melisa 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- data/README.md +11 -0
- data/ext/marisa/bindings/marisa-swig.cxx +253 -0
- data/ext/marisa/bindings/marisa-swig.h +183 -0
- data/ext/marisa/bindings/perl/marisa-swig.cxx +253 -0
- data/ext/marisa/bindings/perl/marisa-swig.h +183 -0
- data/ext/marisa/bindings/perl/marisa-swig_wrap.cxx +5160 -0
- data/ext/marisa/bindings/python/marisa-swig.cxx +253 -0
- data/ext/marisa/bindings/python/marisa-swig.h +183 -0
- data/ext/marisa/bindings/python/marisa-swig_wrap.cxx +6090 -0
- data/ext/marisa/bindings/ruby/extconf.rb +5 -0
- data/ext/marisa/bindings/ruby/marisa-swig.cxx +253 -0
- data/ext/marisa/bindings/ruby/marisa-swig.h +183 -0
- data/ext/marisa/bindings/ruby/marisa-swig_wrap.cxx +4708 -0
- data/ext/marisa/lib/marisa.h +14 -0
- data/ext/marisa/lib/marisa/agent.cc +51 -0
- data/ext/marisa/lib/marisa/agent.h +73 -0
- data/ext/marisa/lib/marisa/base.h +193 -0
- data/ext/marisa/lib/marisa/exception.h +82 -0
- data/ext/marisa/lib/marisa/grimoire/algorithm.h +26 -0
- data/ext/marisa/lib/marisa/grimoire/algorithm/sort.h +196 -0
- data/ext/marisa/lib/marisa/grimoire/intrin.h +115 -0
- data/ext/marisa/lib/marisa/grimoire/io.h +18 -0
- data/ext/marisa/lib/marisa/grimoire/io/mapper.cc +163 -0
- data/ext/marisa/lib/marisa/grimoire/io/mapper.h +67 -0
- data/ext/marisa/lib/marisa/grimoire/io/reader.cc +147 -0
- data/ext/marisa/lib/marisa/grimoire/io/reader.h +66 -0
- data/ext/marisa/lib/marisa/grimoire/io/writer.cc +148 -0
- data/ext/marisa/lib/marisa/grimoire/io/writer.h +65 -0
- data/ext/marisa/lib/marisa/grimoire/trie.h +16 -0
- data/ext/marisa/lib/marisa/grimoire/trie/cache.h +81 -0
- data/ext/marisa/lib/marisa/grimoire/trie/config.h +155 -0
- data/ext/marisa/lib/marisa/grimoire/trie/entry.h +82 -0
- data/ext/marisa/lib/marisa/grimoire/trie/header.h +61 -0
- data/ext/marisa/lib/marisa/grimoire/trie/history.h +65 -0
- data/ext/marisa/lib/marisa/grimoire/trie/key.h +228 -0
- data/ext/marisa/lib/marisa/grimoire/trie/louds-trie.cc +876 -0
- data/ext/marisa/lib/marisa/grimoire/trie/louds-trie.h +134 -0
- data/ext/marisa/lib/marisa/grimoire/trie/range.h +115 -0
- data/ext/marisa/lib/marisa/grimoire/trie/state.h +117 -0
- data/ext/marisa/lib/marisa/grimoire/trie/tail.cc +218 -0
- data/ext/marisa/lib/marisa/grimoire/trie/tail.h +72 -0
- data/ext/marisa/lib/marisa/grimoire/vector.h +18 -0
- data/ext/marisa/lib/marisa/grimoire/vector/bit-vector.cc +826 -0
- data/ext/marisa/lib/marisa/grimoire/vector/bit-vector.h +179 -0
