git.sesse.net Git - casparcg/blob - core/producer/text/utils/texture_font.cpp

   1 #include "..\..\..\StdAfx.h"
   2
   3 #include "texture_atlas.h"
   4 #include "texture_font.h"
   5
   6 #include <map>
   7 #include <ft2build.h>
   8 #include FT_FREETYPE_H
   9 #include FT_GLYPH_H
  10
  11 namespace caspar { namespace core { namespace text {
  12
  13 struct freetype_exception : virtual caspar_exception
  14 {
  15         freetype_exception() {}
  16         explicit freetype_exception(const char* msg) : caspar_exception(msg) {}
  17 };
  18
  19 struct unicode_range
  20 {
  21         unicode_range() : first(0), last(0) {}
  22         unicode_range(int f, int l) : first(f), last(l) {}
  23
  24         int first;
  25         int last;
  26 };
  27
  28 unicode_range get_range(unicode_block block);
  29
  30
  31 struct texture_font::impl
  32 {
  33 private:
  34         struct glyph_info
  35         {
  36                 glyph_info(int w, int h, float l, float t, float r, float b) : width(w), height(h), left(l), top(t), right(r), bottom(b)
  37                 {}
  38
  39                 float left, top, right, bottom;
  40                 int width, height;
  41         };
  42
  43         FT_Library              lib_;
  44         FT_Face                 face_;
  45         texture_atlas   atlas_;
  46         float                   size_;
  47         float                   tracking_;
  48         bool                    normalize_;
  49         std::map<int, glyph_info> glyphs_;
  50
  51 public:
  52         impl::impl(texture_atlas& atlas, const text_info& info, bool normalize_coordinates) : lib_(nullptr), face_(nullptr), atlas_(atlas), size_(info.size), tracking_(info.size*info.tracking/1000.0f), normalize_(normalize_coordinates)
  53         {
  54                 try
  55                 {
  56                         FT_Error err;
  57                         err = FT_Init_FreeType(&lib_);
  58                         if(err) throw freetype_exception("Failed to initialize freetype");
  59
  60                         err = FT_New_Face(lib_, u8(info.font_file).c_str(), 0, &face_);
  61                         if(err) throw freetype_exception("Failed to load font");
  62
  63                         err = FT_Set_Char_Size(face_, static_cast<FT_F26Dot6>(size_*64), 0, 72, 72);
  64                         if(err) throw freetype_exception("Failed to set font size");
  65                 }
  66                 catch(std::exception& ex)
  67                 {
  68                         if(face_ != nullptr)
  69                                 FT_Done_Face(face_);
  70                         if(lib_ != nullptr)
  71                                 FT_Done_FreeType(lib_);
  72
  73                         throw ex;
  74                 }
  75         }
  76
  77         ~impl()
  78         {
  79                 if(face_ != nullptr)
  80                         FT_Done_Face(face_);
  81                 if(lib_ != nullptr)
  82                         FT_Done_FreeType(lib_);
  83         }
  84
  85         int count_glyphs_in_range(unicode_block block)
  86         {
  87                 unicode_range range = get_range(block);
  88
  89                 //TODO: extract info from freetype
  90
  91                 //very pesimistic, assumes a glyph for each charcode
  92                 return range.last - range.first;
  93         }
  94
  95         void impl::load_glyphs(unicode_block block, const color<float>& col)
  96         {
  97                 FT_Error err;
  98                 int flags = FT_LOAD_RENDER | FT_LOAD_FORCE_AUTOHINT | FT_LOAD_TARGET_NORMAL;
  99                 unicode_range range = get_range(block);
 100
 101                 for(int i = range.first; i <= range.last; ++i)
 102                 {
 103                         FT_UInt glyph_index = FT_Get_Char_Index(face_, i);
 104                         if(!glyph_index)        //ignore codes that doesn't have a glyph for now. Might want to map these to a special glyph later.
 105                                 continue;
 106
 107                         err = FT_Load_Glyph(face_, glyph_index, flags);
 108                         if(err) continue;       //igonore glyphs that fail to load
 109
 110                         const FT_Bitmap& bitmap = face_->glyph->bitmap; //shorthand notation
 111
 112                         auto region = atlas_.get_region(bitmap.width+1, bitmap.rows+1);
 113                         if(region.x < 0)
 114                         {
 115                                 //the glyph doesn't fit in the texture-atlas. ignore it for now.
