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 +1,115 @@
|
|
1
|
+
#ifndef MARISA_GRIMOIRE_INTRIN_H_
|
2
|
+
#define MARISA_GRIMOIRE_INTRIN_H_
|
3
|
+
|
4
|
+
#include "marisa/base.h"
|
5
|
+
|
6
|
+
#if defined(__x86_64__) || defined(_M_X64)
|
7
|
+
#define MARISA_X64
|
8
|
+
#elif defined(__i386__) || defined(_M_IX86)
|
9
|
+
#define MARISA_X86
|
10
|
+
#else // defined(__i386__) || defined(_M_IX86)
|
11
|
+
#ifdef MARISA_USE_POPCNT
|
12
|
+
#undef MARISA_USE_POPCNT
|
13
|
+
#endif // MARISA_USE_POPCNT
|
14
|
+
#ifdef MARISA_USE_SSE4A
|
15
|
+
#undef MARISA_USE_SSE4A
|
16
|
+
#endif // MARISA_USE_SSE4A
|
17
|
+
#ifdef MARISA_USE_SSE4
|
18
|
+
#undef MARISA_USE_SSE4
|
19
|
+
#endif // MARISA_USE_SSE4
|
20
|
+
#ifdef MARISA_USE_SSE4_2
|
21
|
+
#undef MARISA_USE_SSE4_2
|
22
|
+
#endif // MARISA_USE_SSE4_2
|
23
|
+
#ifdef MARISA_USE_SSE4_1
|
24
|
+
#undef MARISA_USE_SSE4_1
|
25
|
+
#endif // MARISA_USE_SSE4_1
|
26
|
+
#ifdef MARISA_USE_SSSE3
|
27
|
+
#undef MARISA_USE_SSSE3
|
28
|
+
#endif // MARISA_USE_SSSE3
|
29
|
+
#ifdef MARISA_USE_SSE3
|
30
|
+
#undef MARISA_USE_SSE3
|
31
|
+
#endif // MARISA_USE_SSE3
|
32
|
+
#ifdef MARISA_USE_SSE2
|
33
|
+
#undef MARISA_USE_SSE2
|
34
|
+
#endif // MARISA_USE_SSE2
|
35
|
+
#endif // defined(__i386__) || defined(_M_IX86)
|
36
|
+
|
37
|
+
#ifdef MARISA_USE_POPCNT
|
38
|
+
#ifndef MARISA_USE_SSE3
|
39
|
+
#define MARISA_USE_SSE3
|
40
|
+
#endif // MARISA_USE_SSE3
|
41
|
+
#ifdef _MSC_VER
|
42
|
+
#include <intrin.h>
|
43
|
+
#else // _MSC_VER
|
44
|
+
#include <popcntintrin.h>
|
45
|
+
#endif // _MSC_VER
|
46
|
+
#endif // MARISA_USE_POPCNT
|
47
|
+
|
48
|
+
#ifdef MARISA_USE_SSE4A
|
49
|
+
#ifndef MARISA_USE_SSE3
|
50
|
+
#define MARISA_USE_SSE3
|
51
|
+
#endif // MARISA_USE_SSE3
|
52
|
+
#ifndef MARISA_USE_POPCNT
|
53
|
+
#define MARISA_USE_POPCNT
|
54
|
+
#endif // MARISA_USE_POPCNT
|
55
|
+
#endif // MARISA_USE_SSE4A
|
56
|
+
|
57
|
+
#ifdef MARISA_USE_SSE4
|
58
|
+
#ifndef MARISA_USE_SSE4_2
|
59
|
+
#define MARISA_USE_SSE4_2
|
60
|
+
#endif // MARISA_USE_SSE4_2
|
61
|
+
#endif // MARISA_USE_SSE4
|
62
|
+
|
63
|
+
#ifdef MARISA_USE_SSE4_2
|
64
|
+
#ifndef MARISA_USE_SSE4_1
|
65
|
+
#define MARISA_USE_SSE4_1
|
66
|
+
#endif // MARISA_USE_SSE4_1
|
67
|
+
#ifndef MARISA_USE_POPCNT
|
68
|
+
#define MARISA_USE_POPCNT
|
69
|
+
#endif // MARISA_USE_POPCNT
|
70
|
+
#endif // MARISA_USE_SSE4_2
|
71
|
+
|
72
|
+
#ifdef MARISA_USE_SSE4_1
|
73
|
+
#ifndef MARISA_USE_SSSE3
|
74
|
+
#define MARISA_USE_SSSE3
|
75
|
+
#endif // MARISA_USE_SSSE3
|
76
|
+
#endif // MARISA_USE_SSE4_1
|
77
|
+
|
