tesseract::CHAR_FRAGMENT Class Reference

#include <unicharset.h>

Public Member Functions
void	set_all (const char *unichar, int pos, int total, bool natural)
void	set_unichar (const char *uch)
void	set_pos (int p)
void	set_total (int t)
const char *	get_unichar () const
int	get_pos () const
int	get_total () const
std::string	to_string () const
bool	equals (const char *other_unichar, int other_pos, int other_total) const
bool	equals (const CHAR_FRAGMENT *other) const
bool	is_continuation_of (const CHAR_FRAGMENT *fragment) const
bool	is_beginning () const
bool	is_ending () const
bool	is_natural () const
void	set_natural (bool value)

Static Public Member Functions
static std::string	to_string (const char *unichar, int pos, int total, bool natural)
static CHAR_FRAGMENT *	parse_from_string (const char *str)

Static Public Attributes
static const int	kMinLen = 6
static const int	kMaxLen = 3 + UNICHAR_LEN + 2
static const int	kMaxChunks = 5

Detailed Description

Definition at line 50 of file unicharset.h.

Member Function Documentation

equals() [1/2]

bool tesseract::CHAR_FRAGMENT::equals ( const char *  other_unichar, int  other_pos, int  other_total  ) const

inline

Definition at line 97 of file unicharset.h.

                                            {
    return (strcmp(this->unichar, other_unichar) == 0 &&
            this->pos == other_pos && this->total == other_total);
  }

equals() [2/2]

bool tesseract::CHAR_FRAGMENT::equals ( const CHAR_FRAGMENT *  other ) const

inline

Definition at line 102 of file unicharset.h.

                                                       {
    return this->equals(other->get_unichar(), other->get_pos(),
                        other->get_total());
  }

get_pos()

int tesseract::CHAR_FRAGMENT::get_pos ( ) const

inline

Definition at line 79 of file unicharset.h.

                             {
    return this->pos;
  }

get_total()

int tesseract::CHAR_FRAGMENT::get_total ( ) const

inline

Definition at line 82 of file unicharset.h.

                               {
    return this->total;
  }

get_unichar()

const char * tesseract::CHAR_FRAGMENT::get_unichar ( ) const

inline

Definition at line 76 of file unicharset.h.

                                         {
    return this->unichar;
  }

is_beginning()

bool tesseract::CHAR_FRAGMENT::is_beginning ( ) const

inline

Definition at line 116 of file unicharset.h.

                                   {
    return this->pos == 0;
  }

is_continuation_of()

bool tesseract::CHAR_FRAGMENT::is_continuation_of ( const CHAR_FRAGMENT *  fragment ) const

inline

Definition at line 109 of file unicharset.h.

                                                                      {
    return (strcmp(this->unichar, fragment->get_unichar()) == 0 &&
            this->total == fragment->get_total() &&
            this->pos == fragment->get_pos() + 1);
  }

is_ending()

bool tesseract::CHAR_FRAGMENT::is_ending ( ) const

inline

Definition at line 121 of file unicharset.h.

                                {
    return this->pos == this->total - 1;
  }

is_natural()

bool tesseract::CHAR_FRAGMENT::is_natural ( ) const

inline

Definition at line 128 of file unicharset.h.

                                 {
    return natural;
  }

parse_from_string()

CHAR_FRAGMENT * tesseract::CHAR_FRAGMENT::parse_from_string ( const char *  str )

static

Definition at line 1103 of file unicharset.cpp.

                                                                  {
  const char *ptr = string;
  int len = strlen(string);
  if (len < kMinLen || *ptr != kSeparator) {
    return nullptr; // this string cannot represent a fragment
  }
  ptr++; // move to the next character
  int step = 0;
  while ((ptr + step) < (string + len) && *(ptr + step) != kSeparator) {
    step += UNICHAR::utf8_step(ptr + step);
  }
  if (step == 0 || step > UNICHAR_LEN) {
    return nullptr; // no character for unichar or the character is too long
  }
  char unichar[UNICHAR_LEN + 1];
  strncpy(unichar, ptr, step);
  unichar[step] = '\0'; // null terminate unichar
  ptr += step;          // move to the next fragment separator
  int pos = 0;
  int total = 0;
  bool natural = false;
  char *end_ptr = nullptr;
  for (int i = 0; i < 2; i++) {
    if (ptr > string + len || *ptr != kSeparator) {
      if (i == 1 && *ptr == kNaturalFlag) {
        natural = true;
      } else {
        return nullptr; // Failed to parse fragment representation.
      }
    }
    ptr++; // move to the next character
    i == 0 ? pos = static_cast<int>(strtol(ptr, &end_ptr, 10))
           : total = static_cast<int>(strtol(ptr, &end_ptr, 10));
    ptr = end_ptr;
  }
  if (ptr != string + len) {
    return nullptr; // malformed fragment representation
  }
  auto *fragment = new CHAR_FRAGMENT();
  fragment->set_all(unichar, pos, total, natural);
  return fragment;
}

set_all()

void tesseract::CHAR_FRAGMENT::set_all ( const char *  unichar, int  pos, int  total, bool  natural  )

inline

Definition at line 60 of file unicharset.h.

                                                                             {
    set_unichar(unichar);
    set_pos(pos);
    set_total(total);
    set_natural(natural);
  }

set_natural()

void tesseract::CHAR_FRAGMENT::set_natural ( bool  value )

inline

Definition at line 131 of file unicharset.h.

                               {
    natural = value;
  }

set_pos()

void tesseract::CHAR_FRAGMENT::set_pos ( int  p )

inline

Definition at line 70 of file unicharset.h.

                             {
    this->pos = p;
  }

set_total()

void tesseract::CHAR_FRAGMENT::set_total ( int  t )

inline

Definition at line 73 of file unicharset.h.

                               {
    this->total = t;
  }

set_unichar()

void tesseract::CHAR_FRAGMENT::set_unichar ( const char *  uch )

inline

Definition at line 66 of file unicharset.h.

                                           {
    strncpy(this->unichar, uch, sizeof(this->unichar));
    this->unichar[UNICHAR_LEN] = '\0';
  }

to_string() [1/2]

std::string tesseract::CHAR_FRAGMENT::to_string ( ) const

inline

Definition at line 91 of file unicharset.h.

                              {
    return to_string(unichar, pos, total, natural);
  }

to_string() [2/2]

std::string tesseract::CHAR_FRAGMENT::to_string ( const char *  unichar, int  pos, int  total, bool  natural  )

static

Definition at line 1088 of file unicharset.cpp.

                                                   {
  if (total == 1) {
    return std::string(unichar);
  }
  std::string result;
  result += kSeparator;
  result += unichar;
  char buffer[kMaxLen];
  snprintf(buffer, kMaxLen, "%c%d%c%d", kSeparator, pos,
           natural ? kNaturalFlag : kSeparator, total);
  result += buffer;
  return result;
}

Member Data Documentation

kMaxChunks

const int tesseract::CHAR_FRAGMENT::kMaxChunks = 5

static

Definition at line 57 of file unicharset.h.

kMaxLen

const int tesseract::CHAR_FRAGMENT::kMaxLen = 3 + UNICHAR_LEN + 2

static

Definition at line 55 of file unicharset.h.

kMinLen

const int tesseract::CHAR_FRAGMENT::kMinLen = 6

static

Definition at line 53 of file unicharset.h.

---

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

• /media/home/debian/src/github/tesseract-ocr/tesseract/src/ccutil/unicharset.h
• /media/home/debian/src/github/tesseract-ocr/tesseract/src/ccutil/unicharset.cpp