mirror of https://github.com/explosion/spaCy.git
104 lines
3.3 KiB
Cython
104 lines
3.3 KiB
Cython
from cymem.cymem cimport Pool
|
|
from ._state cimport State
|
|
from ..structs cimport TokenC
|
|
from thinc.typedefs cimport weight_t
|
|
|
|
|
|
cdef weight_t MIN_SCORE = -90000
|
|
|
|
|
|
class OracleError(Exception):
|
|
pass
|
|
|
|
|
|
cdef class TransitionSystem:
|
|
def __init__(self, StringStore string_table, dict labels_by_action):
|
|
self.mem = Pool()
|
|
self.n_moves = sum(len(labels) for labels in labels_by_action.values())
|
|
self._is_valid = <bint*>self.mem.alloc(self.n_moves, sizeof(bint))
|
|
moves = <Transition*>self.mem.alloc(self.n_moves, sizeof(Transition))
|
|
cdef int i = 0
|
|
cdef int label_id
|
|
self.strings = string_table
|
|
for action, label_strs in sorted(labels_by_action.items()):
|
|
for label_str in sorted(label_strs):
|
|
label_id = self.strings[unicode(label_str)] if label_str else 0
|
|
moves[i] = self.init_transition(i, int(action), label_id)
|
|
i += 1
|
|
self.c = moves
|
|
|
|
cdef int initialize_state(self, State* state) except -1:
|
|
pass
|
|
|
|
cdef int finalize_state(self, State* state) except -1:
|
|
pass
|
|
|
|
cdef int preprocess_gold(self, GoldParse gold) except -1:
|
|
raise NotImplementedError
|
|
|
|
cdef Transition lookup_transition(self, object name) except *:
|
|
raise NotImplementedError
|
|
|
|
cdef Transition init_transition(self, int clas, int move, int label) except *:
|
|
raise NotImplementedError
|
|
|
|
cdef Transition best_valid(self, const weight_t* scores, const State* s) except *:
|
|
raise NotImplementedError
|
|
|
|
cdef int set_valid(self, bint* output, const State* state) except -1:
|
|
raise NotImplementedError
|
|
|
|
cdef int set_costs(self, int* output, const State* s, GoldParse gold) except -1:
|
|
cdef int i
|
|
for i in range(self.n_moves):
|
|
output[i] = self.c[i].get_cost(s, &gold.c, self.c[i].label)
|
|
|
|
cdef Transition best_gold(self, const weight_t* scores, const State* s,
|
|
GoldParse gold) except *:
|
|
cdef Transition best
|
|
cdef weight_t score = MIN_SCORE
|
|
cdef int i
|
|
for i in range(self.n_moves):
|
|
cost = self.c[i].get_cost(s, &gold.c, self.c[i].label)
|
|
if scores[i] > score and cost == 0:
|
|
best = self.c[i]
|
|
score = scores[i]
|
|
assert score > MIN_SCORE
|
|
return best
|
|
|
|
|
|
#cdef class PyState:
|
|
# """Provide a Python class for testing purposes."""
|
|
# def __init__(self, GoldParse gold):
|
|
# self.mem = Pool()
|
|
# self.system = EntityRecognition(labels)
|
|
# self._state = init_state(self.mem, tokens, gold.length)
|
|
#
|
|
# def transition(self, name):
|
|
# cdef const Transition* trans = self._transition_by_name(name)
|
|
# trans.do(trans, self._state)
|
|
#
|
|
# def is_valid(self, name):
|
|
# cdef const Transition* trans = self._transition_by_name(name)
|
|
# return _is_valid(trans.move, trans.label, self._state)
|
|
#
|
|
# def is_gold(self, name):
|
|
# cdef const Transition* trans = self._transition_by_name(name)
|
|
# return _get_const(trans, self._state, self._gold)
|
|
#
|
|
# property ent:
|
|
# def __get__(self):
|
|
# pass
|
|
#
|
|
# property n_ents:
|
|
# def __get__(self):
|
|
# pass
|
|
#
|
|
# property i:
|
|
# def __get__(self):
|
|
# pass
|
|
#
|
|
# property open_entity:
|
|
# def __get__(self):
|
|
# return entity_is_open(self._s)
|