tesseract::UNICHAR::const_iterator Class Reference

#include <unichar.h>

Public Member Functions
const_iterator &	operator++ ()
int	operator* () const
int	get_utf8 (char *buf) const
int	utf8_len () const
bool	is_legal () const
const char *	utf8_data () const

Friends
class	UNICHAR
bool	operator== (const CI &lhs, const CI &rhs)
bool	operator!= (const CI &lhs, const CI &rhs)

Detailed Description

Definition at line 105 of file unichar.h.

Member Function Documentation

get_utf8()

int tesseract::UNICHAR::const_iterator::get_utf8

(

char *

buf

)

const

Definition at line 183 of file unichar.cpp.

                                                           {
  ASSERT_HOST(it_ != nullptr);
  const int len = utf8_step(it_);
  if (len == 0) {
    tprintf("WARNING: Illegal UTF8 encountered\n");
    utf8_output[0] = ' ';
    return 1;
  }
  strncpy(utf8_output, it_, len);
  return len;
}

is_legal()

bool tesseract::UNICHAR::const_iterator::is_legal

(

)

const

Definition at line 205 of file unichar.cpp.

                                           {
  return utf8_step(it_) > 0;
}

operator*()

int tesseract::UNICHAR::const_iterator::operator*

(

)

const

Definition at line 172 of file unichar.cpp.

                                           {
  ASSERT_HOST(it_ != nullptr);
  const int len = utf8_step(it_);
  if (len == 0) {
    tprintf("WARNING: Illegal UTF8 encountered\n");
    return ' ';
  }
  UNICHAR uch(it_, len);
  return uch.first_uni();
}

operator++()

UNICHAR::const_iterator & tesseract::UNICHAR::const_iterator::operator++

(

)

Definition at line 158 of file unichar.cpp.

                                                         {
  ASSERT_HOST(it_ != nullptr);
  int step = utf8_step(it_);
  if (step == 0) {
    tprintf("ERROR: Illegal UTF8 encountered.\n");
    for (int i = 0; i < 5 && it_[i] != '\0'; ++i) {
      tprintf("Index %d char = 0x%x\n", i, it_[i]);
    }
    step = 1;
  }
  it_ += step;
  return *this;
}

utf8_data()

const char * tesseract::UNICHAR::const_iterator::utf8_data ( ) const

inline

Definition at line 133 of file unichar.h.

                                  {
      return it_;
    }

utf8_len()

int tesseract::UNICHAR::const_iterator::utf8_len

(

)

const

Definition at line 195 of file unichar.cpp.

                                          {
  ASSERT_HOST(it_ != nullptr);
  const int len = utf8_step(it_);
  if (len == 0) {
    tprintf("WARNING: Illegal UTF8 encountered\n");
    return 1;
  }
  return len;
}

Friends And Related Function Documentation

operator!=

bool operator!= ( const CI &  lhs, const CI &  rhs  )

friend

Definition at line 141 of file unichar.h.

                                                         {
      return !(lhs == rhs);
    }

operator==

bool operator== ( const CI &  lhs, const CI &  rhs  )

friend

Definition at line 138 of file unichar.h.

                                                         {
      return lhs.it_ == rhs.it_;
    }

UNICHAR

friend class UNICHAR

friend

Definition at line 146 of file unichar.h.

---

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

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