Home | History | Annotate | Line # | Download | only in rt
      1      1.1  mrg /**
      2      1.1  mrg  * This code handles decoding UTF strings for foreach_reverse loops.  There are
      3      1.1  mrg  * 6 combinations of conversions between char, wchar, and dchar, and 2 of each
      4      1.1  mrg  * of those.
      5      1.1  mrg  *
      6      1.1  mrg  * Copyright: Copyright Digital Mars 2004 - 2010.
      7  1.1.1.2  mrg  * License:   $(HTTP www.boost.org/LICENSE_1_0.txt, Boost License 1.0).
      8      1.1  mrg  * Authors:   Walter Bright, Sean Kelly
      9  1.1.1.2  mrg  * Source: $(DRUNTIMESRC rt/_aApplyR.d)
     10      1.1  mrg  */
     11      1.1  mrg module rt.aApplyR;
     12      1.1  mrg 
     13  1.1.1.2  mrg import core.internal.utf;
     14      1.1  mrg 
     15      1.1  mrg /**********************************************/
     16      1.1  mrg /* 1 argument versions */
     17      1.1  mrg 
     18      1.1  mrg // dg is D, but _aApplyRcd() is C
     19      1.1  mrg extern (D) alias int delegate(void *) dg_t;
     20      1.1  mrg 
     21      1.1  mrg extern (C) int _aApplyRcd1(in char[] aa, dg_t dg)
     22      1.1  mrg {   int result;
     23      1.1  mrg 
     24      1.1  mrg     debug(apply) printf("_aApplyRcd1(), len = %d\n", aa.length);
     25      1.1  mrg     for (size_t i = aa.length; i != 0; )
     26      1.1  mrg     {   dchar d;
     27      1.1  mrg 
     28      1.1  mrg         i--;
     29      1.1  mrg         d = aa[i];
     30      1.1  mrg         if (d & 0x80)
     31      1.1  mrg         {   char c = cast(char)d;
     32      1.1  mrg             uint j;
     33      1.1  mrg             uint m = 0x3F;
     34      1.1  mrg             d = 0;
     35      1.1  mrg             while ((c & 0xC0) != 0xC0)
     36      1.1  mrg             {   if (i == 0)
     37      1.1  mrg                     onUnicodeError("Invalid UTF-8 sequence", 0);
     38      1.1  mrg                 i--;
     39      1.1  mrg                 d |= (c & 0x3F) << j;
     40      1.1  mrg                 j += 6;
     41      1.1  mrg                 m >>= 1;
     42      1.1  mrg                 c = aa[i];
     43      1.1  mrg             }
     44      1.1  mrg             d |= (c & m) << j;
     45      1.1  mrg         }
     46      1.1  mrg         result = dg(cast(void *)&d);
     47      1.1  mrg         if (result)
     48      1.1  mrg             break;
     49      1.1  mrg     }
     50      1.1  mrg     return result;
     51      1.1  mrg }
     52      1.1  mrg 
     53      1.1  mrg unittest
     54      1.1  mrg {
     55      1.1  mrg     debug(apply) printf("_aApplyRcd1.unittest\n");
     56      1.1  mrg 
     57      1.1  mrg     auto s = "hello"c[];
     58      1.1  mrg     int i;
     59      1.1  mrg 
     60      1.1  mrg     foreach_reverse (dchar d; s)
     61      1.1  mrg     {
     62      1.1  mrg         switch (i)
     63      1.1  mrg         {
     64      1.1  mrg             case 0:     assert(d == 'o'); break;
     65      1.1  mrg             case 1:     assert(d == 'l'); break;
     66      1.1  mrg             case 2:     assert(d == 'l'); break;
     67      1.1  mrg             case 3:     assert(d == 'e'); break;
     68      1.1  mrg             case 4:     assert(d == 'h'); break;
     69      1.1  mrg             default:    assert(0);
     70      1.1  mrg         }
     71      1.1  mrg         i++;
     72      1.1  mrg     }
     73      1.1  mrg     assert(i == 5);
     74      1.1  mrg 
     75      1.1  mrg     s = "a\u1234\U000A0456b";
     76      1.1  mrg     i = 0;
     77      1.1  mrg     foreach_reverse (dchar d; s)
     78      1.1  mrg     {
     79      1.1  mrg         //printf("i = %d, d = %x\n", i, d);
     80      1.1  mrg         switch (i)
     81      1.1  mrg         {
     82      1.1  mrg             case 0:     assert(d == 'b'); break;
     83      1.1  mrg             case 1:     assert(d == '\U000A0456'); break;
     84      1.1  mrg             case 2:     assert(d == '\u1234'); break;
     85      1.1  mrg             case 3:     assert(d == 'a'); break;
     86      1.1  mrg             default:    assert(0);
     87      1.1  mrg         }
     88      1.1  mrg         i++;
     89      1.1  mrg     }
     90      1.1  mrg     assert(i == 4);
     91      1.1  mrg }
     92      1.1  mrg 
     93      1.1  mrg /*****************************/
     94      1.1  mrg 
     95      1.1  mrg extern (C) int _aApplyRwd1(in wchar[] aa, dg_t dg)
     96      1.1  mrg {   int result;
     97      1.1  mrg 
     98      1.1  mrg     debug(apply) printf("_aApplyRwd1(), len = %d\n", aa.length);
     99      1.1  mrg     for (size_t i = aa.length; i != 0; )
    100      1.1  mrg     {   dchar d;
    101      1.1  mrg 
    102      1.1  mrg         i--;
    103      1.1  mrg         d = aa[i];
    104      1.1  mrg         if (d >= 0xDC00 && d <= 0xDFFF)
    105      1.1  mrg         {   if (i == 0)
    106      1.1  mrg                 onUnicodeError("Invalid UTF-16 sequence", 0);
    107      1.1  mrg             i--;
    108      1.1  mrg             d = ((aa[i] - 0xD7C0) << 10) + (d - 0xDC00);
    109      1.1  mrg         }
    110      1.1  mrg         result = dg(cast(void *)&d);
    111      1.1  mrg         if (result)
    112      1.1  mrg             break;
    113      1.1  mrg     }
    114      1.1  mrg     return result;
    115      1.1  mrg }
    116      1.1  mrg 
    117      1.1  mrg unittest
    118      1.1  mrg {
    119      1.1  mrg     debug(apply) printf("_aApplyRwd1.unittest\n");
    120      1.1  mrg 
