src/fccharset.c

   1 /*
   2  * $XFree86: xc/lib/fontconfig/src/fccharset.c,v 1.18 2002/08/22 07:36:44 keithp Exp $
   3  *
   4  * Copyright © 2001 Keith Packard, member of The XFree86 Project, Inc.
   5  *
   6  * Permission to use, copy, modify, distribute, and sell this software and its
   7  * documentation for any purpose is hereby granted without fee, provided that
   8  * the above copyright notice appear in all copies and that both that
   9  * copyright notice and this permission notice appear in supporting
  10  * documentation, and that the name of Keith Packard not be used in
  11  * advertising or publicity pertaining to distribution of the software without
  12  * specific, written prior permission.  Keith Packard makes no
  13  * representations about the suitability of this software for any purpose.  It
  14  * is provided "as is" without express or implied warranty.
  15  *
  16  * KEITH PACKARD DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
  17  * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO
  18  * EVENT SHALL KEITH PACKARD BE LIABLE FOR ANY SPECIAL, INDIRECT OR
  19  * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
  20  * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
  21  * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
  22  * PERFORMANCE OF THIS SOFTWARE.
  23  */
  24
  25 #include <stdlib.h>
  26 #include "fcint.h"
  27
  28 /* #define CHECK */
  29
  30 /* #define CHATTY */
  31
  32 FcCharSet *
  33 FcCharSetCreate (void)
  34 {
  35     FcCharSet   *fcs;
  36
  37     fcs = (FcCharSet *) malloc (sizeof (FcCharSet));
  38     if (!fcs)
  39         return 0;
  40     FcMemAlloc (FC_MEM_CHARSET, sizeof (FcCharSet));
  41     fcs->ref = 1;
  42     fcs->num = 0;
  43     fcs->leaves = 0;
  44     fcs->numbers = 0;
  45     return fcs;
  46 }
  47
  48 FcCharSet *
  49 FcCharSetNew (void);
  50
  51 FcCharSet *
  52 FcCharSetNew (void)
  53 {
  54     return FcCharSetCreate ();
  55 }
  56
  57
  58 void
  59 FcCharSetDestroy (FcCharSet *fcs)
  60 {
  61     int i;
  62     if (fcs->ref == FC_REF_CONSTANT)
  63         return;
  64     if (--fcs->ref > 0)
  65         return;
  66     for (i = 0; i < fcs->num; i++)
  67     {
  68         FcMemFree (FC_MEM_CHARLEAF, sizeof (FcCharLeaf));
  69         free (fcs->leaves[i]);
  70     }
  71     if (fcs->leaves)
  72     {
  73         FcMemFree (FC_MEM_CHARSET, fcs->num * sizeof (FcCharLeaf *));
  74         free (fcs->leaves);
  75     }
  76     if (fcs->numbers)
  77     {
  78         FcMemFree (FC_MEM_CHARSET, fcs->num * sizeof (FcChar16));
  79         free (fcs->numbers);
  80     }
  81     FcMemFree (FC_MEM_CHARSET, sizeof (FcCharSet));
  82     free (fcs);
  83 }
  84
  85 /*
  86  * Locate the leaf containing the specified char, return
  87  * its index if it exists, otherwise return negative of
  88  * the (position + 1) where it should be inserted
  89  */
  90
  91 static int
  92 FcCharSetFindLeafPos (const FcCharSet *fcs, FcChar32 ucs4)
  93 {
  94     FcChar16            *numbers = fcs->numbers;
  95     FcChar16            page;
  96     int                 low = 0;
  97     int                 high = fcs->num - 1;
  98
  99     if (!numbers)
 100         return -1;
 101     ucs4 >>= 8;
 102     while (low <= high)
 103     {
 104         int mid = (low + high) >> 1;
 105         page = numbers[mid];
 106         if (page == ucs4)
 107             return mid;
 108         if (page < ucs4)
 109             low = mid + 1;
 110         else
 111             high = mid - 1;
 112     }
 113     if (high < 0 || (high < fcs->num && numbers[high] < ucs4))
 114         high++;
 115     return -(high + 1);
 116 }
 117
 118 static FcCharLeaf *
 119 FcCharSetFindLeaf (const FcCharSet *fcs, FcChar32 ucs4)
 120 {
 121     int pos = FcCharSetFindLeafPos (fcs, ucs4);
 122     if (pos >= 0)
 123         return fcs->leaves[pos];
 124     return 0;
 125 }
 126
 127 static FcBool
 128 FcCharSetPutLeaf (FcCharSet     *fcs,
 129                   FcChar32      ucs4,
 130                   FcCharLeaf    *leaf,
 131                   int           pos)
 132 {
 133     FcCharLeaf  **leaves;
 134     FcChar16    *numbers;
 135
 136     ucs4 >>= 8;
 137     if (ucs4 >= 0x10000)
 138         return FcFalse;
 139     if (!fcs->leaves)
 140         leaves = malloc (sizeof (FcCharLeaf *));
 141     else
 142         leaves = realloc (fcs->leaves, (fcs->num + 1) * sizeof (FcCharLeaf *));
 143     if (!leaves)
 144         return FcFalse;
 145     if (fcs->num)
 146         FcMemFree (FC_MEM_CHARSET, fcs->num * sizeof (FcCharLeaf *));
 147     FcMemAlloc (FC_MEM_CHARSET, (fcs->num + 1) * sizeof (FcCharLeaf *));
 148     fcs->leaves = leaves;
 149     if (!fcs->numbers)
 150         numbers = malloc (sizeof (FcChar16));
 151     else
 152         numbers = realloc (fcs->numbers, (fcs->num + 1) * sizeof (FcChar16));
 153     if (!numbers)
 154         return FcFalse;
 155     if (fcs->num)
 156         FcMemFree (FC_MEM_CHARSET, fcs->num * sizeof (FcChar16));
 157     FcMemAlloc (FC_MEM_CHARSET, (fcs->num + 1) * sizeof (FcChar16));
 158     fcs->numbers = numbers;
 159
 160     memmove (fcs->leaves + pos + 1, fcs->leaves + pos,
 161              (fcs->num - pos) * sizeof (FcCharLeaf *));
 162     memmove (fcs->numbers + pos + 1, fcs->numbers + pos,
 163              (fcs->num - pos) * sizeof (FcChar16));
 164     fcs->numbers[pos] = (FcChar16) ucs4;
 165     fcs->leaves[pos] = leaf;
 166     fcs->num++;
 167     return FcTrue;
 168 }
 169
 170 /*
 171  * Locate the leaf containing the specified char, creating it
 172  * if desired
 173  */
 174
 175 static FcCharLeaf *
 176 FcCharSetFindLeafCreate (FcCharSet *fcs, FcChar32 ucs4)
 177 {
 178     int                 pos;
 179     FcCharLeaf          *leaf;
 180
 181     pos = FcCharSetFindLeafPos (fcs, ucs4);
 182     if (pos >= 0)
 183         return fcs->leaves[pos];
 184
 185     leaf = calloc (1, sizeof (FcCharLeaf));
 186     if (!leaf)
 187         return 0;
 188
 189     pos = -pos - 1;
 190     if (!FcCharSetPutLeaf (fcs, ucs4, leaf, pos))
 191     {
 192         free (leaf);
 193         return 0;
 194     }
 195     FcMemAlloc (FC_MEM_CHARLEAF, sizeof (FcCharLeaf));
 196     return leaf;
 197 }
 198
 199 static FcBool
 200 FcCharSetInsertLeaf (FcCharSet *fcs, FcChar32 ucs4, FcCharLeaf *leaf)
 201 {
 202     int             pos;
 203
 204     pos = FcCharSetFindLeafPos (fcs, ucs4);
 205     if (pos >= 0)
 206     {
 207         FcMemFree (FC_MEM_CHARLEAF, sizeof (FcCharLeaf));
 208         free (fcs->leaves[pos]);
 209         fcs->leaves[pos] = leaf;
 210         return FcTrue;
 211     }
 212     pos = -pos - 1;
 213     return FcCharSetPutLeaf (fcs, ucs4, leaf, pos);
 214 }
 215
 216 FcBool
 217 FcCharSetAddChar (FcCharSet *fcs, FcChar32 ucs4)
 218 {
 219     FcCharLeaf  *leaf;
 220     FcChar32    *b;
 221
 222     if (fcs->ref == FC_REF_CONSTANT)
 223         return FcFalse;
 224     leaf = FcCharSetFindLeafCreate (fcs, ucs4);
 225     if (!leaf)
 226         return FcFalse;
 227     b = &leaf->map[(ucs4 & 0xff) >> 5];
 228     *b |= (1 << (ucs4 & 0x1f));
 229     return FcTrue;
 230 }
 231
 232 /*
 233  * An iterator for the leaves of a charset
 234  */
 235
 236 typedef struct _fcCharSetIter {
 237     FcCharLeaf      *leaf;
 238     FcChar32        ucs4;
 239     int             pos;
 240 } FcCharSetIter;
 241
 242 /*
 243  * Set iter->leaf to the leaf containing iter->ucs4 or higher
 244  */
 245
 246 static void
 247 FcCharSetIterSet (const FcCharSet *fcs, FcCharSetIter *iter)
 248 {
 249     int             pos = FcCharSetFindLeafPos (fcs, iter->ucs4);
 250
 251     if (pos < 0)
 252     {
 253         pos = -pos - 1;
 254         if (pos == fcs->num)
 255         {
 256             iter->ucs4 = ~0;
 257             iter->leaf = 0;
 258             return;
 259         }
 260         iter->ucs4 = (FcChar32) fcs->numbers[pos] << 8;
 261     }
 262     iter->leaf = fcs->leaves[pos];
 263     iter->pos = pos;
 264 #ifdef CHATTY
 265     printf ("set %08x: %08x\n", iter->ucs4, (FcChar32) iter->leaf);
 266 #endif
 267 }
 268
 269 static void
 270 FcCharSetIterNext (const FcCharSet *fcs, FcCharSetIter *iter)
 271 {
 272     int pos = iter->pos + 1;
 273     if (pos >= fcs->num)
 274     {
 275         iter->ucs4 = ~0;
 276         iter->leaf = 0;
 277     }
 278     else
 279     {
 280         iter->ucs4 = (FcChar32) fcs->numbers[pos] << 8;
 281         iter->leaf = fcs->leaves[pos];
 282         iter->pos = pos;
 283     }
 284 }
 285
 286 #ifdef CHATTY
 287 static void
 288 FcCharSetDump (const FcCharSet *fcs)
 289 {
 290     int         pos;
 291
 292     printf ("fcs %08x:\n", (FcChar32) fcs);
 293     for (pos = 0; pos < fcs->num; pos++)
 294     {
 295         FcCharLeaf      *leaf = fcs->leaves[pos];
 296         FcChar32        ucs4 = (FcChar32) fcs->numbers[pos] << 8;
 297
 298         printf ("    %08x: %08x\n", ucs4, (FcChar32) leaf);
 299     }
 300 }
 301 #endif
 302
 303 static void
 304 FcCharSetIterStart (const FcCharSet *fcs, FcCharSetIter *iter)
 305 {
 306 #ifdef CHATTY
 307     FcCharSetDump (fcs);
 308 #endif
 309     iter->ucs4 = 0;
 310     FcCharSetIterSet (fcs, iter);
 311 }
 312
 313 FcCharSet *
 314 FcCharSetCopy (FcCharSet *src)
 315 {
 316     if (src->ref != FC_REF_CONSTANT)
 317         src->ref++;
 318     return src;
 319 }
 320
 321 FcBool
 322 FcCharSetEqual (const FcCharSet *a, const FcCharSet *b)
 323 {
 324     FcCharSetIter   ai, bi;
 325     int             i;
 326
 327     if (a == b)
 328         return FcTrue;
 329     for (FcCharSetIterStart (a, &ai), FcCharSetIterStart (b, &bi);
 330          ai.leaf && bi.leaf;
 331          FcCharSetIterNext (a, &ai), FcCharSetIterNext (b, &bi))
 332     {
 333         if (ai.ucs4 != bi.ucs4)
 334             return FcFalse;
 335         for (i = 0; i < 256/32; i++)
 336             if (ai.leaf->map[i] != bi.leaf->map[i])
 337                 return FcFalse;
 338     }
 339     return ai.leaf == bi.leaf;
 340 }
 341
 342 static FcBool
 343 FcCharSetAddLeaf (FcCharSet     *fcs,
 344                   FcChar32      ucs4,
 345                   FcCharLeaf    *leaf)
 346 {
 347     FcCharLeaf   *new = FcCharSetFindLeafCreate (fcs, ucs4);
 348     if (!new)
 349         return FcFalse;
 350     *new = *leaf;
 351     return FcTrue;