78
|
+
#ifdef MARISA_USE_SSSE3
|
79
|
+
#ifndef MARISA_USE_SSE3
|
80
|
+
#define MARISA_USE_SSE3
|
81
|
+
#endif // MARISA_USE_SSE3
|
82
|
+
#ifdef MARISA_X64
|
83
|
+
#define MARISA_X64_SSSE3
|
84
|
+
#else // MARISA_X64
|
85
|
+
#define MARISA_X86_SSSE3
|
86
|
+
#endif // MAIRSA_X64
|
87
|
+
#include <tmmintrin.h>
|
88
|
+
#endif // MARISA_USE_SSSE3
|
89
|
+
|
90
|
+
#ifdef MARISA_USE_SSE3
|
91
|
+
#ifndef MARISA_USE_SSE2
|
92
|
+
#define MARISA_USE_SSE2
|
93
|
+
#endif // MARISA_USE_SSE2
|
94
|
+
#endif // MARISA_USE_SSE3
|
95
|
+
|
96
|
+
#ifdef MARISA_USE_SSE2
|
97
|
+
#ifdef MARISA_X64
|
98
|
+
#define MARISA_X64_SSE2
|
99
|
+
#else // MARISA_X64
|
100
|
+
#define MARISA_X86_SSE2
|
101
|
+
#endif // MAIRSA_X64
|
102
|
+
#include <emmintrin.h>
|
103
|
+
#endif // MARISA_USE_SSE2
|
104
|
+
|
105
|
+
#ifdef _MSC_VER
|
106
|
+
#if MARISA_WORD_SIZE == 64
|
107
|
+
#include <intrin.h>
|
108
|
+
#pragma intrinsic(_BitScanForward64)
|
109
|
+
#else // MARISA_WORD_SIZE == 64
|
110
|
+
#include <intrin.h>
|
111
|
+
#pragma intrinsic(_BitScanForward)
|
112
|
+
#endif // MARISA_WORD_SIZE == 64
|
113
|
+
#endif // _MSC_VER
|
114
|
+
|
115
|
+
#endif // MARISA_GRIMOIRE_INTRIN_H_
|
@@ -0,0 +1,18 @@
|
|
1
|
+
#ifndef MARISA_GRIMOIRE_IO_H_
|
2
|
+
#define MARISA_GRIMOIRE_IO_H_
|
3
|
+
|
4
|
+
#include "marisa/grimoire/io/mapper.h"
|
5
|
+
#include "marisa/grimoire/io/reader.h"
|
6
|
+
#include "marisa/grimoire/io/writer.h"
|
7
|
+
|
8
|
+
namespace marisa {
|
9
|
+
namespace grimoire {
|
10
|
+
|
11
|
+
using io::Mapper;
|
12
|
+
using io::Reader;
|
13
|
+
using io::Writer;
|
14
|
+
|
15
|
+
} // namespace grimoire
|
16
|
+
} // namespace marisa
|
17
|
+
|
18
|
+
#endif // MARISA_GRIMOIRE_IO_H_
|
@@ -0,0 +1,163 @@
|
|
1
|
+
#if (defined _WIN32) || (defined _WIN64)
|
2
|
+
#include <sys/types.h>
|
3
|
+
#include <sys/stat.h>
|
4
|
+
#include <windows.h>
|
5
|
+
#else // (defined _WIN32) || (defined _WIN64)
|
6
|
+
#include <sys/mman.h>
|
7
|
+
#include <sys/stat.h>
|
8
|
+
#include <sys/types.h>
|
9
|
+
#include <fcntl.h>
|
10
|
+
#include <unistd.h>
|
11
|
+
#endif // (defined _WIN32) || (defined _WIN64)
|
12
|
+
|
13
|
+
#include "marisa/grimoire/io/mapper.h"
|
14
|
+
|
15
|
+
namespace marisa {
|
16
|
+
namespace grimoire {
|
17
|
+
namespace io {
|
18
|
+
|
19
|
+
#if (defined _WIN32) || (defined _WIN64)
|
20
|
+
Mapper::Mapper()
|
21
|
+
: ptr_(NULL), origin_(NULL), avail_(0), size_(0),
|
22
|
+
file_(NULL), map_(NULL) {}
|
23
|
+
#else // (defined _WIN32) || (defined _WIN64)
|
24
|
+
Mapper::Mapper()
|
25
|
+
