src/hb-ot-shaper-hebrew.cc

   1 /*
   2  * Copyright © 2010,2012  Google, Inc.
   3  *
   4  *  This is part of HarfBuzz, a text shaping library.
   5  *
   6  * Permission is hereby granted, without written agreement and without
   7  * license or royalty fees, to use, copy, modify, and distribute this
   8  * software and its documentation for any purpose, provided that the
   9  * above copyright notice and the following two paragraphs appear in
  10  * all copies of this software.
  11  *
  12  * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
  13  * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
  14  * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
  15  * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
  16  * DAMAGE.
  17  *
  18  * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
  19  * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
  20  * FITNESS FOR A PARTICULAR PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS
  21  * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
  22  * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
  23  *
  24  * Google Author(s): Behdad Esfahbod
  25  */
  26
  27 #include "hb.hh"
  28
  29 #ifndef HB_NO_OT_SHAPE
  30
  31 #include "hb-ot-shaper.hh"
  32
  33
  34 static bool
  35 compose_hebrew (const hb_ot_shape_normalize_context_t *c,
  36                 hb_codepoint_t  a,
  37                 hb_codepoint_t  b,
  38                 hb_codepoint_t *ab)
  39 {
  40   /* Hebrew presentation-form shaping.
  41    * https://bugzilla.mozilla.org/show_bug.cgi?id=728866
  42    * Hebrew presentation forms with dagesh, for characters U+05D0..05EA;
  43    * Note that some letters do not have a dagesh presForm encoded.
  44    */
  45   static const hb_codepoint_t sDageshForms[0x05EAu - 0x05D0u + 1] = {
  46     0xFB30u, /* ALEF */
  47     0xFB31u, /* BET */
  48     0xFB32u, /* GIMEL */
  49     0xFB33u, /* DALET */
  50     0xFB34u, /* HE */
  51     0xFB35u, /* VAV */
  52     0xFB36u, /* ZAYIN */
  53     0x0000u, /* HET */
  54     0xFB38u, /* TET */
  55     0xFB39u, /* YOD */
  56     0xFB3Au, /* FINAL KAF */
  57     0xFB3Bu, /* KAF */
  58     0xFB3Cu, /* LAMED */
  59     0x0000u, /* FINAL MEM */
  60     0xFB3Eu, /* MEM */
  61     0x0000u, /* FINAL NUN */
  62     0xFB40u, /* NUN */
  63     0xFB41u, /* SAMEKH */
  64     0x0000u, /* AYIN */
  65     0xFB43u, /* FINAL PE */
  66     0xFB44u, /* PE */
  67     0x0000u, /* FINAL TSADI */
  68     0xFB46u, /* TSADI */
  69     0xFB47u, /* QOF */
  70     0xFB48u, /* RESH */
  71     0xFB49u, /* SHIN */
  72     0xFB4Au /* TAV */
  73   };
  74
  75   bool found = (bool) c->unicode->compose (a, b, ab);
  76
  77 #ifdef HB_NO_OT_SHAPER_HEBREW_FALLBACK
  78   return found;
  79 #endif
  80
  81   if (!found && !c->plan->has_gpos_mark)
  82   {
  83       /* Special-case Hebrew presentation forms that are excluded from
  84        * standard normalization, but wanted for old fonts. */
  85       switch (b) {
  86       case 0x05B4u: /* HIRIQ */
  87           if (a == 0x05D9u) { /* YOD */
  88               *ab = 0xFB1Du;
  89               found = true;
  90           }
  91           break;
  92       case 0x05B7u: /* PATAH */
  93           if (a == 0x05F2u) { /* YIDDISH YOD YOD */
  94               *ab = 0xFB1Fu;
  95               found = true;
  96           } else if (a == 0x05D0u) { /* ALEF */
  97               *ab = 0xFB2Eu;
  98               found = true;
  99           }
 100           break;
 101       case 0x05B8u: /* QAMATS */
 102           if (a == 0x05D0u) { /* ALEF */
 103               *ab = 0xFB2Fu;
 104               found = true;