    121      1.1  mrg     auto s = "hello"w[];
    122      1.1  mrg     int i;
    123      1.1  mrg 
    124      1.1  mrg     foreach_reverse (dchar d; s)
    125      1.1  mrg     {
    126      1.1  mrg         switch (i)
    127      1.1  mrg         {
    128      1.1  mrg             case 0:     assert(d == 'o'); break;
    129      1.1  mrg             case 1:     assert(d == 'l'); break;
    130      1.1  mrg             case 2:     assert(d == 'l'); break;
    131      1.1  mrg             case 3:     assert(d == 'e'); break;
    132      1.1  mrg             case 4:     assert(d == 'h'); break;
    133      1.1  mrg             default:    assert(0);
    134      1.1  mrg         }
    135      1.1  mrg         i++;
    136      1.1  mrg     }
    137      1.1  mrg     assert(i == 5);
    138      1.1  mrg 
    139      1.1  mrg     s = "a\u1234\U000A0456b";
    140      1.1  mrg     i = 0;
    141      1.1  mrg     foreach_reverse (dchar d; s)
    142      1.1  mrg     {
    143      1.1  mrg         //printf("i = %d, d = %x\n", i, d);
    144      1.1  mrg         switch (i)
    145      1.1  mrg         {
    146      1.1  mrg             case 0:     assert(d == 'b'); break;
    147      1.1  mrg             case 1:     assert(d == '\U000A0456'); break;
    148      1.1  mrg             case 2:     assert(d == '\u1234'); break;
    149      1.1  mrg             case 3:     assert(d == 'a'); break;
    150      1.1  mrg             default:    assert(0);
    151      1.1  mrg         }
    152      1.1  mrg         i++;
    153      1.1  mrg     }
    154      1.1  mrg     assert(i == 4);
    155      1.1  mrg }
    156      1.1  mrg 
    157      1.1  mrg /*****************************/
    158      1.1  mrg 
    159      1.1  mrg extern (C) int _aApplyRcw1(in char[] aa, dg_t dg)
    160      1.1  mrg {   int result;
    161      1.1  mrg 
    162      1.1  mrg     debug(apply) printf("_aApplyRcw1(), len = %d\n", aa.length);
    163      1.1  mrg     for (size_t i = aa.length; i != 0; )
    164      1.1  mrg     {   dchar d;
    165      1.1  mrg         wchar w;
    166      1.1  mrg 
    167      1.1  mrg         i--;
    168      1.1  mrg         w = aa[i];
    169      1.1  mrg         if (w & 0x80)
    170      1.1  mrg         {   char c = cast(char)w;
    171      1.1  mrg             uint j;
    172      1.1  mrg             uint m = 0x3F;
    173      1.1  mrg             d = 0;
    174      1.1  mrg             while ((c & 0xC0) != 0xC0)
    175      1.1  mrg             {   if (i == 0)
    176      1.1  mrg                     onUnicodeError("Invalid UTF-8 sequence", 0);
    177      1.1  mrg                 i--;
    178      1.1  mrg                 d |= (c & 0x3F) << j;
    179      1.1  mrg                 j += 6;
    180      1.1  mrg                 m >>= 1;
    181      1.1  mrg                 c = aa[i];
    182      1.1  mrg             }
    183      1.1  mrg             d |= (c & m) << j;
    184      1.1  mrg 
    185      1.1  mrg             if (d <= 0xFFFF)
    186      1.1  mrg                 w = cast(wchar) d;
    187      1.1  mrg             else
    188      1.1  mrg             {
    189      1.1  mrg                 w = cast(wchar) ((((d - 0x10000) >> 10) & 0x3FF) + 0xD800);
    190      1.1  mrg                 result = dg(cast(void *)&w);
    191      1.1  mrg                 if (result)
    192      1.1  mrg                     break;
    193      1.1  mrg                 w = cast(wchar) (((d - 0x10000) & 0x3FF) + 0xDC00);
    194      1.1  mrg             }
    195      1.1  mrg         }
    196      1.1  mrg         result = dg(cast(void *)&w);
    197      1.1  mrg         if (result)
    198      1.1  mrg             break;
    199      1.1  mrg     }
    200      1.1  mrg     return result;
    201      1.1  mrg }
    202      1.1  mrg 
    203      1.1  mrg unittest
    204      1.1  mrg {
    205      1.1  mrg     debug(apply) printf("_aApplyRcw1.unittest\n");
    206      1.1  mrg 
    207      1.1  mrg     auto s = "hello"c[];
    208      1.1  mrg     int i;
    209      1.1  mrg 
    210      1.1  mrg     foreach_reverse (wchar d; s)
    211      1.1  mrg     {
    212      1.1  mrg         switch (i)
    213      1.1  mrg         {
    214      1.1  mrg             case 0:     assert(d == 'o'); break;
    215      1.1  mrg             case 1:     assert(d == 'l'); break;
    216      1.1  mrg             case 2:     assert(d == 'l'); break;
    217      1.1  mrg             case 3:     assert(d == 'e'); break;
    218      1.1  mrg             case 4:     assert(d == 'h'); break;
    219      1.1  mrg             default:    assert(0);
    220      1.1  mrg         }
    221      1.1  mrg         i++;
    222      1.1  mrg     }
    223      1.1  mrg     assert(i == 5);
    224      1.1  mrg 
    225      1.1  mrg     s = "a\u1234\U000A0456b";
    226      1.1  mrg     i = 0;
    227      1.1  mrg     foreach_reverse (wchar d; s)
    228      1.1  mrg     {
    229      1.1  mrg         //printf("i = %d, d = %x\n", i, d);
    230      1.1  mrg         switch (i)
    231      1.1  mrg         {
    232      1.1  mrg             case 0:     assert(d == 'b'); break;
    233      1.1  mrg             case 1:     assert(d == 0xDA41); break;
    234      1.1  mrg             case 2:     assert(d == 0xDC56); break;
    235      1.1  mrg             case 3:     assert(d == 0x1234); break;
    236      1.1  mrg             case 4:     assert(d == 'a'); break;
    237      1.1  mrg             default:    assert(0);
    238      1.1  mrg         }
