melisa 0.1.0

data/ext/marisa/lib/marisa/keyset.h

@@ -0,0 +1,80 @@
+#ifndef MARISA_KEYSET_H_
+#define MARISA_KEYSET_H_
+
+#include "marisa/key.h"
+
+namespace marisa {
+
+class Keyset {
+ public:
+  enum {
+    BASE_BLOCK_SIZE  = 4096,
+    EXTRA_BLOCK_SIZE = 1024,
+    KEY_BLOCK_SIZE   = 256
+  };
+
+  Keyset();
+
+  void push_back(const Key &key);
+  void push_back(const Key &key, char end_marker);
+
+  void push_back(const char *str);
+  void push_back(const char *ptr, std::size_t length, float weight = 1.0);
+
+  const Key &operator[](std::size_t i) const {
+    MARISA_DEBUG_IF(i >= size_, MARISA_BOUND_ERROR);
+    return key_blocks_[i / KEY_BLOCK_SIZE][i % KEY_BLOCK_SIZE];
+  }
+  Key &operator[](std::size_t i) {
+    MARISA_DEBUG_IF(i >= size_, MARISA_BOUND_ERROR);
+    return key_blocks_[i / KEY_BLOCK_SIZE][i % KEY_BLOCK_SIZE];
+  }
+
+  std::size_t num_keys() const {
+    return size_;
+  }
+
+  bool empty() const {
+    return size_ == 0;
+  }
+  std::size_t size() const {
+    return size_;
+  }
+  std::size_t total_length() const {
+    return total_length_;
+  }
+
+  void reset();
+
+  void clear();
+  void swap(Keyset &rhs);
+
+ private:
+  scoped_array<scoped_array<char> > base_blocks_;
+  std::size_t base_blocks_size_;
+  std::size_t base_blocks_capacity_;
+  scoped_array<scoped_array<char> > extra_blocks_;
+  std::size_t extra_blocks_size_;
+  std::size_t extra_blocks_capacity_;
+  scoped_array<scoped_array<Key> > key_blocks_;
+  std::size_t key_blocks_size_;
+  std::size_t key_blocks_capacity_;
+  char *ptr_;
+  std::size_t avail_;
+  std::size_t size_;
+  std::size_t total_length_;
+
+  char *reserve(std::size_t size);
+
+  void append_base_block();
+  void append_extra_block(std::size_t size);
+  void append_key_block();
+
+  // Disallows copy and assignment.
+  Keyset(const Keyset &);
+  Keyset &operator=(const Keyset &);
+};
+
+}  // namespace marisa
+
+#endif  // MARISA_KEYSET_H_

data/ext/marisa/lib/marisa/query.h

@@ -0,0 +1,71 @@
+#ifndef MARISA_QUERY_H_
+#define MARISA_QUERY_H_
+
+#include "marisa/base.h"
+
+namespace marisa {
+
+class Query {
+ public:
+  Query() : ptr_(NULL), length_(0), id_(0) {}
+  Query(const Query &query)
+      : ptr_(query.ptr_), length_(query.length_), id_(query.id_) {}
+
+  Query &operator=(const Query &query) {
+    ptr_ = query.ptr_;
+    length_ = query.length_;
+    id_ = query.id_;
+    return *this;
+  }
+
+  char operator[](std::size_t i) const {
+    MARISA_DEBUG_IF(i >= length_, MARISA_BOUND_ERROR);
+    return ptr_[i];
+  }
+
+  void set_str(const char *str) {
+    MARISA_DEBUG_IF(str == NULL, MARISA_NULL_ERROR);
+    std::size_t length = 0;
+    while (str[length] != '\0') {
+      ++length;
+    }
+    ptr_ = str;
+    length_ = length;
+  }
+  void set_str(const char *ptr, std::size_t length) {
+    MARISA_DEBUG_IF((ptr == NULL) && (length != 0), MARISA_NULL_ERROR);
+    ptr_ = ptr;
+    length_ = length;
+  }
+  void set_id(std::size_t id) {
+    id_ = id;
+  }
+
+  const char *ptr() const {
+    return ptr_;
+  }
+  std::size_t length() const {
+    return length_;
+  }
+  std::size_t id() const {
+    return id_;
+  }
+
+  void clear() {
+    Query().swap(*this);
+  }
+  void swap(Query &rhs) {
+    marisa::swap(ptr_, rhs.ptr_);
+    marisa::swap(length_, rhs.length_);
+    marisa::swap(id_, rhs.id_);
+  }
+
+ private:
+  const char *ptr_;
+  std::size_t length_;
+  std::size_t id_;
+};
+
+}  // namespace marisa
+
+#endif  // MARISA_QUERY_H_