 116                                 //we might want to restart with a bigger atlas in the future
 117                                 continue;
 118                         }
 119
 120                         atlas_.set_region(region.x, region.y, bitmap.width, bitmap.rows, bitmap.buffer, bitmap.pitch, col);
 121                         glyphs_.insert(std::pair<int, glyph_info>(i, glyph_info(bitmap.width, bitmap.rows,
 122                                                                                 region.x / (float)atlas_.width(),
 123                                                                                 region.y / (float)atlas_.height(),
 124                                                                                 (region.x + bitmap.width) / (float)atlas_.width(),
 125                                                                                 (region.y + bitmap.rows) / (float)atlas_.height())));
 126                 }
 127         }
 128
 129         std::vector<float> create_vertex_stream(const std::wstring& str, int x, int y, int parent_width, int parent_height, string_metrics* metrics)
 130         {
 131                 //TODO: detect glyphs that aren't in the atlas and load them (and maybe that entire unicode_block on the fly
 132
 133                 std::vector<float> result(16*str.length(), 0);
 134
 135                 bool use_kerning = (face_->face_flags & FT_FACE_FLAG_KERNING) == FT_FACE_FLAG_KERNING;
 136                 int index = 0;
 137                 FT_UInt previous = 0;
 138                 float pos_x = (float)x;
 139                 float pos_y = (float)y;
 140
 141                 int maxBearingY = 0;
 142                 int maxProtrudeUnderY = 0;
 143                 int maxHeight = 0;
 144
 145                 auto end = str.end();
 146                 for(auto it = str.begin(); it != end; ++it, ++index)
 147                 {
 148                         auto glyph_it = glyphs_.find(*it);
 149                         if(glyph_it != glyphs_.end())
 150                         {
 151                                 const glyph_info& coords = glyph_it->second;
 152
 153                                 FT_UInt glyph_index = FT_Get_Char_Index(face_, (*it));
 154
 155                                 if(use_kerning && previous && glyph_index)
 156                                 {
 157                                         FT_Vector delta;
 158                                         FT_Get_Kerning(face_, previous, glyph_index, FT_KERNING_DEFAULT, &delta);
 159
 160                                         pos_x += delta.x / 64.0f;
 161                                 }
 162
 163                                 FT_Load_Glyph(face_, glyph_index, FT_LOAD_NO_BITMAP | FT_LOAD_FORCE_AUTOHINT | FT_LOAD_TARGET_NORMAL);
 164
 165                                 float left = (pos_x + face_->glyph->metrics.horiBearingX/64.0f) / parent_width ;
 166                                 float right = ((pos_x + face_->glyph->metrics.horiBearingX/64.0f) + coords.width) / parent_width;
 167
 168                                 float top = (pos_y - face_->glyph->metrics.horiBearingY/64.0f) / parent_height;
 169                                 float bottom = ((pos_y - face_->glyph->metrics.horiBearingY/64.0f) + coords.height) / parent_height;
 170
 171                                 //vertex 1 top left
 172                                 result[index*16 + 0] = left;                    //vertex.x
 173                                 result[index*16 + 1] = top;                             //vertex.y
 174                                 result[index*16 + 2] = coords.left;             //texcoord.r
 175                                 result[index*16 + 3] = coords.top;              //texcoord.s
 176
 177                                 //vertex 2 top right
 178                                 result[index*16 + 4] = right;                   //vertex.x
 179                                 result[index*16 + 5] = top;                             //vertex.y
 180                                 result[index*16 + 6] = coords.right;    //texcoord.r
 181                                 result[index*16 + 7] = coords.top;              //texcoord.s
 182
 183                                 //vertex 3 bottom right
 184                                 result[index*16 + 8] = right;                   //vertex.x
 185                                 result[index*16 + 9] = bottom;                  //vertex.y
 186                                 result[index*16 + 10] = coords.right;   //texcoord.r
 187                                 result[index*16 + 11] = coords.bottom;  //texcoord.s
 188
 189                                 //vertex 4 bottom left
 190                                 result[index*16 + 12] = left;                   //vertex.x
 191                                 result[index*16 + 13] = bottom;                 //vertex.y
 192                                 result[index*16 + 14] = coords.left;    //texcoord.r
 193                                 result[index*16 + 15] = coords.bottom;  //texcoord.s
 194
 195                                 int bearingY = face_->glyph->metrics.horiBearingY >> 6;
 196
 197                                 if(bearingY > maxBearingY)
 198                                         maxBearingY = bearingY;
 199
 200                                 int protrudeUnderY = coords.height - bearingY;
 201
 202                                 if (protrudeUnderY > maxProtrudeUnderY)
 203                                         maxProtrudeUnderY = protrudeUnderY;
 204
 205                                 if (maxBearingY + maxProtrudeUnderY > maxHeight)
 206                                         maxHeight = maxBearingY + maxProtrudeUnderY;
 207
 208                                 pos_x += face_->glyph->advance.x / 64.0f;
 209                                 pos_x += tracking_;
 210                                 previous = glyph_index;
 211                         }
 212                         else
 213                         {
 214                                 //TODO: maybe we should try to load the glyph on the fly if it is missing.
