mirror of
https://github.com/explosion/spaCy.git
synced 2025-05-31 11:13:26 +03:00
* Removing happax stuff. Added length
This commit is contained in:
parent
fc7c10d7f8
commit
edd38a84b1
|
@ -138,7 +138,7 @@ cdef class Language:
|
||||||
return <Lexeme_addr>word_ptr
|
return <Lexeme_addr>word_ptr
|
||||||
|
|
||||||
cdef int _happax_to_vocab(self, StringHash hashed, Lexeme_addr word_ptr):
|
cdef int _happax_to_vocab(self, StringHash hashed, Lexeme_addr word_ptr):
|
||||||
self.vocab[0][hashed] = <Lexeme_addr>word_ptr
|
self.vocab[0][hashed] = word_ptr
|
||||||
self.happax.erase(hashed)
|
self.happax.erase(hashed)
|
||||||
|
|
||||||
cdef Lexeme* _add(self, StringHash hashed, unicode string, int split, size_t length):
|
cdef Lexeme* _add(self, StringHash hashed, unicode string, int split, size_t length):
|
||||||
|
@ -146,7 +146,8 @@ cdef class Language:
|
||||||
word = self.init_lexeme(string, hashed, split, length)
|
word = self.init_lexeme(string, hashed, split, length)
|
||||||
cdef Lexeme* clobbered = <Lexeme*>self.happax.insert(hashed, <size_t>word)
|
cdef Lexeme* clobbered = <Lexeme*>self.happax.insert(hashed, <size_t>word)
|
||||||
if clobbered != NULL:
|
if clobbered != NULL:
|
||||||
free(clobbered)
|
#free(clobbered)
|
||||||
|
pass
|
||||||
self.bacov[hashed] = string
|
self.bacov[hashed] = string
|
||||||
return word
|
return word
|
||||||
|
|
||||||
|
@ -220,6 +221,7 @@ cdef class Language:
|
||||||
|
|
||||||
cdef int length = len(lex)
|
cdef int length = len(lex)
|
||||||
|
|
||||||
|
orth.length = length
|
||||||
orth.flags = set_orth_flags(lex, length)
|
orth.flags = set_orth_flags(lex, length)
|
||||||
|
|
||||||
cdef unicode last3 = substr(lex, length - 3, length, length)
|
cdef unicode last3 = substr(lex, length - 3, length, length)
|
||||||
|
|
Loading…
Reference in New Issue
Block a user