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
|
//
// Zinnia: Online hand recognition system with machine learning
//
// $Id: recognizer.cpp 27 2010-05-09 05:34:05Z taku-ku $;
//
// Copyright(C) 2008 Taku Kudo <taku@chasen.org>
//
#ifdef HAVE_CONFIG_H
# include "config.h"
#endif
#include <iostream>
#include <vector>
#include <cmath>
#include <algorithm>
#include <functional>
#include "common.h"
#include "mmap.h"
#include "zinnia.h"
#include "feature.h"
namespace {
static inline char *read_ptr(char **ptr, size_t size) {
char *r = *ptr;
*ptr += size;
return r;
}
template <typename T>
inline void read_static(char **ptr, T *value) {
char *r = read_ptr(ptr, sizeof(T));
memcpy(value, r, sizeof(T));
}
#ifndef WORDS_LITENDIAN
template <>
inline void read_static<unsigned int>(char **ptr, unsigned int *value) {
unsigned char *buf = reinterpret_cast<unsigned char *>(*ptr);
*value = (buf[0]) | (buf[1] << 8) | (buf[2] << 16) | (buf[3] << 24);
*ptr += 4;
}
template <>
inline void read_static<float>(char **ptr, float *value) {
unsigned int x;
read_static<unsigned int>(ptr, &x);
memcpy(value, &x, sizeof(x));
}
#endif
}
namespace zinnia {
class ResultImpl: public Result {
public:
void add(const char *character, float score) {
results_.push_back(std::make_pair(score, character));
}
void clear() { results_.clear(); }
const char *value(size_t i) const {
return (i >= results_.size()) ? 0 : results_[i].second;
}
float score(size_t i) const {
return (i >= results_.size()) ? -1 : results_[i].first;
}
size_t size() const { return results_.size(); }
ResultImpl() {}
virtual ~ResultImpl() {}
private:
std::vector<std::pair<float, const char*> > results_;
};
class RecognizerImpl: public Recognizer {
public:
bool open(const char *filename);
bool open(const char *ptr, size_t size);
bool close();
size_t size() const { return model_.size(); }
const char *value(size_t i) const;
Result* classify(const Character &character,
size_t nbest) const;
const char *what() { return what_.str(); }
explicit RecognizerImpl() {}
virtual ~RecognizerImpl() { close(); }
private:
struct Model {
const char *character;
float bias;
const FeatureNode *x;
};
Mmap<char> mmap_;
std::vector<Model> model_;
whatlog what_;
};
const char *RecognizerImpl::value(size_t i) const {
return (i >= model_.size()) ? 0 : model_[i].character;
}
bool RecognizerImpl::open(const char *filename) {
CHECK_FALSE(mmap_.open(filename))
<< "no such file or directory: " << filename;
model_.clear();
return open(mmap_.begin(), mmap_.file_size());
}
bool RecognizerImpl::open(const char *p, size_t ptr_size) {
char *ptr = const_cast<char *>(p);
const char *begin = ptr;
const char *end = ptr + ptr_size;
unsigned int version = 0;
unsigned int magic = 0;
read_static<unsigned int>(&ptr, &magic);
CHECK_CLOSE_FALSE((magic ^ DIC_MAGIC_ID) == ptr_size)
<< "model file is broken";
read_static<unsigned int>(&ptr, &version);
CHECK_CLOSE_FALSE(version == DIC_VERSION)
<< "incompatible version: " << version;
unsigned int size = 0;
read_static<unsigned int>(&ptr, &size);
model_.resize(size);
for (size_t i = 0; i < size; ++i) {
Model &m = model_[i];
m.character = read_ptr(&ptr, 16);
CHECK_CLOSE_FALSE(ptr < end) << "model file is broken";
float bias = 0.0;
read_static<float>(&ptr, &bias);
m.bias = bias;
m.x = const_cast<const FeatureNode *>
(reinterpret_cast<FeatureNode *>(ptr));
size_t len = 0;
for (const FeatureNode *x = m.x; x->index != -1; ++x) ++len;
CHECK_CLOSE_FALSE(ptr < end) << "model file is broken";
ptr += sizeof(FeatureNode) * (len + 1);
}
CHECK_FALSE(static_cast<size_t>(ptr - begin) == ptr_size)
<< "size of model file is invalid";
return true;
}
bool RecognizerImpl::close() {
mmap_.close();
model_.clear();
return true;
}
Result *RecognizerImpl::classify(const Character &character,
size_t nbest) const {
if (model_.empty() || nbest <= 0) {
return 0;
}
Features feature;
if (!feature.read(character)) {
return 0;
}
const FeatureNode *x = feature.get();
std::vector<std::pair<float, const char*> > results(size());
for (size_t i = 0; i < model_.size(); ++i) {
results[i].first = model_[i].bias + dot(model_[i].x, x);
results[i].second = model_[i].character;
}
nbest = _min(nbest, results.size());
std::partial_sort(results.begin(),
results.begin() + nbest, results.end(),
std::greater<std::pair<float, const char*> >());
ResultImpl *result = new ResultImpl;
for (size_t i = 0; i < nbest; ++i)
result->add(results[i].second, results[i].first);
return result;
}
Recognizer* Recognizer::create() {
return new RecognizerImpl;
}
Recognizer* createRecognizer() {
return new RecognizerImpl;
}
}
|