    239      1.1  mrg         i++;
    240      1.1  mrg     }
    241      1.1  mrg     assert(i == 5);
    242      1.1  mrg }
    243      1.1  mrg 
    244      1.1  mrg /*****************************/
    245      1.1  mrg 
    246      1.1  mrg extern (C) int _aApplyRwc1(in wchar[] aa, dg_t dg)
    247      1.1  mrg {   int result;
    248      1.1  mrg 
    249      1.1  mrg     debug(apply) printf("_aApplyRwc1(), len = %d\n", aa.length);
    250      1.1  mrg     for (size_t i = aa.length; i != 0; )
    251      1.1  mrg     {   dchar d;
    252      1.1  mrg         char c;
    253      1.1  mrg 
    254      1.1  mrg         i--;
    255      1.1  mrg         d = aa[i];
    256      1.1  mrg         if (d >= 0xDC00 && d <= 0xDFFF)
    257      1.1  mrg         {   if (i == 0)
    258      1.1  mrg                 onUnicodeError("Invalid UTF-16 sequence", 0);
    259      1.1  mrg             i--;
    260      1.1  mrg             d = ((aa[i] - 0xD7C0) << 10) + (d - 0xDC00);
    261      1.1  mrg         }
    262      1.1  mrg 
    263      1.1  mrg         if (d & ~0x7F)
    264      1.1  mrg         {
    265      1.1  mrg             char[4] buf = void;
    266      1.1  mrg 
    267      1.1  mrg             auto b = toUTF8(buf, d);
    268      1.1  mrg             foreach (char c2; b)
    269      1.1  mrg             {
    270      1.1  mrg                 result = dg(cast(void *)&c2);
    271      1.1  mrg                 if (result)
    272      1.1  mrg                     return result;
    273      1.1  mrg             }
    274      1.1  mrg             continue;
    275      1.1  mrg         }
    276      1.1  mrg         c = cast(char)d;
    277      1.1  mrg         result = dg(cast(void *)&c);
    278      1.1  mrg         if (result)
    279      1.1  mrg             break;
    280      1.1  mrg     }
    281      1.1  mrg     return result;
    282      1.1  mrg }
    283      1.1  mrg 
    284      1.1  mrg unittest
    285      1.1  mrg {
    286      1.1  mrg     debug(apply) printf("_aApplyRwc1.unittest\n");
    287      1.1  mrg 
    288      1.1  mrg     auto s = "hello"w[];
    289      1.1  mrg     int i;
    290      1.1  mrg 
    291      1.1  mrg     foreach_reverse (char d; s)
    292      1.1  mrg     {
    293      1.1  mrg         switch (i)
    294      1.1  mrg         {
    295      1.1  mrg             case 0:     assert(d == 'o'); break;
    296      1.1  mrg             case 1:     assert(d == 'l'); break;
    297      1.1  mrg             case 2:     assert(d == 'l'); break;
    298      1.1  mrg             case 3:     assert(d == 'e'); break;
    299      1.1  mrg             case 4:     assert(d == 'h'); break;
    300      1.1  mrg             default:    assert(0);
    301      1.1  mrg         }
    302      1.1  mrg         i++;
    303      1.1  mrg     }
    304      1.1  mrg     assert(i == 5);
    305      1.1  mrg 
    306      1.1  mrg     s = "a\u1234\U000A0456b";
    307      1.1  mrg     i = 0;
    308      1.1  mrg     foreach_reverse (char d; s)
    309      1.1  mrg     {
    310      1.1  mrg         //printf("i = %d, d = %x\n", i, d);
    311      1.1  mrg         switch (i)
    312      1.1  mrg         {
    313      1.1  mrg             case 0:     assert(d == 'b'); break;
    314      1.1  mrg             case 1:     assert(d == 0xF2); break;
    315      1.1  mrg             case 2:     assert(d == 0xA0); break;
    316      1.1  mrg             case 3:     assert(d == 0x91); break;
    317      1.1  mrg             case 4:     assert(d == 0x96); break;
    318      1.1  mrg             case 5:     assert(d == 0xE1); break;
    319      1.1  mrg             case 6:     assert(d == 0x88); break;
    320      1.1  mrg             case 7:     assert(d == 0xB4); break;
    321      1.1  mrg             case 8:     assert(d == 'a'); break;
    322      1.1  mrg             default:    assert(0);
    323      1.1  mrg         }
    324      1.1  mrg         i++;
    325      1.1  mrg     }
    326      1.1  mrg     assert(i == 9);
    327      1.1  mrg }
    328      1.1  mrg 
    329      1.1  mrg /*****************************/
    330      1.1  mrg 
    331      1.1  mrg extern (C) int _aApplyRdc1(in dchar[] aa, dg_t dg)
    332      1.1  mrg {   int result;
    333      1.1  mrg 
    334      1.1  mrg     debug(apply) printf("_aApplyRdc1(), len = %d\n", aa.length);
    335      1.1  mrg     for (size_t i = aa.length; i != 0;)
    336      1.1  mrg     {   dchar d = aa[--i];
    337      1.1  mrg         char c;
    338      1.1  mrg 
    339      1.1  mrg         if (d & ~0x7F)
    340      1.1  mrg         {
    341      1.1  mrg             char[4] buf = void;
    342      1.1  mrg 
    343      1.1  mrg             auto b = toUTF8(buf, d);
    344      1.1  mrg             foreach (char c2; b)
    345      1.1  mrg             {
    346      1.1  mrg                 result = dg(cast(void *)&c2);
    347      1.1  mrg                 if (result)
    348      1.1  mrg                     return result;
    349      1.1  mrg             }
    350      1.1  mrg             continue;
    351      1.1  mrg         }
    352      1.1  mrg         else
    353      1.1  mrg         {
    354      1.1  mrg             c = cast(char)d;
    355      1.1  mrg         }
    356      1.1  mrg         result = dg(cast(void *)&c);
    357      1.1  mrg         if (result)
    358      1.1  mrg             break;
    359      1.1  mrg     }
    360      1.1  mrg     return result;
    361      1.1  mrg }
    362      1.1  mrg 
    363      1.1  mrg unittest
    364      1.1  mrg {