: ptr_(NULL), origin_(MAP_FAILED), avail_(0), size_(0), fd_(-1) {}
|
26
|
+
#endif // (defined _WIN32) || (defined _WIN64)
|
27
|
+
|
28
|
+
#if (defined _WIN32) || (defined _WIN64)
|
29
|
+
Mapper::~Mapper() {
|
30
|
+
if (origin_ != NULL) {
|
31
|
+
::UnmapViewOfFile(origin_);
|
32
|
+
}
|
33
|
+
|
34
|
+
if (map_ != NULL) {
|
35
|
+
::CloseHandle(map_);
|
36
|
+
}
|
37
|
+
|
38
|
+
if (file_ != NULL) {
|
39
|
+
::CloseHandle(file_);
|
40
|
+
}
|
41
|
+
}
|
42
|
+
#else // (defined _WIN32) || (defined _WIN64)
|
43
|
+
Mapper::~Mapper() {
|
44
|
+
if (origin_ != MAP_FAILED) {
|
45
|
+
::munmap(origin_, size_);
|
46
|
+
}
|
47
|
+
|
48
|
+
if (fd_ != -1) {
|
49
|
+
::close(fd_);
|
50
|
+
}
|
51
|
+
}
|
52
|
+
#endif // (defined _WIN32) || (defined _WIN64)
|
53
|
+
|
54
|
+
void Mapper::open(const char *filename) {
|
55
|
+
MARISA_THROW_IF(filename == NULL, MARISA_NULL_ERROR);
|
56
|
+
|
57
|
+
Mapper temp;
|
58
|
+
temp.open_(filename);
|
59
|
+
swap(temp);
|
60
|
+
}
|
61
|
+
|
62
|
+
void Mapper::open(const void *ptr, std::size_t size) {
|
63
|
+
MARISA_THROW_IF((ptr == NULL) && (size != 0), MARISA_NULL_ERROR);
|
64
|
+
|
65
|
+
Mapper temp;
|
66
|
+
temp.open_(ptr, size);
|
67
|
+
swap(temp);
|
68
|
+
}
|
69
|
+
|
70
|
+
void Mapper::seek(std::size_t size) {
|
71
|
+
MARISA_THROW_IF(!is_open(), MARISA_STATE_ERROR);
|
72
|
+
MARISA_THROW_IF(size > avail_, MARISA_IO_ERROR);
|
73
|
+
|
74
|
+
map_data(size);
|
75
|
+
}
|
76
|
+
|
77
|
+
bool Mapper::is_open() const {
|
78
|
+
return ptr_ != NULL;
|
79
|
+
}
|
80
|
+
|
81
|
+
void Mapper::clear() {
|
82
|
+
Mapper().swap(*this);
|
83
|
+
}
|
84
|
+
|
85
|
+
void Mapper::swap(Mapper &rhs) {
|
86
|
+
marisa::swap(ptr_, rhs.ptr_);
|
87
|
+
marisa::swap(avail_, rhs.avail_);
|
88
|
+
marisa::swap(origin_, rhs.origin_);
|
89
|
+
marisa::swap(size_, rhs.size_);
|
90
|
+
#if (defined _WIN32) || (defined _WIN64)
|
91
|
+
marisa::swap(file_, rhs.file_);
|
92
|
+
marisa::swap(map_, rhs.map_);
|
93
|
+
#else // (defined _WIN32) || (defined _WIN64)
|
94
|
+
marisa::swap(fd_, rhs.fd_);
|
95
|
+
#endif // (defined _WIN32) || (defined _WIN64)
|
96
|
+
}
|
97
|
+
|
98
|
+
const void *Mapper::map_data(std::size_t size) {
|
99
|
+
MARISA_THROW_IF(!is_open(), MARISA_STATE_ERROR);
|
100
|
+
MARISA_THROW_IF(size > avail_, MARISA_IO_ERROR);
|
101
|
+
|
102
|
+
const char * const data = static_cast<const char *>(ptr_);
|
103
|
+
ptr_ = data + size;
|
104
|
+
avail_ -= size;
|
105
|
+
return data;
|
106
|
+
}
|
107
|
+
|
108
|
+
#if (defined _WIN32) || (defined _WIN64)
|
109
|
+
#ifdef __MSVCRT_VERSION__
|
110
|
+