- data/ext/marisa/lib/marisa/grimoire/vector/flat-vector.h +205 -0
- data/ext/marisa/lib/marisa/grimoire/vector/pop-count.h +110 -0
- data/ext/marisa/lib/marisa/grimoire/vector/rank-index.h +82 -0
- data/ext/marisa/lib/marisa/grimoire/vector/vector.h +256 -0
- data/ext/marisa/lib/marisa/iostream.h +18 -0
- data/ext/marisa/lib/marisa/key.h +85 -0
- data/ext/marisa/lib/marisa/keyset.cc +181 -0
- data/ext/marisa/lib/marisa/keyset.h +80 -0
- data/ext/marisa/lib/marisa/query.h +71 -0
- data/ext/marisa/lib/marisa/scoped-array.h +48 -0
- data/ext/marisa/lib/marisa/scoped-ptr.h +52 -0
- data/ext/marisa/lib/marisa/stdio.h +15 -0
- data/ext/marisa/lib/marisa/trie.cc +249 -0
- data/ext/marisa/lib/marisa/trie.h +64 -0
- data/ext/marisa/tests/base-test.cc +309 -0
- data/ext/marisa/tests/io-test.cc +252 -0
- data/ext/marisa/tests/marisa-assert.h +26 -0
- data/ext/marisa/tests/marisa-test.cc +388 -0
- data/ext/marisa/tests/trie-test.cc +507 -0
- data/ext/marisa/tests/vector-test.cc +466 -0
- data/ext/marisa/tools/cmdopt.cc +298 -0
- data/ext/marisa/tools/cmdopt.h +58 -0
- data/ext/marisa/tools/marisa-benchmark.cc +418 -0
- data/ext/marisa/tools/marisa-build.cc +206 -0
- data/ext/marisa/tools/marisa-common-prefix-search.cc +143 -0
- data/ext/marisa/tools/marisa-dump.cc +151 -0
- data/ext/marisa/tools/marisa-lookup.cc +110 -0
- data/ext/marisa/tools/marisa-predictive-search.cc +143 -0
- data/ext/marisa/tools/marisa-reverse-lookup.cc +110 -0
- data/lib/melisa.rb +7 -0
- data/lib/melisa/base_config_flags.rb +76 -0
- data/lib/melisa/bytes_trie.rb +55 -0
- data/lib/melisa/int_trie.rb +14 -0
- data/lib/melisa/search.rb +55 -0
- data/lib/melisa/trie.rb +96 -0
- data/lib/melisa/version.rb +3 -0
- data/melisa.gemspec +36 -0
- data/spec/base_config_flags_spec.rb +73 -0
- data/spec/bytes_trie_spec.rb +16 -0
- data/spec/int_trie_spec.rb +16 -0
- data/spec/search_spec.rb +29 -0
- data/spec/spec_helper.rb +1 -0
- data/spec/trie_spec.rb +30 -0
- metadata +207 -0
@@ -0,0 +1,66 @@
|
|
1
|
+
#ifndef MARISA_GRIMOIRE_IO_READER_H_
|
2
|
+
#define MARISA_GRIMOIRE_IO_READER_H_
|
3
|
+
|
4
|
+
#include <cstdio>
|
5
|
+
#include <iostream>
|
6
|
+
|
7
|
+
#include "marisa/base.h"
|
8
|
+
|
9
|
+
namespace marisa {
|
10
|
+
namespace grimoire {
|
11
|
+
namespace io {
|
12
|
+
|
13
|
+
class Reader {
|
14
|
+
public:
|
15
|
+
Reader();
|
16
|
+
~Reader();
|
17
|
+
|
18
|
+
void open(const char *filename);
|
19
|
+
void open(std::FILE *file);
|
20
|
+
void open(int fd);
|
21
|
+