 215                         }
 216                 }
 217
 218                 if(normalize_)
 219                 {
 220                         float ratio_x = parent_width/(pos_x - x);
 221                         float ratio_y = parent_height/(float)(maxHeight);
 222                         for(index = 0; index < result.size(); index += 4)
 223                         {
 224                                 result[index + 0] *= ratio_x;
 225                                 result[index + 1] *= ratio_y;
 226                         }
 227                 }
 228
 229                 if(metrics != nullptr)
 230                 {
 231                         metrics->width = (int)(pos_x - x + 0.5f);
 232                         metrics->bearingY = maxBearingY;
 233                         metrics->height = maxHeight;
 234                         metrics->protrudeUnderY = maxProtrudeUnderY;
 235                 }
 236                 return result;
 237         }
 238
 239         string_metrics measure_string(const std::wstring& str)
 240         {
 241                 string_metrics result;
 242
 243                 bool use_kerning = (face_->face_flags & FT_FACE_FLAG_KERNING) == FT_FACE_FLAG_KERNING;
 244                 int index = 0;
 245                 FT_UInt previous = 0;
 246                 float pos_x = 0;
 247 //              float pos_y = 0;
 248
 249                 auto end = str.end();
 250                 for(auto it = str.begin(); it != end; ++it, ++index)
 251                 {
 252                         auto glyph_it = glyphs_.find(*it);
 253                         if(glyph_it != glyphs_.end())
 254                         {
 255                                 const glyph_info& coords = glyph_it->second;
 256
 257                                 FT_UInt glyph_index = FT_Get_Char_Index(face_, (*it));
 258
 259                                 if(use_kerning && previous && glyph_index)
 260                                 {
 261                                         FT_Vector delta;
 262                                         FT_Get_Kerning(face_, previous, glyph_index, FT_KERNING_DEFAULT, &delta);
 263
 264                                         pos_x += delta.x / 64.0f;
 265                                 }
 266
 267                                 FT_Load_Glyph(face_, glyph_index, FT_LOAD_NO_BITMAP | FT_LOAD_FORCE_AUTOHINT | FT_LOAD_TARGET_NORMAL);
 268
 269                                 int bearingY = face_->glyph->metrics.horiBearingY >> 6;
 270                                 if(bearingY > result.bearingY)
 271                                         result.bearingY = bearingY;
 272
 273                                 int protrudeUnderY = coords.height - bearingY;
 274
 275                                 if (protrudeUnderY > result.protrudeUnderY)
 276                                         result.protrudeUnderY = protrudeUnderY;
 277
 278                                 if (result.bearingY + result.protrudeUnderY > result.height)
 279                                          result.height = result.bearingY + result.protrudeUnderY;
 280
 281                                 pos_x += face_->glyph->advance.x / 64.0f;
 282                                 previous = glyph_index;
 283                         }
 284                 }
 285
 286                 result.width = (int)(pos_x+.5f);
 287                 return result;
 288         }
 289 };
 290
 291 texture_font::texture_font(texture_atlas& atlas, const text_info& info, bool normalize_coordinates) : impl_(new impl(atlas, info, normalize_coordinates)) {}
 292 void texture_font::load_glyphs(unicode_block range, const color<float>& col) { impl_->load_glyphs(range, col); }
 293 std::vector<float> texture_font::create_vertex_stream(const std::wstring& str, int x, int y, int parent_width, int parent_height, string_metrics* metrics) { return impl_->create_vertex_stream(str, x, y, parent_width, parent_height, metrics); }
 294 string_metrics texture_font::measure_string(const std::wstring& str) { return impl_->measure_string(str); }
 295
 296 unicode_range get_range(unicode_block block)
 297 {