#if __MSVCRT_VERSION__ >= 0x0601
|
111
|
+
#define MARISA_HAS_STAT64
|
112
|
+
#endif // __MSVCRT_VERSION__ >= 0x0601
|
113
|
+
#endif // __MSVCRT_VERSION__
|
114
|
+
void Mapper::open_(const char *filename) {
|
115
|
+
#ifdef MARISA_HAS_STAT64
|
116
|
+
struct __stat64 st;
|
117
|
+
MARISA_THROW_IF(::_stat64(filename, &st) != 0, MARISA_IO_ERROR);
|
118
|
+
#else // MARISA_HAS_STAT64
|
119
|
+
struct _stat st;
|
120
|
+
MARISA_THROW_IF(::_stat(filename, &st) != 0, MARISA_IO_ERROR);
|
121
|
+
#endif // MARISA_HAS_STAT64
|
122
|
+
MARISA_THROW_IF((UInt64)st.st_size > MARISA_SIZE_MAX, MARISA_SIZE_ERROR);
|
123
|
+
size_ = (std::size_t)st.st_size;
|
124
|
+
|
125
|
+
file_ = ::CreateFileA(filename, GENERIC_READ, FILE_SHARE_READ,
|
126
|
+
NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL);
|
127
|
+
MARISA_THROW_IF(file_ == INVALID_HANDLE_VALUE, MARISA_IO_ERROR);
|
128
|
+
|
129
|
+
map_ = ::CreateFileMapping(file_, NULL, PAGE_READONLY, 0, 0, NULL);
|
130
|
+
MARISA_THROW_IF(map_ == NULL, MARISA_IO_ERROR);
|
131
|
+
|
132
|
+
origin_ = ::MapViewOfFile(map_, FILE_MAP_READ, 0, 0, 0);
|
133
|
+
MARISA_THROW_IF(origin_ == NULL, MARISA_IO_ERROR);
|
134
|
+
|
135
|
+
ptr_ = static_cast<const char *>(origin_);
|
136
|
+
avail_ = size_;
|
137
|
+
}
|
138
|
+
#else // (defined _WIN32) || (defined _WIN64)
|
139
|
+
void Mapper::open_(const char *filename) {
|
140
|
+
struct stat st;
|
141
|
+
MARISA_THROW_IF(::stat(filename, &st) != 0, MARISA_IO_ERROR);
|
142
|
+
MARISA_THROW_IF((UInt64)st.st_size > MARISA_SIZE_MAX, MARISA_SIZE_ERROR);
|
143
|
+
size_ = (std::size_t)st.st_size;
|
144
|
+
|
145
|
+
fd_ = ::open(filename, O_RDONLY);
|
146
|
+
MARISA_THROW_IF(fd_ == -1, MARISA_IO_ERROR);
|
147
|
+
|
148
|
+
origin_ = ::mmap(NULL, size_, PROT_READ, MAP_SHARED, fd_, 0);
|
149
|
+
MARISA_THROW_IF(origin_ == MAP_FAILED, MARISA_IO_ERROR);
|
150
|
+
|
151
|
+
ptr_ = static_cast<const char *>(origin_);
|
152
|
+
avail_ = size_;
|
153
|
+
}
|
154
|
+
#endif // (defined _WIN32) || (defined _WIN64)
|
155
|
+
|
156
|
+
void Mapper::open_(const void *ptr, std::size_t size) {
|
157
|
+
ptr_ = ptr;
|
158
|
+
avail_ = size;
|
159
|
+
}
|
160
|
+
|
161
|
+
} // namespace io
|
162
|
+
} // namespace grimoire
|
163
|
+
} // namespace marisa
|
@@ -0,0 +1,67 @@
|
|
1
|
+
#ifndef MARISA_GRIMOIRE_IO_MAPPER_H_
|
2
|
+
#define MARISA_GRIMOIRE_IO_MAPPER_H_
|
3
|
+
|
4
|
+
#include <cstdio>
|
5
|
+
|
6
|
+
#include "marisa/base.h"
|
7
|
+
|
8
|
+
namespace marisa {
|
9
|
+
namespace grimoire {
|
10
|
+
namespace io {
|
11
|
+
|
12
|
+
class Mapper {
|
13
|
+