void open(std::istream &stream);
|
22
|
+
|
23
|
+
template <typename T>
|
24
|
+
void read(T *obj) {
|
25
|
+
MARISA_THROW_IF(obj == NULL, MARISA_NULL_ERROR);
|
26
|
+
read_data(obj, sizeof(T));
|
27
|
+
}
|
28
|
+
|
29
|
+
template <typename T>
|
30
|
+
void read(T *objs, std::size_t num_objs) {
|
31
|
+
MARISA_THROW_IF((objs == NULL) && (num_objs != 0), MARISA_NULL_ERROR);
|
32
|
+
MARISA_THROW_IF(num_objs > (MARISA_SIZE_MAX / sizeof(T)),
|
33
|
+
MARISA_SIZE_ERROR);
|
34
|
+
read_data(objs, sizeof(T) * num_objs);
|
35
|
+
}
|
36
|
+
|
37
|
+
void seek(std::size_t size);
|
38
|
+
|
39
|
+
bool is_open() const;
|
40
|
+
|
41
|
+
void clear();
|
42
|
+
void swap(Reader &rhs);
|
43
|
+
|
44
|
+
private:
|
45
|
+
std::FILE *file_;
|
46
|
+
int fd_;
|
47
|
+
std::istream *stream_;
|
48
|
+
bool needs_fclose_;
|
49
|
+
|
50
|
+
void open_(const char *filename);
|
51
|
+
void open_(std::FILE *file);
|
52
|
+
void open_(int fd);
|
53
|
+
void open_(std::istream &stream);
|
54
|
+
|
55
|
+
void read_data(void *buf, std::size_t size);
|
56
|
+
|
57
|
+
// Disallows copy and assignment.
|
58
|
+
Reader(const Reader &);
|
59
|
+
Reader &operator=(const Reader &);
|
60
|
+
};
|
61
|
+
|
62
|
+
} // namespace io
|
63
|
+
} // namespace grimoire
|
64
|
+
} // namespace marisa
|
65
|
+
|
66
|
+
#endif // MARISA_GRIMOIRE_IO_READER_H_
|
@@ -0,0 +1,148 @@
|
|
1
|
+
#include <stdio.h>
|
2
|
+
|
3
|
+
#ifdef _WIN32
|
4
|
+
#include <io.h>
|
5
|
+
#else // _WIN32
|
6
|
+
#include <unistd.h>
|
7
|
+
#endif // _WIN32
|
8
|
+
|
9
|
+
#include <limits>
|
10
|
+
|
11
|
+
#include "marisa/grimoire/io/writer.h"
|
12
|
+
|
13
|
+
namespace marisa {
|
14
|
+
namespace grimoire {
|
15
|
+
namespace io {
|
16
|
+
|
17
|
+
Writer::Writer()
|
18
|
+
: file_(NULL), fd_(-1), stream_(NULL), needs_fclose_(false) {}
|
19
|
+
|
20
|
+
Writer::~Writer() {
|
21
|
+
if (needs_fclose_) {
|
22
|
+
::fclose(file_);
|
23
|
+
}
|
24
|
+
}
|
25
|
+
|
26
|
+
void Writer::open(const char *filename) {
|
27
|
+
MARISA_THROW_IF(filename == NULL, MARISA_NULL_ERROR);
|
28
|
+
|
29
|
+
Writer temp;
|
30
|
+
temp.open_(filename);
|
31
|
+
swap(temp);
|
32
|
+
}
|
33
|
+
|
34
|
+
void Writer::open(std::FILE *file) {
|
35
|
+
MARISA_THROW_IF(file == NULL, MARISA_NULL_ERROR);
|
36
|
+
|
37
|
+
Writer temp;
|
38
|
+
temp.open_(file);
|
39
|
+
swap(temp);
|
40
|
+
}
|
41
|
+
|
42
|
+
void Writer::open(int fd) {
|
43
|
+
MARISA_THROW_IF(fd == -1, MARISA_CODE_ERROR);
|
44
|
+
|
45
|
+
Writer temp;
|
46
|
+
temp.open_(fd);
|
47
|
