src/hb-ot-shaper-arabic-fallback.hh

   1 /*
   2  * Copyright © 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 #ifndef HB_OT_SHAPER_ARABIC_FALLBACK_HH
  28 #define HB_OT_SHAPER_ARABIC_FALLBACK_HH
  29
  30 #include "hb.hh"
  31
  32 #include "hb-ot-shape.hh"
  33 #include "hb-ot-layout-gsub-table.hh"
  34
  35
  36 /* Features ordered the same as the entries in shaping_table rows,
  37  * followed by rlig.  Don't change.
  38  *
  39  * We currently support one subtable per lookup, and one lookup
  40  * per feature.  But we allow duplicate features, so we use that!
  41  */
  42 static const hb_tag_t arabic_fallback_features[] =
  43 {
  44   HB_TAG('i','n','i','t'),
  45   HB_TAG('m','e','d','i'),
  46   HB_TAG('f','i','n','a'),
  47   HB_TAG('i','s','o','l'),
  48   HB_TAG('r','l','i','g'),
  49   HB_TAG('r','l','i','g'),
  50   HB_TAG('r','l','i','g'),
  51 };
  52
  53 static OT::SubstLookup *
  54 arabic_fallback_synthesize_lookup_single (const hb_ot_shape_plan_t *plan HB_UNUSED,
  55                                           hb_font_t *font,
  56                                           unsigned int feature_index)
  57 {
  58   OT::HBGlyphID16 glyphs[SHAPING_TABLE_LAST - SHAPING_TABLE_FIRST + 1];
  59   OT::HBGlyphID16 substitutes[SHAPING_TABLE_LAST - SHAPING_TABLE_FIRST + 1];
  60   unsigned int num_glyphs = 0;
  61
  62   /* Populate arrays */
  63   for (hb_codepoint_t u = SHAPING_TABLE_FIRST; u < SHAPING_TABLE_LAST + 1; u++)
  64   {
  65     hb_codepoint_t s = shaping_table[u - SHAPING_TABLE_FIRST][feature_index];
  66     hb_codepoint_t u_glyph, s_glyph;
  67
  68     if (!s ||
  69         !hb_font_get_glyph (font, u, 0, &u_glyph) ||
  70         !hb_font_get_glyph (font, s, 0, &s_glyph) ||
  71         u_glyph == s_glyph ||
  72         u_glyph > 0xFFFFu || s_glyph > 0xFFFFu)
  73       continue;
  74
  75     glyphs[num_glyphs] = u_glyph;
  76     substitutes[num_glyphs] = s_glyph;
  77
  78     num_glyphs++;
  79   }
  80
  81   if (!num_glyphs)
  82     return nullptr;
  83
  84   /* Bubble-sort or something equally good!
  85    * May not be good-enough for presidential candidate interviews, but good-enough for us... */
  86   hb_stable_sort (&glyphs[0], num_glyphs,
  87                   (int(*)(const OT::HBUINT16*, const OT::HBUINT16 *)) OT::HBGlyphID16::cmp,
  88                   &substitutes[0]);
  89
  90
  91   /* Each glyph takes four bytes max, and there's some overhead. */
  92   char buf[(SHAPING_TABLE_LAST - SHAPING_TABLE_FIRST + 1) * 4 + 128];
  93   hb_serialize_context_t c (buf, sizeof (buf));
  94   OT::SubstLookup *lookup = c.start_serialize<OT::SubstLookup> ();
  95   bool ret = lookup->serialize_single (&c,
  96                                        OT::LookupFlag::IgnoreMarks,
  97                                        hb_sorted_array (glyphs, num_glyphs),
  98                                        hb_array (substitutes, num_glyphs));
  99   c.end_serialize ();
 100
 101   return ret && !c.in_error () ? c.copy<OT::SubstLookup> () : nullptr;
 102 }
 103
 104 template <typename T>
 105 static OT::SubstLookup *
 106 arabic_fallback_synthesize_lookup_ligature (const hb_ot_shape_plan_t *plan HB_UNUSED,
 107                                             hb_font_t *font,
 108                                             const T &ligature_table,
 109                                             unsigned lookup_flags)
 110 {
 111   OT::HBGlyphID16 first_glyphs[ARRAY_LENGTH_CONST (ligature_table)];
 112   unsigned int first_glyphs_indirection[ARRAY_LENGTH_CONST (ligature_table)];
 113   unsigned int ligature_per_first_glyph_count_list[ARRAY_LENGTH_CONST (first_glyphs)];
 114   unsigned int num_first_glyphs = 0;
