2014-12-02 12:48:05 +00:00
|
|
|
import numpy as np
|
|
|
|
|
cimport numpy as np
|
|
|
|
|
|
2014-10-22 14:57:59 +00:00
|
|
|
from cymem.cymem cimport Pool
|
2014-12-04 09:46:55 +00:00
|
|
|
from thinc.typedefs cimport atom_t
|
2014-10-22 14:57:59 +00:00
|
|
|
|
2014-10-23 13:59:17 +00:00
|
|
|
from .lexeme cimport Lexeme
|
2014-12-08 10:12:15 +00:00
|
|
|
|
2014-12-03 00:05:15 +00:00
|
|
|
from .typedefs cimport flags_t
|
2014-10-23 13:59:17 +00:00
|
|
|
from .utf8string cimport StringStore
|
2014-12-08 10:12:15 +00:00
|
|
|
from libc.stdint cimport uint8_t, uint16_t
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
cdef struct Morphology:
|
|
|
|
|
uint8_t number
|
|
|
|
|
uint8_t tenspect # Tense/aspect/voice
|
|
|
|
|
uint8_t mood
|
|
|
|
|
uint8_t gender
|
|
|
|
|
uint8_t person
|
|
|
|
|
uint8_t case
|
|
|
|
|
uint8_t misc
|
|
|
|
|
|
2014-10-23 13:59:17 +00:00
|
|
|
|
2014-12-05 04:56:14 +00:00
|
|
|
cdef struct TokenC:
|
|
|
|
|
const Lexeme* lex
|
2014-12-08 10:12:15 +00:00
|
|
|
Morphology morph
|
2014-12-05 04:56:14 +00:00
|
|
|
int idx
|
|
|
|
|
int pos
|
2014-12-08 10:12:15 +00:00
|
|
|
int lemma
|
2014-12-05 04:56:14 +00:00
|
|
|
int sense
|
|
|
|
|
|
|
|
|
|
|
2014-09-15 01:22:40 +00:00
|
|
|
cdef class Tokens:
|
2014-10-22 14:57:59 +00:00
|
|
|
cdef Pool mem
|
2014-10-23 13:59:17 +00:00
|
|
|
cdef StringStore _string_store
|
2014-10-22 14:57:59 +00:00
|
|
|
|
2014-12-05 04:56:14 +00:00
|
|
|
cdef TokenC* data
|
2014-10-22 14:57:59 +00:00
|
|
|
|
|
|
|
|
cdef int length
|
|
|
|
|
cdef int max_length
|
2014-10-14 04:21:03 +00:00
|
|
|
|
2014-12-03 04:44:25 +00:00
|
|
|
cdef int extend(self, int i, const Lexeme* const* lexemes, int n) except -1
|
|
|
|
|
cdef int push_back(self, int i, const Lexeme* lexeme) except -1
|
2014-12-03 00:05:15 +00:00
|
|
|
cpdef int set_tag(self, int i, int tag_type, int tag) except -1
|
2014-10-23 13:59:17 +00:00
|
|
|
|
2014-12-04 09:46:55 +00:00
|
|
|
cpdef np.ndarray[long, ndim=2] get_array(self, list features)
|
2014-12-02 12:48:05 +00:00
|
|
|
|
2014-10-23 13:59:17 +00:00
|
|
|
|
|
|
|
|
cdef class Token:
|
|
|
|
|
cdef StringStore _string_store
|
|
|
|
|
cdef public int i
|
|
|
|
|
cdef public int idx
|
|
|
|
|
cdef public int pos
|
2014-12-08 10:12:15 +00:00
|
|
|
cdef int lemma
|
2014-10-23 13:59:17 +00:00
|
|
|
|
2014-10-31 06:44:39 +00:00
|
|
|
cdef public atom_t id
|
2014-10-23 13:59:17 +00:00
|
|
|
cdef public atom_t cluster
|
|
|
|
|
cdef public atom_t length
|
2014-10-31 06:44:39 +00:00
|
|
|
cdef public atom_t postype
|
|
|
|
|
cdef public atom_t sensetype
|
2014-10-23 13:59:17 +00:00
|
|
|
|
2014-10-29 12:19:38 +00:00
|
|
|
cdef public atom_t sic
|
2014-10-23 13:59:17 +00:00
|
|
|
cdef public atom_t norm
|
|
|
|
|
cdef public atom_t shape
|
|
|
|
|
cdef public atom_t asciied
|
|
|
|
|
cdef public atom_t prefix
|
|
|
|
|
cdef public atom_t suffix
|
|
|
|
|
|
|
|
|
|
cdef public float prob
|
|
|
|
|
|
2014-12-03 00:05:15 +00:00
|
|
|
cdef public flags_t flags
|