 352 }
 353
 354 static FcCharSet *
 355 FcCharSetOperate (const FcCharSet   *a,
 356                   const FcCharSet   *b,
 357                   FcBool            (*overlap) (FcCharLeaf          *result,
 358                                                 const FcCharLeaf    *al,
 359                                                 const FcCharLeaf    *bl),
 360                   FcBool        aonly,
 361                   FcBool        bonly)
 362 {
 363     FcCharSet       *fcs;
 364     FcCharSetIter   ai, bi;
 365
 366     fcs = FcCharSetCreate ();
 367     if (!fcs)
 368         goto bail0;
 369     FcCharSetIterStart (a, &ai);
 370     FcCharSetIterStart (b, &bi);
 371     while ((ai.leaf || (bonly && bi.leaf)) && (bi.leaf || (aonly && ai.leaf)))
 372     {
 373         if (ai.ucs4 < bi.ucs4)
 374         {
 375             if (aonly)
 376             {
 377                 if (!FcCharSetAddLeaf (fcs, ai.ucs4, ai.leaf))
 378                     goto bail1;
 379                 FcCharSetIterNext (a, &ai);
 380             }
 381             else
 382             {
 383                 ai.ucs4 = bi.ucs4;
 384                 FcCharSetIterSet (a, &ai);
 385             }
 386         }
 387         else if (bi.ucs4 < ai.ucs4 )
 388         {
 389             if (bonly)
 390             {
 391                 if (!FcCharSetAddLeaf (fcs, bi.ucs4, bi.leaf))
 392                     goto bail1;
 393                 FcCharSetIterNext (b, &bi);
 394             }
 395             else
 396             {
 397                 bi.ucs4 = ai.ucs4;
 398                 FcCharSetIterSet (b, &bi);
 399             }
 400         }
 401         else
 402         {
 403             FcCharLeaf  leaf;
 404
 405             if ((*overlap) (&leaf, ai.leaf, bi.leaf))
 406             {
 407                 if (!FcCharSetAddLeaf (fcs, ai.ucs4, &leaf))
 408                     goto bail1;
 409             }
 410             FcCharSetIterNext (a, &ai);
 411             FcCharSetIterNext (b, &bi);
 412         }
 413     }
 414     return fcs;
 415 bail1:
 416     FcCharSetDestroy (fcs);
 417 bail0:
 418     return 0;
 419 }
 420
 421 static FcBool
 422 FcCharSetIntersectLeaf (FcCharLeaf *result,
 423                         const FcCharLeaf *al,
 424                         const FcCharLeaf *bl)
 425 {
 426     int     i;
 427     FcBool  nonempty = FcFalse;
 428
 429     for (i = 0; i < 256/32; i++)
 430         if ((result->map[i] = al->map[i] & bl->map[i]))
 431             nonempty = FcTrue;
 432     return nonempty;
 433 }
 434
 435 FcCharSet *
 436 FcCharSetIntersect (const FcCharSet *a, const FcCharSet *b)
 437 {
 438     return FcCharSetOperate (a, b, FcCharSetIntersectLeaf, FcFalse, FcFalse);
 439 }
 440
 441 static FcBool
 442 FcCharSetUnionLeaf (FcCharLeaf *result,
 443                     const FcCharLeaf *al,
 444                     const FcCharLeaf *bl)
 445 {
 446     int i;
 447
 448     for (i = 0; i < 256/32; i++)
 449         result->map[i] = al->map[i] | bl->map[i];
 450     return FcTrue;
 451 }
 452
 453 FcCharSet *
 454 FcCharSetUnion (const FcCharSet *a, const FcCharSet *b)
 455 {
 456     return FcCharSetOperate (a, b, FcCharSetUnionLeaf, FcTrue, FcTrue);
 457 }
 458
 459 static FcBool
 460 FcCharSetSubtractLeaf (FcCharLeaf *result,
 461                        const FcCharLeaf *al,
 462                        const FcCharLeaf *bl)
 463 {
 464     int     i;
 465     FcBool  nonempty = FcFalse;
 466
 467     for (i = 0; i < 256/32; i++)
 468         if ((result->map[i] = al->map[i] & ~bl->map[i]))
 469             nonempty = FcTrue;
 470     return nonempty;
 471 }
 472
 473 FcCharSet *
 474 FcCharSetSubtract (const FcCharSet *a, const FcCharSet *b)
 475 {
 476     return FcCharSetOperate (a, b, FcCharSetSubtractLeaf, FcTrue, FcFalse);
 477 }
 478
 479 FcBool
 480 FcCharSetHasChar (const FcCharSet *fcs, FcChar32 ucs4)
 481 {
 482     FcCharLeaf  *leaf = FcCharSetFindLeaf (fcs, ucs4);
 483     if (!leaf)
 484         return FcFalse;
 485     return (leaf->map[(ucs4 & 0xff) >> 5] & (1 << (ucs4 & 0x1f))) != 0;
 486 }
 487
 488 static FcChar32
 489 FcCharSetPopCount (FcChar32 c1)
 490 {
 491     /* hackmem 169 */
 492     FcChar32    c2 = (c1 >> 1) & 033333333333;
 493     c2 = c1 - c2 - ((c2 >> 1) & 033333333333);
 494     return (((c2 + (c2 >> 3)) & 030707070707) % 077);
 495 }
 496
 497 FcChar32
 498 FcCharSetIntersectCount (const FcCharSet *a, const FcCharSet *b)
 499 {
 500     FcCharSetIter   ai, bi;
 501     FcChar32        count = 0;
 502
 503     FcCharSetIterStart (a, &ai);
 504     FcCharSetIterStart (b, &bi);
 505     while (ai.leaf && bi.leaf)
 506     {
 507         if (ai.ucs4 == bi.ucs4)
 508         {
 509             FcChar32    *am = ai.leaf->map;
 510             FcChar32    *bm = bi.leaf->map;
 511             int         i = 256/32;
 512             while (i--)
 513                 count += FcCharSetPopCount (*am++ & *bm++);
 514             FcCharSetIterNext (a, &ai);
 515         }
 516         else if (ai.ucs4 < bi.ucs4)
 517         {
 518             ai.ucs4 = bi.ucs4;
 519             FcCharSetIterSet (a, &ai);
 520         }
 521         if (bi.ucs4 < ai.ucs4)
 522         {
 523             bi.ucs4 = ai.ucs4;
 524             FcCharSetIterSet (b, &bi);
 525         }
 526     }
 527     return count;
 528 }
 529
 530 FcChar32
 531 FcCharSetCount (const FcCharSet *a)
 532 {
 533     FcCharSetIter   ai;
 534     FcChar32        count = 0;
 535
 536     for (FcCharSetIterStart (a, &ai); ai.leaf; FcCharSetIterNext (a, &ai))
 537     {
 538         int                 i = 256/32;
 539         FcChar32            *am = ai.leaf->map;
 540
 541         while (i--)
 542             count += FcCharSetPopCount (*am++);
 543     }
 544     return count;
 545 }
 546
 547 FcChar32
 548 FcCharSetSubtractCount (const FcCharSet *a, const FcCharSet *b)
 549 {
 550     FcCharSetIter   ai, bi;
 551     FcChar32        count = 0;
 552
 553     FcCharSetIterStart (a, &ai);
 554     FcCharSetIterStart (b, &bi);
 555     while (ai.leaf)
 556     {
 557         if (ai.ucs4 <= bi.ucs4)
 558         {
 559             FcChar32    *am = ai.leaf->map;
 560             int         i = 256/32;
 561             if (ai.ucs4 == bi.ucs4)
 562             {
 563                 FcChar32        *bm = bi.leaf->map;;
 564                 while (i--)
 565                     count += FcCharSetPopCount (*am++ & ~*bm++);
 566             }
 567             else
 568             {
 569                 while (i--)
 570                     count += FcCharSetPopCount (*am++);
 571             }
 572             FcCharSetIterNext (a, &ai);
 573         }
 574         else if (bi.leaf)
 575         {
 576             bi.ucs4 = ai.ucs4;
 577             FcCharSetIterSet (b, &bi);
 578         }
 579     }
 580     return count;
 581 }
 582
 583 /*
 584  * return FcTrue iff a is a subset of b
 585  */
 586 FcBool
 587 FcCharSetIsSubset (const FcCharSet *a, const FcCharSet *b)
 588 {
 589     int         ai, bi;
 590     FcChar16    an, bn;
 591
 592     if (a == b) return FcTrue;
 593     bi = 0;
 594     ai = 0;
 595     while (ai < a->num && bi < b->num)
 596     {
 597         an = a->numbers[ai];
 598         bn = b->numbers[bi];
 599         if (an == bn)
 600         {
 601             FcChar32    *am = a->leaves[ai]->map;
 602             FcChar32    *bm = b->leaves[bi]->map;
 603
 604             if (am != bm)
 605             {
 606                 int     i = 256/32;
 607                 while (i--)
 608                     if (*am++ & ~*bm++)
 609                         return FcFalse;
 610             }
 611             ai++;
 612             bi++;
 613         }
 614         else if (an < bn)
 615             return FcFalse;
 616         else
 617         {
 618             int     low = bi + 1;
 619             int     high = b->num - 1;
 620
 621             while (low <= high)
 622             {
 623                 int mid = (low + high) >> 1;
 624                 bn = b->numbers[mid];
 625                 if (bn == an)
 626                 {
 627                     high = mid;
 628                     break;
 629                 }
 630                 if (bn < an)
 631                     low = mid + 1;
 632                 else
 633                     high = mid - 1;
 634             }
 635             bi = high;
 636             while (bi < b->num && b->numbers[bi] < an)
 637                 bi++;
 638         }
 639     }
 640     return FcTrue;
 641 }
 642
 643 /*
 644  * These two functions efficiently walk the entire charmap for
 645  * other software (like pango) that want their own copy
 646  */
 647
 648 FcChar32
 649 FcCharSetNextPage (const FcCharSet  *a,
 650                    FcChar32         map[FC_CHARSET_MAP_SIZE],
 651                    FcChar32         *next)
 652 {
 653     FcCharSetIter   ai;
 654     FcChar32        page;
 655
 656     ai.ucs4 = *next;
 657     FcCharSetIterSet (a, &ai);
 658     if (!ai.leaf)
 659         return FC_CHARSET_DONE;
 660
 661     /*
 662      * Save current information
 663      */
 664     page = ai.ucs4;
 665     memcpy (map, ai.leaf->map, sizeof (ai.leaf->map));
 666     /*
 667      * Step to next page
 668      */
 669     FcCharSetIterNext (a, &ai);
 670     *next = ai.ucs4;
 671
 672     return page;
 673 }
 674
 675 FcChar32
 676 FcCharSetFirstPage (const FcCharSet *a,
 677                     FcChar32        map[FC_CHARSET_MAP_SIZE],
 678                     FcChar32        *next)
 679 {
 680     *next = 0;
 681     return FcCharSetNextPage (a, map, next);
 682 }
 683
 684 /*
 685  * old coverage API, rather hard to use correctly
 686  */
 687 FcChar32
 688 FcCharSetCoverage (const FcCharSet *a, FcChar32 page, FcChar32 *result);
 689
 690 FcChar32
 691 FcCharSetCoverage (const FcCharSet *a, FcChar32 page, FcChar32 *result)
 692 {
 693     FcCharSetIter   ai;
 694
 695     ai.ucs4 = page;
 696     FcCharSetIterSet (a, &ai);
 697     if (!ai.leaf)
 698     {
 699         memset (result, '\0', 256 / 8);
 700         page = 0;
 701     }
 702     else
 703     {
 704         memcpy (result, ai.leaf->map, sizeof (ai.leaf->map));
 705         FcCharSetIterNext (a, &ai);
 706         page = ai.ucs4;
 707     }
 708     return page;
 709 }
 710
 711 /*
 712  * ASCII representation of charsets.