+
swap(temp);
|
48
|
+
}
|
49
|
+
|
50
|
+
void Writer::open(std::ostream &stream) {
|
51
|
+
Writer temp;
|
52
|
+
temp.open_(stream);
|
53
|
+
swap(temp);
|
54
|
+
}
|
55
|
+
|
56
|
+
void Writer::clear() {
|
57
|
+
Writer().swap(*this);
|
58
|
+
}
|
59
|
+
|
60
|
+
void Writer::swap(Writer &rhs) {
|
61
|
+
marisa::swap(file_, rhs.file_);
|
62
|
+
marisa::swap(fd_, rhs.fd_);
|
63
|
+
marisa::swap(stream_, rhs.stream_);
|
64
|
+
marisa::swap(needs_fclose_, rhs.needs_fclose_);
|
65
|
+
}
|
66
|
+
|
67
|
+
void Writer::seek(std::size_t size) {
|
68
|
+
MARISA_THROW_IF(!is_open(), MARISA_STATE_ERROR);
|
69
|
+
if (size == 0) {
|
70
|
+
return;
|
71
|
+
} else if (size <= 16) {
|
72
|
+
const char buf[16] = {};
|
73
|
+
write_data(buf, size);
|
74
|
+
} else {
|
75
|
+
const char buf[1024] = {};
|
76
|
+
do {
|
77
|
+
const std::size_t count = (size < sizeof(buf)) ? size : sizeof(buf);
|
78
|
+
write_data(buf, count);
|
79
|
+
size -= count;
|
80
|
+
} while (size != 0);
|
81
|
+
}
|
82
|
+
}
|
83
|
+
|
84
|
+
bool Writer::is_open() const {
|
85
|
+
return (file_ != NULL) || (fd_ != -1) || (stream_ != NULL);
|
86
|
+
}
|
87
|
+
|
88
|
+
void Writer::open_(const char *filename) {
|
89
|
+
std::FILE *file = NULL;
|
90
|
+
#ifdef _MSC_VER
|
91
|
+
MARISA_THROW_IF(::fopen_s(&file, filename, "wb") != 0, MARISA_IO_ERROR);
|
92
|
+
#else // _MSC_VER
|
93
|
+
file = ::fopen(filename, "wb");
|
94
|
+
MARISA_THROW_IF(file == NULL, MARISA_IO_ERROR);
|
95
|
+
#endif // _MSC_VER
|
96
|
+
file_ = file;
|
97
|
+
needs_fclose_ = true;
|
98
|
+
}
|
99
|
+
|
100
|
+
void Writer::open_(std::FILE *file) {
|
101
|
+
file_ = file;
|
102
|
+
}
|
103
|
+
|
104
|
+
void Writer::open_(int fd) {
|
105
|
+
fd_ = fd;
|
106
|
+
}
|
107
|
+
|
108
|
+
void Writer::open_(std::ostream &stream) {
|
109
|
+
stream_ = &stream;
|
110
|
+
}
|
111
|
+
|
112
|
+
void Writer::write_data(const void *data, std::size_t size) {
|
113
|
+
MARISA_THROW_IF(!is_open(), MARISA_STATE_ERROR);
|
114
|
+
if (size == 0) {
|
115
|
+
return;
|
116
|
+
} else if (fd_ != -1) {
|
117
|
+
while (size != 0) {
|
118
|
+
#ifdef _WIN32
|
119
|
+
static const std::size_t CHUNK_SIZE =
|
120
|
+
std::numeric_limits<int>::max();
|
121
|
+
const unsigned int count = (size < CHUNK_SIZE) ? size : CHUNK_SIZE;
|
122
|
+
const int size_written = ::_write(fd_, data, count);
|
123
|
+
#else // _WIN32
|
124
|
+
static const std::size_t CHUNK_SIZE =
|
125
|
+
std::numeric_limits< ::ssize_t>::max();
|
126
|
+