data/ext/marisa/lib/marisa/scoped-array.h

@@ -0,0 +1,48 @@
+#ifndef MARISA_SCOPED_ARRAY_H_
+#define MARISA_SCOPED_ARRAY_H_
+
+#include "marisa/base.h"
+
+namespace marisa {
+
+template <typename T>
+class scoped_array {
+ public:
+  scoped_array() : array_(NULL) {}
+  explicit scoped_array(T *array) : array_(array) {}
+
+  ~scoped_array() {
+    delete [] array_;
+  }
+
+  void reset(T *array = NULL) {
+    MARISA_THROW_IF((array != NULL) && (array == array_), MARISA_RESET_ERROR);
+    scoped_array(array).swap(*this);
+  }
+
+  T &operator[](std::size_t i) const {
+    MARISA_DEBUG_IF(array_ == NULL, MARISA_STATE_ERROR);
+    return array_[i];
+  }
+  T *get() const {
+    return array_;
+  }
+
+  void clear() {
+    scoped_array().swap(*this);
+  }
+  void swap(scoped_array &rhs) {
+    marisa::swap(array_, rhs.array_);
+  }
+
+ private:
+  T *array_;
+
+  // Disallows copy and assignment.
+  scoped_array(const scoped_array &);
+  scoped_array &operator=(const scoped_array &);
+};
+
+}  // namespace marisa
+
+#endif  // MARISA_SCOPED_ARRAY_H_

data/ext/marisa/lib/marisa/scoped-ptr.h

@@ -0,0 +1,52 @@
+#ifndef MARISA_SCOPED_PTR_H_
+#define MARISA_SCOPED_PTR_H_
+
+#include "marisa/base.h"
+
+namespace marisa {
+
+template <typename T>
+class scoped_ptr {
+ public:
+  scoped_ptr() : ptr_(NULL) {}
+  explicit scoped_ptr(T *ptr) : ptr_(ptr) {}
+
+  ~scoped_ptr() {
+    delete ptr_;
+  }
+
+  void reset(T *ptr = NULL) {
+    MARISA_THROW_IF((ptr != NULL) && (ptr == ptr_), MARISA_RESET_ERROR);
+    scoped_ptr(ptr).swap(*this);
+  }
+
+  T &operator*() const {
+    MARISA_DEBUG_IF(ptr_ == NULL, MARISA_STATE_ERROR);
+    return *ptr_;
+  }
+  T *operator->() const {
+    MARISA_DEBUG_IF(ptr_ == NULL, MARISA_STATE_ERROR);
+    return ptr_;
+  }
+  T *get() const {
+    return ptr_;
+  }
+
+  void clear() {
+    scoped_ptr().swap(*this);
+  }
+  void swap(scoped_ptr &rhs) {
+    marisa::swap(ptr_, rhs.ptr_);
+  }
+
+ private:
+  T *ptr_;
+
+  // Disallows copy and assignment.
+  scoped_ptr(const scoped_ptr &);
+  scoped_ptr &operator=(const scoped_ptr &);
+};
+
+}  // namespace marisa
+
+#endif  // MARISA_SCOPED_PTR_H_

data/ext/marisa/lib/marisa/stdio.h

@@ -0,0 +1,15 @@
+#ifndef MARISA_STDIO_H_
+#define MARISA_STDIO_H_
+
+#include <cstdio>
+
+namespace marisa {
+
+class Trie;
+
+void fread(std::FILE *file, Trie *trie);
+void fwrite(std::FILE *file, const Trie &trie);
+
+}  // namespace marisa
+
+#endif  // MARISA_STDIO_H_