 713  *
 714  * Each leaf is represented as 9 32-bit values, the code of the first character followed
 715  * by 8 32 bit values for the leaf itself.  Each value is encoded as 5 ASCII characters,
 716  * only 85 different values are used to avoid control characters as well as the other
 717  * characters used to encode font names.  85**5 > 2^32 so things work out, but
 718  * it's not exactly human readable output.  As a special case, 0 is encoded as a space
 719  */
 720
 721 static const unsigned char      charToValue[256] = {
 722     /*     "" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 723     /*   "\b" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 724     /* "\020" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 725     /* "\030" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 726     /*    " " */ 0xff,  0x00,  0xff,  0x01,  0x02,  0x03,  0x04,  0xff,
 727     /*    "(" */ 0x05,  0x06,  0x07,  0x08,  0xff,  0xff,  0x09,  0x0a,
 728     /*    "0" */ 0x0b,  0x0c,  0x0d,  0x0e,  0x0f,  0x10,  0x11,  0x12,
 729     /*    "8" */ 0x13,  0x14,  0xff,  0x15,  0x16,  0xff,  0x17,  0x18,
 730     /*    "@" */ 0x19,  0x1a,  0x1b,  0x1c,  0x1d,  0x1e,  0x1f,  0x20,
 731     /*    "H" */ 0x21,  0x22,  0x23,  0x24,  0x25,  0x26,  0x27,  0x28,
 732     /*    "P" */ 0x29,  0x2a,  0x2b,  0x2c,  0x2d,  0x2e,  0x2f,  0x30,
 733     /*    "X" */ 0x31,  0x32,  0x33,  0x34,  0xff,  0x35,  0x36,  0xff,
 734     /*    "`" */ 0xff,  0x37,  0x38,  0x39,  0x3a,  0x3b,  0x3c,  0x3d,
 735     /*    "h" */ 0x3e,  0x3f,  0x40,  0x41,  0x42,  0x43,  0x44,  0x45,
 736     /*    "p" */ 0x46,  0x47,  0x48,  0x49,  0x4a,  0x4b,  0x4c,  0x4d,
 737     /*    "x" */ 0x4e,  0x4f,  0x50,  0x51,  0x52,  0x53,  0x54,  0xff,
 738     /* "\200" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 739     /* "\210" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 740     /* "\220" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 741     /* "\230" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 742     /* "\240" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 743     /* "\250" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 744     /* "\260" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 745     /* "\270" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 746     /* "\300" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 747     /* "\310" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 748     /* "\320" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 749     /* "\330" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 750     /* "\340" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 751     /* "\350" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 752     /* "\360" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 753     /* "\370" */ 0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,  0xff,
 754 };
 755
 756 static const FcChar8 valueToChar[0x55] = {
 757     /* 0x00 */ '!', '#', '$', '%', '&', '(', ')', '*',
 758     /* 0x08 */ '+', '.', '/', '0', '1', '2', '3', '4',
 759     /* 0x10 */ '5', '6', '7', '8', '9', ';', '<', '>',
 760     /* 0x18 */ '?', '@', 'A', 'B', 'C', 'D', 'E', 'F',
 761     /* 0x20 */ 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N',
 762     /* 0x28 */ 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V',
 763     /* 0x30 */ 'W', 'X', 'Y', 'Z', '[', ']', '^', 'a',
 764     /* 0x38 */ 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i',
 765     /* 0x40 */ 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q',
 766     /* 0x48 */ 'r', 's', 't', 'u', 'v', 'w', 'x', 'y',
 767     /* 0x50 */ 'z', '{', '|', '}', '~',
 768 };
 769
 770 static FcChar8 *
 771 FcCharSetParseValue (FcChar8 *string, FcChar32 *value)
 772 {
 773     int         i;
 774     FcChar32    v;
 775     FcChar32    c;
 776
 777     if (*string == ' ')
 778     {
 779         v = 0;
 780         string++;
 781     }
 782     else
 783     {
 784         v = 0;
 785         for (i = 0; i < 5; i++)
 786         {
 787             if (!(c = (FcChar32) (unsigned char) *string++))
 788                 return 0;
 789             c = charToValue[c];
 790             if (c == 0xff)
 791                 return 0;
 792             v = v * 85 + c;
 793         }
 794     }
 795     *value = v;
 796     return string;
 797 }
 798
 799 static FcBool
 800 FcCharSetUnparseValue (FcStrBuf *buf, FcChar32 value)
 801 {
 802     int     i;
 803     if (value == 0)
 804     {
 805         return FcStrBufChar (buf, ' ');
 806     }
 807     else
 808     {
 809         FcChar8 string[6];
 810         FcChar8 *s = string + 5;
 811         string[5] = '\0';
 812         for (i = 0; i < 5; i++)
 813         {
 814             *--s = valueToChar[value % 85];
 815             value /= 85;
 816         }
 817         for (i = 0; i < 5; i++)
 818             if (!FcStrBufChar (buf, *s++))
 819                 return FcFalse;
 820     }
 821     return FcTrue;
 822 }
 823
 824 typedef struct _FcCharLeafEnt FcCharLeafEnt;
 825
 826 struct _FcCharLeafEnt {
 827     FcCharLeafEnt   *next;
 828     FcChar32        hash;
 829     FcCharLeaf      leaf;
 830 };
 831
 832 #define FC_CHAR_LEAF_BLOCK      (4096 / sizeof (FcCharLeafEnt))
 833
 834 static FcCharLeafEnt *
 835 FcCharLeafEntCreate (void)
 836 {
 837     static FcCharLeafEnt    *block;
 838     static int              remain;
 839
 840     if (!remain)
 841     {
 842         block = malloc (FC_CHAR_LEAF_BLOCK * sizeof (FcCharLeafEnt));
 843         if (!block)
 844             return 0;
 845         FcMemAlloc (FC_MEM_CHARLEAF, FC_CHAR_LEAF_BLOCK * sizeof (FcCharLeafEnt));
 846         remain = FC_CHAR_LEAF_BLOCK;
 847     }
 848     remain--;
 849     return block++;
 850 }
 851
 852 #define FC_CHAR_LEAF_HASH_SIZE  257
 853
 854 static FcChar32
 855 FcCharLeafHash (FcCharLeaf *leaf)
 856 {
 857     FcChar32    hash = 0;
 858     int         i;
 859
 860     for (i = 0; i < 256/32; i++)
 861         hash = ((hash << 1) | (hash >> 31)) ^ leaf->map[i];
 862     return hash;
 863 }
 864
 865 static int      FcCharLeafTotal;
 866 static int      FcCharLeafUsed;
 867
 868 static FcCharLeaf *
 869 FcCharSetFreezeLeaf (FcCharLeaf *leaf)
 870 {
 871     static FcCharLeafEnt        *hashTable[FC_CHAR_LEAF_HASH_SIZE];
 872     FcChar32                    hash = FcCharLeafHash (leaf);
 873     FcCharLeafEnt               **bucket = &hashTable[hash % FC_CHAR_LEAF_HASH_SIZE];
 874     FcCharLeafEnt               *ent;
 875
 876     FcCharLeafTotal++;
 877     for (ent = *bucket; ent; ent = ent->next)
 878     {
 879         if (ent->hash == hash && !memcmp (&ent->leaf, leaf, sizeof (FcCharLeaf)))
 880             return &ent->leaf;
 881     }
 882
 883     ent = FcCharLeafEntCreate();
 884     if (!ent)
 885         return 0;
 886     FcCharLeafUsed++;
 887     ent->leaf = *leaf;
 888     ent->hash = hash;
 889     ent->next = *bucket;
 890     *bucket = ent;
 891     return &ent->leaf;
 892 }
 893
 894 typedef struct _FcCharSetEnt FcCharSetEnt;
 895
 896 struct _FcCharSetEnt {
 897     FcCharSetEnt        *next;
 898     FcChar32            hash;
 899     FcCharSet           set;
 900 };
 901
 902 #define FC_CHAR_SET_HASH_SIZE    67
 903
 904 static FcChar32
 905 FcCharSetHash (FcCharSet *fcs)
 906 {
 907     FcChar32    hash = 0;
 908     FcChar32    *p;
 909     int         i;
 910
 911     /* hash in leaves */
 912     p = (FcChar32 *) fcs->leaves;
 913     for (i = 0; i < fcs->num * sizeof (FcCharLeaf *) / sizeof (FcChar32); i++)
 914         hash = ((hash << 1) | (hash >> 31)) ^ *p++;
 915     /* hash in numbers */
 916     for (i = 0; i < fcs->num; i++)
 917         hash = ((hash << 1) | (hash >> 31)) ^ fcs->numbers[i];
 918     return hash;
 919 }
 920
 921 static int      FcCharSetTotal;
 922 static int      FcCharSetUsed;
 923 static int      FcCharSetTotalEnts, FcCharSetUsedEnts;
 924
 925 static FcCharSet *
 926 FcCharSetFreezeBase (FcCharSet *fcs)