public:
|
14
|
+
Mapper();
|
15
|
+
~Mapper();
|
16
|
+
|
17
|
+
void open(const char *filename);
|
18
|
+
void open(const void *ptr, std::size_t size);
|
19
|
+
|
20
|
+
template <typename T>
|
21
|
+
void map(T *obj) {
|
22
|
+
MARISA_THROW_IF(obj == NULL, MARISA_NULL_ERROR);
|
23
|
+
*obj = *static_cast<const T *>(map_data(sizeof(T)));
|
24
|
+
}
|
25
|
+
|
26
|
+
template <typename T>
|
27
|
+
void map(const T **objs, std::size_t num_objs) {
|
28
|
+
MARISA_THROW_IF((objs == NULL) && (num_objs != 0), MARISA_NULL_ERROR);
|
29
|
+
MARISA_THROW_IF(num_objs > (MARISA_SIZE_MAX / sizeof(T)),
|
30
|
+
MARISA_SIZE_ERROR);
|
31
|
+
*objs = static_cast<const T *>(map_data(sizeof(T) * num_objs));
|
32
|
+
}
|
33
|
+
|
34
|
+
void seek(std::size_t size);
|
35
|
+
|
36
|
+
bool is_open() const;
|
37
|
+
|
38
|
+
void clear();
|
39
|
+
void swap(Mapper &rhs);
|
40
|
+
|
41
|
+
private:
|
42
|
+
const void *ptr_;
|
43
|
+
void *origin_;
|
44
|
+
std::size_t avail_;
|
45
|
+
std::size_t size_;
|
46
|
+
#if (defined _WIN32) || (defined _WIN64)
|
47
|
+
void *file_;
|
48
|
+
void *map_;
|
49
|
+
#else // (defined _WIN32) || (defined _WIN64)
|
50
|
+
int fd_;
|
51
|
+
#endif // (defined _WIN32) || (defined _WIN64)
|
52
|
+
|
53
|
+
void open_(const char *filename);
|
54
|
+
void open_(const void *ptr, std::size_t size);
|
55
|
+
|
56
|
+
const void *map_data(std::size_t size);
|
57
|
+
|
58
|
+
// Disallows copy and assignment.
|
59
|
+
Mapper(const Mapper &);
|
60
|
+
Mapper &operator=(const Mapper &);
|
61
|
+
};
|
62
|
+
|
63
|
+
} // namespace io
|
64
|
+
} // namespace grimoire
|
65
|
+
} // namespace marisa
|
66
|
+
|
67
|
+
#endif // MARISA_GRIMOIRE_IO_MAPPER_H_
|
@@ -0,0 +1,147 @@
|
|
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/reader.h"
|
12
|
+
|
13
|
+
namespace marisa {
|
14
|
+
namespace grimoire {
|
15
|
+
namespace io {
|
16
|
+
|
17
|
+
Reader::Reader()
|
18
|
+
: file_(NULL), fd_(-1), stream_(NULL), needs_fclose_(false) {}
|
19
|
+
|
20
|
+
Reader::~Reader() {
|
21
|
+
if (needs_fclose_) {
|
22
|
+
::fclose(file_);
|
23
|
+
}
|
24
|
+
}
|
25
|
+
|
26
|
+
void Reader::open(const char *filename) {
|
27
|
+
MARISA_THROW_IF(filename == NULL, MARISA_NULL_ERROR);
|
28
|
+
|
29
|
+
Reader temp;
|
30
|
+
temp.open_(filename);
|
31
|
+
swap(temp);
|
32
|
+
}
|
33
|
+
|
34
|
+
void Reader::open(std::FILE *file) {
|
35
|
+
MARISA_THROW_IF(file == NULL, MARISA_NULL_ERROR);
|
36
|
+
|
37
|
+
Reader temp;
|
38
|
+
temp.open_(file);
|
39
|
+
swap(temp);
|
40
|
+
}
|
41
|
+
|
42
|
+
void Reader::open(int fd) {
|
43
|
+