const ::size_t count = (size < CHUNK_SIZE) ? size : CHUNK_SIZE;
|
127
|
+
const ::ssize_t size_written = ::write(fd_, data, count);
|
128
|
+
#endif // _WIN32
|
129
|
+
MARISA_THROW_IF(size_written <= 0, MARISA_IO_ERROR);
|
130
|
+
data = static_cast<const char *>(data) + size_written;
|
131
|
+
size -= size_written;
|
132
|
+
}
|
133
|
+
} else if (file_ != NULL) {
|
134
|
+
MARISA_THROW_IF(::fwrite(data, 1, size, file_) != size, MARISA_IO_ERROR);
|
135
|
+
MARISA_THROW_IF(::fflush(file_) != 0, MARISA_IO_ERROR);
|
136
|
+
} else if (stream_ != NULL) {
|
137
|
+
try {
|
138
|
+
MARISA_THROW_IF(!stream_->write(static_cast<const char *>(data), size),
|
139
|
+
MARISA_IO_ERROR);
|
140
|
+
} catch (const std::ios_base::failure &) {
|
141
|
+
MARISA_THROW(MARISA_IO_ERROR, "std::ios_base::failure");
|
142
|
+
}
|
143
|
+
}
|
144
|
+
}
|
145
|
+
|
146
|
+
} // namespace io
|
147
|
+
} // namespace grimoire
|
148
|
+
} // namespace marisa
|
@@ -0,0 +1,65 @@
|
|
1
|
+
#ifndef MARISA_GRIMOIRE_IO_WRITER_H_
|
2
|
+
#define MARISA_GRIMOIRE_IO_WRITER_H_
|
3
|
+
|
4
|
+
#include <cstdio>
|
5
|
+
#include <iostream>
|
6
|
+
|
7
|
+
#include "marisa/base.h"
|
8
|
+
|
9
|
+
namespace marisa {
|
10
|
+
namespace grimoire {
|
11
|
+
namespace io {
|
12
|
+
|
13
|
+
class Writer {
|
14
|
+
public:
|
15
|
+
Writer();
|
16
|
+
~Writer();
|
17
|
+
|
18
|
+
void open(const char *filename);
|
19
|
+
void open(std::FILE *file);
|
20
|
+
void open(int fd);
|
21
|
+
void open(std::ostream &stream);
|
22
|
+
|
23
|
+
template <typename T>
|
24
|
+
void write(const T &obj) {
|
25
|
+
write_data(&obj, sizeof(T));
|
26
|
+
}
|
27
|
+
|
28
|
+
template <typename T>
|
29
|
+
void write(const T *objs, std::size_t num_objs) {
|
30
|
+
MARISA_THROW_IF((objs == NULL) && (num_objs != 0), MARISA_NULL_ERROR);
|
31
|
+
MARISA_THROW_IF(num_objs > (MARISA_SIZE_MAX / sizeof(T)),
|
32
|
+
MARISA_SIZE_ERROR);
|
33
|
+
write_data(objs, sizeof(T) * num_objs);
|
34
|
+
}
|
35
|
+
|
36
|
+
void seek(std::size_t size);
|
37
|
+
|
38
|
+
bool is_open() const;
|
39
|
+
|
40
|
+
void clear();
|
41
|
+
void swap(Writer &rhs);
|
42
|
+
|
43
|
+
private:
|
44
|
+
std::FILE *file_;
|
45
|
+
int fd_;
|
46
|
+
std::ostream *stream_;
|
47
|
+
bool needs_fclose_;
|
48
|
+
|
49
|
+
void open_(const char *filename);
|
50
|
+
void open_(std::FILE *file);
|
51
|
+
void open_(int fd);
|
52
|
+
void open_(std::ostream &stream);
|
53
|
+
|
54
|
+
void write_data(const void *data, std::size_t size);
|
55
|
+
|
56
|
+
// Disallows copy and assignment.