    365      1.1  mrg     debug(apply) printf("_aApplyRdc1.unittest\n");
    366      1.1  mrg 
    367      1.1  mrg     auto s = "hello"d[];
    368      1.1  mrg     int i;
    369      1.1  mrg 
    370      1.1  mrg     foreach_reverse (char d; s)
    371      1.1  mrg     {
    372      1.1  mrg         switch (i)
    373      1.1  mrg         {
    374      1.1  mrg             case 0:     assert(d == 'o'); break;
    375      1.1  mrg             case 1:     assert(d == 'l'); break;
    376      1.1  mrg             case 2:     assert(d == 'l'); break;
    377      1.1  mrg             case 3:     assert(d == 'e'); break;
    378      1.1  mrg             case 4:     assert(d == 'h'); break;
    379      1.1  mrg             default:    assert(0);
    380      1.1  mrg         }
    381      1.1  mrg         i++;
    382      1.1  mrg     }
    383      1.1  mrg     assert(i == 5);
    384      1.1  mrg 
    385      1.1  mrg     s = "a\u1234\U000A0456b";
    386      1.1  mrg     i = 0;
    387      1.1  mrg     foreach_reverse (char d; s)
    388      1.1  mrg     {
    389      1.1  mrg         //printf("i = %d, d = %x\n", i, d);
    390      1.1  mrg         switch (i)
    391      1.1  mrg         {
    392      1.1  mrg             case 0:     assert(d == 'b'); break;
    393      1.1  mrg             case 1:     assert(d == 0xF2); break;
    394      1.1  mrg             case 2:     assert(d == 0xA0); break;
    395      1.1  mrg             case 3:     assert(d == 0x91); break;
    396      1.1  mrg             case 4:     assert(d == 0x96); break;
    397      1.1  mrg             case 5:     assert(d == 0xE1); break;
    398      1.1  mrg             case 6:     assert(d == 0x88); break;
    399      1.1  mrg             case 7:     assert(d == 0xB4); break;
    400      1.1  mrg             case 8:     assert(d == 'a'); break;
    401      1.1  mrg             default:    assert(0);
    402      1.1  mrg         }
    403      1.1  mrg         i++;
    404      1.1  mrg     }
    405      1.1  mrg     assert(i == 9);
    406      1.1  mrg }
    407      1.1  mrg 
    408      1.1  mrg /*****************************/
    409      1.1  mrg 
    410      1.1  mrg extern (C) int _aApplyRdw1(in dchar[] aa, dg_t dg)
    411      1.1  mrg {   int result;
    412      1.1  mrg 
    413      1.1  mrg     debug(apply) printf("_aApplyRdw1(), len = %d\n", aa.length);
    414      1.1  mrg     for (size_t i = aa.length; i != 0; )
    415      1.1  mrg     {   dchar d = aa[--i];
    416      1.1  mrg         wchar w;
    417      1.1  mrg 
    418      1.1  mrg         if (d <= 0xFFFF)
    419      1.1  mrg             w = cast(wchar) d;
    420      1.1  mrg         else
    421      1.1  mrg         {
    422      1.1  mrg             w = cast(wchar) ((((d - 0x10000) >> 10) & 0x3FF) + 0xD800);
    423      1.1  mrg             result = dg(cast(void *)&w);
    424      1.1  mrg             if (result)
    425      1.1  mrg                 break;
    426      1.1  mrg             w = cast(wchar) (((d - 0x10000) & 0x3FF) + 0xDC00);
    427      1.1  mrg         }
    428      1.1  mrg         result = dg(cast(void *)&w);
    429      1.1  mrg         if (result)
    430      1.1  mrg             break;
    431      1.1  mrg     }
    432      1.1  mrg     return result;
    433      1.1  mrg }
    434      1.1  mrg 
    435      1.1  mrg unittest
    436      1.1  mrg {
    437      1.1  mrg     debug(apply) printf("_aApplyRdw1.unittest\n");
    438      1.1  mrg 
    439      1.1  mrg     auto s = "hello"d[];
    440      1.1  mrg     int i;
    441      1.1  mrg 
    442      1.1  mrg     foreach_reverse (wchar d; s)
    443      1.1  mrg     {
    444      1.1  mrg         switch (i)
    445      1.1  mrg         {
    446      1.1  mrg             case 0:     assert(d == 'o'); break;
    447      1.1  mrg             case 1:     assert(d == 'l'); break;
    448      1.1  mrg             case 2:     assert(d == 'l'); break;
    449      1.1  mrg             case 3:     assert(d == 'e'); break;
    450      1.1  mrg             case 4:     assert(d == 'h'); break;
    451      1.1  mrg             default:    assert(0);
    452      1.1  mrg         }
    453      1.1  mrg         i++;
    454      1.1  mrg     }
    455      1.1  mrg     assert(i == 5);
    456      1.1  mrg 
    457      1.1  mrg     s = "a\u1234\U000A0456b";
    458      1.1  mrg     i = 0;
    459      1.1  mrg     foreach_reverse (wchar d; s)
    460      1.1  mrg     {
    461      1.1  mrg         //printf("i = %d, d = %x\n", i, d);
    462      1.1  mrg         switch (i)
    463      1.1  mrg         {
    464      1.1  mrg             case 0:     assert(d == 'b'); break;
    465      1.1  mrg             case 1:     assert(d == 0xDA41); break;
    466      1.1  mrg             case 2:     assert(d == 0xDC56); break;
    467      1.1  mrg             case 3:     assert(d == 0x1234); break;
    468      1.1  mrg             case 4:     assert(d == 'a'); break;
    469      1.1  mrg             default:    assert(0);
    470      1.1  mrg         }
    471      1.1  mrg         i++;
    472      1.1  mrg     }
    473      1.1  mrg     assert(i == 5);
    474      1.1  mrg }
    475      1.1  mrg 
    476      1.1  mrg 
    477      1.1  mrg /****************************************************************************/
    478      1.1  mrg /* 2 argument versions */
    479      1.1  mrg 
    480      1.1  mrg // dg is D, but _aApplyRcd2() is C
    481      1.1  mrg extern (D) alias int delegate(void *, void *) dg2_t;
    482      1.1  mrg 