 298         switch(block)
 299         {
 300                 case Basic_Latin: return unicode_range(0x0000, 0x007F);
 301                 case Latin_1_Supplement: return unicode_range(0x0080, 0x00FF);
 302                 case Latin_Extended_A: return           unicode_range(0x0100, 0x017F);
 303                 case Latin_Extended_B: return           unicode_range(0x0180, 0x024F);
 304                 case IPA_Extensions: return             unicode_range(0x0250, 0x02AF);
 305                 case Spacing_Modifier_Letters: return           unicode_range(0x02B0, 0x02FF);
 306                 case Combining_Diacritical_Marks: return                unicode_range(0x0300, 0x036F);
 307                 case Greek_and_Coptic: return           unicode_range(0x0370, 0x03FF);
 308                 case Cyrillic: return           unicode_range(0x0400, 0x04FF);
 309                 case Cyrillic_Supplement: return                unicode_range(0x0500, 0x052F);
 310                 case Armenian: return           unicode_range(0x0530, 0x058F);
 311                 case Hebrew: return             unicode_range(0x0590, 0x05FF);
 312                 case Arabic: return             unicode_range(0x0600, 0x06FF);
 313                 case Syriac: return             unicode_range(0x0700, 0x074F);
 314                 case Arabic_Supplement: return          unicode_range(0x0750, 0x077F);
 315                 case Thaana: return             unicode_range(0x0780, 0x07BF);
 316                 case NKo: return                unicode_range(0x07C0, 0x07FF);
 317                 case Samaritan: return          unicode_range(0x0800, 0x083F);
 318                 case Mandaic: return            unicode_range(0x0840, 0x085F);
 319                 case Arabic_Extended_A: return          unicode_range(0x08A0, 0x08FF);
 320                 case Devanagari: return         unicode_range(0x0900, 0x097F);
 321                 case Bengali: return            unicode_range(0x0980, 0x09FF);
 322                 case Gurmukhi: return           unicode_range(0x0A00, 0x0A7F);
 323                 case Gujarati: return           unicode_range(0x0A80, 0x0AFF);
 324                 case Oriya: return              unicode_range(0x0B00, 0x0B7F);
 325                 case Tamil: return              unicode_range(0x0B80, 0x0BFF);
 326                 case Telugu: return             unicode_range(0x0C00, 0x0C7F);
 327                 case Kannada: return            unicode_range(0x0C80, 0x0CFF);
 328                 case Malayalam: return          unicode_range(0x0D00, 0x0D7F);
 329                 case Sinhala: return            unicode_range(0x0D80, 0x0DFF);
 330                 case Thai: return               unicode_range(0x0E00, 0x0E7F);
 331                 case Lao: return                unicode_range(0x0E80, 0x0EFF);
 332                 case Tibetan: return            unicode_range(0x0F00, 0x0FFF);
 333                 case Myanmar: return            unicode_range(0x1000, 0x109F);
 334                 case Georgian: return           unicode_range(0x10A0, 0x10FF);
 335                 case Hangul_Jamo: return                unicode_range(0x1100, 0x11FF);
 336                 case Ethiopic: return           unicode_range(0x1200, 0x137F);
 337                 case Ethiopic_Supplement: return                unicode_range(0x1380, 0x139F);
 338                 case Cherokee: return           unicode_range(0x13A0, 0x13FF);
 339                 case Unified_Canadian_Aboriginal_Syllabics: return              unicode_range(0x1400, 0x167F);
 340                 case Ogham: return              unicode_range(0x1680, 0x169F);
 341                 case Runic: return              unicode_range(0x16A0, 0x16FF);
 342                 case Tagalog: return            unicode_range(0x1700, 0x171F);
 343                 case Hanunoo: return            unicode_range(0x1720, 0x173F);
 344                 case Buhid: return              unicode_range(0x1740, 0x175F);
 345                 case Tagbanwa: return           unicode_range(0x1760, 0x177F);
 346                 case Khmer: return              unicode_range(0x1780, 0x17FF);
