1
1
mirror of https://github.com/mawww/kakoune.git synced 2024-12-11 12:44:59 +03:00
kakoune/src/unicode.hh

86 lines
1.7 KiB
C++
Raw Normal View History

#ifndef unicode_hh_INCLUDED
#define unicode_hh_INCLUDED
#include <wctype.h>
#include <wchar.h>
#include <locale>
namespace Kakoune
{
2015-04-29 15:51:15 +03:00
using Codepoint = char32_t;
inline bool is_eol(Codepoint c)
{
return c == '\n';
}
2013-11-18 02:54:26 +04:00
inline bool is_horizontal_blank(Codepoint c)
{
return c == ' ' or c == '\t';
}
inline bool is_blank(Codepoint c)
{
return c == ' ' or c == '\t' or c == '\n';
}
2013-12-14 18:49:10 +04:00
enum WordType { Word, WORD };
template<WordType word_type = Word>
inline bool is_word(Codepoint c)
{
return c == '_' or iswalnum((wchar_t)c);
2013-12-14 18:49:10 +04:00
}
template<>
inline bool is_word<WORD>(Codepoint c)
{
return not is_horizontal_blank(c) and not is_eol(c);
2013-12-14 18:49:10 +04:00
}
inline bool is_punctuation(Codepoint c)
{
return not (is_word(c) or is_horizontal_blank(c) or is_eol(c));
2013-12-14 18:49:10 +04:00
}
2015-11-15 16:24:39 +03:00
inline bool is_basic_alpha(Codepoint c)
{
return (c >= 'a' and c <= 'z') or (c >= 'A' and c <= 'Z');
}
2016-09-25 12:55:57 +03:00
inline size_t codepoint_width(Codepoint c)
{
return wcwidth((wchar_t)c);
}
2013-12-14 18:49:10 +04:00
enum class CharCategories
{
Blank,
EndOfLine,
Word,
Punctuation,
};
template<WordType word_type = Word>
inline CharCategories categorize(Codepoint c)
{
if (is_eol(c))
return CharCategories::EndOfLine;
if (is_horizontal_blank(c))
2013-12-14 18:49:10 +04:00
return CharCategories::Blank;
if (word_type == WORD or is_word(c))
return CharCategories::Word;
return CharCategories::Punctuation;
2013-12-14 18:49:10 +04:00
}
inline Codepoint to_lower(Codepoint cp) { return towlower((wchar_t)cp); }
inline Codepoint to_upper(Codepoint cp) { return towupper((wchar_t)cp); }
inline char to_lower(char c) { return c >= 'A' and c <= 'Z' ? c - 'A' + 'a' : c; }
inline char to_upper(char c) { return c >= 'a' and c <= 'z' ? c - 'a' + 'A' : c; }
}
#endif // unicode_hh_INCLUDED