 927 {
 928     static FcCharSetEnt *hashTable[FC_CHAR_SET_HASH_SIZE];
 929     FcChar32            hash = FcCharSetHash (fcs);
 930     FcCharSetEnt        **bucket = &hashTable[hash % FC_CHAR_SET_HASH_SIZE];
 931     FcCharSetEnt        *ent;
 932     int                 size;
 933
 934     FcCharSetTotal++;
 935     FcCharSetTotalEnts += fcs->num;
 936     for (ent = *bucket; ent; ent = ent->next)
 937     {
 938         if (ent->hash == hash &&
 939             ent->set.num == fcs->num &&
 940             !memcmp (ent->set.leaves, fcs->leaves,
 941                      fcs->num * sizeof (FcCharLeaf *)) &&
 942             !memcmp (ent->set.numbers, fcs->numbers,
 943                      fcs->num * sizeof (FcChar16)))
 944         {
 945             return &ent->set;
 946         }
 947     }
 948
 949     size = (sizeof (FcCharSetEnt) +
 950             fcs->num * sizeof (FcCharLeaf *) +
 951             fcs->num * sizeof (FcChar16));
 952     ent = malloc (size);
 953     if (!ent)
 954         return 0;
 955     FcMemAlloc (FC_MEM_CHARSET, size);
 956     FcCharSetUsed++;
 957     FcCharSetUsedEnts += fcs->num;
 958
 959     ent->set.ref = FC_REF_CONSTANT;
 960     ent->set.num = fcs->num;
 961     if (fcs->num)
 962     {
 963         ent->set.leaves = (FcCharLeaf **) (ent + 1);
 964         ent->set.numbers = (FcChar16 *) (ent->set.leaves + fcs->num);
 965         memcpy (ent->set.leaves, fcs->leaves, fcs->num * sizeof (FcCharLeaf *));
 966         memcpy (ent->set.numbers, fcs->numbers, fcs->num * sizeof (FcChar16));
 967     }
 968     else
 969     {
 970         ent->set.leaves = 0;
 971         ent->set.numbers = 0;
 972     }
 973
 974     ent->hash = hash;
 975     ent->next = *bucket;
 976     *bucket = ent;
 977     return &ent->set;
 978 }
 979
 980 FcCharSet *
 981 FcCharSetFreeze (FcCharSet *fcs)
 982 {
 983     FcCharSet   *b;
 984     FcCharSet   *n = 0;
 985     FcCharLeaf  *l;
 986     int         i;
 987
 988     b = FcCharSetCreate ();
 989     if (!b)
 990         goto bail0;
 991     for (i = 0; i < fcs->num; i++)
 992     {
 993         l = FcCharSetFreezeLeaf (fcs->leaves[i]);
 994         if (!l)
 995             goto bail1;
 996         if (!FcCharSetInsertLeaf (b, fcs->numbers[i] << 8, l))
 997             goto bail1;
 998     }
 999     n = FcCharSetFreezeBase (b);
1000 bail1:
1001     if (b->leaves)
1002     {
1003         FcMemFree (FC_MEM_CHARSET, b->num * sizeof (FcCharLeaf *));
1004         free (b->leaves);
1005     }
1006     if (b->numbers)
1007     {
1008         FcMemFree (FC_MEM_CHARSET, b->num * sizeof (FcChar16));
1009         free (b->numbers);
1010     }
1011     FcMemFree (FC_MEM_CHARSET, sizeof (FcCharSet));
1012     free (b);
1013 bail0:
1014     return n;
1015 }
1016
1017 FcCharSet *
1018 FcNameParseCharSet (FcChar8 *string)
1019 {
1020     FcCharSet   *c, *n = 0;
1021     FcChar32    ucs4;
1022     FcCharLeaf  *leaf;
1023     FcCharLeaf  temp;
1024     FcChar32    bits;
1025     int         i;
1026
1027     c = FcCharSetCreate ();
1028     if (!c)
1029         goto bail0;
1030     while (*string)
1031     {
1032         string = FcCharSetParseValue (string, &ucs4);
1033         if (!string)
1034             goto bail1;
1035         bits = 0;
1036         for (i = 0; i < 256/32; i++)
1037         {
1038             string = FcCharSetParseValue (string, &temp.map[i]);
1039             if (!string)
1040                 goto bail1;
1041             bits |= temp.map[i];
1042         }
1043         if (bits)
1044         {
1045             leaf = FcCharSetFreezeLeaf (&temp);
1046             if (!leaf)
1047                 goto bail1;
1048             if (!FcCharSetInsertLeaf (c, ucs4, leaf))
1049                 goto bail1;
1050         }
1051     }
1052 #ifdef CHATTY
1053     printf ("          %8s %8s %8s %8s\n", "total", "totalmem", "new", "newmem");
1054     printf ("Leaves:   %8d %8d %8d %8d\n",
1055             FcCharLeafTotal, sizeof (FcCharLeaf) * FcCharLeafTotal,
1056             FcCharLeafUsed, sizeof (FcCharLeaf) * FcCharLeafUsed);
1057     printf ("Charsets: %8d %8d %8d %8d\n",
1058             FcCharSetTotal, sizeof (FcCharSet) * FcCharSetTotal,
1059             FcCharSetUsed, sizeof (FcCharSet) * FcCharSetUsed);
1060     printf ("Tables:   %8d %8d %8d %8d\n",
1061             FcCharSetTotalEnts, FcCharSetTotalEnts * (sizeof (FcCharLeaf *) + sizeof (FcChar16)),
1062             FcCharSetUsedEnts, FcCharSetUsedEnts * (sizeof (FcCharLeaf *) + sizeof (FcChar16)));
1063     printf ("Total:    %8s %8d %8s %8d\n",
1064             "",
1065             sizeof (FcCharLeaf) * FcCharLeafTotal +
1066             sizeof (FcCharSet) * FcCharSetTotal +
1067             FcCharSetTotalEnts * (sizeof (FcCharLeaf *) + sizeof (FcChar16)),
1068             "",
1069             sizeof (FcCharLeaf) * FcCharLeafUsed +
1070             sizeof (FcCharSet) * FcCharSetUsed +
1071             FcCharSetUsedEnts * (sizeof (FcCharLeaf *) + sizeof (FcChar16)));
1072 #endif
1073     n = FcCharSetFreezeBase (c);
1074 bail1:
1075     if (c->leaves)
1076     {
1077         FcMemFree (FC_MEM_CHARSET, c->num * sizeof (FcCharLeaf *));
1078         free (c->leaves);
1079     }
1080     if (c->numbers)
1081     {
1082         FcMemFree (FC_MEM_CHARSET, c->num * sizeof (FcChar16));
1083         free (c->numbers);
1084     }
1085     FcMemFree (FC_MEM_CHARSET, sizeof (FcCharSet));
1086     free (c);
1087 bail0:
1088     return n;
1089 }
1090
1091 FcBool
1092 FcNameUnparseCharSet (FcStrBuf *buf, const FcCharSet *c)
1093 {
1094     FcCharSetIter   ci;
1095     int             i;
1096 #ifdef CHECK
1097     int             len = buf->len;
1098 #endif
1099
1100     for (FcCharSetIterStart (c, &ci);
1101          ci.leaf;
1102          FcCharSetIterNext (c, &ci))
1103     {
1104         if (!FcCharSetUnparseValue (buf, ci.ucs4))
1105             return FcFalse;
1106         for (i = 0; i < 256/32; i++)
1107             if (!FcCharSetUnparseValue (buf, ci.leaf->map[i]))
1108                 return FcFalse;
1109     }
1110 #ifdef CHECK
1111     {
1112         FcCharSet       *check;
1113         FcChar32        missing;
1114         FcCharSetIter   ci, checki;
1115
1116         /* null terminate for parser */
1117         FcStrBufChar (buf, '\0');
1118         /* step back over null for life after test */
1119         buf->len--;
1120         check = FcNameParseCharSet (buf->buf + len);
1121         FcCharSetIterStart (c, &ci);
1122         FcCharSetIterStart (check, &checki);
1123         while (ci.leaf || checki.leaf)
1124         {
1125             if (ci.ucs4 < checki.ucs4)
1126             {
1127                 printf ("Missing leaf node at 0x%x\n", ci.ucs4);
1128                 FcCharSetIterNext (c, &ci);
1129             }
1130             else if (checki.ucs4 < ci.ucs4)
1131             {
1132                 printf ("Extra leaf node at 0x%x\n", checki.ucs4);
1133                 FcCharSetIterNext (check, &checki);
1134             }
1135             else
1136             {
1137                 int         i = 256/32;
1138                 FcChar32    *cm = ci.leaf->map;
1139                 FcChar32    *checkm = checki.leaf->map;
1140
1141                 for (i = 0; i < 256; i += 32)
1142                 {
1143                     if (*cm != *checkm)
1144                         printf ("Mismatching sets at 0x%08x: 0x%08x != 0x%08x\n",
1145                                 ci.ucs4 + i, *cm, *checkm);
1146                     cm++;
1147                     checkm++;
1148                 }
1149                 FcCharSetIterNext (c, &ci);
1150                 FcCharSetIterNext (check, &checki);
1151             }
1152         }
1153         if ((missing = FcCharSetSubtractCount (c, check)))
1154             printf ("%d missing in reparsed result\n", missing);
1155         if ((missing = FcCharSetSubtractCount (check, c)))
1156             printf ("%d extra in reparsed result\n", missing);