|
57
|
+
Writer(const Writer &);
|
58
|
+
Writer &operator=(const Writer &);
|
59
|
+
};
|
60
|
+
|
61
|
+
} // namespace io
|
62
|
+
} // namespace grimoire
|
63
|
+
} // namespace marisa
|
64
|
+
|
65
|
+
#endif // MARISA_GRIMOIRE_IO_WRITER_H_
|
@@ -0,0 +1,16 @@
|
|
1
|
+
#ifndef MARISA_GRIMOIRE_TRIE_H_
|
2
|
+
#define MARISA_GRIMOIRE_TRIE_H_
|
3
|
+
|
4
|
+
#include "marisa/grimoire/trie/state.h"
|
5
|
+
#include "marisa/grimoire/trie/louds-trie.h"
|
6
|
+
|
7
|
+
namespace marisa {
|
8
|
+
namespace grimoire {
|
9
|
+
|
10
|
+
using trie::State;
|
11
|
+
using trie::LoudsTrie;
|
12
|
+
|
13
|
+
} // namespace grimoire
|
14
|
+
} // namespace marisa
|
15
|
+
|
16
|
+
#endif // MARISA_GRIMOIRE_TRIE_H_
|
@@ -0,0 +1,81 @@
|
|
1
|
+
#ifndef MARISA_GRIMOIRE_TRIE_CACHE_H_
|
2
|
+
#define MARISA_GRIMOIRE_TRIE_CACHE_H_
|
3
|
+
|
4
|
+
#include <cfloat>
|
5
|
+
|
6
|
+
#include "marisa/base.h"
|
7
|
+
|
8
|
+
namespace marisa {
|
9
|
+
namespace grimoire {
|
10
|
+
namespace trie {
|
11
|
+
|
12
|
+
class Cache {
|
13
|
+
public:
|
14
|
+
Cache() : parent_(0), child_(0), union_() {
|
15
|
+
union_.weight = FLT_MIN;
|
16
|
+
}
|
17
|
+
Cache(const Cache &cache)
|
18
|
+
: parent_(cache.parent_), child_(cache.child_), union_(cache.union_) {}
|
19
|
+
|
20
|
+
Cache &operator=(const Cache &cache) {
|
21
|
+
parent_ = cache.parent_;
|
22
|
+
child_ = cache.child_;
|
23
|
+
union_ = cache.union_;
|
24
|
+
return *this;
|
25
|
+
}
|
26
|
+
|
27
|
+
void set_parent(std::size_t parent) {
|
28
|
+
MARISA_DEBUG_IF(parent > MARISA_UINT32_MAX, MARISA_SIZE_ERROR);
|
29
|
+
parent_ = (UInt32)parent;
|
30
|
+
}
|
31
|
+
void set_child(std::size_t child) {
|
32
|
+
MARISA_DEBUG_IF(child > MARISA_UINT32_MAX, MARISA_SIZE_ERROR);
|
33
|
+
child_ = (UInt32)child;
|
34
|
+
}
|
35
|
+
void set_base(UInt8 base) {
|
36
|
+
union_.link = (union_.link & ~0xFFU) | base;
|
37
|
+
}
|
38
|
+
void set_extra(std::size_t extra) {
|
39
|
+
MARISA_DEBUG_IF(extra > (MARISA_UINT32_MAX >> 8), MARISA_SIZE_ERROR);
|
40
|
+
union_.link = (UInt32)((union_.link & 0xFFU) | (extra << 8));
|
41
|
+
}
|
42
|
+
void set_weight(float weight) {
|
43
|
+
union_.weight = weight;
|
44
|
+
}
|
45
|
+
|
46
|
+
std::size_t parent() const {
|
47
|
+
return parent_;
|
48
|
+
}
|
49
|
+
std::size_t child() const {
|
50
|
+
return child_;
|
51
|
+
}
|
52
|
+
UInt8 base() const {
|
53
|
+
return (UInt8)(union_.link & 0xFFU);
|
54
|
+
}
|
55
|
+
std::size_t extra() const {
|
56
|
+
return union_.link >> 8;