data/ext/marisa/lib/marisa/trie.cc

@@ -0,0 +1,249 @@
+#include "marisa/stdio.h"
+#include "marisa/iostream.h"
+#include "marisa/trie.h"
+#include "marisa/grimoire/trie.h"
+
+namespace marisa {
+
+Trie::Trie() : trie_() {}
+
+Trie::~Trie() {}
+
+void Trie::build(Keyset &keyset, int config_flags) {
+  scoped_ptr<grimoire::LoudsTrie> temp(new (std::nothrow) grimoire::LoudsTrie);
+  MARISA_THROW_IF(temp.get() == NULL, MARISA_MEMORY_ERROR);
+
+  temp->build(keyset, config_flags);
+  trie_.swap(temp);
+}
+
+void Trie::mmap(const char *filename) {
+  MARISA_THROW_IF(filename == NULL, MARISA_NULL_ERROR);
+
+  scoped_ptr<grimoire::LoudsTrie> temp(new (std::nothrow) grimoire::LoudsTrie);
+  MARISA_THROW_IF(temp.get() == NULL, MARISA_MEMORY_ERROR);
+
+  grimoire::Mapper mapper;
+  mapper.open(filename);
+  temp->map(mapper);
+  trie_.swap(temp);
+}
+
+void Trie::map(const void *ptr, std::size_t size) {
+  MARISA_THROW_IF((ptr == NULL) && (size != 0), MARISA_NULL_ERROR);
+
+  scoped_ptr<grimoire::LoudsTrie> temp(new (std::nothrow) grimoire::LoudsTrie);
+  MARISA_THROW_IF(temp.get() == NULL, MARISA_MEMORY_ERROR);
+
+  grimoire::Mapper mapper;
+  mapper.open(ptr, size);
+  temp->map(mapper);
+  trie_.swap(temp);
+}
+
+void Trie::load(const char *filename) {
+  MARISA_THROW_IF(filename == NULL, MARISA_NULL_ERROR);
+
+  scoped_ptr<grimoire::LoudsTrie> temp(new (std::nothrow) grimoire::LoudsTrie);
+  MARISA_THROW_IF(temp.get() == NULL, MARISA_MEMORY_ERROR);
+
+  grimoire::Reader reader;
+  reader.open(filename);
+  temp->read(reader);
+  trie_.swap(temp);
+}
+
+void Trie::read(int fd) {
+  MARISA_THROW_IF(fd == -1, MARISA_CODE_ERROR);
+
+  scoped_ptr<grimoire::LoudsTrie> temp(new (std::nothrow) grimoire::LoudsTrie);
+  MARISA_THROW_IF(temp.get() == NULL, MARISA_MEMORY_ERROR);
+
+  grimoire::Reader reader;
+  reader.open(fd);
+  temp->read(reader);
+  trie_.swap(temp);
+}
+
+void Trie::save(const char *filename) const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  MARISA_THROW_IF(filename == NULL, MARISA_NULL_ERROR);
+
+  grimoire::Writer writer;
+  writer.open(filename);
+  trie_->write(writer);
+}
+
+void Trie::write(int fd) const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  MARISA_THROW_IF(fd == -1, MARISA_CODE_ERROR);
+
+  grimoire::Writer writer;
+  writer.open(fd);
+  trie_->write(writer);
+}
+
+bool Trie::lookup(Agent &agent) const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  if (!agent.has_state()) {
+    agent.init_state();
+  }
+  return trie_->lookup(agent);
+}
+
+void Trie::reverse_lookup(Agent &agent) const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  if (!agent.has_state()) {
+    agent.init_state();
+  }
+  trie_->reverse_lookup(agent);
+}
+
+bool Trie::common_prefix_search(Agent &agent) const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  if (!agent.has_state()) {
+    agent.init_state();
+  }
+  return trie_->common_prefix_search(agent);
+}
+
+bool Trie::predictive_search(Agent &agent) const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  if (!agent.has_state()) {
+    agent.init_state();
+  }
+  return trie_->predictive_search(agent);
+}
+
+std::size_t Trie::num_tries() const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  return trie_->num_tries();
+}
+
