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