explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2024-12-25 17:36:30 +03:00
Code Issues Packages Projects Releases Wiki Activity

* WordTree in use. Need to reform the way chunks are handled. Should be properly one Lexeme per word, with split points being the things that are cached.

Browse Source
This commit is contained in:
Matthew Honnibal 2014-08-16 20:10:22 +02:00
parent 34b68a18ab
commit b94c9b72c9
2 changed files with 22 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
spacy/spacy.pxd
View File

@ -29,20 +29,20 @@ from spacy._hashing cimport WordTree
cdef class Language:
cdef object name
cdef WordTree vocab
cdef Vocab* distri
cdef Vocab* ortho
cdef WordTree distri
cdef WordTree ortho
cdef dict bacov
cpdef Tokens tokenize(self, unicode text)
cdef Lexeme_addr lookup(self, unicode string) except 0
cdef Lexeme_addr lookup_chunk(self, unicode string) except 0
cdef Orthography* lookup_orth(self, StringHash key, unicode lex) except NULL
cdef Distribution* lookup_dist(self, StringHash key) except NULL
cdef Orthography* lookup_orth(self, unicode lex) except NULL
cdef Distribution* lookup_dist(self, unicode lex) except NULL
cdef Lexeme* new_lexeme(self, unicode key, unicode lex) except NULL
cdef Orthography* new_orth(self, StringHash hashed, unicode lex) except NULL
cdef Distribution* new_dist(self, StringHash key) except NULL
cdef Orthography* new_orth(self, unicode lex) except NULL
cdef Distribution* new_dist(self, unicode lex) except NULL
cdef unicode unhash(self, StringHash hashed)

34
spacy/spacy.pyx
View File

@ -65,10 +65,8 @@ cdef class Language:
self.name = name
self.bacov = {}
self.vocab = WordTree(0, 5)
self.ortho = new Vocab()
self.distri = new Vocab()
self.distri[0].set_empty_key(0)
self.ortho[0].set_empty_key(0)
self.ortho = WordTree(0, 5)
self.distri = WordTree(0, 5)
self.load_tokenization(util.read_tokenization(name))
cpdef Tokens tokenize(self, unicode characters):
@ -125,16 +123,16 @@ cdef class Language:
word_ptr = self.new_lexeme(string, string)
return <Lexeme_addr>word_ptr
cdef Orthography* lookup_orth(self, StringHash hashed, unicode lex):
cdef Orthography* orth = <Orthography*>self.ortho[0][hashed]
cdef Orthography* lookup_orth(self, unicode lex):
cdef Orthography* orth = <Orthography*>self.ortho.get(lex)
if orth == NULL:
orth = self.new_orth(hashed, lex)
orth = self.new_orth(lex)
return orth
cdef Distribution* lookup_dist(self, StringHash hashed):
cdef Distribution* dist = <Distribution*>self.distri[0][hashed]
cdef Distribution* lookup_dist(self, unicode lex):
cdef Distribution* dist = <Distribution*>self.distri.get(lex)
if dist == NULL:
dist = self.new_dist(hashed)
dist = self.new_dist(lex)
return dist
cdef Lexeme* new_lexeme(self, unicode key, unicode string) except NULL:
@ -143,12 +141,12 @@ cdef class Language:
word.lex = hash(string)
self.bacov[word.lex] = string
self.bacov[word.sic] = key
word.orth = self.lookup_orth(word.lex, string)
word.dist = self.lookup_dist(word.lex)
word.orth = self.lookup_orth(string)
word.dist = self.lookup_dist(string)
self.vocab.set(key, <size_t>word)
return word
return word
cdef Orthography* new_orth(self, StringHash hashed, unicode lex) except NULL:
cdef Orthography* new_orth(self, unicode lex) except NULL:
cdef unicode last3
cdef unicode norm
cdef unicode shape
@ -160,7 +158,7 @@ cdef class Language:
orth.length = length
orth.flags = set_orth_flags(lex, orth.length)
orth.norm = hashed
orth.norm = hash(lex)
last3 = substr(lex, length - 3, length, length)
orth.last3 = hash(last3)
norm = get_normalized(lex, length)
@ -172,12 +170,12 @@ cdef class Language:
self.bacov[orth.norm] = norm
self.bacov[orth.shape] = shape
self.ortho[0][hashed] = <size_t>orth
self.ortho.set(lex, <size_t>orth)
return orth
cdef Distribution* new_dist(self, StringHash hashed) except NULL:
cdef Distribution* new_dist(self, unicode lex) except NULL:
dist = <Distribution*>calloc(1, sizeof(Distribution))
self.distri[0][hashed] = <size_t>dist
self.distri.set(lex, <size_t>dist)
return dist
cdef unicode unhash(self, StringHash hash_value):