    483      1.1  mrg extern (C) int _aApplyRcd2(in char[] aa, dg2_t dg)
    484      1.1  mrg {   int result;
    485      1.1  mrg     size_t i;
    486      1.1  mrg     size_t len = aa.length;
    487      1.1  mrg 
    488      1.1  mrg     debug(apply) printf("_aApplyRcd2(), len = %d\n", len);
    489      1.1  mrg     for (i = len; i != 0; )
    490      1.1  mrg     {   dchar d;
    491      1.1  mrg 
    492      1.1  mrg         i--;
    493      1.1  mrg         d = aa[i];
    494      1.1  mrg         if (d & 0x80)
    495      1.1  mrg         {   char c = cast(char)d;
    496      1.1  mrg             uint j;
    497      1.1  mrg             uint m = 0x3F;
    498      1.1  mrg             d = 0;
    499      1.1  mrg             while ((c & 0xC0) != 0xC0)
    500      1.1  mrg             {   if (i == 0)
    501      1.1  mrg                     onUnicodeError("Invalid UTF-8 sequence", 0);
    502      1.1  mrg                 i--;
    503      1.1  mrg                 d |= (c & 0x3F) << j;
    504      1.1  mrg                 j += 6;
    505      1.1  mrg                 m >>= 1;
    506      1.1  mrg                 c = aa[i];
    507      1.1  mrg             }
    508      1.1  mrg             d |= (c & m) << j;
    509      1.1  mrg         }
    510      1.1  mrg         result = dg(&i, cast(void *)&d);
    511      1.1  mrg         if (result)
    512      1.1  mrg             break;
    513      1.1  mrg     }
    514      1.1  mrg     return result;
    515      1.1  mrg }
    516      1.1  mrg 
    517      1.1  mrg unittest
    518      1.1  mrg {
    519      1.1  mrg     debug(apply) printf("_aApplyRcd2.unittest\n");
    520      1.1  mrg 
    521      1.1  mrg     auto s = "hello"c[];
    522      1.1  mrg     int i;
    523      1.1  mrg 
    524      1.1  mrg     foreach_reverse (k, dchar d; s)
    525      1.1  mrg     {
    526      1.1  mrg         assert(k == 4 - i);
    527      1.1  mrg         switch (i)
    528      1.1  mrg         {
    529      1.1  mrg             case 0:     assert(d == 'o'); break;
    530      1.1  mrg             case 1:     assert(d == 'l'); break;
    531      1.1  mrg             case 2:     assert(d == 'l'); break;
    532      1.1  mrg             case 3:     assert(d == 'e'); break;
    533      1.1  mrg             case 4:     assert(d == 'h'); break;
    534      1.1  mrg             default:    assert(0);
    535      1.1  mrg         }
    536      1.1  mrg         i++;
    537      1.1  mrg     }
    538      1.1  mrg     assert(i == 5);
    539      1.1  mrg 
    540      1.1  mrg     s = "a\u1234\U000A0456b";
    541      1.1  mrg     i = 0;
    542      1.1  mrg     foreach_reverse (k, dchar d; s)
    543      1.1  mrg     {
    544      1.1  mrg         //printf("i = %d, k = %d, d = %x\n", i, k, d);
    545      1.1  mrg         switch (i)
    546      1.1  mrg         {
    547      1.1  mrg             case 0:     assert(d == 'b'); assert(k == 8); break;
    548      1.1  mrg             case 1:     assert(d == '\U000A0456'); assert(k == 4); break;
    549      1.1  mrg             case 2:     assert(d == '\u1234'); assert(k == 1); break;
    550      1.1  mrg             case 3:     assert(d == 'a'); assert(k == 0); break;
    551      1.1  mrg             default:    assert(0);
    552      1.1  mrg         }
    553      1.1  mrg         i++;
    554      1.1  mrg     }
    555      1.1  mrg     assert(i == 4);
    556      1.1  mrg }
    557      1.1  mrg 
    558      1.1  mrg /*****************************/
    559      1.1  mrg 
    560      1.1  mrg extern (C) int _aApplyRwd2(in wchar[] aa, dg2_t dg)
    561      1.1  mrg {   int result;
    562      1.1  mrg 
    563      1.1  mrg     debug(apply) printf("_aApplyRwd2(), len = %d\n", aa.length);
    564      1.1  mrg     for (size_t i = aa.length; i != 0; )
    565      1.1  mrg     {   dchar d;
    566      1.1  mrg 
    567      1.1  mrg         i--;
    568      1.1  mrg         d = aa[i];
    569      1.1  mrg         if (d >= 0xDC00 && d <= 0xDFFF)
    570      1.1  mrg         {   if (i == 0)
    571      1.1  mrg                 onUnicodeError("Invalid UTF-16 sequence", 0);
    572      1.1  mrg             i--;
    573      1.1  mrg             d = ((aa[i] - 0xD7C0) << 10) + (d - 0xDC00);
    574      1.1  mrg         }
    575      1.1  mrg         result = dg(&i, cast(void *)&d);
    576      1.1  mrg         if (result)
    577      1.1  mrg             break;
    578      1.1  mrg     }
    579      1.1  mrg     return result;
    580      1.1  mrg }
    581      1.1  mrg 
    582      1.1  mrg unittest
    583      1.1  mrg {
    584      1.1  mrg     debug(apply) printf("_aApplyRwd2.unittest\n");
    585      1.1  mrg 
    586      1.1  mrg     auto s = "hello"w[];
    587      1.1  mrg     int i;
    588      1.1  mrg 
    589      1.1  mrg     foreach_reverse (k, dchar d; s)
    590      1.1  mrg     {
    591      1.1  mrg         //printf("i = %d, k = %d, d = %x\n", i, k, d);
    592      1.1  mrg         assert(k == 4 - i);
    593      1.1  mrg         switch (i)
    594      1.1  mrg         {
    595      1.1  mrg             case 0:     assert(d == 'o'); break;
    596      1.1  mrg             case 1:     assert(d == 'l'); break;
    597      1.1  mrg             case 2:     assert(d == 'l'); break;
    598      1.1  mrg             case 3:     assert(d == 'e'); break;
    599      1.1  mrg             case 4:     assert(d == 'h'); break;
    600      1.1  mrg             default:    assert(0);
    601      1.1  mrg         }
    602      1.1  mrg         i++;
