1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
|
cimport hypergraph
cimport kbest
cdef class Hypergraph:
cdef hypergraph.Hypergraph* hg
cdef MT19937* rng
def __dealloc__(self):
del self.hg
if self.rng != NULL:
del self.rng
def viterbi(self):
cdef vector[WordID] trans
hypergraph.ViterbiESentence(self.hg[0], &trans)
return unicode(GetString(trans).c_str(), 'utf8')
def viterbi_trees(self):
f_tree = unicode(hypergraph.ViterbiFTree(self.hg[0]).c_str(), 'utf8')
e_tree = unicode(hypergraph.ViterbiETree(self.hg[0]).c_str(), 'utf8')
return (f_tree, e_tree)
def viterbi_features(self):
cdef SparseVector fmap = SparseVector.__new__(SparseVector)
fmap.vector = new FastSparseVector[weight_t](hypergraph.ViterbiFeatures(self.hg[0]))
return fmap
def viterbi_joshua(self):
return unicode(hypergraph.JoshuaVisualizationString(self.hg[0]).c_str(), 'utf8')
def kbest(self, size):
cdef kbest.KBestDerivations[vector[WordID], kbest.ESentenceTraversal]* derivations = new kbest.KBestDerivations[vector[WordID], kbest.ESentenceTraversal](self.hg[0], size)
cdef kbest.KBestDerivations[vector[WordID], kbest.ESentenceTraversal].Derivation* derivation
cdef unsigned k
try:
for k in range(size):
derivation = derivations.LazyKthBest(self.hg.nodes_.size() - 1, k)
if not derivation: break
yield unicode(GetString(derivation._yield).c_str(), 'utf8')
finally:
del derivations
def kbest_trees(self, size):
cdef kbest.KBestDerivations[vector[WordID], kbest.FTreeTraversal]* f_derivations = new kbest.KBestDerivations[vector[WordID], kbest.FTreeTraversal](self.hg[0], size)
cdef kbest.KBestDerivations[vector[WordID], kbest.FTreeTraversal].Derivation* f_derivation
cdef kbest.KBestDerivations[vector[WordID], kbest.ETreeTraversal]* e_derivations = new kbest.KBestDerivations[vector[WordID], kbest.ETreeTraversal](self.hg[0], size)
cdef kbest.KBestDerivations[vector[WordID], kbest.ETreeTraversal].Derivation* e_derivation
cdef unsigned k
try:
for k in range(size):
f_derivation = f_derivations.LazyKthBest(self.hg.nodes_.size() - 1, k)
e_derivation = e_derivations.LazyKthBest(self.hg.nodes_.size() - 1, k)
if not f_derivation or not e_derivation: break
f_tree = unicode(GetString(f_derivation._yield).c_str(), 'utf8')
e_tree = unicode(GetString(e_derivation._yield).c_str(), 'utf8')
yield (f_tree, e_tree)
finally:
del f_derivations
del e_derivations
def kbest_features(self, size):
cdef kbest.KBestDerivations[FastSparseVector[weight_t], kbest.FeatureVectorTraversal]* derivations = new kbest.KBestDerivations[FastSparseVector[weight_t], kbest.FeatureVectorTraversal](self.hg[0], size)
cdef kbest.KBestDerivations[FastSparseVector[weight_t], kbest.FeatureVectorTraversal].Derivation* derivation
cdef SparseVector fmap
cdef unsigned k
try:
for k in range(size):
derivation = derivations.LazyKthBest(self.hg.nodes_.size() - 1, k)
if not derivation: break
fmap = SparseVector.__new__(SparseVector)
fmap.vector = new FastSparseVector[weight_t](derivation._yield)
yield fmap
finally:
del derivations
def sample(self, unsigned n):
cdef vector[hypergraph.Hypothesis]* hypos = new vector[hypergraph.Hypothesis]()
if self.rng == NULL:
self.rng = new MT19937()
hypergraph.sample_hypotheses(self.hg[0], n, self.rng, hypos)
cdef unsigned k
try:
for k in range(hypos.size()):
yield unicode(GetString(hypos[0][k].words).c_str(), 'utf8')
finally:
del hypos
def intersect(self, Lattice lat):
return hypergraph.Intersect(lat.lattice[0], self.hg)
def prune(self, beam_alpha=0, density=0, **kwargs):
cdef hypergraph.EdgeMask* preserve_mask = NULL
if 'csplit_preserve_full_word' in kwargs:
preserve_mask = new hypergraph.EdgeMask(self.hg.edges_.size())