 347                 case Mongolian: return          unicode_range(0x1800, 0x18AF);
 348                 case Unified_Canadian_Aboriginal_Syllabics_Extended: return             unicode_range(0x18B0, 0x18FF);
 349                 case Limbu: return              unicode_range(0x1900, 0x194F);
 350                 case Tai_Le: return             unicode_range(0x1950, 0x197F);
 351                 case New_Tai_Lue: return                unicode_range(0x1980, 0x19DF);
 352                 case Khmer_Symbols: return              unicode_range(0x19E0, 0x19FF);
 353                 case Buginese: return           unicode_range(0x1A00, 0x1A1F);
 354                 case Tai_Tham: return           unicode_range(0x1A20, 0x1AAF);
 355                 case Balinese: return           unicode_range(0x1B00, 0x1B7F);
 356                 case Sundanese: return          unicode_range(0x1B80, 0x1BBF);
 357                 case Batak: return              unicode_range(0x1BC0, 0x1BFF);
 358                 case Lepcha: return             unicode_range(0x1C00, 0x1C4F);
 359                 case Ol_Chiki: return           unicode_range(0x1C50, 0x1C7F);
 360                 case Sundanese_Supplement: return               unicode_range(0x1CC0, 0x1CCF);
 361                 case Vedic_Extensions: return           unicode_range(0x1CD0, 0x1CFF);
 362                 case Phonetic_Extensions: return                unicode_range(0x1D00, 0x1D7F);
 363                 case Phonetic_Extensions_Supplement: return             unicode_range(0x1D80, 0x1DBF);
 364                 case Combining_Diacritical_Marks_Supplement: return             unicode_range(0x1DC0, 0x1DFF);
 365                 case Latin_Extended_Additional: return          unicode_range(0x1E00, 0x1EFF);
 366                 case Greek_Extended: return             unicode_range(0x1F00, 0x1FFF);
 367                 case General_Punctuation: return                unicode_range(0x2000, 0x206F);
 368                 case Superscripts_and_Subscripts: return                unicode_range(0x2070, 0x209F);
 369                 case Currency_Symbols: return           unicode_range(0x20A0, 0x20CF);
 370                 case Combining_Diacritical_Marks_for_Symbols: return            unicode_range(0x20D0, 0x20FF);
 371                 case Letterlike_Symbols: return         unicode_range(0x2100, 0x214F);
 372                 case Number_Forms: return               unicode_range(0x2150, 0x218F);
 373                 case Arrows: return             unicode_range(0x2190, 0x21FF);
 374                 case Mathematical_Operators: return             unicode_range(0x2200, 0x22FF);
 375                 case Miscellaneous_Technical: return            unicode_range(0x2300, 0x23FF);
 376                 case Control_Pictures: return           unicode_range(0x2400, 0x243F);
 377                 case Optical_Character_Recognition: return              unicode_range(0x2440, 0x245F);
 378                 case Enclosed_Alphanumerics: return             unicode_range(0x2460, 0x24FF);
 379                 case Box_Drawing: return                unicode_range(0x2500, 0x257F);
 380                 case Block_Elements: return             unicode_range(0x2580, 0x259F);
 381                 case Geometric_Shapes: return           unicode_range(0x25A0, 0x25FF);
 382                 case Miscellaneous_Symbols: return              unicode_range(0x2600, 0x26FF);
 383                 case Dingbats: return           unicode_range(0x2700, 0x27BF);
 384                 case Miscellaneous_Mathematical_Symbols_A: return               unicode_range(0x27C0, 0x27EF);
 385                 case Supplemental_Arrows_A: return              unicode_range(0x27F0, 0x27FF);
 386                 case Braille_Patterns: return           unicode_range(0x2800, 0x28FF);
 387                 case Supplemental_Arrows_B: return              unicode_range(0x2900, 0x297F);
 388                 case Miscellaneous_Mathematical_Symbols_B: return               unicode_range(0x2980, 0x29FF);
 389                 case Supplemental_Mathematical_Operators: return                unicode_range(0x2A00, 0x2AFF);