|
57
|
+
}
|
58
|
+
char label() const {
|
59
|
+
return (char)base();
|
60
|
+
}
|
61
|
+
std::size_t link() const {
|
62
|
+
return union_.link;
|
63
|
+
}
|
64
|
+
float weight() const {
|
65
|
+
return union_.weight;
|
66
|
+
}
|
67
|
+
|
68
|
+
private:
|
69
|
+
UInt32 parent_;
|
70
|
+
UInt32 child_;
|
71
|
+
union Union {
|
72
|
+
UInt32 link;
|
73
|
+
float weight;
|
74
|
+
} union_;
|
75
|
+
};
|
76
|
+
|
77
|
+
} // namespace trie
|
78
|
+
} // namespace grimoire
|
79
|
+
} // namespace marisa
|
80
|
+
|
81
|
+
#endif // MARISA_GRIMOIRE_TRIE_CACHE_H_
|
@@ -0,0 +1,155 @@
|
|
1
|
+
#ifndef MARISA_GRIMOIRE_TRIE_CONFIG_H_
|
2
|
+
#define MARISA_GRIMOIRE_TRIE_CONFIG_H_
|
3
|
+
|
4
|
+
#include "marisa/base.h"
|
5
|
+
|
6
|
+
namespace marisa {
|
7
|
+
namespace grimoire {
|
8
|
+
namespace trie {
|
9
|
+
|
10
|
+
class Config {
|
11
|
+
public:
|
12
|
+
Config()
|
13
|
+
: num_tries_(MARISA_DEFAULT_NUM_TRIES),
|
14
|
+
cache_level_(MARISA_DEFAULT_CACHE),
|
15
|
+
tail_mode_(MARISA_DEFAULT_TAIL),
|
16
|
+
node_order_(MARISA_DEFAULT_ORDER) {}
|
17
|
+
|
18
|
+
void parse(int config_flags) {
|
19
|
+
Config temp;
|
20
|
+
temp.parse_(config_flags);
|
21
|
+
swap(temp);
|
22
|
+
}
|
23
|
+
|
24
|
+
int flags() const {
|
25
|
+
return (int)num_tries_ | tail_mode_ | node_order_;
|
26
|
+
}
|
27
|
+
|
28
|
+
std::size_t num_tries() const {
|
29
|
+
return num_tries_;
|
30
|
+
}
|
31
|
+
CacheLevel cache_level() const {
|
32
|
+
return cache_level_;
|
33
|
+
}
|
34
|
+
TailMode tail_mode() const {
|
35
|
+
return tail_mode_;
|
36
|
+
}
|
37
|
+
NodeOrder node_order() const {
|
38
|
+
return node_order_;
|
39
|
+
}
|
40
|
+
|
41
|
+
void clear() {
|
42
|
+
Config().swap(*this);
|
43
|
+
}
|
44
|
+
void swap(Config &rhs) {
|
45
|
+
marisa::swap(num_tries_, rhs.num_tries_);
|
46
|
+
marisa::swap(cache_level_, rhs.cache_level_);
|
47
|
+
marisa::swap(tail_mode_, rhs.tail_mode_);
|
48
|
+
marisa::swap(node_order_, rhs.node_order_);
|
49
|
+
}
|
50
|
+
|
51
|
+
private:
|
52
|
+
std::size_t num_tries_;
|
53
|
+
CacheLevel cache_level_;
|
54
|
+
TailMode tail_mode_;
|
55
|
+
NodeOrder node_order_;
|
56
|
+
|
57
|
+
void parse_(int config_flags) {
|
58
|
+
MARISA_THROW_IF((config_flags & ~MARISA_CONFIG_MASK) != 0,
|
59
|
+
MARISA_CODE_ERROR);
|
60
|
+
|
61
|
+
parse_num_tries(config_flags);
|
62
|
+
parse_cache_level(config_flags);
|
63
|
+
parse_tail_mode(config_flags);
|
64
|
+
parse_node_order(config_flags);
|
65
|
+
}
|
66
|
+
|
67
|
+
