2014-12-16 11:44:43 +00:00
|
|
|
from libc.stdint cimport uint32_t
|
|
|
|
|
|
2015-01-05 06:54:13 +00:00
|
|
|
from cython.view cimport array as cvarray
|
2014-12-02 12:48:05 +00:00
|
|
|
|
2014-10-22 14:57:59 +00:00
|
|
|
from cymem.cymem cimport Pool
|
2014-12-21 20:25:43 +00:00
|
|
|
from thinc.typedefs cimport atom_t
|
2014-10-22 14:57:59 +00:00
|
|
|
|
2015-01-11 23:26:22 +00:00
|
|
|
from .typedefs cimport flags_t, attr_id_t, attr_t, univ_tag_t
|
|
|
|
|
from .structs cimport Morphology, TokenC, LexemeC
|
2014-12-21 20:25:43 +00:00
|
|
|
from .vocab cimport Vocab
|
2014-12-19 20:03:26 +00:00
|
|
|
from .strings cimport StringStore
|
2014-12-05 04:56:14 +00:00
|
|
|
|
|
|
|
|
|
2015-01-11 23:26:22 +00:00
|
|
|
ctypedef const LexemeC* const_Lexeme_ptr
|
2014-12-09 05:50:01 +00:00
|
|
|
ctypedef TokenC* TokenC_ptr
|
|
|
|
|
|
|
|
|
|
ctypedef fused LexemeOrToken:
|
|
|
|
|
const_Lexeme_ptr
|
|
|
|
|
TokenC_ptr
|
|
|
|
|
|
|
|
|
|
|
2015-01-11 23:26:22 +00:00
|
|
|
cdef attr_t get_lex_attr(const LexemeC* lex, attr_id_t feat_name) nogil
|
2014-12-24 06:42:00 +00:00
|
|
|
cdef attr_t get_token_attr(const TokenC* lex, attr_id_t feat_name) nogil
|
|
|
|
|
|
2015-01-11 23:26:22 +00:00
|
|
|
cdef inline bint check_flag(const LexemeC* lexeme, attr_id_t flag_id) nogil:
|
2014-12-24 06:42:00 +00:00
|
|
|
return lexeme.flags & (1 << flag_id)
|
|
|
|
|
|
|
|
|
|
|
2014-09-15 01:22:40 +00:00
|
|
|
cdef class Tokens:
|
2014-10-22 14:57:59 +00:00
|
|
|
cdef Pool mem
|
2014-12-21 20:25:43 +00:00
|
|
|
cdef Vocab vocab
|
2014-12-09 21:09:32 +00:00
|
|
|
cdef list tag_names
|
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
|
|
|
|
2014-12-09 05:50:01 +00:00
|
|
|
cdef int push_back(self, int i, LexemeOrToken lex_or_tok) except -1
|
2014-10-23 13:59:17 +00:00
|
|
|
|
2015-01-05 06:54:13 +00:00
|
|
|
cpdef long[:,:] to_array(self, object features)
|
2014-12-02 12:48:05 +00:00
|
|
|
|
2014-10-23 13:59:17 +00:00
|
|
|
|
|
|
|
|
cdef class Token:
|
2015-01-11 23:26:22 +00:00
|
|
|
cdef cvarray vec
|
|
|
|
|
|
|
|
|
|
cdef readonly flags_t flags
|
|
|
|
|
|
|
|
|
|
cdef readonly attr_t id
|
|
|
|
|
cdef readonly attr_t sic
|
|
|
|
|
cdef readonly attr_t dense
|
|
|
|
|
cdef readonly attr_t shape
|
|
|
|
|
cdef readonly attr_t prefix
|
|
|
|
|
cdef readonly attr_t suffix
|
|
|
|
|
|
|
|
|
|
cdef readonly attr_t length
|
|
|
|
|
cdef readonly attr_t cluster
|
|
|
|
|
cdef readonly attr_t pos_type
|
|
|
|
|
|
|
|
|
|
cdef readonly float prob
|
|
|
|
|
cdef readonly float sentiment
|
|
|
|
|
|
|
|
|
|
cdef readonly Morphology morph
|
|
|
|
|
cdef readonly univ_tag_t pos
|
|
|
|
|
cdef readonly int fine_pos
|
|
|
|
|
cdef readonly int idx
|
|
|
|
|
cdef readonly int lemma
|
|
|
|
|
cdef readonly int sense
|
|
|
|
|
cdef readonly int dep_tag
|
|
|
|
|
|
|
|
|
|
cdef readonly int head_offset
|
|
|
|
|
cdef readonly uint32_t l_kids
|
|
|
|
|
cdef readonly uint32_t r_kids
|