1157         FcCharSetDestroy (check);
1158     }
1159 #endif
1160
1161     return FcTrue;
1162 }
1163
1164 /*
1165  * Figure out whether the available freetype has FT_Get_Next_Char
1166  */
1167
1168 #if FREETYPE_MAJOR > 2
1169 # define HAS_NEXT_CHAR
1170 #else
1171 # if FREETYPE_MAJOR == 2
1172 #  if FREETYPE_MINOR > 0
1173 #   define HAS_NEXT_CHAR
1174 #  else
1175 #   if FREETYPE_MINOR == 0
1176 #    if FREETYPE_PATCH >= 9
1177 #     define HAS_NEXT_CHAR
1178 #    endif
1179 #   endif
1180 #  endif
1181 # endif
1182 #endif
1183
1184 /*
1185  * For our purposes, this approximation is sufficient
1186  */
1187 #ifndef HAS_NEXT_CHAR
1188 #define FT_Get_First_Char(face, gi) ((*(gi) = 1), 1)
1189 #define FT_Get_Next_Char(face, ucs4, gi) ((ucs4) >= 0xffffff ? \
1190                                           (*(gi) = 0), 0 : \
1191                                           (*(gi) = 1), (ucs4) + 1)
1192 #warning "No FT_Get_Next_Char"
1193 #endif
1194
1195 typedef struct _FcCharEnt {
1196     FcChar16        bmp;
1197     unsigned char   encode;
1198 } FcCharEnt;
1199
1200 struct _FcCharMap {
1201     const FcCharEnt *ent;
1202     int             nent;
1203 };
1204
1205 typedef struct _FcFontDecode {
1206     FT_Encoding     encoding;
1207     const FcCharMap *map;
1208     FcChar32        max;
1209 } FcFontDecode;
1210
1211 static const FcCharEnt AppleRomanEnt[] = {
1212     { 0x0020, 0x20 }, /* SPACE */
1213     { 0x0021, 0x21 }, /* EXCLAMATION MARK */
1214     { 0x0022, 0x22 }, /* QUOTATION MARK */
1215     { 0x0023, 0x23 }, /* NUMBER SIGN */
1216     { 0x0024, 0x24 }, /* DOLLAR SIGN */
1217     { 0x0025, 0x25 }, /* PERCENT SIGN */
1218     { 0x0026, 0x26 }, /* AMPERSAND */
1219     { 0x0027, 0x27 }, /* APOSTROPHE */
1220     { 0x0028, 0x28 }, /* LEFT PARENTHESIS */
1221     { 0x0029, 0x29 }, /* RIGHT PARENTHESIS */
1222     { 0x002A, 0x2A }, /* ASTERISK */
1223     { 0x002B, 0x2B }, /* PLUS SIGN */
1224     { 0x002C, 0x2C }, /* COMMA */
1225     { 0x002D, 0x2D }, /* HYPHEN-MINUS */
1226     { 0x002E, 0x2E }, /* FULL STOP */
1227     { 0x002F, 0x2F }, /* SOLIDUS */
1228     { 0x0030, 0x30 }, /* DIGIT ZERO */
1229     { 0x0031, 0x31 }, /* DIGIT ONE */
1230     { 0x0032, 0x32 }, /* DIGIT TWO */
1231     { 0x0033, 0x33 }, /* DIGIT THREE */
1232     { 0x0034, 0x34 }, /* DIGIT FOUR */
1233     { 0x0035, 0x35 }, /* DIGIT FIVE */
1234     { 0x0036, 0x36 }, /* DIGIT SIX */
1235     { 0x0037, 0x37 }, /* DIGIT SEVEN */
1236     { 0x0038, 0x38 }, /* DIGIT EIGHT */
1237     { 0x0039, 0x39 }, /* DIGIT NINE */
1238     { 0x003A, 0x3A }, /* COLON */
1239     { 0x003B, 0x3B }, /* SEMICOLON */
1240     { 0x003C, 0x3C }, /* LESS-THAN SIGN */
1241     { 0x003D, 0x3D }, /* EQUALS SIGN */
1242     { 0x003E, 0x3E }, /* GREATER-THAN SIGN */
1243     { 0x003F, 0x3F }, /* QUESTION MARK */
1244     { 0x0040, 0x40 }, /* COMMERCIAL AT */
1245     { 0x0041, 0x41 }, /* LATIN CAPITAL LETTER A */
1246     { 0x0042, 0x42 }, /* LATIN CAPITAL LETTER B */
1247     { 0x0043, 0x43 }, /* LATIN CAPITAL LETTER C */
1248     { 0x0044, 0x44 }, /* LATIN CAPITAL LETTER D */
1249     { 0x0045, 0x45 }, /* LATIN CAPITAL LETTER E */
1250     { 0x0046, 0x46 }, /* LATIN CAPITAL LETTER F */
1251     { 0x0047, 0x47 }, /* LATIN CAPITAL LETTER G */
1252     { 0x0048, 0x48 }, /* LATIN CAPITAL LETTER H */
1253     { 0x0049, 0x49 }, /* LATIN CAPITAL LETTER I */
1254     { 0x004A, 0x4A }, /* LATIN CAPITAL LETTER J */
1255     { 0x004B, 0x4B }, /* LATIN CAPITAL LETTER K */
1256     { 0x004C, 0x4C }, /* LATIN CAPITAL LETTER L */
1257     { 0x004D, 0x4D }, /* LATIN CAPITAL LETTER M */
1258     { 0x004E, 0x4E }, /* LATIN CAPITAL LETTER N */
1259     { 0x004F, 0x4F }, /* LATIN CAPITAL LETTER O */
1260     { 0x0050, 0x50 }, /* LATIN CAPITAL LETTER P */
1261     { 0x0051, 0x51 }, /* LATIN CAPITAL LETTER Q */
1262     { 0x0052, 0x52 }, /* LATIN CAPITAL LETTER R */
1263     { 0x0053, 0x53 }, /* LATIN CAPITAL LETTER S */
1264     { 0x0054, 0x54 }, /* LATIN CAPITAL LETTER T */
1265     { 0x0055, 0x55 }, /* LATIN CAPITAL LETTER U */
1266     { 0x0056, 0x56 }, /* LATIN CAPITAL LETTER V */
1267     { 0x0057, 0x57 }, /* LATIN CAPITAL LETTER W */
1268     { 0x0058, 0x58 }, /* LATIN CAPITAL LETTER X */
1269     { 0x0059, 0x59 }, /* LATIN CAPITAL LETTER Y */
1270     { 0x005A, 0x5A }, /* LATIN CAPITAL LETTER Z */
1271     { 0x005B, 0x5B }, /* LEFT SQUARE BRACKET */
1272     { 0x005C, 0x5C }, /* REVERSE SOLIDUS */
1273     { 0x005D, 0x5D }, /* RIGHT SQUARE BRACKET */
1274     { 0x005E, 0x5E }, /* CIRCUMFLEX ACCENT */
1275     { 0x005F, 0x5F }, /* LOW LINE */
1276     { 0x0060, 0x60 }, /* GRAVE ACCENT */
1277     { 0x0061, 0x61 }, /* LATIN SMALL LETTER A */
1278     { 0x0062, 0x62 }, /* LATIN SMALL LETTER B */
1279     { 0x0063, 0x63 }, /* LATIN SMALL LETTER C */
1280     { 0x0064, 0x64 }, /* LATIN SMALL LETTER D */
1281     { 0x0065, 0x65 }, /* LATIN SMALL LETTER E */
1282     { 0x0066, 0x66 }, /* LATIN SMALL LETTER F */
1283     { 0x0067, 0x67 }, /* LATIN SMALL LETTER G */
1284     { 0x0068, 0x68 }, /* LATIN SMALL LETTER H */
1285     { 0x0069, 0x69 }, /* LATIN SMALL LETTER I */
1286     { 0x006A, 0x6A }, /* LATIN SMALL LETTER J */
1287     { 0x006B, 0x6B }, /* LATIN SMALL LETTER K */
1288     { 0x006C, 0x6C }, /* LATIN SMALL LETTER L */
1289     { 0x006D, 0x6D }, /* LATIN SMALL LETTER M */
1290     { 0x006E, 0x6E }, /* LATIN SMALL LETTER N */
1291     { 0x006F, 0x6F }, /* LATIN SMALL LETTER O */
1292     { 0x0070, 0x70 }, /* LATIN SMALL LETTER P */
1293     { 0x0071, 0x71 }, /* LATIN SMALL LETTER Q */
1294     { 0x0072, 0x72 }, /* LATIN SMALL LETTER R */
1295     { 0x0073, 0x73 }, /* LATIN SMALL LETTER S */
1296     { 0x0074, 0x74 }, /* LATIN SMALL LETTER T */
1297     { 0x0075, 0x75 }, /* LATIN SMALL LETTER U */
1298     { 0x0076, 0x76 }, /* LATIN SMALL LETTER V */
1299     { 0x0077, 0x77 }, /* LATIN SMALL LETTER W */
1300     { 0x0078, 0x78 }, /* LATIN SMALL LETTER X */
1301     { 0x0079, 0x79 }, /* LATIN SMALL LETTER Y */
1302     { 0x007A, 0x7A }, /* LATIN SMALL LETTER Z */
1303     { 0x007B, 0x7B }, /* LEFT CURLY BRACKET */
1304     { 0x007C, 0x7C }, /* VERTICAL LINE */
1305     { 0x007D, 0x7D }, /* RIGHT CURLY BRACKET */
1306     { 0x007E, 0x7E }, /* TILDE */
1307     { 0x00A0, 0xCA }, /* NO-BREAK SPACE */
1308     { 0x00A1, 0xC1 }, /* INVERTED EXCLAMATION MARK */
1309     { 0x00A2, 0xA2 }, /* CENT SIGN */
1310     { 0x00A3, 0xA3 }, /* POUND SIGN */
1311     { 0x00A5, 0xB4 }, /* YEN SIGN */
1312     { 0x00A7, 0xA4 }, /* SECTION SIGN */
1313     { 0x00A8, 0xAC }, /* DIAERESIS */
1314     { 0x00A9, 0xA9 }, /* COPYRIGHT SIGN */
1315     { 0x00AA, 0xBB }, /* FEMININE ORDINAL INDICATOR */
1316     { 0x00AB, 0xC7 }, /* LEFT-POINTING DOUBLE ANGLE QUOTATION MARK */
1317     { 0x00AC, 0xC2 }, /* NOT SIGN */
1318     { 0x00AE, 0xA8 }, /* REGISTERED SIGN */
1319     { 0x00AF, 0xF8 }, /* MACRON */
1320     { 0x00B0, 0xA1 }, /* DEGREE SIGN */
1321     { 0x00B1, 0xB1 }, /* PLUS-MINUS SIGN */
1322     { 0x00B4, 0xAB }, /* ACUTE ACCENT */
1323     { 0x00B5, 0xB5 }, /* MICRO SIGN */
1324     { 0x00B6, 0xA6 }, /* PILCROW SIGN */
1325     { 0x00B7, 0xE1 }, /* MIDDLE DOT */
1326     { 0x00B8, 0xFC }, /* CEDILLA */
1327     { 0x00BA, 0xBC }, /* MASCULINE ORDINAL INDICATOR */
1328     { 0x00BB, 0xC8 }, /* RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK */
1329     { 0x00BF, 0xC0 }, /* INVERTED QUESTION MARK */
1330     { 0x00C0, 0xCB }, /* LATIN CAPITAL LETTER A WITH GRAVE */
1331     { 0x00C1, 0xE7 }, /* LATIN CAPITAL LETTER A WITH ACUTE */
1332     { 0x00C2, 0xE5 }, /* LATIN CAPITAL LETTER A WITH CIRCUMFLEX */
1333     { 0x00C3, 0xCC }, /* LATIN CAPITAL LETTER A WITH TILDE */
1334     { 0x00C4, 0x80 }, /* LATIN CAPITAL LETTER A WITH DIAERESIS */
1335     { 0x00C5, 0x81 }, /* LATIN CAPITAL LETTER A WITH RING ABOVE */
1336     { 0x00C6, 0xAE }, /* LATIN CAPITAL LETTER AE */
1337     { 0x00C7, 0x82 }, /* LATIN CAPITAL LETTER C WITH CEDILLA */
