smash::utf8 Namespace Reference

Functions
std::string	fill_left (const std::string &s, size_t width, char fill=' ')
	Fill string with characters to the left until the given width is reached. More...
std::string	fill_right (const std::string &s, size_t width, char fill=' ')
	Fill string with characters to the right until the given width is reached. More...
std::string	fill_both (const std::string &s, size_t width, char fill=' ')
	Fill string with characters at both sides until the given width is reached. More...
template<typename octet_type >
uint8_t	mask8 (octet_type oc)
	Extract the first byte from a given value. More...
template<typename octet_iterator >
std::iterator_traits< octet_iterator >::difference_type	sequence_length (octet_iterator lead_it)
	Given an iterator to the beginning of a UTF-8 sequence, return the length of the next UTF-8 code point. More...
static size_t	adjust (const std::string &s, size_t width)
	Adjust filling width by taking the size of unicode characters into account. More...

Function Documentation

fill_left()

std::string smash::utf8::fill_left	(	const std::string &	s,
		size_t	width,
		char	fill = ' '
	)

Fill string with characters to the left until the given width is reached.

Parameters

[in]	s	Input string.
[in]	width	Total width of output string.
[in]	fill	Filling character.

Returns

Padded string.

Definition at line 48 of file stringfunctions.cc.

                                                                 {
  width = adjust(s, width - s.size());
  if (width > 0) {
    return std::string(width, fill) + s;
  }
  return s;
}

fill_right()

std::string smash::utf8::fill_right	(	const std::string &	s,
		size_t	width,
		char	fill = ' '
	)

Fill string with characters to the right until the given width is reached.

Parameters

[in]	s	Input string.
[in]	width	Total width of output string.
[in]	fill	Filling character.

Returns

Padded string.

Definition at line 56 of file stringfunctions.cc.

                                                                  {
  width = adjust(s, width - s.size());
  if (width > 0) {
    return s + std::string(width, fill);
  }
  return s;
}

fill_both()

std::string smash::utf8::fill_both	(	const std::string &	s,
		size_t	width,
		char	fill = ' '
	)

Fill string with characters at both sides until the given width is reached.

Parameters

[in]	s	Input string.
[in]	width	Total width of output string.
[in]	fill	Filling character.

Returns

Padded string.

Definition at line 64 of file stringfunctions.cc.

                                                                 {
  width = adjust(s, width - s.size());
  if (width > 0) {
    const int l = width / 2;
    const int r = width - l;
    return std::string(l, fill) + s + std::string(r, fill);
  }
  return s;
}

mask8()

template<typename octet_type >

uint8_t smash::utf8::mask8 ( octet_type  oc )

inline

Extract the first byte from a given value.

This function was taken from the Boost-licensed library UTF8-CPP. See http://utfcpp.sourceforge.net/.

Template Parameters

octet_type

Type for one byte

Definition at line 109 of file stringfunctions.h.

                                    {
  return static_cast<uint8_t>(0xff & oc);
}

sequence_length()

template<typename octet_iterator >

std::iterator_traits<octet_iterator>::difference_type smash::utf8::sequence_length ( octet_iterator  lead_it )

inline

Given an iterator to the beginning of a UTF-8 sequence, return the length of the next UTF-8 code point.

This function was taken from the Boost-licensed library UTF8-CPP. See http://utfcpp.sourceforge.net/.

Definition at line 122 of file stringfunctions.h.

                                        {
  uint8_t lead = mask8(*lead_it);
  if (lead < 0x80)
    return 1;
  else if ((lead >> 5) == 0x6)
    return 2;
  else if ((lead >> 4) == 0xe)
    return 3;
  else if ((lead >> 3) == 0x1e)
    return 4;
  else
    return 0;
}

adjust()

static size_t smash::utf8::adjust ( const std::string &  s, size_t  width  )

inlinestatic

Adjust filling width by taking the size of unicode characters into account.

This is necessary, because UTF-8 characters can be represented by more than byte.

Parameters

[in]	s	String to be filled.
[in]	width	Width (in bytes) to be adjusted.

Returns

Adjusted width.

Definition at line 28 of file stringfunctions.cc.

                                                              {
  for (unsigned char c : s) {
    if (c >= 0xFC) {
      width += 5;
    } else if (c >= 0xF8) {
      width += 4;
    } else if (c >= 0xF0) {
      width += 3;
    } else if (c >= 0xE0) {
      width += 2;
    } else if (c == 0xCC || c == 0xCD) {
      // combining character (2 Bytes) - doesn't appear at all
      width += 2;
    } else if (c >= 0xC0) {
      width += 1;
    }
  }
  return width;
}