*><void*>self.get_by_orth(self.mem, orth)
lexeme.vector = <float*>self.mem.alloc(self.vectors_length, sizeof(float))
for i, val_str in enumerate(pieces):