 115
 116   /* We know that all our ligatures have the same number of components. */
 117   OT::HBGlyphID16 ligature_list[ARRAY_LENGTH_CONST (first_glyphs) * ARRAY_LENGTH_CONST(ligature_table[0].ligatures)];
 118   unsigned int component_count_list[ARRAY_LENGTH_CONST (ligature_list)];
 119   OT::HBGlyphID16 component_list[ARRAY_LENGTH_CONST (ligature_list) *
 120                                  ARRAY_LENGTH_CONST (ligature_table[0].ligatures[0].components)];
 121   unsigned int num_ligatures = 0;
 122   unsigned int num_components = 0;
 123
 124   /* Populate arrays */
 125
 126   /* Sort out the first-glyphs */
 127   for (unsigned int first_glyph_idx = 0; first_glyph_idx < ARRAY_LENGTH (first_glyphs); first_glyph_idx++)
 128   {
 129     hb_codepoint_t first_u = ligature_table[first_glyph_idx].first;
 130     hb_codepoint_t first_glyph;
 131     if (!hb_font_get_glyph (font, first_u, 0, &first_glyph))
 132       continue;
 133     first_glyphs[num_first_glyphs] = first_glyph;
 134     ligature_per_first_glyph_count_list[num_first_glyphs] = 0;
 135     first_glyphs_indirection[num_first_glyphs] = first_glyph_idx;
 136     num_first_glyphs++;
 137   }
 138   hb_stable_sort (&first_glyphs[0], num_first_glyphs,
 139                   (int(*)(const OT::HBUINT16*, const OT::HBUINT16 *)) OT::HBGlyphID16::cmp,
 140                   &first_glyphs_indirection[0]);
 141
 142   /* Now that the first-glyphs are sorted, walk again, populate ligatures. */
 143   for (unsigned int i = 0; i < num_first_glyphs; i++)
 144   {
 145     unsigned int first_glyph_idx = first_glyphs_indirection[i];
 146
 147     for (unsigned int ligature_idx = 0; ligature_idx < ARRAY_LENGTH (ligature_table[0].ligatures); ligature_idx++)
 148     {
 149       hb_codepoint_t ligature_u = ligature_table[first_glyph_idx].ligatures[ligature_idx].ligature;
 150       hb_codepoint_t ligature_glyph;
 151       if (!hb_font_get_glyph (font, ligature_u, 0, &ligature_glyph))
 152         continue;
 153
 154       const auto &components = ligature_table[first_glyph_idx].ligatures[ligature_idx].components;
 155       unsigned component_count = ARRAY_LENGTH_CONST (components);
 156
 157       bool matched = true;
 158       for (unsigned j = 0; j < component_count; j++)
 159       {
 160         hb_codepoint_t component_u   = ligature_table[first_glyph_idx].ligatures[ligature_idx].components[j];
 161         hb_codepoint_t component_glyph;
 162         if (!component_u ||
 163             !hb_font_get_nominal_glyph (font, component_u, &component_glyph))
 164         {
 165           matched = false;
 166           break;
 167         }
 168
 169         component_list[num_components++] = component_glyph;
 170       }
 171       if (!matched)
 172         continue;
 173
 174       component_count_list[num_ligatures] = 1 + component_count;
 175       ligature_list[num_ligatures] = ligature_glyph;
 176
 177       ligature_per_first_glyph_count_list[i]++;
 178
 179       num_ligatures++;
 180     }
 181   }
 182
 183   if (!num_ligatures)
 184     return nullptr;
 185
 186
 187   /* 16 bytes per ligature ought to be enough... */
 188   char buf[ARRAY_LENGTH_CONST (ligature_list) * 16 + 128];
 189   hb_serialize_context_t c (buf, sizeof (buf));
 190   OT::SubstLookup *lookup = c.start_serialize<OT::SubstLookup> ();
 191   bool ret = lookup->serialize_ligature (&c,
 192                                          lookup_flags,
 193                                          hb_sorted_array (first_glyphs, num_first_glyphs),
 194                                          hb_array (ligature_per_first_glyph_count_list, num_first_glyphs),
 195                                          hb_array (ligature_list, num_ligatures),