+std::size_t Trie::num_keys() const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  return trie_->num_keys();
+}
+
+std::size_t Trie::num_nodes() const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  return trie_->num_nodes();
+}
+
+TailMode Trie::tail_mode() const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  return trie_->tail_mode();
+}
+
+NodeOrder Trie::node_order() const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  return trie_->node_order();
+}
+
+bool Trie::empty() const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  return trie_->empty();
+}
+
+std::size_t Trie::size() const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  return trie_->size();
+}
+
+std::size_t Trie::total_size() const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  return trie_->total_size();
+}
+
+std::size_t Trie::io_size() const {
+  MARISA_THROW_IF(trie_.get() == NULL, MARISA_STATE_ERROR);
+  return trie_->io_size();
+}
+
+void Trie::clear() {
+  Trie().swap(*this);
+}
+
+void Trie::swap(Trie &rhs) {
+  trie_.swap(rhs.trie_);
+}
+
+}  // namespace marisa
+
+#include <iostream>
+
+namespace marisa {
+
+class TrieIO {
+ public:
+  static void fread(std::FILE *file, Trie *trie) {
+    MARISA_THROW_IF(trie == NULL, MARISA_NULL_ERROR);
+
+    scoped_ptr<grimoire::LoudsTrie> temp(
+        new (std::nothrow) grimoire::LoudsTrie);
+    MARISA_THROW_IF(temp.get() == NULL, MARISA_MEMORY_ERROR);
+
+    grimoire::Reader reader;
+    reader.open(file);
+    temp->read(reader);
+    trie->trie_.swap(temp);
+  }
+  static void fwrite(std::FILE *file, const Trie &trie) {
+    MARISA_THROW_IF(file == NULL, MARISA_NULL_ERROR);
+    MARISA_THROW_IF(trie.trie_.get() == NULL, MARISA_STATE_ERROR);
+    grimoire::Writer writer;
+    writer.open(file);
+    trie.trie_->write(writer);
+  }
+
+  static std::istream &read(std::istream &stream, Trie *trie) {
+    MARISA_THROW_IF(trie == NULL, MARISA_NULL_ERROR);
+
+    scoped_ptr<grimoire::LoudsTrie> temp(
+        new (std::nothrow) grimoire::LoudsTrie);
+    MARISA_THROW_IF(temp.get() == NULL, MARISA_MEMORY_ERROR);
+
+    grimoire::Reader reader;
+    reader.open(stream);
+    temp->read(reader);
+    trie->trie_.swap(temp);
+    return stream;
+  }
+  static std::ostream &write(std::ostream &stream, const Trie &trie) {
+    MARISA_THROW_IF(trie.trie_.get() == NULL, MARISA_STATE_ERROR);
+    grimoire::Writer writer;
+    writer.open(stream);
+    trie.trie_->write(writer);
+    return stream;
+  }
+};
+
+void fread(std::FILE *file, Trie *trie) {
+  MARISA_THROW_IF(file == NULL, MARISA_NULL_ERROR);
+  MARISA_THROW_IF(trie == NULL, MARISA_NULL_ERROR);
+  TrieIO::fread(file, trie);
+}
+
+void fwrite(std::FILE *file, const Trie &trie) {
+  MARISA_THROW_IF(file == NULL, MARISA_NULL_ERROR);
+  TrieIO::fwrite(file, trie);
+}
+
+std::istream &read(std::istream &stream, Trie *trie) {
+  MARISA_THROW_IF(trie == NULL, MARISA_NULL_ERROR);
+  return TrieIO::read(stream, trie);
+}
+
+std::ostream &write(std::ostream &stream, const Trie &trie) {
+  return TrieIO::write(stream, trie);
+}
+
+std::istream &operator>>(std::istream &stream, Trie &trie) {
+  return read(stream, &trie);
+}
+
+std::ostream &operator<<(std::ostream &stream, const Trie &trie) {
+  return write(stream, trie);
+}
+
+}  // namespace marisa