MARISA_THROW_IF(fd == -1, MARISA_CODE_ERROR);
|
44
|
+
|
45
|
+
Reader temp;
|
46
|
+
temp.open_(fd);
|
47
|
+
swap(temp);
|
48
|
+
}
|
49
|
+
|
50
|
+
void Reader::open(std::istream &stream) {
|
51
|
+
Reader temp;
|
52
|
+
temp.open_(stream);
|
53
|
+
swap(temp);
|
54
|
+
}
|
55
|
+
|
56
|
+
void Reader::clear() {
|
57
|
+
Reader().swap(*this);
|
58
|
+
}
|
59
|
+
|
60
|
+
void Reader::swap(Reader &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 Reader::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
|
+
char buf[16];
|
73
|
+
read_data(buf, size);
|
74
|
+
} else {
|
75
|
+
char buf[1024];
|
76
|
+
while (size != 0) {
|
77
|
+
const std::size_t count = (size < sizeof(buf)) ? size : sizeof(buf);
|
78
|
+
read_data(buf, count);
|
79
|
+
size -= count;
|
80
|
+
}
|
81
|
+
}
|
82
|
+
}
|
83
|
+
|
84
|
+
bool Reader::is_open() const {
|
85
|
+
return (file_ != NULL) || (fd_ != -1) || (stream_ != NULL);
|
86
|
+
}
|
87
|
+
|
88
|
+
void Reader::open_(const char *filename) {
|
89
|
+
std::FILE *file = NULL;
|
90
|
+
#ifdef _MSC_VER
|
91
|
+
MARISA_THROW_IF(::fopen_s(&file, filename, "rb") != 0, MARISA_IO_ERROR);
|
92
|
+
#else // _MSC_VER
|
93
|
+
file = ::fopen(filename, "rb");
|
94
|
+
MARISA_THROW_IF(file == NULL, MARISA_IO_ERROR);
|
95
|
+
#endif // _MSC_VER
|
96
|
+
file_ = file;
|
97
|
+
needs_fclose_ = true;
|
98
|
+
}
|
99
|
+
|
100
|
+
void Reader::open_(std::FILE *file) {
|
101
|
+
file_ = file;
|
102
|
+
}
|
103
|
+
|
104
|
+
void Reader::open_(int fd) {
|
105
|
+
fd_ = fd;
|
106
|
+
}
|
107
|
+
|
108
|
+
void Reader::open_(std::istream &stream) {
|
109
|
+
stream_ = &stream;
|
110
|
+
}
|
111
|
+
|
112
|
+
void Reader::read_data(void *buf, 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_read = ::_read(fd_, buf, 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_read = ::read(fd_, buf, count);
|
128
|
+
#endif // _WIN32
|
129
|
+
MARISA_THROW_IF(size_read <= 0, MARISA_IO_ERROR);
|
130
|
+
buf = static_cast<char *>(buf) + size_read;
|
131
|
+
size -= size_read;
|
132
|
+
}
|
133
|
+
} else if (file_ != NULL) {
|
134
|
+
MARISA_THROW_IF(::fread(buf, 1, size, file_) != size, MARISA_IO_ERROR);
|
135
|
+
} else if (stream_ != NULL) {
|
136
|
+
try {
|
137
|
+
MARISA_THROW_IF(!stream_->read(static_cast<char *>(buf), size),
|
138
|
+
MARISA_IO_ERROR);
|
139
|
+
} catch (const std::ios_base::failure &) {
|
140
|
+
MARISA_THROW(MARISA_IO_ERROR, "std::ios_base::failure");
|
141
|
+
}
|
142
|
+
}
|
143
|
+
}
|
144
|
+
|
145
|
+
} // namespace io
|
146
|
+
} // namespace grimoire
|
147
|
+
} // namespace marisa
|