 390                 case Miscellaneous_Symbols_and_Arrows: return           unicode_range(0x2B00, 0x2BFF);
 391                 case Glagolitic: return         unicode_range(0x2C00, 0x2C5F);
 392                 case Latin_Extended_C: return           unicode_range(0x2C60, 0x2C7F);
 393                 case Coptic: return             unicode_range(0x2C80, 0x2CFF);
 394                 case Georgian_Supplement: return                unicode_range(0x2D00, 0x2D2F);
 395                 case Tifinagh: return           unicode_range(0x2D30, 0x2D7F);
 396                 case Ethiopic_Extended: return          unicode_range(0x2D80, 0x2DDF);
 397                 case Cyrillic_Extended_A: return                unicode_range(0x2DE0, 0x2DFF);
 398                 case Supplemental_Punctuation: return           unicode_range(0x2E00, 0x2E7F);
 399                 case CJK_Radicals_Supplement: return            unicode_range(0x2E80, 0x2EFF);
 400                 case Kangxi_Radicals: return            unicode_range(0x2F00, 0x2FDF);
 401                 case Ideographic_Description_Characters: return         unicode_range(0x2FF0, 0x2FFF);
 402                 case CJK_Symbols_and_Punctuation: return                unicode_range(0x3000, 0x303F);
 403                 case Hiragana: return           unicode_range(0x3040, 0x309F);
 404                 case Katakana: return           unicode_range(0x30A0, 0x30FF);
 405                 case Bopomofo: return           unicode_range(0x3100, 0x312F);
 406                 case Hangul_Compatibility_Jamo: return          unicode_range(0x3130, 0x318F);
 407                 case Kanbun: return             unicode_range(0x3190, 0x319F);
 408                 case Bopomofo_Extended: return          unicode_range(0x31A0, 0x31BF);
 409                 case CJK_Strokes: return                unicode_range(0x31C0, 0x31EF);
 410                 case Katakana_Phonetic_Extensions: return               unicode_range(0x31F0, 0x31FF);
 411                 case Enclosed_CJK_Letters_and_Months: return            unicode_range(0x3200, 0x32FF);
 412                 case CJK_Compatibility: return          unicode_range(0x3300, 0x33FF);
 413                 case CJK_Unified_Ideographs_Extension_A: return         unicode_range(0x3400, 0x4DBF);
 414                 case Yijing_Hexagram_Symbols: return            unicode_range(0x4DC0, 0x4DFF);
 415                 case CJK_Unified_Ideographs: return             unicode_range(0x4E00, 0x9FFF);
 416                 case Yi_Syllables: return               unicode_range(0xA000, 0xA48F);
 417                 case Yi_Radicals: return                unicode_range(0xA490, 0xA4CF);
 418                 case Lisu: return               unicode_range(0xA4D0, 0xA4FF);
 419                 case Vai: return                unicode_range(0xA500, 0xA63F);
 420                 case Cyrillic_Extended_B: return                unicode_range(0xA640, 0xA69F);
 421                 case Bamum: return              unicode_range(0xA6A0, 0xA6FF);
 422                 case Modifier_Tone_Letters: return              unicode_range(0xA700, 0xA71F);
 423                 case Latin_Extended_D: return           unicode_range(0xA720, 0xA7FF);
 424                 case Syloti_Nagri: return               unicode_range(0xA800, 0xA82F);
 425                 case Common_Indic_Number_Forms: return          unicode_range(0xA830, 0xA83F);
 426                 case Phags_pa: return           unicode_range(0xA840, 0xA87F);
 427                 case Saurashtra: return         unicode_range(0xA880, 0xA8DF);
 428                 case Devanagari_Extended: return                unicode_range(0xA8E0, 0xA8FF);
 429                 case Kayah_Li: return           unicode_range(0xA900, 0xA92F);
 430                 case Rejang: return             unicode_range(0xA930, 0xA95F);
 431                 case Hangul_Jamo_Extended_A: return             unicode_range(0xA960, 0xA97F);
 432                 case Javanese: return           unicode_range(0xA980, 0xA9DF);
 433                 case Cham: return               unicode_range(0xAA00, 0xAA5F);
 434                 case Myanmar_Extended_A: return         unicode_range(0xAA60, 0xAA7F);