1338     { 0x00C8, 0xE9 }, /* LATIN CAPITAL LETTER E WITH GRAVE */
1339     { 0x00C9, 0x83 }, /* LATIN CAPITAL LETTER E WITH ACUTE */
1340     { 0x00CA, 0xE6 }, /* LATIN CAPITAL LETTER E WITH CIRCUMFLEX */
1341     { 0x00CB, 0xE8 }, /* LATIN CAPITAL LETTER E WITH DIAERESIS */
1342     { 0x00CC, 0xED }, /* LATIN CAPITAL LETTER I WITH GRAVE */
1343     { 0x00CD, 0xEA }, /* LATIN CAPITAL LETTER I WITH ACUTE */
1344     { 0x00CE, 0xEB }, /* LATIN CAPITAL LETTER I WITH CIRCUMFLEX */
1345     { 0x00CF, 0xEC }, /* LATIN CAPITAL LETTER I WITH DIAERESIS */
1346     { 0x00D1, 0x84 }, /* LATIN CAPITAL LETTER N WITH TILDE */
1347     { 0x00D2, 0xF1 }, /* LATIN CAPITAL LETTER O WITH GRAVE */
1348     { 0x00D3, 0xEE }, /* LATIN CAPITAL LETTER O WITH ACUTE */
1349     { 0x00D4, 0xEF }, /* LATIN CAPITAL LETTER O WITH CIRCUMFLEX */
1350     { 0x00D5, 0xCD }, /* LATIN CAPITAL LETTER O WITH TILDE */
1351     { 0x00D6, 0x85 }, /* LATIN CAPITAL LETTER O WITH DIAERESIS */
1352     { 0x00D8, 0xAF }, /* LATIN CAPITAL LETTER O WITH STROKE */
1353     { 0x00D9, 0xF4 }, /* LATIN CAPITAL LETTER U WITH GRAVE */
1354     { 0x00DA, 0xF2 }, /* LATIN CAPITAL LETTER U WITH ACUTE */
1355     { 0x00DB, 0xF3 }, /* LATIN CAPITAL LETTER U WITH CIRCUMFLEX */
1356     { 0x00DC, 0x86 }, /* LATIN CAPITAL LETTER U WITH DIAERESIS */
1357     { 0x00DF, 0xA7 }, /* LATIN SMALL LETTER SHARP S */
1358     { 0x00E0, 0x88 }, /* LATIN SMALL LETTER A WITH GRAVE */
1359     { 0x00E1, 0x87 }, /* LATIN SMALL LETTER A WITH ACUTE */
1360     { 0x00E2, 0x89 }, /* LATIN SMALL LETTER A WITH CIRCUMFLEX */
1361     { 0x00E3, 0x8B }, /* LATIN SMALL LETTER A WITH TILDE */
1362     { 0x00E4, 0x8A }, /* LATIN SMALL LETTER A WITH DIAERESIS */
1363     { 0x00E5, 0x8C }, /* LATIN SMALL LETTER A WITH RING ABOVE */
1364     { 0x00E6, 0xBE }, /* LATIN SMALL LETTER AE */
1365     { 0x00E7, 0x8D }, /* LATIN SMALL LETTER C WITH CEDILLA */
1366     { 0x00E8, 0x8F }, /* LATIN SMALL LETTER E WITH GRAVE */
1367     { 0x00E9, 0x8E }, /* LATIN SMALL LETTER E WITH ACUTE */
1368     { 0x00EA, 0x90 }, /* LATIN SMALL LETTER E WITH CIRCUMFLEX */
1369     { 0x00EB, 0x91 }, /* LATIN SMALL LETTER E WITH DIAERESIS */
1370     { 0x00EC, 0x93 }, /* LATIN SMALL LETTER I WITH GRAVE */
1371     { 0x00ED, 0x92 }, /* LATIN SMALL LETTER I WITH ACUTE */
1372     { 0x00EE, 0x94 }, /* LATIN SMALL LETTER I WITH CIRCUMFLEX */
1373     { 0x00EF, 0x95 }, /* LATIN SMALL LETTER I WITH DIAERESIS */
1374     { 0x00F1, 0x96 }, /* LATIN SMALL LETTER N WITH TILDE */
1375     { 0x00F2, 0x98 }, /* LATIN SMALL LETTER O WITH GRAVE */
1376     { 0x00F3, 0x97 }, /* LATIN SMALL LETTER O WITH ACUTE */
1377     { 0x00F4, 0x99 }, /* LATIN SMALL LETTER O WITH CIRCUMFLEX */
1378     { 0x00F5, 0x9B }, /* LATIN SMALL LETTER O WITH TILDE */
1379     { 0x00F6, 0x9A }, /* LATIN SMALL LETTER O WITH DIAERESIS */
1380     { 0x00F7, 0xD6 }, /* DIVISION SIGN */
1381     { 0x00F8, 0xBF }, /* LATIN SMALL LETTER O WITH STROKE */
1382     { 0x00F9, 0x9D }, /* LATIN SMALL LETTER U WITH GRAVE */
1383     { 0x00FA, 0x9C }, /* LATIN SMALL LETTER U WITH ACUTE */
1384     { 0x00FB, 0x9E }, /* LATIN SMALL LETTER U WITH CIRCUMFLEX */
1385     { 0x00FC, 0x9F }, /* LATIN SMALL LETTER U WITH DIAERESIS */
1386     { 0x00FF, 0xD8 }, /* LATIN SMALL LETTER Y WITH DIAERESIS */
1387     { 0x0131, 0xF5 }, /* LATIN SMALL LETTER DOTLESS I */
1388     { 0x0152, 0xCE }, /* LATIN CAPITAL LIGATURE OE */
1389     { 0x0153, 0xCF }, /* LATIN SMALL LIGATURE OE */
1390     { 0x0178, 0xD9 }, /* LATIN CAPITAL LETTER Y WITH DIAERESIS */
1391     { 0x0192, 0xC4 }, /* LATIN SMALL LETTER F WITH HOOK */
1392     { 0x02C6, 0xF6 }, /* MODIFIER LETTER CIRCUMFLEX ACCENT */
1393     { 0x02C7, 0xFF }, /* CARON */
1394     { 0x02D8, 0xF9 }, /* BREVE */
1395     { 0x02D9, 0xFA }, /* DOT ABOVE */
1396     { 0x02DA, 0xFB }, /* RING ABOVE */
1397     { 0x02DB, 0xFE }, /* OGONEK */
1398     { 0x02DC, 0xF7 }, /* SMALL TILDE */
1399     { 0x02DD, 0xFD }, /* DOUBLE ACUTE ACCENT */
1400     { 0x03A9, 0xBD }, /* GREEK CAPITAL LETTER OMEGA */
1401     { 0x03C0, 0xB9 }, /* GREEK SMALL LETTER PI */
1402     { 0x2013, 0xD0 }, /* EN DASH */
1403     { 0x2014, 0xD1 }, /* EM DASH */
1404     { 0x2018, 0xD4 }, /* LEFT SINGLE QUOTATION MARK */
1405     { 0x2019, 0xD5 }, /* RIGHT SINGLE QUOTATION MARK */
1406     { 0x201A, 0xE2 }, /* SINGLE LOW-9 QUOTATION MARK */
1407     { 0x201C, 0xD2 }, /* LEFT DOUBLE QUOTATION MARK */
1408     { 0x201D, 0xD3 }, /* RIGHT DOUBLE QUOTATION MARK */
1409     { 0x201E, 0xE3 }, /* DOUBLE LOW-9 QUOTATION MARK */
1410     { 0x2020, 0xA0 }, /* DAGGER */
1411     { 0x2021, 0xE0 }, /* DOUBLE DAGGER */
1412     { 0x2022, 0xA5 }, /* BULLET */
1413     { 0x2026, 0xC9 }, /* HORIZONTAL ELLIPSIS */
1414     { 0x2030, 0xE4 }, /* PER MILLE SIGN */
1415     { 0x2039, 0xDC }, /* SINGLE LEFT-POINTING ANGLE QUOTATION MARK */
1416     { 0x203A, 0xDD }, /* SINGLE RIGHT-POINTING ANGLE QUOTATION MARK */
1417     { 0x2044, 0xDA }, /* FRACTION SLASH */
1418     { 0x20AC, 0xDB }, /* EURO SIGN */
1419     { 0x2122, 0xAA }, /* TRADE MARK SIGN */
1420     { 0x2202, 0xB6 }, /* PARTIAL DIFFERENTIAL */
1421     { 0x2206, 0xC6 }, /* INCREMENT */
1422     { 0x220F, 0xB8 }, /* N-ARY PRODUCT */
1423     { 0x2211, 0xB7 }, /* N-ARY SUMMATION */
1424     { 0x221A, 0xC3 }, /* SQUARE ROOT */
1425     { 0x221E, 0xB0 }, /* INFINITY */
1426     { 0x222B, 0xBA }, /* INTEGRAL */
1427     { 0x2248, 0xC5 }, /* ALMOST EQUAL TO */
1428     { 0x2260, 0xAD }, /* NOT EQUAL TO */
1429     { 0x2264, 0xB2 }, /* LESS-THAN OR EQUAL TO */
1430     { 0x2265, 0xB3 }, /* GREATER-THAN OR EQUAL TO */
1431     { 0x25CA, 0xD7 }, /* LOZENGE */
1432     { 0xF8FF, 0xF0 }, /* Apple logo */
1433     { 0xFB01, 0xDE }, /* LATIN SMALL LIGATURE FI */
1434     { 0xFB02, 0xDF }, /* LATIN SMALL LIGATURE FL */
1435 };
1436
1437 static const FcCharMap AppleRoman = {
1438     AppleRomanEnt,
1439     sizeof (AppleRomanEnt) / sizeof (AppleRomanEnt[0])
1440 };
1441
1442 static const FcCharEnt AdobeSymbolEnt[] = {
1443     { 0x0020, 0x20 }, /* SPACE  # space */
1444     { 0x0021, 0x21 }, /* EXCLAMATION MARK       # exclam */
1445     { 0x0023, 0x23 }, /* NUMBER SIGN    # numbersign */
1446     { 0x0025, 0x25 }, /* PERCENT SIGN   # percent */
1447     { 0x0026, 0x26 }, /* AMPERSAND      # ampersand */
1448     { 0x0028, 0x28 }, /* LEFT PARENTHESIS       # parenleft */
1449     { 0x0029, 0x29 }, /* RIGHT PARENTHESIS      # parenright */
1450     { 0x002B, 0x2B }, /* PLUS SIGN      # plus */
1451     { 0x002C, 0x2C }, /* COMMA  # comma */
1452     { 0x002E, 0x2E }, /* FULL STOP      # period */
1453     { 0x002F, 0x2F }, /* SOLIDUS        # slash */
1454     { 0x0030, 0x30 }, /* DIGIT ZERO     # zero */
1455     { 0x0031, 0x31 }, /* DIGIT ONE      # one */
1456     { 0x0032, 0x32 }, /* DIGIT TWO      # two */
1457     { 0x0033, 0x33 }, /* DIGIT THREE    # three */
1458     { 0x0034, 0x34 }, /* DIGIT FOUR     # four */
1459     { 0x0035, 0x35 }, /* DIGIT FIVE     # five */
1460     { 0x0036, 0x36 }, /* DIGIT SIX      # six */
1461     { 0x0037, 0x37 }, /* DIGIT SEVEN    # seven */
1462     { 0x0038, 0x38 }, /* DIGIT EIGHT    # eight */
1463     { 0x0039, 0x39 }, /* DIGIT NINE     # nine */
1464     { 0x003A, 0x3A }, /* COLON  # colon */
1465     { 0x003B, 0x3B }, /* SEMICOLON      # semicolon */
1466     { 0x003C, 0x3C }, /* LESS-THAN SIGN # less */
1467     { 0x003D, 0x3D }, /* EQUALS SIGN    # equal */
1468     { 0x003E, 0x3E }, /* GREATER-THAN SIGN      # greater */
1469     { 0x003F, 0x3F }, /* QUESTION MARK  # question */
1470     { 0x005B, 0x5B }, /* LEFT SQUARE BRACKET    # bracketleft */
1471     { 0x005D, 0x5D }, /* RIGHT SQUARE BRACKET   # bracketright */
1472     { 0x005F, 0x5F }, /* LOW LINE       # underscore */
1473     { 0x007B, 0x7B }, /* LEFT CURLY BRACKET     # braceleft */
1474     { 0x007C, 0x7C }, /* VERTICAL LINE  # bar */
1475     { 0x007D, 0x7D }, /* RIGHT CURLY BRACKET    # braceright */
1476     { 0x00A0, 0x20 }, /* NO-BREAK SPACE # space */