void parse_num_tries(int config_flags) {
|
68
|
+
const int num_tries = config_flags & MARISA_NUM_TRIES_MASK;
|
69
|
+
if (num_tries != 0) {
|
70
|
+
num_tries_ = num_tries;
|
71
|
+
}
|
72
|
+
}
|
73
|
+
|
74
|
+
void parse_cache_level(int config_flags) {
|
75
|
+
switch (config_flags & MARISA_CACHE_LEVEL_MASK) {
|
76
|
+
case 0: {
|
77
|
+
cache_level_ = MARISA_DEFAULT_CACHE;
|
78
|
+
break;
|
79
|
+
}
|
80
|
+
case MARISA_HUGE_CACHE: {
|
81
|
+
cache_level_ = MARISA_HUGE_CACHE;
|
82
|
+
break;
|
83
|
+
}
|
84
|
+
case MARISA_LARGE_CACHE: {
|
85
|
+
cache_level_ = MARISA_LARGE_CACHE;
|
86
|
+
break;
|
87
|
+
}
|
88
|
+
case MARISA_NORMAL_CACHE: {
|
89
|
+
cache_level_ = MARISA_NORMAL_CACHE;
|
90
|
+
break;
|
91
|
+
}
|
92
|
+
case MARISA_SMALL_CACHE: {
|
93
|
+
cache_level_ = MARISA_SMALL_CACHE;
|
94
|
+
break;
|
95
|
+
}
|
96
|
+
case MARISA_TINY_CACHE: {
|
97
|
+
cache_level_ = MARISA_TINY_CACHE;
|
98
|
+
break;
|
99
|
+
}
|
100
|
+
default: {
|
101
|
+
MARISA_THROW(MARISA_CODE_ERROR, "undefined cache level");
|
102
|
+
}
|
103
|
+
}
|
104
|
+
}
|
105
|
+
|
106
|
+
void parse_tail_mode(int config_flags) {
|
107
|
+
switch (config_flags & MARISA_TAIL_MODE_MASK) {
|
108
|
+
case 0: {
|
109
|
+
tail_mode_ = MARISA_DEFAULT_TAIL;
|
110
|
+
break;
|
111
|
+
}
|
112
|
+
case MARISA_TEXT_TAIL: {
|
113
|
+
tail_mode_ = MARISA_TEXT_TAIL;
|
114
|
+
break;
|
115
|
+
}
|
116
|
+
case MARISA_BINARY_TAIL: {
|
117
|
+
tail_mode_ = MARISA_BINARY_TAIL;
|
118
|
+
break;
|
119
|
+
}
|
120
|
+
default: {
|
121
|
+
MARISA_THROW(MARISA_CODE_ERROR, "undefined tail mode");
|
122
|
+
}
|
123
|
+
}
|
124
|
+
}
|
125
|
+
|
126
|
+
void parse_node_order(int config_flags) {
|
127
|
+
switch (config_flags & MARISA_NODE_ORDER_MASK) {
|
128
|
+
case 0: {
|
129
|
+
node_order_ = MARISA_DEFAULT_ORDER;
|
130
|
+
break;
|
131
|
+
}
|
132
|
+
case MARISA_LABEL_ORDER: {
|
133
|
+
node_order_ = MARISA_LABEL_ORDER;
|
134
|
+
break;
|
135
|
+
}
|
136
|
+
case MARISA_WEIGHT_ORDER: {
|
137
|
+
node_order_ = MARISA_WEIGHT_ORDER;
|
138
|
+
break;
|
139
|
+
}
|
140
|
+
default: {
|
141
|
+
MARISA_THROW(MARISA_CODE_ERROR, "undefined node order");
|
142
|
+
}
|
143
|
+
}
|
144
|
+
}
|
145
|
+
|
146
|
+
// Disallows copy and assignment.
|
147
|
+
Config(const Config &);
|
148
|
+
Config &operator=(const Config &);
|
149
|
+
};
|
150
|
+
|
151
|
+
} // namespace trie
|
152
|
+
} // namespace grimoire
|
153
|
+
} // namespace marisa
|
154
|
+
|
155
|
+
#endif // MARISA_GRIMOIRE_TRIE_CONFIG_H_
|