2015-06-09 02:39:54 +03:00
|
|
|
from libc.string cimport memcpy, memset
|
|
|
|
|
|
|
|
from cymem.cymem cimport Pool
|
|
|
|
|
2015-06-10 00:23:28 +03:00
|
|
|
from ..structs cimport TokenC, Entity
|
2015-06-09 02:39:54 +03:00
|
|
|
|
2015-06-09 22:20:14 +03:00
|
|
|
from ..vocab cimport EMPTY_LEXEME
|
2015-06-09 02:39:54 +03:00
|
|
|
|
|
|
|
|
|
|
|
cdef class StateClass:
|
|
|
|
cdef Pool mem
|
|
|
|
cdef int* _stack
|
|
|
|
cdef int* _buffer
|
2015-06-10 11:13:03 +03:00
|
|
|
cdef bint* shifted
|
2015-06-09 02:39:54 +03:00
|
|
|
cdef TokenC* _sent
|
2015-06-10 00:23:28 +03:00
|
|
|
cdef Entity* _ents
|
2015-06-09 22:20:14 +03:00
|
|
|
cdef TokenC _empty_token
|
2015-06-09 02:39:54 +03:00
|
|
|
cdef int length
|
|
|
|
cdef int _s_i
|
|
|
|
cdef int _b_i
|
2015-06-10 00:23:28 +03:00
|
|
|
cdef int _e_i
|
2015-06-10 13:33:55 +03:00
|
|
|
cdef int _break
|
2015-06-09 02:39:54 +03:00
|
|
|
|
2015-06-10 05:20:23 +03:00
|
|
|
@staticmethod
|
|
|
|
cdef inline StateClass init(const TokenC* sent, int length):
|
|
|
|
cdef StateClass self = StateClass(length)
|
|
|
|
cdef int i
|
|
|
|
for i in range(length):
|
|
|
|
self._sent[i] = sent[i]
|
|
|
|
self._buffer[i] = i
|
2015-06-23 05:13:09 +03:00
|
|
|
for i in range(length, length + 5):
|
|
|
|
self._sent[i].lex = &EMPTY_LEXEME
|
2015-06-10 05:20:23 +03:00
|
|
|
return self
|
|
|
|
|
2015-06-10 08:22:33 +03:00
|
|
|
cdef inline int S(self, int i) nogil:
|
|
|
|
if i >= self._s_i:
|
|
|
|
return -1
|
|
|
|
return self._stack[self._s_i - (i+1)]
|
|
|
|
|
|
|
|
cdef inline int B(self, int i) nogil:
|
|
|
|
if (i + self._b_i) >= self.length:
|
|
|
|
return -1
|
|
|
|
return self._buffer[self._b_i + i]
|
2015-06-09 02:39:54 +03:00
|
|
|
|
|
|
|
cdef int H(self, int i) nogil
|
2015-06-10 00:23:28 +03:00
|
|
|
cdef int E(self, int i) nogil
|
2015-06-09 02:39:54 +03:00
|
|
|
|
|
|
|
cdef int L(self, int i, int idx) nogil
|
|
|
|
cdef int R(self, int i, int idx) nogil
|
|
|
|
|
2015-06-09 22:20:14 +03:00
|
|
|
cdef const TokenC* S_(self, int i) nogil
|
|
|
|
cdef const TokenC* B_(self, int i) nogil
|
|
|
|
|
|
|
|
cdef const TokenC* H_(self, int i) nogil
|
2015-06-10 00:23:28 +03:00
|
|
|
cdef const TokenC* E_(self, int i) nogil
|
2015-06-09 22:20:14 +03:00
|
|
|
|
|
|
|
cdef const TokenC* L_(self, int i, int idx) nogil
|
|
|
|
cdef const TokenC* R_(self, int i, int idx) nogil
|
|
|
|
|
|
|
|
cdef const TokenC* safe_get(self, int i) nogil
|
|
|
|
|
2015-06-09 02:39:54 +03:00
|
|
|
cdef bint empty(self) nogil
|
2015-06-10 00:23:28 +03:00
|
|
|
|
|
|
|
cdef bint entity_is_open(self) nogil
|
2015-06-09 02:39:54 +03:00
|
|
|
|
|
|
|
cdef bint eol(self) nogil
|
2015-06-10 15:08:30 +03:00
|
|
|
|
|
|
|
cdef bint at_break(self) nogil
|
2015-06-09 02:39:54 +03:00
|
|
|
|
|
|
|
cdef bint is_final(self) nogil
|
|
|
|
|
|
|
|
cdef bint has_head(self, int i) nogil
|
|
|
|
|
2015-06-09 22:20:14 +03:00
|
|
|
cdef int n_L(self, int i) nogil
|
|
|
|
|
|
|
|
cdef int n_R(self, int i) nogil
|
|
|
|
|
2015-06-09 02:39:54 +03:00
|
|
|
cdef bint stack_is_connected(self) nogil
|
|
|
|
|
|
|
|
cdef int stack_depth(self) nogil
|
|
|
|
|
|
|
|
cdef int buffer_length(self) nogil
|
|
|
|
|
|
|
|
cdef void push(self) nogil
|
|
|
|
|
|
|
|
cdef void pop(self) nogil
|
2015-06-10 11:13:03 +03:00
|
|
|
|
|
|
|
cdef void unshift(self) nogil
|
2015-06-09 02:39:54 +03:00
|
|
|
|
|
|
|
cdef void add_arc(self, int head, int child, int label) nogil
|
|
|
|
|
|
|
|
cdef void del_arc(self, int head, int child) nogil
|
2015-06-10 00:23:28 +03:00
|
|
|
|
|
|
|
cdef void open_ent(self, int label) nogil
|
|
|
|
|
|
|
|
cdef void close_ent(self) nogil
|
|
|
|
|
|
|
|
cdef void set_ent_tag(self, int i, int ent_iob, int ent_type) nogil
|
2015-06-09 02:39:54 +03:00
|
|
|
|
2015-06-10 13:33:55 +03:00
|
|
|
cdef void set_break(self, int i) nogil
|
2015-06-09 02:39:54 +03:00
|
|
|
|
|
|
|
cdef void clone(self, StateClass src) nogil
|
2015-06-10 15:08:30 +03:00
|
|
|
|
|
|
|
cdef void fast_forward(self) nogil
|