1477     { 0x00AC, 0xD8 }, /* NOT SIGN       # logicalnot */
1478     { 0x00B0, 0xB0 }, /* DEGREE SIGN    # degree */
1479     { 0x00B1, 0xB1 }, /* PLUS-MINUS SIGN        # plusminus */
1480     { 0x00B5, 0x6D }, /* MICRO SIGN     # mu */
1481     { 0x00D7, 0xB4 }, /* MULTIPLICATION SIGN    # multiply */
1482     { 0x00F7, 0xB8 }, /* DIVISION SIGN  # divide */
1483     { 0x0192, 0xA6 }, /* LATIN SMALL LETTER F WITH HOOK # florin */
1484     { 0x0391, 0x41 }, /* GREEK CAPITAL LETTER ALPHA     # Alpha */
1485     { 0x0392, 0x42 }, /* GREEK CAPITAL LETTER BETA      # Beta */
1486     { 0x0393, 0x47 }, /* GREEK CAPITAL LETTER GAMMA     # Gamma */
1487     { 0x0394, 0x44 }, /* GREEK CAPITAL LETTER DELTA     # Delta */
1488     { 0x0395, 0x45 }, /* GREEK CAPITAL LETTER EPSILON   # Epsilon */
1489     { 0x0396, 0x5A }, /* GREEK CAPITAL LETTER ZETA      # Zeta */
1490     { 0x0397, 0x48 }, /* GREEK CAPITAL LETTER ETA       # Eta */
1491     { 0x0398, 0x51 }, /* GREEK CAPITAL LETTER THETA     # Theta */
1492     { 0x0399, 0x49 }, /* GREEK CAPITAL LETTER IOTA      # Iota */
1493     { 0x039A, 0x4B }, /* GREEK CAPITAL LETTER KAPPA     # Kappa */
1494     { 0x039B, 0x4C }, /* GREEK CAPITAL LETTER LAMDA     # Lambda */
1495     { 0x039C, 0x4D }, /* GREEK CAPITAL LETTER MU        # Mu */
1496     { 0x039D, 0x4E }, /* GREEK CAPITAL LETTER NU        # Nu */
1497     { 0x039E, 0x58 }, /* GREEK CAPITAL LETTER XI        # Xi */
1498     { 0x039F, 0x4F }, /* GREEK CAPITAL LETTER OMICRON   # Omicron */
1499     { 0x03A0, 0x50 }, /* GREEK CAPITAL LETTER PI        # Pi */
1500     { 0x03A1, 0x52 }, /* GREEK CAPITAL LETTER RHO       # Rho */
1501     { 0x03A3, 0x53 }, /* GREEK CAPITAL LETTER SIGMA     # Sigma */
1502     { 0x03A4, 0x54 }, /* GREEK CAPITAL LETTER TAU       # Tau */
1503     { 0x03A5, 0x55 }, /* GREEK CAPITAL LETTER UPSILON   # Upsilon */
1504     { 0x03A6, 0x46 }, /* GREEK CAPITAL LETTER PHI       # Phi */
1505     { 0x03A7, 0x43 }, /* GREEK CAPITAL LETTER CHI       # Chi */
1506     { 0x03A8, 0x59 }, /* GREEK CAPITAL LETTER PSI       # Psi */
1507     { 0x03A9, 0x57 }, /* GREEK CAPITAL LETTER OMEGA     # Omega */
1508     { 0x03B1, 0x61 }, /* GREEK SMALL LETTER ALPHA       # alpha */
1509     { 0x03B2, 0x62 }, /* GREEK SMALL LETTER BETA        # beta */
1510     { 0x03B3, 0x67 }, /* GREEK SMALL LETTER GAMMA       # gamma */
1511     { 0x03B4, 0x64 }, /* GREEK SMALL LETTER DELTA       # delta */
1512     { 0x03B5, 0x65 }, /* GREEK SMALL LETTER EPSILON     # epsilon */
1513     { 0x03B6, 0x7A }, /* GREEK SMALL LETTER ZETA        # zeta */
1514     { 0x03B7, 0x68 }, /* GREEK SMALL LETTER ETA # eta */
1515     { 0x03B8, 0x71 }, /* GREEK SMALL LETTER THETA       # theta */
1516     { 0x03B9, 0x69 }, /* GREEK SMALL LETTER IOTA        # iota */
1517     { 0x03BA, 0x6B }, /* GREEK SMALL LETTER KAPPA       # kappa */
1518     { 0x03BB, 0x6C }, /* GREEK SMALL LETTER LAMDA       # lambda */
1519     { 0x03BC, 0x6D }, /* GREEK SMALL LETTER MU  # mu */
1520     { 0x03BD, 0x6E }, /* GREEK SMALL LETTER NU  # nu */
1521     { 0x03BE, 0x78 }, /* GREEK SMALL LETTER XI  # xi */
1522     { 0x03BF, 0x6F }, /* GREEK SMALL LETTER OMICRON     # omicron */
1523     { 0x03C0, 0x70 }, /* GREEK SMALL LETTER PI  # pi */
1524     { 0x03C1, 0x72 }, /* GREEK SMALL LETTER RHO # rho */
1525     { 0x03C2, 0x56 }, /* GREEK SMALL LETTER FINAL SIGMA # sigma1 */
1526     { 0x03C3, 0x73 }, /* GREEK SMALL LETTER SIGMA       # sigma */
1527     { 0x03C4, 0x74 }, /* GREEK SMALL LETTER TAU # tau */
1528     { 0x03C5, 0x75 }, /* GREEK SMALL LETTER UPSILON     # upsilon */
1529     { 0x03C6, 0x66 }, /* GREEK SMALL LETTER PHI # phi */
1530     { 0x03C7, 0x63 }, /* GREEK SMALL LETTER CHI # chi */
1531     { 0x03C8, 0x79 }, /* GREEK SMALL LETTER PSI # psi */
1532     { 0x03C9, 0x77 }, /* GREEK SMALL LETTER OMEGA       # omega */
1533     { 0x03D1, 0x4A }, /* GREEK THETA SYMBOL     # theta1 */
1534     { 0x03D2, 0xA1 }, /* GREEK UPSILON WITH HOOK SYMBOL # Upsilon1 */
1535     { 0x03D5, 0x6A }, /* GREEK PHI SYMBOL       # phi1 */
1536     { 0x03D6, 0x76 }, /* GREEK PI SYMBOL        # omega1 */
1537     { 0x2022, 0xB7 }, /* BULLET # bullet */
1538     { 0x2026, 0xBC }, /* HORIZONTAL ELLIPSIS    # ellipsis */
1539     { 0x2032, 0xA2 }, /* PRIME  # minute */
1540     { 0x2033, 0xB2 }, /* DOUBLE PRIME   # second */
1541     { 0x2044, 0xA4 }, /* FRACTION SLASH # fraction */
1542     { 0x20AC, 0xA0 }, /* EURO SIGN      # Euro */
1543     { 0x2111, 0xC1 }, /* BLACK-LETTER CAPITAL I # Ifraktur */
1544     { 0x2118, 0xC3 }, /* SCRIPT CAPITAL P       # weierstrass */
1545     { 0x211C, 0xC2 }, /* BLACK-LETTER CAPITAL R # Rfraktur */
1546     { 0x2126, 0x57 }, /* OHM SIGN       # Omega */
1547     { 0x2135, 0xC0 }, /* ALEF SYMBOL    # aleph */
1548     { 0x2190, 0xAC }, /* LEFTWARDS ARROW        # arrowleft */
1549     { 0x2191, 0xAD }, /* UPWARDS ARROW  # arrowup */
1550     { 0x2192, 0xAE }, /* RIGHTWARDS ARROW       # arrowright */
1551     { 0x2193, 0xAF }, /* DOWNWARDS ARROW        # arrowdown */
1552     { 0x2194, 0xAB }, /* LEFT RIGHT ARROW       # arrowboth */
1553     { 0x21B5, 0xBF }, /* DOWNWARDS ARROW WITH CORNER LEFTWARDS  # carriagereturn */
1554     { 0x21D0, 0xDC }, /* LEFTWARDS DOUBLE ARROW # arrowdblleft */
1555     { 0x21D1, 0xDD }, /* UPWARDS DOUBLE ARROW   # arrowdblup */
1556     { 0x21D2, 0xDE }, /* RIGHTWARDS DOUBLE ARROW        # arrowdblright */
1557     { 0x21D3, 0xDF }, /* DOWNWARDS DOUBLE ARROW # arrowdbldown */
1558     { 0x21D4, 0xDB }, /* LEFT RIGHT DOUBLE ARROW        # arrowdblboth */
1559     { 0x2200, 0x22 }, /* FOR ALL        # universal */
1560     { 0x2202, 0xB6 }, /* PARTIAL DIFFERENTIAL   # partialdiff */
1561     { 0x2203, 0x24 }, /* THERE EXISTS   # existential */
1562     { 0x2205, 0xC6 }, /* EMPTY SET      # emptyset */
1563     { 0x2206, 0x44 }, /* INCREMENT      # Delta */
1564     { 0x2207, 0xD1 }, /* NABLA  # gradient */
1565     { 0x2208, 0xCE }, /* ELEMENT OF     # element */
1566     { 0x2209, 0xCF }, /* NOT AN ELEMENT OF      # notelement */
1567     { 0x220B, 0x27 }, /* CONTAINS AS MEMBER     # suchthat */
1568     { 0x220F, 0xD5 }, /* N-ARY PRODUCT  # product */
1569     { 0x2211, 0xE5 }, /* N-ARY SUMMATION        # summation */
1570     { 0x2212, 0x2D }, /* MINUS SIGN     # minus */
1571     { 0x2215, 0xA4 }, /* DIVISION SLASH # fraction */
1572     { 0x2217, 0x2A }, /* ASTERISK OPERATOR      # asteriskmath */
1573     { 0x221A, 0xD6 }, /* SQUARE ROOT    # radical */
1574     { 0x221D, 0xB5 }, /* PROPORTIONAL TO        # proportional */
1575     { 0x221E, 0xA5 }, /* INFINITY       # infinity */
1576     { 0x2220, 0xD0 }, /* ANGLE  # angle */
1577     { 0x2227, 0xD9 }, /* LOGICAL AND    # logicaland */
1578     { 0x2228, 0xDA }, /* LOGICAL OR     # logicalor */
1579     { 0x2229, 0xC7 }, /* INTERSECTION   # intersection */
1580     { 0x222A, 0xC8 }, /* UNION  # union */
1581     { 0x222B, 0xF2 }, /* INTEGRAL       # integral */
1582     { 0x2234, 0x5C }, /* THEREFORE      # therefore */
1583     { 0x223C, 0x7E }, /* TILDE OPERATOR # similar */
1584     { 0x2245, 0x40 }, /* APPROXIMATELY EQUAL TO # congruent */
1585     { 0x2248, 0xBB }, /* ALMOST EQUAL TO        # approxequal */
1586     { 0x2260, 0xB9 }, /* NOT EQUAL TO   # notequal */
1587     { 0x2261, 0xBA }, /* IDENTICAL TO   # equivalence */
1588     { 0x2264, 0xA3 }, /* LESS-THAN OR EQUAL TO  # lessequal */
1589     { 0x2265, 0xB3 }, /* GREATER-THAN OR EQUAL TO       # greaterequal */
1590     { 0x2282, 0xCC }, /* SUBSET OF      # propersubset */
1591     { 0x2283, 0xC9 }, /* SUPERSET OF    # propersuperset */
1592     { 0x2284, 0xCB }, /* NOT A SUBSET OF        # notsubset */
1593     { 0x2286, 0xCD }, /* SUBSET OF OR EQUAL TO  # reflexsubset */
1594     { 0x2287, 0xCA }, /* SUPERSET OF OR EQUAL TO        # reflexsuperset */
1595     { 0x2295, 0xC5 }, /* CIRCLED PLUS   # circleplus */
1596     { 0x2297, 0xC4 }, /* CIRCLED TIMES  # circlemultiply */
1597     { 0x22A5, 0x5E }, /* UP TACK        # perpendicular */