    603      1.1  mrg     }
    604      1.1  mrg     assert(i == 5);
    605      1.1  mrg 
    606      1.1  mrg     s = "a\u1234\U000A0456b";
    607      1.1  mrg     i = 0;
    608      1.1  mrg     foreach_reverse (k, dchar d; s)
    609      1.1  mrg     {
    610      1.1  mrg         //printf("i = %d, k = %d, d = %x\n", i, k, d);
    611      1.1  mrg         switch (i)
    612      1.1  mrg         {
    613      1.1  mrg             case 0:     assert(k == 4); assert(d == 'b'); break;
    614      1.1  mrg             case 1:     assert(k == 2); assert(d == '\U000A0456'); break;
    615      1.1  mrg             case 2:     assert(k == 1); assert(d == '\u1234'); break;
    616      1.1  mrg             case 3:     assert(k == 0); assert(d == 'a'); break;
    617      1.1  mrg             default:    assert(0);
    618      1.1  mrg         }
    619      1.1  mrg         i++;
    620      1.1  mrg     }
    621      1.1  mrg     assert(i == 4);
    622      1.1  mrg }
    623      1.1  mrg 
    624      1.1  mrg /*****************************/
    625      1.1  mrg 
    626      1.1  mrg extern (C) int _aApplyRcw2(in char[] aa, dg2_t dg)
    627      1.1  mrg {   int result;
    628      1.1  mrg 
    629      1.1  mrg     debug(apply) printf("_aApplyRcw2(), len = %d\n", aa.length);
    630      1.1  mrg     for (size_t i = aa.length; i != 0; )
    631      1.1  mrg     {   dchar d;
    632      1.1  mrg         wchar w;
    633      1.1  mrg 
    634      1.1  mrg         i--;
    635      1.1  mrg         w = aa[i];
    636      1.1  mrg         if (w & 0x80)
    637      1.1  mrg         {   char c = cast(char)w;
    638      1.1  mrg             uint j;
    639      1.1  mrg             uint m = 0x3F;
    640      1.1  mrg             d = 0;
    641      1.1  mrg             while ((c & 0xC0) != 0xC0)
    642      1.1  mrg             {   if (i == 0)
    643      1.1  mrg                     onUnicodeError("Invalid UTF-8 sequence", 0);
    644      1.1  mrg                 i--;
    645      1.1  mrg                 d |= (c & 0x3F) << j;
    646      1.1  mrg                 j += 6;
    647      1.1  mrg                 m >>= 1;
    648      1.1  mrg                 c = aa[i];
    649      1.1  mrg             }
    650      1.1  mrg             d |= (c & m) << j;
    651      1.1  mrg 
    652      1.1  mrg             if (d <= 0xFFFF)
    653      1.1  mrg                 w = cast(wchar) d;
    654      1.1  mrg             else
    655      1.1  mrg             {
    656      1.1  mrg                 w = cast(wchar) ((((d - 0x10000) >> 10) & 0x3FF) + 0xD800);
    657      1.1  mrg                 result = dg(&i, cast(void *)&w);
    658      1.1  mrg                 if (result)
    659      1.1  mrg                     break;
    660      1.1  mrg                 w = cast(wchar) (((d - 0x10000) & 0x3FF) + 0xDC00);
    661      1.1  mrg             }
    662      1.1  mrg         }
    663      1.1  mrg         result = dg(&i, cast(void *)&w);
    664      1.1  mrg         if (result)
    665      1.1  mrg             break;
    666      1.1  mrg     }
    667      1.1  mrg     return result;
    668      1.1  mrg }
    669      1.1  mrg 
    670      1.1  mrg unittest
    671      1.1  mrg {
    672      1.1  mrg     debug(apply) printf("_aApplyRcw2.unittest\n");
    673      1.1  mrg 
    674      1.1  mrg     auto s = "hello"c[];
    675      1.1  mrg     int i;
    676      1.1  mrg 
    677      1.1  mrg     foreach_reverse (k, wchar d; s)
    678      1.1  mrg     {
    679      1.1  mrg         //printf("i = %d, k = %d, d = %x\n", i, k, d);
    680      1.1  mrg         assert(k == 4 - i);
    681      1.1  mrg         switch (i)
    682      1.1  mrg         {
    683      1.1  mrg             case 0:     assert(d == 'o'); break;
    684      1.1  mrg             case 1:     assert(d == 'l'); break;
    685      1.1  mrg             case 2:     assert(d == 'l'); break;
    686      1.1  mrg             case 3:     assert(d == 'e'); break;
    687      1.1  mrg             case 4:     assert(d == 'h'); break;
    688      1.1  mrg             default:    assert(0);
    689      1.1  mrg         }
    690      1.1  mrg         i++;
    691      1.1  mrg     }
    692      1.1  mrg     assert(i == 5);
    693      1.1  mrg 
    694      1.1  mrg     s = "a\u1234\U000A0456b";
    695      1.1  mrg     i = 0;
    696      1.1  mrg     foreach_reverse (k, wchar d; s)
    697      1.1  mrg     {
    698      1.1  mrg         //printf("i = %d, k = %d, d = %x\n", i, k, d);
    699      1.1  mrg         switch (i)
    700      1.1  mrg         {
    701      1.1  mrg             case 0:     assert(k == 8); assert(d == 'b'); break;
    702      1.1  mrg             case 1:     assert(k == 4); assert(d == 0xDA41); break;
    703      1.1  mrg             case 2:     assert(k == 4); assert(d == 0xDC56); break;
    704      1.1  mrg             case 3:     assert(k == 1); assert(d == 0x1234); break;
    705      1.1  mrg             case 4:     assert(k == 0); assert(d == 'a'); break;
    706      1.1  mrg             default:    assert(0);
    707      1.1  mrg         }
    708      1.1  mrg         i++;
    709      1.1  mrg     }
    710      1.1  mrg     assert(i == 5);
    711      1.1  mrg }
    712      1.1  mrg 
    713      1.1  mrg /*****************************/
    714      1.1  mrg 
    715      1.1  mrg extern (C) int _aApplyRwc2(in wchar[] aa, dg2_t dg)
    716      1.1  mrg {   int result;
    717      1.1  mrg 
    718      1.1  mrg     debug(apply) printf("_aApplyRwc2(), len = %d\n", aa.length);
