src/unicode.h

   1 /* unicode.h
   2  *  Copyright (C) 2001-2006, The Perl Foundation.
   3  *  SVN Info
   4  *     $Id$
   5  *  Overview:
   6  *     Unicode support header
   7  *  Data Structure and Algorithms:
   8  *  History:
   9  *  Notes:
  10  *  References:
  11  */
  12
  13 #if !defined(PARROT_UNICODE_H_GUARD)
  14 #define PARROT_UNICODE_H_GUARD
  15
  16 typedef unsigned char utf8_t;
  17 typedef unsigned short utf16_t;
  18 typedef unsigned long utf32_t;
  19
  20 #define UNICODE_SURROGATE_FIRST         0xD800u
  21 #define UNICODE_SURROGATE_LAST          0xDFFFu
  22 #define UNICODE_HIGH_SURROGATE_FIRST    0xD800u
  23 #define UNICODE_HIGH_SURROGATE_LAST     0xDBFFu
  24 #define UNICODE_HIGH_SURROGATE_SHIFT    10
  25 #define UNICODE_LOW_SURROGATE_FIRST     0xDC00u
  26 #define UNICODE_LOW_SURROGATE_LAST      0xDFFFu
  27 #define UNICODE_LOW_SURROGATE_MASK      0x3FFu
  28
  29 #define UNICODE_IS_SURROGATE(c)        ((c) >= UNICODE_SURROGATE_FIRST && \
  30                                         (c) <= UNICODE_SURROGATE_LAST)
  31 #define UNICODE_IS_HIGH_SURROGATE(c)   ((c) >= UNICODE_HIGH_SURROGATE_FIRST && \
  32                                         (c) <= UNICODE_HIGH_SURROGATE_LAST)
  33 #define UNICODE_IS_LOW_SURROGATE(c)    ((c) >= UNICODE_LOW_SURROGATE_FIRST && \
  34                                         (c) <= UNICODE_LOW_SURROGATE_LAST)
  35 #define UNICODE_IS_INVARIANT(c)        ((c) <  0x80u)
  36
  37 #define UNICODE_HIGH_SURROGATE(c) \
  38   ((((c) - 0x10000u) >> UNICODE_HIGH_SURROGATE_SHIFT) + UNICODE_HIGH_SURROGATE_FIRST)
  39 #define UNICODE_LOW_SURROGATE(c) \
  40   ((((c) - 0x10000u) & UNICODE_LOW_SURROGATE_MASK) + UNICODE_LOW_SURROGATE_FIRST)
  41 #define UNICODE_DECODE_SURROGATE(high,low) \
  42   ((((high) - UNICODE_HIGH_SURROGATE_FIRST) << UNICODE_HIGH_SURROGATE_SHIFT) + \
  43     ((low) - UNICODE_LOW_SURROGATE_FIRST) + 0x10000u)
  44
  45 #define UNISKIP(uv) ( (uv) < 0x80    ? 1 : \
  46                       (uv) < 0x800   ? 2 : \
  47                       (uv) < 0x10000 ? 3 : 4 )
  48
  49 #define UTF16SKIP(s) ( UNICODE_IS_HIGH_SURROGATE(*(s)) ? 2 : 1 )
  50
  51 /*
  52
  53  The following table is from Unicode 3.1.
  54
  55  Code Points            1st Byte  2nd Byte  3rd Byte  4th Byte
  56
  57    U+0000..U+007F       00..7F
  58    U+0080..U+07FF       C2..DF    80..BF
  59    U+0800..U+0FFF       E0        A0..BF    80..BF
  60    U+1000..U+FFFF       E1..EF    80..BF    80..BF
  61   U+10000..U+3FFFF      F0        90..BF    80..BF    80..BF
  62   U+40000..U+FFFFF      F1..F3    80..BF    80..BF    80..BF
  63  U+100000..U+10FFFF     F4        80..8F    80..BF    80..BF
  64
  65  */
  66
  67 #define UTF8_IS_START(c)                ((c) >= 0xC0u && (c) <= 0xFDu)
  68 #define UTF8_IS_CONTINUATION(c)         ((c) >= 0x80u && (c) <= 0xBFu)
  69 #define UTF8_IS_CONTINUED(c)            ((c) &  0x80u)
  70
  71 #define UTF8_START_MARK(len) (len == 1 ? 0 : 0x7Eu << (7-len))
  72 #define UTF8_START_MASK(len) (len == 1 ? 0x7Fu : 0x1Fu >> (len-2))
  73
  74 #define UTF8_CONTINUATION_MARK           0x80u
  75 #define UTF8_ACCUMULATION_SHIFT          6
  76 #define UTF8_CONTINUATION_MASK           0x3Fu
  77 #define UTF8_ACCUMULATE(old, new)       (((old) << UTF8_ACCUMULATION_SHIFT) | ((new) & UTF8_CONTINUATION_MASK))
  78
  79 extern const char Parrot_utf8skip[256];
  80
  81 #define UTF8SKIP(s) Parrot_utf8skip[*(s)]
  82
  83 #define UTF8_MAXLEN 4
  84 #define UTF16_MAXLEN 4
  85
  86 #endif /* PARROT_UNICODE_H_GUARD */
  87
  88
  89 /*
  90  * Local variables:
  91  *   c-file-style: "parrot"
  92  * End:
  93  * vim: expandtab shiftwidth=4:
  94  */