chiark - git - mdw - disorder/blob - libtests/t-utf8.c

   1 /*
   2  * This file is part of DisOrder.
   3  * Copyright (C) 2005, 2007, 2008 Richard Kettlewell
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful, but
  11  * WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  * General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program; if not, write to the Free Software
  17  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
  18  * USA
  19  */
  20 #include "test.h"
  21
  22 static void test_utf8(void) {
  23   /* Test validutf8, convert to UCS-4, check the answer is right,
  24    * convert back to UTF-8, check we got to where we started */
  25 #define U8(CHARS, WORDS) do {                   \
  26   uint32_t *w = ucs4parse(WORDS);               \
  27   uint32_t *ucs;                                \
  28   char *u8;                                     \
  29                                                 \
  30   insist(validutf8(CHARS));                     \
  31   ucs = utf8_to_utf32(CHARS, strlen(CHARS), 0); \
  32   insist(ucs != 0);                             \
  33   insist(!utf32_cmp(w, ucs));                   \
  34   u8 = utf32_to_utf8(ucs, utf32_len(ucs), 0);   \
  35   insist(u8 != 0);                              \
  36   check_string(u8, CHARS);                      \
  37 } while(0)
  38
  39 #define validutf8(S) utf8_valid((S), strlen(S))
  40
  41   /* empty string */
  42
  43   U8("", "");
  44
  45   /* ASCII characters */
  46
  47   U8(" !\"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~",
  48      "0x20 0x21 0x22 0x23 0x24 0x25 0x26 0x27 0x28 0x29 0x2a 0x2b 0x2c 0x2d "
  49      "0x2e 0x2f 0x30 0x31 0x32 0x33 0x34 0x35 0x36 0x37 0x38 0x39 0x3a "
  50      "0x3b 0x3c 0x3d 0x3e 0x3f 0x40 0x41 0x42 0x43 0x44 0x45 0x46 0x47 "
  51      "0x48 0x49 0x4a 0x4b 0x4c 0x4d 0x4e 0x4f 0x50 0x51 0x52 0x53 0x54 "
  52      "0x55 0x56 0x57 0x58 0x59 0x5a 0x5b 0x5c 0x5d 0x5e 0x5f 0x60 0x61 "
  53      "0x62 0x63 0x64 0x65 0x66 0x67 0x68 0x69 0x6a 0x6b 0x6c 0x6d 0x6e "
  54      "0x6f 0x70 0x71 0x72 0x73 0x74 0x75 0x76 0x77 0x78 0x79 0x7a 0x7b "
  55      "0x7c 0x7d 0x7e");
  56   U8("\001\002\003\004\005\006\007\010\011\012\013\014\015\016\017\020\021\022\023\024\025\026\027\030\031\032\033\034\035\036\037\177",
  57      "0x1 0x2 0x3 0x4 0x5 0x6 0x7 0x8 0x9 0xa 0xb 0xc 0xd 0xe 0xf 0x10 "
  58      "0x11 0x12 0x13 0x14 0x15 0x16 0x17 0x18 0x19 0x1a 0x1b 0x1c 0x1d "
  59      "0x1e 0x1f 0x7f");
  60
  61   /* from RFC3629 */
  62
  63   /* UTF8-2      = %xC2-DF UTF8-tail */
  64   insist(!validutf8("\xC0\x80"));
  65   insist(!validutf8("\xC1\x80"));
  66   insist(!validutf8("\xC2\x7F"));
  67   U8("\xC2\x80", "0x80");
  68   U8("\xDF\xBF", "0x7FF");
  69   insist(!validutf8("\xDF\xC0"));
  70
  71   /*  UTF8-3      = %xE0 %xA0-BF UTF8-tail / %xE1-EC 2( UTF8-tail ) /
  72    *                %xED %x80-9F UTF8-tail / %xEE-EF 2( UTF8-tail )
  73    */
  74   insist(!validutf8("\xE0\x9F\x80"));
  75   U8("\xE0\xA0\x80", "0x800");
  76   U8("\xE0\xBF\xBF", "0xFFF");
  77   insist(!validutf8("\xE0\xC0\xBF"));
  78
  79   insist(!validutf8("\xE1\x80\x7F"));
  80   U8("\xE1\x80\x80", "0x1000");
  81   U8("\xEC\xBF\xBF", "0xCFFF");
  82   insist(!validutf8("\xEC\xC0\xBF"));
  83
  84   U8("\xED\x80\x80", "0xD000");
  85   U8("\xED\x9F\xBF", "0xD7FF");
  86   insist(!validutf8("\xED\xA0\xBF"));
  87
  88   insist(!validutf8("\xEE\x7f\x80"));
  89   U8("\xEE\x80\x80", "0xE000");
  90   U8("\xEF\xBF\xBF", "0xFFFF");
  91   insist(!validutf8("\xEF\xC0\xBF"));
  92
  93   /*  UTF8-4      = %xF0 %x90-BF 2( UTF8-tail ) / %xF1-F3 3( UTF8-tail ) /
  94    *                %xF4 %x80-8F 2( UTF8-tail )
  95    */
  96   insist(!validutf8("\xF0\x8F\x80\x80"));
  97   U8("\xF0\x90\x80\x80", "0x10000");
  98   U8("\xF0\xBF\xBF\xBF", "0x3FFFF");
  99   insist(!validutf8("\xF0\xC0\x80\x80"));
 100
 101   insist(!validutf8("\xF1\x80\x80\x7F"));
 102   U8("\xF1\x80\x80\x80", "0x40000");
 103   U8("\xF3\xBF\xBF\xBF", "0xFFFFF");
 104   insist(!validutf8("\xF3\xC0\x80\x80"));
 105
 106   insist(!validutf8("\xF4\x80\x80\x7F"));
 107   U8("\xF4\x80\x80\x80", "0x100000");
 108   U8("\xF4\x8F\xBF\xBF", "0x10FFFF");
 109   insist(!validutf8("\xF4\x90\x80\x80"));
 110   insist(!validutf8("\xF4\x80\xFF\x80"));
 111
 112   /* miscellaneous non-UTF-8 rubbish */
 113   insist(!validutf8("\x80"));
 114   insist(!validutf8("\xBF"));
 115   insist(!validutf8("\xC0"));
 116   insist(!validutf8("\xC0\x7F"));
 117   insist(!validutf8("\xC0\xC0"));
 118   insist(!validutf8("\xE0"));
 119   insist(!validutf8("\xE0\x7F"));
 120   insist(!validutf8("\xE0\xC0"));
 121   insist(!validutf8("\xE0\x80"));
 122   insist(!validutf8("\xE0\x80\x7f"));
 123   insist(!validutf8("\xE0\x80\xC0"));
 124   insist(!validutf8("\xF0"));
 125   insist(!validutf8("\xF0\x7F"));
 126   insist(!validutf8("\xF0\xC0"));
 127   insist(!validutf8("\xF0\x80"));
 128   insist(!validutf8("\xF0\x80\x7f"));
 129   insist(!validutf8("\xF0\x80\xC0"));
 130   insist(!validutf8("\xF0\x80\x80\x7f"));
 131   insist(!validutf8("\xF0\x80\x80\xC0"));
 132   insist(!validutf8("\xF5\x80\x80\x80"));
 133   insist(!validutf8("\xF8"));
 134 }
 135
 136 TEST(utf8);
 137
 138 /*
 139 Local Variables:
 140 c-basic-offset:2
 141 comment-column:40
 142 fill-column:79
 143 indent-tabs-mode:nil
 144 End:
 145 */