    719      1.1  mrg     for (size_t i = aa.length; i != 0; )
    720      1.1  mrg     {   dchar d;
    721      1.1  mrg         char c;
    722      1.1  mrg 
    723      1.1  mrg         i--;
    724      1.1  mrg         d = aa[i];
    725      1.1  mrg         if (d >= 0xDC00 && d <= 0xDFFF)
    726      1.1  mrg         {   if (i == 0)
    727      1.1  mrg                 onUnicodeError("Invalid UTF-16 sequence", 0);
    728      1.1  mrg             i--;
    729      1.1  mrg             d = ((aa[i] - 0xD7C0) << 10) + (d - 0xDC00);
    730      1.1  mrg         }
    731      1.1  mrg 
    732      1.1  mrg         if (d & ~0x7F)
    733      1.1  mrg         {
    734      1.1  mrg             char[4] buf = void;
    735      1.1  mrg 
    736      1.1  mrg             auto b = toUTF8(buf, d);
    737      1.1  mrg             foreach (char c2; b)
    738      1.1  mrg             {
    739      1.1  mrg                 result = dg(&i, cast(void *)&c2);
    740      1.1  mrg                 if (result)
    741      1.1  mrg                     return result;
    742      1.1  mrg             }
    743      1.1  mrg             continue;
    744      1.1  mrg         }
    745      1.1  mrg         c = cast(char)d;
    746      1.1  mrg         result = dg(&i, cast(void *)&c);
    747      1.1  mrg         if (result)
    748      1.1  mrg             break;
    749      1.1  mrg     }
    750      1.1  mrg     return result;
    751      1.1  mrg }
    752      1.1  mrg 
    753      1.1  mrg unittest
    754      1.1  mrg {
    755      1.1  mrg     debug(apply) printf("_aApplyRwc2.unittest\n");
    756      1.1  mrg 
    757      1.1  mrg     auto s = "hello"w[];
    758      1.1  mrg     int i;
    759      1.1  mrg 
    760      1.1  mrg     foreach_reverse (k, char d; s)
    761      1.1  mrg     {
    762      1.1  mrg         //printf("i = %d, k = %d, d = %x\n", i, k, d);
    763      1.1  mrg         assert(k == 4 - i);
    764      1.1  mrg         switch (i)
    765      1.1  mrg         {
    766      1.1  mrg             case 0:     assert(d == 'o'); break;
    767      1.1  mrg             case 1:     assert(d == 'l'); break;
    768      1.1  mrg             case 2:     assert(d == 'l'); break;
    769      1.1  mrg             case 3:     assert(d == 'e'); break;
    770      1.1  mrg             case 4:     assert(d == 'h'); break;
    771      1.1  mrg             default:    assert(0);
    772      1.1  mrg         }
    773      1.1  mrg         i++;
    774      1.1  mrg     }
    775      1.1  mrg     assert(i == 5);
    776      1.1  mrg 
    777      1.1  mrg     s = "a\u1234\U000A0456b";
    778      1.1  mrg     i = 0;
    779      1.1  mrg     foreach_reverse (k, char d; s)
    780      1.1  mrg     {
    781      1.1  mrg         //printf("i = %d, k = %d, d = %x\n", i, k, d);
    782      1.1  mrg         switch (i)
    783      1.1  mrg         {
    784      1.1  mrg             case 0:     assert(k == 4); assert(d == 'b'); break;
    785      1.1  mrg             case 1:     assert(k == 2); assert(d == 0xF2); break;
    786      1.1  mrg             case 2:     assert(k == 2); assert(d == 0xA0); break;
    787      1.1  mrg             case 3:     assert(k == 2); assert(d == 0x91); break;
    788      1.1  mrg             case 4:     assert(k == 2); assert(d == 0x96); break;
    789      1.1  mrg             case 5:     assert(k == 1); assert(d == 0xE1); break;
    790      1.1  mrg             case 6:     assert(k == 1); assert(d == 0x88); break;
    791      1.1  mrg             case 7:     assert(k == 1); assert(d == 0xB4); break;
    792      1.1  mrg             case 8:     assert(k == 0); assert(d == 'a'); break;
    793      1.1  mrg             default:    assert(0);
    794      1.1  mrg         }
    795      1.1  mrg         i++;
    796      1.1  mrg     }
    797      1.1  mrg     assert(i == 9);
    798      1.1  mrg }
    799      1.1  mrg 
    800      1.1  mrg /*****************************/
    801      1.1  mrg 
    802      1.1  mrg extern (C) int _aApplyRdc2(in dchar[] aa, dg2_t dg)
    803      1.1  mrg {   int result;
    804      1.1  mrg 
    805      1.1  mrg     debug(apply) printf("_aApplyRdc2(), len = %d\n", aa.length);
    806      1.1  mrg     for (size_t i = aa.length; i != 0; )
    807      1.1  mrg     {   dchar d = aa[--i];
    808      1.1  mrg         char c;
    809      1.1  mrg 
    810      1.1  mrg         if (d & ~0x7F)
    811      1.1  mrg         {
    812      1.1  mrg             char[4] buf = void;
    813      1.1  mrg 
    814      1.1  mrg             auto b = toUTF8(buf, d);
    815      1.1  mrg             foreach (char c2; b)
    816      1.1  mrg             {
    817      1.1  mrg                 result = dg(&i, cast(void *)&c2);
    818      1.1  mrg                 if (result)
    819      1.1  mrg                     return result;
    820      1.1  mrg             }
    821      1.1  mrg             continue;
    822      1.1  mrg         }
    823      1.1  mrg         else
    824      1.1  mrg         {   c = cast(char)d;
    825      1.1  mrg         }
    826      1.1  mrg         result = dg(&i, cast(void *)&c);
    827      1.1  mrg         if (result)
    828      1.1  mrg             break;
    829      1.1  mrg     }
    830      1.1  mrg     return result;
    831      1.1  mrg }
    832      1.1  mrg 
    833      1.1  mrg unittest
    834      1.1  mrg {
    835      1.1  mrg     debug(apply) printf("_aApplyRdc2.unittest\n");
    836      1.1  mrg 
    837      1.1  mrg     auto s = "hello"d[];