 196                                          hb_array (component_count_list, num_ligatures),
 197                                          hb_array (component_list, num_components));
 198   c.end_serialize ();
 199
 200   return ret && !c.in_error () ? c.copy<OT::SubstLookup> () : nullptr;
 201 }
 202
 203 static OT::SubstLookup *
 204 arabic_fallback_synthesize_lookup (const hb_ot_shape_plan_t *plan,
 205                                    hb_font_t *font,
 206                                    unsigned int feature_index)
 207 {
 208   if (feature_index < 4)
 209     return arabic_fallback_synthesize_lookup_single (plan, font, feature_index);
 210   else
 211   {
 212     switch (feature_index) {
 213       case 4: return arabic_fallback_synthesize_lookup_ligature (plan, font, ligature_3_table, OT::LookupFlag::IgnoreMarks);
 214       case 5: return arabic_fallback_synthesize_lookup_ligature (plan, font, ligature_table, OT::LookupFlag::IgnoreMarks);
 215       case 6: return arabic_fallback_synthesize_lookup_ligature (plan, font, ligature_mark_table, 0);
 216     }
 217   }
 218   assert (false);
 219   return nullptr;
 220 }
 221
 222 #define ARABIC_FALLBACK_MAX_LOOKUPS ARRAY_LENGTH_CONST (arabic_fallback_features)
 223
 224 struct arabic_fallback_plan_t
 225 {
 226   unsigned int num_lookups;
 227   bool free_lookups;
 228
 229   hb_mask_t mask_array[ARABIC_FALLBACK_MAX_LOOKUPS];
 230   OT::SubstLookup *lookup_array[ARABIC_FALLBACK_MAX_LOOKUPS];
 231   OT::hb_ot_layout_lookup_accelerator_t *accel_array[ARABIC_FALLBACK_MAX_LOOKUPS];
 232 };
 233
 234 #if defined(_WIN32) && !defined(HB_NO_WIN1256)
 235 #define HB_WITH_WIN1256
 236 #endif
 237
 238 #ifdef HB_WITH_WIN1256
 239 #include "hb-ot-shaper-arabic-win1256.hh"
 240 #endif
 241
 242 struct ManifestLookup
 243 {
 244   public:
 245   OT::Tag tag;
 246   OT::Offset16To<OT::SubstLookup> lookupOffset;
 247   public:
 248   DEFINE_SIZE_STATIC (6);
 249 };
 250 typedef OT::Array16Of<ManifestLookup> Manifest;
 251
 252 static bool
 253 arabic_fallback_plan_init_win1256 (arabic_fallback_plan_t *fallback_plan HB_UNUSED,
 254                                    const hb_ot_shape_plan_t *plan HB_UNUSED,
 255                                    hb_font_t *font HB_UNUSED)
 256 {
 257 #ifdef HB_WITH_WIN1256
 258   /* Does this font look like it's Windows-1256-encoded? */
 259   hb_codepoint_t g;
 260   if (!(hb_font_get_glyph (font, 0x0627u, 0, &g) && g == 199 /* ALEF */ &&
 261         hb_font_get_glyph (font, 0x0644u, 0, &g) && g == 225 /* LAM */ &&
 262         hb_font_get_glyph (font, 0x0649u, 0, &g) && g == 236 /* ALEF MAKSURA */ &&
 263         hb_font_get_glyph (font, 0x064Au, 0, &g) && g == 237 /* YEH */ &&
 264         hb_font_get_glyph (font, 0x0652u, 0, &g) && g == 250 /* SUKUN */))
 265     return false;
 266
 267   const Manifest &manifest = reinterpret_cast<const Manifest&> (arabic_win1256_gsub_lookups.manifest);
 268   static_assert (sizeof (arabic_win1256_gsub_lookups.manifestData) <=
 269                  ARABIC_FALLBACK_MAX_LOOKUPS * sizeof (ManifestLookup), "");
 270
 271   unsigned j = 0;
 272   unsigned int count = manifest.len;
 273   for (unsigned int i = 0; i < count; i++)
 274   {
 275     fallback_plan->mask_array[j] = plan->map.get_1_mask (manifest[i].tag);
 276     if (fallback_plan->mask_array[j])
 277     {
 278       fallback_plan->lookup_array[j] = const_cast<OT::SubstLookup*> (&(&manifest+manifest[i].lookupOffset));
 279       if (fallback_plan->lookup_array[j])
 280       {
 281         fallback_plan->accel_array[j] = OT::hb_ot_layout_lookup_accelerator_t::create (*fallback_plan->lookup_array[j]);
 282         j++;
 283       }
 284     }
 285   }
