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 std::string FixInvalidUTF8(const std::string &text);
  27
  28 extern int UTF8BytesOfLead[256];
  29 void UTF8BytesOfLeadInitialise();
  30
  31 inline bool UTF8IsTrailByte(int ch) {
  32         return (ch >= 0x80) && (ch < 0xc0);
  33 }
  34
  35 inline bool UTF8IsAscii(int ch) {
  36         return ch < 0x80;
  37 }
  38
  39 enum { UTF8MaskWidth=0x7, UTF8MaskInvalid=0x8 };
  40 int UTF8Classify(const unsigned char *us, int len);
  41
  42 // Similar to UTF8Classify but returns a length of 1 for invalid bytes
  43 // instead of setting the invalid flag
  44 int UTF8DrawBytes(const unsigned char *us, int len);
  45
  46 // Line separator is U+2028 \xe2\x80\xa8
  47 // Paragraph separator is U+2029 \xe2\x80\xa9
  48 const int UTF8SeparatorLength = 3;
  49 inline bool UTF8IsSeparator(const unsigned char *us) {
  50         return (us[0] == 0xe2) && (us[1] == 0x80) && ((us[2] == 0xa8) || (us[2] == 0xa9));
  51 }
  52
  53 // NEL is U+0085 \xc2\x85
  54 const int UTF8NELLength = 2;
  55 inline bool UTF8IsNEL(const unsigned char *us) {
  56         return (us[0] == 0xc2) && (us[1] == 0x85);
  57 }
  58
  59 enum { SURROGATE_LEAD_FIRST = 0xD800 };
  60 enum { SURROGATE_LEAD_LAST = 0xDBFF };
  61 enum { SURROGATE_TRAIL_FIRST = 0xDC00 };
  62 enum { SURROGATE_TRAIL_LAST = 0xDFFF };
  63 enum { SUPPLEMENTAL_PLANE_FIRST = 0x10000 };
  64
  65 inline unsigned int UTF16CharLength(wchar_t uch) {
  66         return ((uch >= SURROGATE_LEAD_FIRST) && (uch <= SURROGATE_LEAD_LAST)) ? 2 : 1;
  67 }
  68
  69 #ifdef SCI_NAMESPACE
  70 }
  71 #endif
  72
  73 #endif