    838      1.1  mrg     int i;
    839      1.1  mrg 
    840      1.1  mrg     foreach_reverse (k, char d; s)
    841      1.1  mrg     {
    842      1.1  mrg         //printf("i = %d, k = %d, d = %x\n", i, k, d);
    843      1.1  mrg         assert(k == 4 - i);
    844      1.1  mrg         switch (i)
    845      1.1  mrg         {
    846      1.1  mrg             case 0:     assert(d == 'o'); break;
    847      1.1  mrg             case 1:     assert(d == 'l'); break;
    848      1.1  mrg             case 2:     assert(d == 'l'); break;
    849      1.1  mrg             case 3:     assert(d == 'e'); break;
    850      1.1  mrg             case 4:     assert(d == 'h'); break;
    851      1.1  mrg             default:    assert(0);
    852      1.1  mrg         }
    853      1.1  mrg         i++;
    854      1.1  mrg     }
    855      1.1  mrg     assert(i == 5);
    856      1.1  mrg 
    857      1.1  mrg     s = "a\u1234\U000A0456b";
    858      1.1  mrg     i = 0;
    859      1.1  mrg     foreach_reverse (k, char d; s)
    860      1.1  mrg     {
    861      1.1  mrg         //printf("i = %d, k = %d, d = %x\n", i, k, d);
    862      1.1  mrg         switch (i)
    863      1.1  mrg         {
    864      1.1  mrg             case 0:     assert(k == 3); assert(d == 'b'); break;
    865      1.1  mrg             case 1:     assert(k == 2); assert(d == 0xF2); break;
    866      1.1  mrg             case 2:     assert(k == 2); assert(d == 0xA0); break;
    867      1.1  mrg             case 3:     assert(k == 2); assert(d == 0x91); break;
    868      1.1  mrg             case 4:     assert(k == 2); assert(d == 0x96); break;
    869      1.1  mrg             case 5:     assert(k == 1); assert(d == 0xE1); break;
    870      1.1  mrg             case 6:     assert(k == 1); assert(d == 0x88); break;
    871      1.1  mrg             case 7:     assert(k == 1); assert(d == 0xB4); break;
    872      1.1  mrg             case 8:     assert(k == 0); assert(d == 'a'); break;
    873      1.1  mrg             default:    assert(0);
    874      1.1  mrg         }
    875      1.1  mrg         i++;
    876      1.1  mrg     }
    877      1.1  mrg     assert(i == 9);
    878      1.1  mrg }
    879      1.1  mrg 
    880      1.1  mrg /*****************************/
    881      1.1  mrg 
    882      1.1  mrg extern (C) int _aApplyRdw2(in dchar[] aa, dg2_t dg)
    883      1.1  mrg {   int result;
    884      1.1  mrg 
    885      1.1  mrg     debug(apply) printf("_aApplyRdw2(), len = %d\n", aa.length);
    886      1.1  mrg     for (size_t i = aa.length; i != 0; )
    887      1.1  mrg     {   dchar d = aa[--i];
    888      1.1  mrg         wchar w;
    889      1.1  mrg 
    890      1.1  mrg         if (d <= 0xFFFF)
    891      1.1  mrg             w = cast(wchar) d;
    892      1.1  mrg         else
    893      1.1  mrg         {
    894      1.1  mrg             w = cast(wchar) ((((d - 0x10000) >> 10) & 0x3FF) + 0xD800);
    895      1.1  mrg             result = dg(&i, cast(void *)&w);
    896      1.1  mrg             if (result)
    897      1.1  mrg                 break;
    898      1.1  mrg             w = cast(wchar) (((d - 0x10000) & 0x3FF) + 0xDC00);
    899      1.1  mrg         }
    900      1.1  mrg         result = dg(&i, cast(void *)&w);
    901      1.1  mrg         if (result)
    902      1.1  mrg             break;
    903      1.1  mrg     }
    904      1.1  mrg     return result;
    905      1.1  mrg }
    906      1.1  mrg 
    907      1.1  mrg unittest
    908      1.1  mrg {
    909      1.1  mrg     debug(apply) printf("_aApplyRdw2.unittest\n");
    910      1.1  mrg 
    911      1.1  mrg     auto s = "hello"d[];
    912      1.1  mrg     int i;
    913      1.1  mrg 
    914      1.1  mrg     foreach_reverse (k, wchar d; s)
    915      1.1  mrg     {
    916      1.1  mrg         //printf("i = %d, k = %d, d = %x\n", i, k, d);
    917      1.1  mrg         assert(k == 4 - i);
    918      1.1  mrg         switch (i)
    919      1.1  mrg         {
    920      1.1  mrg             case 0:     assert(d == 'o'); break;
    921      1.1  mrg             case 1:     assert(d == 'l'); break;
    922      1.1  mrg             case 2:     assert(d == 'l'); break;
    923      1.1  mrg             case 3:     assert(d == 'e'); break;
    924      1.1  mrg             case 4:     assert(d == 'h'); break;
    925      1.1  mrg             default:    assert(0);
    926      1.1  mrg         }
    927      1.1  mrg         i++;
    928      1.1  mrg     }
    929      1.1  mrg     assert(i == 5);
    930      1.1  mrg 
    931      1.1  mrg     s = "a\u1234\U000A0456b";
    932      1.1  mrg     i = 0;
    933      1.1  mrg     foreach_reverse (k, wchar d; s)
    934      1.1  mrg     {
    935      1.1  mrg         //printf("i = %d, k = %d, d = %x\n", i, k, d);
    936      1.1  mrg         switch (i)
    937      1.1  mrg         {
    938      1.1  mrg             case 0:     assert(k == 3); assert(d == 'b'); break;
    939      1.1  mrg             case 1:     assert(k == 2); assert(d == 0xDA41); break;
    940      1.1  mrg             case 2:     assert(k == 2); assert(d == 0xDC56); break;
    941      1.1  mrg             case 3:     assert(k == 1); assert(d == 0x1234); break;
    942      1.1  mrg             case 4:     assert(k == 0); assert(d == 'a'); break;
    943      1.1  mrg             default:    assert(0);
    944      1.1  mrg         }
    945      1.1  mrg         i++;
    946      1.1  mrg     }
    947      1.1  mrg     assert(i == 5);
    948      1.1  mrg }
    949