 286
 287   fallback_plan->num_lookups = j;
 288   fallback_plan->free_lookups = false;
 289
 290   return j > 0;
 291 #else
 292   return false;
 293 #endif
 294 }
 295
 296 static bool
 297 arabic_fallback_plan_init_unicode (arabic_fallback_plan_t *fallback_plan,
 298                                    const hb_ot_shape_plan_t *plan,
 299                                    hb_font_t *font)
 300 {
 301   static_assert ((ARRAY_LENGTH_CONST (arabic_fallback_features) <= ARABIC_FALLBACK_MAX_LOOKUPS), "");
 302   unsigned int j = 0;
 303   for (unsigned int i = 0; i < ARRAY_LENGTH(arabic_fallback_features) ; i++)
 304   {
 305     fallback_plan->mask_array[j] = plan->map.get_1_mask (arabic_fallback_features[i]);
 306     if (fallback_plan->mask_array[j])
 307     {
 308       fallback_plan->lookup_array[j] = arabic_fallback_synthesize_lookup (plan, font, i);
 309       if (fallback_plan->lookup_array[j])
 310       {
 311         fallback_plan->accel_array[j] = OT::hb_ot_layout_lookup_accelerator_t::create (*fallback_plan->lookup_array[j]);
 312         j++;
 313       }
 314     }
 315   }
 316
 317   fallback_plan->num_lookups = j;
 318   fallback_plan->free_lookups = true;
 319
 320   return j > 0;
 321 }
 322
 323 static arabic_fallback_plan_t *
 324 arabic_fallback_plan_create (const hb_ot_shape_plan_t *plan,
 325                              hb_font_t *font)
 326 {
 327   arabic_fallback_plan_t *fallback_plan = (arabic_fallback_plan_t *) hb_calloc (1, sizeof (arabic_fallback_plan_t));
 328   if (unlikely (!fallback_plan))
 329     return const_cast<arabic_fallback_plan_t *> (&Null (arabic_fallback_plan_t));
 330
 331   fallback_plan->num_lookups = 0;
 332   fallback_plan->free_lookups = false;
 333
 334   /* Try synthesizing GSUB table using Unicode Arabic Presentation Forms,
 335    * in case the font has cmap entries for the presentation-forms characters. */
 336   if (arabic_fallback_plan_init_unicode (fallback_plan, plan, font))
 337     return fallback_plan;
 338
 339   /* See if this looks like a Windows-1256-encoded font.  If it does, use a
 340    * hand-coded GSUB table. */
 341   if (arabic_fallback_plan_init_win1256 (fallback_plan, plan, font))
 342     return fallback_plan;
 343
 344   assert (fallback_plan->num_lookups == 0);
 345   hb_free (fallback_plan);
 346   return const_cast<arabic_fallback_plan_t *> (&Null (arabic_fallback_plan_t));
 347 }
 348
 349 static void
 350 arabic_fallback_plan_destroy (arabic_fallback_plan_t *fallback_plan)
 351 {
 352   if (!fallback_plan || fallback_plan->num_lookups == 0)
 353     return;
 354
 355   for (unsigned int i = 0; i < fallback_plan->num_lookups; i++)
 356     if (fallback_plan->lookup_array[i])
 357     {
 358       hb_free (fallback_plan->accel_array[i]);
 359       if (fallback_plan->free_lookups)
 360         hb_free (fallback_plan->lookup_array[i]);
 361     }
 362
 363   hb_free (fallback_plan);
 364 }
 365
 366 static void
 367 arabic_fallback_plan_shape (arabic_fallback_plan_t *fallback_plan,
 368                             hb_font_t *font,
 369                             hb_buffer_t *buffer)
 370 {
 371   OT::hb_ot_apply_context_t c (0, font, buffer, hb_blob_get_empty ());
 372   for (unsigned int i = 0; i < fallback_plan->num_lookups; i++)
 373     if (fallback_plan->lookup_array[i]) {
 374       c.set_lookup_mask (fallback_plan->mask_array[i]);
 375       if (fallback_plan->accel_array[i])
 376         hb_ot_layout_substitute_lookup (&c,
 377                                         *fallback_plan->lookup_array[i],
 378                                         *fallback_plan->accel_array[i]);
 379     }
 380 }
 381
 382
 383 #endif /* HB_OT_SHAPER_ARABIC_FALLBACK_HH */