preserve_mask[0][hypergraph.GetFullWordEdgeIndex(self.hg[0])] = True
self.hg.PruneInsideOutside(beam_alpha, density, preserve_mask, False, 1, False)
if preserve_mask:
del preserve_mask
def lattice(self): # TODO direct hg -> lattice conversion in cdec
cdef str plf = hypergraph.AsPLF(self.hg[0], True).c_str()
return Lattice(eval(plf))
def reweight(self, weights):
if isinstance(weights, SparseVector):
self.hg.Reweight((<SparseVector> weights).vector[0])
elif isinstance(weights, DenseVector):
self.hg.Reweight((<DenseVector> weights).vector[0])
else:
raise TypeError('cannot reweight hypergraph with %s' % type(weights))
property edges:
def __get__(self):
cdef unsigned i
for i in range(self.hg.edges_.size()):
yield HypergraphEdge().init(self.hg, i)
property nodes:
def __get__(self):
cdef unsigned i
for i in range(self.hg.nodes_.size()):
yield HypergraphNode().init(self.hg, i)
property goal:
def __get__(self):
return HypergraphNode().init(self.hg, self.hg.GoalNode())
property npaths:
def __get__(self):
return self.hg.NumberOfPaths()
def inside_outside(self):
cdef FastSparseVector[prob_t]* result = new FastSparseVector[prob_t]()
cdef prob_t z = hypergraph.InsideOutside(self.hg[0], result)
result[0] /= z
cdef SparseVector vector = SparseVector.__new__(SparseVector)
vector.vector = new FastSparseVector[double]()
cdef FastSparseVector[prob_t].const_iterator* it = new FastSparseVector[prob_t].const_iterator(result[0], False)
cdef unsigned i
for i in range(result.size()):
vector.vector.set_value(it[0].ptr().first, log(it[0].ptr().second))
pinc(it[0]) # ++it
del it
del result
return vector
cdef class HypergraphEdge:
cdef hypergraph.Hypergraph* hg
cdef hypergraph.HypergraphEdge* edge
cdef public TRule trule
cdef init(self, hypergraph.Hypergraph* hg, unsigned i):
self.hg = hg
self.edge = &hg.edges_[i]
self.trule = TRule.__new__(TRule)
self.trule.rule = new shared_ptr[grammar.TRule](self.edge.rule_)
return self
def __len__(self):
return self.edge.tail_nodes_.size()
property head_node:
def __get__(self):
return HypergraphNode().init(self.hg, self.edge.head_node_)
property tail_nodes:
def __get__(self):
cdef unsigned i
for i in range(self.edge.tail_nodes_.size()):
yield HypergraphNode().init(self.hg, self.edge.tail_nodes_[i])
property span:
def __get__(self):
return (self.edge.i_, self.edge.j_)
property feature_values:
def __get__(self):
cdef SparseVector vector = SparseVector.__new__(SparseVector)
vector.vector = new FastSparseVector[double](self.edge.feature_values_)
return vector
property prob:
def __get__(self):
return self.edge.edge_prob_.as_float()
def __richcmp__(HypergraphEdge x, HypergraphEdge y, int op):
if op == 2: # ==
return x.edge == y.edge
elif op == 3: # !=
return not (x == y)
raise NotImplemented('comparison not implemented for HypergraphEdge')
cdef class HypergraphNode:
cdef hypergraph.Hypergraph* hg
cdef hypergraph.HypergraphNode* node
cdef init(self, hypergraph.Hypergraph* hg, unsigned i):
self.hg = hg
self.node = &hg.nodes_[i]
return self
property in_edges:
def __get__(self):
cdef unsigned i
for i in range(self.node.in_edges_.size()):
yield HypergraphEdge().init(self.hg, self.node.in_edges_[i])
property out_edges:
def __get__(self):
cdef unsigned i
for i in range(self.node.out_edges_.size()):
yield HypergraphEdge().init(self.hg, self.node.out_edges_[i])
property span:
def __get__(self):
return next(self.in_edges).span
property cat:
def __get__(self):
if self.node.cat_:
return TDConvert(-self.node.cat_)
def __richcmp__(HypergraphNode x, HypergraphNode y, int op):
if op == 2: # ==
return x.node == y.node
elif op == 3: # !=
return not (x == y)
raise NotImplemented('comparison not implemented for HypergraphNode')
|