 105           }
 106           break;
 107       case 0x05B9u: /* HOLAM */
 108           if (a == 0x05D5u) { /* VAV */
 109               *ab = 0xFB4Bu;
 110               found = true;
 111           }
 112           break;
 113       case 0x05BCu: /* DAGESH */
 114           if (a >= 0x05D0u && a <= 0x05EAu) {
 115               *ab = sDageshForms[a - 0x05D0u];
 116               found = (*ab != 0);
 117           } else if (a == 0xFB2Au) { /* SHIN WITH SHIN DOT */
 118               *ab = 0xFB2Cu;
 119               found = true;
 120           } else if (a == 0xFB2Bu) { /* SHIN WITH SIN DOT */
 121               *ab = 0xFB2Du;
 122               found = true;
 123           }
 124           break;
 125       case 0x05BFu: /* RAFE */
 126           switch (a) {
 127           case 0x05D1u: /* BET */
 128               *ab = 0xFB4Cu;
 129               found = true;
 130               break;
 131           case 0x05DBu: /* KAF */
 132               *ab = 0xFB4Du;
 133               found = true;
 134               break;
 135           case 0x05E4u: /* PE */
 136               *ab = 0xFB4Eu;
 137               found = true;
 138               break;
 139           }
 140           break;
 141       case 0x05C1u: /* SHIN DOT */
 142           if (a == 0x05E9u) { /* SHIN */
 143               *ab = 0xFB2Au;
 144               found = true;
 145           } else if (a == 0xFB49u) { /* SHIN WITH DAGESH */
 146               *ab = 0xFB2Cu;
 147               found = true;
 148           }
 149           break;
 150       case 0x05C2u: /* SIN DOT */
 151           if (a == 0x05E9u) { /* SHIN */
 152               *ab = 0xFB2Bu;
 153               found = true;
 154           } else if (a == 0xFB49u) { /* SHIN WITH DAGESH */
 155               *ab = 0xFB2Du;
 156               found = true;
 157           }
 158           break;
 159       }
 160   }
 161
 162   return found;
 163 }
 164
 165 static void
 166 reorder_marks_hebrew (const hb_ot_shape_plan_t *plan HB_UNUSED,
 167                       hb_buffer_t              *buffer,
 168                       unsigned int              start,
 169                       unsigned int              end)
 170 {
 171   hb_glyph_info_t *info = buffer->info;
 172
 173   for (unsigned i = start + 2; i < end; i++)
 174   {
 175     unsigned c0 = info_cc (info[i - 2]);
 176     unsigned c1 = info_cc (info[i - 1]);
 177     unsigned c2 = info_cc (info[i - 0]);
 178
 179     if ((c0 == HB_MODIFIED_COMBINING_CLASS_CCC17 || c0 == HB_MODIFIED_COMBINING_CLASS_CCC18) /* patach or qamats */ &&
 180         (c1 == HB_MODIFIED_COMBINING_CLASS_CCC10 || c1 == HB_MODIFIED_COMBINING_CLASS_CCC14) /* sheva or hiriq */ &&
 181         (c2 == HB_MODIFIED_COMBINING_CLASS_CCC22 || c2 == HB_UNICODE_COMBINING_CLASS_BELOW) /* meteg or below */)
 182     {
 183       buffer->merge_clusters (i - 1, i + 1);
 184       hb_swap (info[i - 1], info[i]);
 185       break;
 186     }
 187   }
 188
 189
 190 }
 191
 192 const hb_ot_shaper_t _hb_ot_shaper_hebrew =
 193 {
 194   nullptr, /* collect_features */
 195   nullptr, /* override_features */
 196   nullptr, /* data_create */
 197   nullptr, /* data_destroy */
 198   nullptr, /* preprocess_text */
 199   nullptr, /* postprocess_glyphs */
 200   nullptr, /* decompose */
 201   compose_hebrew,
 202   nullptr, /* setup_masks */
 203   reorder_marks_hebrew,
 204   HB_TAG ('h','e','b','r'), /* gpos_tag. https://github.com/harfbuzz/harfbuzz/issues/347#issuecomment-267838368 */
 205   HB_OT_SHAPE_NORMALIZATION_MODE_DEFAULT,
 206   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_LATE,
 207   true, /* fallback_position */
 208 };
 209
 210
 211 #endif