UNICHARMAP Class Reference

#include <unicharmap.h>

Public Member Functions
	UNICHARMAP ()
	~UNICHARMAP ()
void	insert (const char *const unichar_repr, UNICHAR_ID id)
UNICHAR_ID	unichar_to_id (const char *const unichar_repr, int length) const
bool	contains (const char *const unichar_repr, int length) const
int	minmatch (const char *const unichar_repr) const
void	clear ()

Detailed Description

Definition at line 27 of file unicharmap.h.

Constructor & Destructor Documentation

UNICHARMAP()

UNICHARMAP::UNICHARMAP

(

)

Definition at line 23 of file unicharmap.cpp.

                       :
nodes(nullptr) {
}

~UNICHARMAP()

UNICHARMAP::~UNICHARMAP

(

)

Definition at line 27 of file unicharmap.cpp.

                        {
  delete[] nodes;
}

Member Function Documentation

clear()

void UNICHARMAP::clear

(

)

Definition at line 115 of file unicharmap.cpp.

                       {
  delete[] nodes;
  nodes = nullptr;
}

contains()

bool UNICHARMAP::contains	(	const char *const	unichar_repr,
		int	length
	)		const

Definition at line 79 of file unicharmap.cpp.

                                            {
  if (unichar_repr == nullptr || *unichar_repr == '\0') return false;
  if (length <= 0 || length > UNICHAR_LEN) return false;
  int index = 0;
  if (unichar_repr[index] == '\0') return false;
  UNICHARMAP_NODE* current_nodes = nodes;
 
  while (current_nodes != nullptr && index + 1 < length &&
         unichar_repr[index + 1] != '\0') {
    current_nodes =
        current_nodes[static_cast<unsigned char>(unichar_repr[index])].children;
    ++index;
  }
  return current_nodes != nullptr &&
         (index + 1 >= length || unichar_repr[index + 1] == '\0') &&
         current_nodes[static_cast<unsigned char>(unichar_repr[index])].id >= 0;
}

insert()

void UNICHARMAP::insert	(	const char *const	unichar_repr,
		UNICHAR_ID	id
	)

Definition at line 56 of file unicharmap.cpp.

                                                                     {
  const char* current_char = unichar_repr;
  if (*current_char == '\0') return;
  UNICHARMAP_NODE** current_nodes_pointer = &nodes;
  do {
    if (*current_nodes_pointer == nullptr)
      *current_nodes_pointer = new UNICHARMAP_NODE[256];
    if (current_char[1] == '\0') {
      (*current_nodes_pointer)
          [static_cast<unsigned char>(*current_char)].id = id;
      return;
    }
    current_nodes_pointer =
        &((*current_nodes_pointer)
          [static_cast<unsigned char>(*current_char)].children);
    ++current_char;
  } while (true);
}

minmatch()

int UNICHARMAP::minmatch

(

const char *const

unichar_repr

)

const

Definition at line 100 of file unicharmap.cpp.

                                                             {
  const char* current_char = unichar_repr;
  if (*current_char == '\0') return 0;
  UNICHARMAP_NODE* current_nodes = nodes;
 
  while (current_nodes != nullptr && *current_char != '\0') {
    if (current_nodes[static_cast<unsigned char>(*current_char)].id >= 0)
      return current_char + 1 - unichar_repr;
    current_nodes =
        current_nodes[static_cast<unsigned char>(*current_char)].children;
    ++current_char;
  }
  return 0;
}

unichar_to_id()

UNICHAR_ID UNICHARMAP::unichar_to_id	(	const char *const	unichar_repr,
		int	length
	)		const

Definition at line 34 of file unicharmap.cpp.

                                                       {
  UNICHARMAP_NODE* current_nodes = nodes;
 
  assert(*unichar_repr != '\0');
  assert(length > 0 && length <= UNICHAR_LEN);
 
  int index = 0;
  if (length <= 0 || unichar_repr[index] == '\0') return INVALID_UNICHAR_ID;
  do {
    if (index + 1 >= length || unichar_repr[index + 1] == '\0')
      return current_nodes[static_cast<unsigned char>(unichar_repr[index])].id;
    current_nodes =
        current_nodes[static_cast<unsigned char>(unichar_repr[index])].children;
    ++index;
  } while (true);
}

---

The documentation for this class was generated from the following files:

• /tesseract/src/ccutil/unicharmap.h
• /tesseract/src/ccutil/unicharmap.cpp