 435                 case Tai_Viet: return           unicode_range(0xAA80, 0xAADF);
 436                 case Meetei_Mayek_Extensions: return            unicode_range(0xAAE0, 0xAAFF);
 437                 case Ethiopic_Extended_A: return                unicode_range(0xAB00, 0xAB2F);
 438                 case Meetei_Mayek: return               unicode_range(0xABC0, 0xABFF);
 439                 case Hangul_Syllables: return           unicode_range(0xAC00, 0xD7AF);
 440                 case Hangul_Jamo_Extended_B: return             unicode_range(0xD7B0, 0xD7FF);
 441                 case High_Surrogates: return            unicode_range(0xD800, 0xDB7F);
 442                 case High_Private_Use_Surrogates: return                unicode_range(0xDB80, 0xDBFF);
 443                 case Low_Surrogates: return             unicode_range(0xDC00, 0xDFFF);
 444                 case Private_Use_Area: return           unicode_range(0xE000, 0xF8FF);
 445                 case CJK_Compatibility_Ideographs: return               unicode_range(0xF900, 0xFAFF);
 446                 case Alphabetic_Presentation_Forms: return              unicode_range(0xFB00, 0xFB4F);
 447                 case Arabic_Presentation_Forms_A: return                unicode_range(0xFB50, 0xFDFF);
 448                 case Variation_Selectors: return                unicode_range(0xFE00, 0xFE0F);
 449                 case Vertical_Forms: return             unicode_range(0xFE10, 0xFE1F);
 450                 case Combining_Half_Marks: return               unicode_range(0xFE20, 0xFE2F);
 451                 case CJK_Compatibility_Forms: return            unicode_range(0xFE30, 0xFE4F);
 452                 case Small_Form_Variants: return                unicode_range(0xFE50, 0xFE6F);
 453                 case Arabic_Presentation_Forms_B: return                unicode_range(0xFE70, 0xFEFF);
 454                 case Halfwidth_and_Fullwidth_Forms: return              unicode_range(0xFF00, 0xFFEF);
 455                 case Specials: return           unicode_range(0xFFF0, 0xFFFF);
 456                 case Linear_B_Syllabary: return         unicode_range(0x10000, 0x1007F);
 457                 case Linear_B_Ideograms: return         unicode_range(0x10080, 0x100FF);
 458                 case Aegean_Numbers: return             unicode_range(0x10100, 0x1013F);
 459                 case Ancient_Greek_Numbers: return              unicode_range(0x10140, 0x1018F);
 460                 case Ancient_Symbols: return            unicode_range(0x10190, 0x101CF);
 461                 case Phaistos_Disc: return              unicode_range(0x101D0, 0x101FF);
 462                 case Lycian: return             unicode_range(0x10280, 0x1029F);
 463                 case Carian: return             unicode_range(0x102A0, 0x102DF);
 464                 case Old_Italic: return         unicode_range(0x10300, 0x1032F);
 465                 case Gothic: return             unicode_range(0x10330, 0x1034F);
 466                 case Ugaritic: return           unicode_range(0x10380, 0x1039F);
 467                 case Old_Persian: return                unicode_range(0x103A0, 0x103DF);
 468                 case Deseret: return            unicode_range(0x10400, 0x1044F);
 469                 case Shavian: return            unicode_range(0x10450, 0x1047F);
 470                 case Osmanya: return            unicode_range(0x10480, 0x104AF);
 471                 case Cypriot_Syllabary: return          unicode_range(0x10800, 0x1083F);
 472                 case Imperial_Aramaic: return           unicode_range(0x10840, 0x1085F);
 473                 case Phoenician: return         unicode_range(0x10900, 0x1091F);
 474                 case Lydian: return             unicode_range(0x10920, 0x1093F);
 475                 case Meroitic_Hieroglyphs: return               unicode_range(0x10980, 0x1099F);
 476                 case Meroitic_Cursive: return           unicode_range(0x109A0, 0x109FF);
 477                 case Kharoshthi: return         unicode_range(0x10A00, 0x10A5F);
 478                 case Old_South_Arabian: return          unicode_range(0x10A60, 0x10A7F);
