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