ext/scintilla/src/UniConversion.h

   1 // Scintilla source code edit control
   2 /** @file UniConversion.h
   3  ** Functions to handle UTF-8 and UTF-16 strings.
   4  **/
   5 // Copyright 1998-2001 by Neil Hodgson <neilh@scintilla.org>
   6 // The License.txt file describes the conditions under which this software may be distributed.
   7
   8 #ifndef UNICONVERSION_H
   9 #define UNICONVERSION_H
  10
  11 #ifdef SCI_NAMESPACE
  12 namespace Scintilla {
  13 #endif
  14
  15 const int UTF8MaxBytes = 4;
  16
  17 const int unicodeReplacementChar = 0xFFFD;
  18
  19 unsigned int UTF8Length(const wchar_t *uptr, unsigned int tlen);
  20 void UTF8FromUTF16(const wchar_t *uptr, unsigned int tlen, char *putf, unsigned int len);
  21 unsigned int UTF8CharLength(unsigned char ch);
  22 size_t UTF16Length(const char *s, size_t len);
  23 size_t UTF16FromUTF8(const char *s, size_t len, wchar_t *tbuf, size_t tlen);
  24 unsigned int UTF32FromUTF8(const char *s, unsigned int len, unsigned int *tbuf, unsigned int tlen);
  25 unsigned int UTF16FromUTF32Character(unsigned int val, wchar_t *tbuf);
  26
  27 extern int UTF8BytesOfLead[256];
  28 void UTF8BytesOfLeadInitialise();
  29
  30 inline bool UTF8IsTrailByte(int ch) {
  31         return (ch >= 0x80) && (ch < 0xc0);
  32 }
  33
  34 inline bool UTF8IsAscii(int ch) {
  35         return ch < 0x80;
  36 }
  37
  38 enum { UTF8MaskWidth=0x7, UTF8MaskInvalid=0x8 };
  39 int UTF8Classify(const unsigned char *us, int len);
  40
  41 // Similar to UTF8Classify but returns a length of 1 for invalid bytes
  42 // instead of setting the invalid flag
  43 int UTF8DrawBytes(const unsigned char *us, int len);
  44
  45 // Line separator is U+2028 \xe2\x80\xa8
  46 // Paragraph separator is U+2029 \xe2\x80\xa9
  47 const int UTF8SeparatorLength = 3;
  48 inline bool UTF8IsSeparator(const unsigned char *us) {
  49         return (us[0] == 0xe2) && (us[1] == 0x80) && ((us[2] == 0xa8) || (us[2] == 0xa9));
  50 }
  51
  52 // NEL is U+0085 \xc2\x85
  53 const int UTF8NELLength = 2;
  54 inline bool UTF8IsNEL(const unsigned char *us) {
  55         return (us[0] == 0xc2) && (us[1] == 0x85);
  56 }
  57
  58 #ifdef SCI_NAMESPACE
  59 }
  60 #endif
  61
  62 #endif