 479                 case Avestan: return            unicode_range(0x10B00, 0x10B3F);
 480                 case Inscriptional_Parthian: return             unicode_range(0x10B40, 0x10B5F);
 481                 case Inscriptional_Pahlavi: return              unicode_range(0x10B60, 0x10B7F);
 482                 case Old_Turkic: return         unicode_range(0x10C00, 0x10C4F);
 483                 case Rumi_Numeral_Symbols: return               unicode_range(0x10E60, 0x10E7F);
 484                 case Brahmi: return             unicode_range(0x11000, 0x1107F);
 485                 case Kaithi: return             unicode_range(0x11080, 0x110CF);
 486                 case Sora_Sompeng: return               unicode_range(0x110D0, 0x110FF);
 487                 case Chakma: return             unicode_range(0x11100, 0x1114F);
 488                 case Sharada: return            unicode_range(0x11180, 0x111DF);
 489                 case Takri: return              unicode_range(0x11680, 0x116CF);
 490                 case Cuneiform: return          unicode_range(0x12000, 0x123FF);
 491                 case Cuneiform_Numbers_and_Punctuation: return          unicode_range(0x12400, 0x1247F);
 492                 case Egyptian_Hieroglyphs: return               unicode_range(0x13000, 0x1342F);
 493                 case Bamum_Supplement: return           unicode_range(0x16800, 0x16A3F);
 494                 case Miao: return               unicode_range(0x16F00, 0x16F9F);
 495                 case Kana_Supplement: return            unicode_range(0x1B000, 0x1B0FF);
 496                 case Byzantine_Musical_Symbols: return          unicode_range(0x1D000, 0x1D0FF);
 497                 case Musical_Symbols: return            unicode_range(0x1D100, 0x1D1FF);
 498                 case Ancient_Greek_Musical_Notation: return             unicode_range(0x1D200, 0x1D24F);
 499                 case Tai_Xuan_Jing_Symbols: return              unicode_range(0x1D300, 0x1D35F);
 500                 case Counting_Rod_Numerals: return              unicode_range(0x1D360, 0x1D37F);
 501                 case Mathematical_Alphanumeric_Symbols: return          unicode_range(0x1D400, 0x1D7FF);
 502                 case Arabic_Mathematical_Alphabetic_Symbols: return             unicode_range(0x1EE00, 0x1EEFF);
 503                 case Mahjong_Tiles: return              unicode_range(0x1F000, 0x1F02F);
 504                 case Domino_Tiles: return               unicode_range(0x1F030, 0x1F09F);
 505                 case Playing_Cards: return              unicode_range(0x1F0A0, 0x1F0FF);
 506                 case Enclosed_Alphanumeric_Supplement: return           unicode_range(0x1F100, 0x1F1FF);
 507                 case Enclosed_Ideographic_Supplement: return            unicode_range(0x1F200, 0x1F2FF);
 508                 case Miscellaneous_Symbols_And_Pictographs: return              unicode_range(0x1F300, 0x1F5FF);
 509                 case Emoticons: return          unicode_range(0x1F600, 0x1F64F);
 510                 case Transport_And_Map_Symbols: return          unicode_range(0x1F680, 0x1F6FF);
 511                 case Alchemical_Symbols: return         unicode_range(0x1F700, 0x1F77F);
 512                 case CJK_Unified_Ideographs_Extension_B: return         unicode_range(0x20000, 0x2A6DF);
 513                 case CJK_Unified_Ideographs_Extension_C: return         unicode_range(0x2A700, 0x2B73F);
 514                 case CJK_Unified_Ideographs_Extension_D: return         unicode_range(0x2B740, 0x2B81F);
 515                 case CJK_Compatibility_Ideographs_Supplement: return            unicode_range(0x2F800, 0x2FA1F);
 516                 case Tags: return               unicode_range(0xE0000, 0xE007F);
 517                 case Variation_Selectors_Supplement: return             unicode_range(0xE0100, 0xE01EF);
 518                 case Supplementary_Private_Use_Area_A: return           unicode_range(0xF0000, 0xFFFFF);
 519                 case Supplementary_Private_Use_Area_B: return           unicode_range(0x100000, 0x10FFFF);
 520         }
 521         return unicode_range(0,0);
 522 }
 523
 524 }}}