1598     { 0x22C5, 0xD7 }, /* DOT OPERATOR   # dotmath */
1599     { 0x2320, 0xF3 }, /* TOP HALF INTEGRAL      # integraltp */
1600     { 0x2321, 0xF5 }, /* BOTTOM HALF INTEGRAL   # integralbt */
1601     { 0x2329, 0xE1 }, /* LEFT-POINTING ANGLE BRACKET    # angleleft */
1602     { 0x232A, 0xF1 }, /* RIGHT-POINTING ANGLE BRACKET   # angleright */
1603     { 0x25CA, 0xE0 }, /* LOZENGE        # lozenge */
1604     { 0x2660, 0xAA }, /* BLACK SPADE SUIT       # spade */
1605     { 0x2663, 0xA7 }, /* BLACK CLUB SUIT        # club */
1606     { 0x2665, 0xA9 }, /* BLACK HEART SUIT       # heart */
1607     { 0x2666, 0xA8 }, /* BLACK DIAMOND SUIT     # diamond */
1608     { 0xF6D9, 0xD3 }, /* COPYRIGHT SIGN SERIF   # copyrightserif (CUS) */
1609     { 0xF6DA, 0xD2 }, /* REGISTERED SIGN SERIF  # registerserif (CUS) */
1610     { 0xF6DB, 0xD4 }, /* TRADE MARK SIGN SERIF  # trademarkserif (CUS) */
1611     { 0xF8E5, 0x60 }, /* RADICAL EXTENDER       # radicalex (CUS) */
1612     { 0xF8E6, 0xBD }, /* VERTICAL ARROW EXTENDER        # arrowvertex (CUS) */
1613     { 0xF8E7, 0xBE }, /* HORIZONTAL ARROW EXTENDER      # arrowhorizex (CUS) */
1614     { 0xF8E8, 0xE2 }, /* REGISTERED SIGN SANS SERIF     # registersans (CUS) */
1615     { 0xF8E9, 0xE3 }, /* COPYRIGHT SIGN SANS SERIF      # copyrightsans (CUS) */
1616     { 0xF8EA, 0xE4 }, /* TRADE MARK SIGN SANS SERIF     # trademarksans (CUS) */
1617     { 0xF8EB, 0xE6 }, /* LEFT PAREN TOP # parenlefttp (CUS) */
1618     { 0xF8EC, 0xE7 }, /* LEFT PAREN EXTENDER    # parenleftex (CUS) */
1619     { 0xF8ED, 0xE8 }, /* LEFT PAREN BOTTOM      # parenleftbt (CUS) */
1620     { 0xF8EE, 0xE9 }, /* LEFT SQUARE BRACKET TOP        # bracketlefttp (CUS) */
1621     { 0xF8EF, 0xEA }, /* LEFT SQUARE BRACKET EXTENDER   # bracketleftex (CUS) */
1622     { 0xF8F0, 0xEB }, /* LEFT SQUARE BRACKET BOTTOM     # bracketleftbt (CUS) */
1623     { 0xF8F1, 0xEC }, /* LEFT CURLY BRACKET TOP # bracelefttp (CUS) */
1624     { 0xF8F2, 0xED }, /* LEFT CURLY BRACKET MID # braceleftmid (CUS) */
1625     { 0xF8F3, 0xEE }, /* LEFT CURLY BRACKET BOTTOM      # braceleftbt (CUS) */
1626     { 0xF8F4, 0xEF }, /* CURLY BRACKET EXTENDER # braceex (CUS) */
1627     { 0xF8F5, 0xF4 }, /* INTEGRAL EXTENDER      # integralex (CUS) */
1628     { 0xF8F6, 0xF6 }, /* RIGHT PAREN TOP        # parenrighttp (CUS) */
1629     { 0xF8F7, 0xF7 }, /* RIGHT PAREN EXTENDER   # parenrightex (CUS) */
1630     { 0xF8F8, 0xF8 }, /* RIGHT PAREN BOTTOM     # parenrightbt (CUS) */
1631     { 0xF8F9, 0xF9 }, /* RIGHT SQUARE BRACKET TOP       # bracketrighttp (CUS) */
1632     { 0xF8FA, 0xFA }, /* RIGHT SQUARE BRACKET EXTENDER  # bracketrightex (CUS) */
1633     { 0xF8FB, 0xFB }, /* RIGHT SQUARE BRACKET BOTTOM    # bracketrightbt (CUS) */
1634     { 0xF8FC, 0xFC }, /* RIGHT CURLY BRACKET TOP        # bracerighttp (CUS) */
1635     { 0xF8FD, 0xFD }, /* RIGHT CURLY BRACKET MID        # bracerightmid (CUS) */
1636     { 0xF8FE, 0xFE }, /* RIGHT CURLY BRACKET BOTTOM     # bracerightbt (CUS) */
1637 };
1638
1639 static const FcCharMap AdobeSymbol = {
1640     AdobeSymbolEnt,
1641     sizeof (AdobeSymbolEnt) / sizeof (AdobeSymbolEnt[0]),
1642 };
1643
1644 static const FcFontDecode fcFontDecoders[] = {
1645     { ft_encoding_unicode,      0,              (1 << 21) - 1 },
1646     { ft_encoding_symbol,       &AdobeSymbol,   (1 << 16) - 1 },
1647     { ft_encoding_apple_roman,  &AppleRoman,    (1 << 16) - 1 },
1648 };
1649
1650 #define NUM_DECODE  (sizeof (fcFontDecoders) / sizeof (fcFontDecoders[0]))
1651
1652 FcChar32
1653 FcFreeTypeUcs4ToPrivate (FcChar32 ucs4, const FcCharMap *map)
1654 {
1655     int         low, high, mid;
1656     FcChar16    bmp;
1657
1658     low = 0;
1659     high = map->nent - 1;
1660     if (ucs4 < map->ent[low].bmp || map->ent[high].bmp < ucs4)
1661         return ~0;
1662     while (low <= high)
1663     {
1664         mid = (high + low) >> 1;
1665         bmp = map->ent[mid].bmp;
1666         if (ucs4 == bmp)
1667             return (FT_ULong) map->ent[mid].encode;
1668         if (ucs4 < bmp)
1669             high = mid - 1;
1670         else
1671             low = mid + 1;
1672     }
1673     return ~0;
1674 }
1675
1676 FcChar32
1677 FcFreeTypePrivateToUcs4 (FcChar32 private, const FcCharMap *map)
1678 {
1679     int     i;
1680
1681     for (i = 0; i < map->nent; i++)
1682         if (map->ent[i].encode == private)
1683             return (FcChar32) map->ent[i].bmp;
1684     return ~0;
1685 }
1686
1687 const FcCharMap *
1688 FcFreeTypeGetPrivateMap (FT_Encoding encoding)
1689 {
1690     int i;
1691
1692     for (i = 0; i < NUM_DECODE; i++)
1693         if (fcFontDecoders[i].encoding == encoding)
1694             return fcFontDecoders[i].map;
1695     return 0;
1696 }
1697
1698 /*
1699  * Map a UCS4 glyph to a glyph index.  Use all available encoding
1700  * tables to try and find one that works.  This information is expected
1701  * to be cached by higher levels, so performance isn't critical
1702  */
1703
1704 FT_UInt
1705 FcFreeTypeCharIndex (FT_Face face, FcChar32 ucs4)
1706 {
1707     int             initial, offset, decode;
1708     FT_UInt         glyphindex;
1709     FcChar32        charcode;
1710
1711     initial = 0;
1712     /*
1713      * Find the current encoding
1714      */
1715     if (face->charmap)
1716     {
1717         for (; initial < NUM_DECODE; initial++)
1718             if (fcFontDecoders[initial].encoding == face->charmap->encoding)
1719                 break;
1720         if (initial == NUM_DECODE)
1721             initial = 0;
1722     }
1723     /*
1724      * Check each encoding for the glyph, starting with the current one
1725      */
1726     for (offset = 0; offset < NUM_DECODE; offset++)
1727     {
1728         decode = (initial + offset) % NUM_DECODE;
1729         if (!face->charmap || face->charmap->encoding != fcFontDecoders[decode].encoding)
1730             if (FT_Select_Charmap (face, fcFontDecoders[decode].encoding) != 0)
1731                 continue;
1732         if (fcFontDecoders[decode].map)
1733         {
1734             charcode = FcFreeTypeUcs4ToPrivate (ucs4, fcFontDecoders[decode].map);
1735             if (charcode == ~0)
1736                 continue;
1737         }
1738         else
1739             charcode = ucs4;
1740         glyphindex = FT_Get_Char_Index (face, (FT_ULong) charcode);
1741         if (glyphindex)
1742             return glyphindex;
1743     }
1744     return 0;
1745 }
1746
1747 static FcBool
1748 FcFreeTypeCheckGlyph (FT_Face face, FcChar32 ucs4,
1749                       FT_UInt glyph, FcBlanks *blanks)
1750 {
1751     FT_Int          load_flags = FT_LOAD_NO_SCALE | FT_LOAD_NO_HINTING;
1752     FT_GlyphSlot    slot;
1753
1754     /*
1755      * When using scalable fonts, only report those glyphs
1756      * which can be scaled; otherwise those fonts will
1757      * only be available at some sizes, and never when
1758      * transformed.  Avoid this by simply reporting bitmap-only
1759      * glyphs as missing
1760      */
1761     if (face->face_flags & FT_FACE_FLAG_SCALABLE)
1762         load_flags |= FT_LOAD_NO_BITMAP;
1763
1764     if (FT_Load_Glyph (face, glyph, load_flags))
1765         return FcFalse;
1766
1767     slot = face->glyph;
1768     if (!glyph)
1769         return FcFalse;
1770
1771     switch (slot->format) {
1772     case ft_glyph_format_bitmap:
1773         /*
1774          * Bitmaps are assumed to be reasonable; if
1775          * this proves to be a rash assumption, this
1776          * code can be easily modified
1777          */
1778         return FcTrue;
1779     case ft_glyph_format_outline:
1780         /*
1781          * Glyphs with contours are always OK
1782          */
1783         if (slot->outline.n_contours != 0)
1784             return FcTrue;
1785         /*
1786          * Glyphs with no contours are only OK if
1787          * they're members of the Blanks set specified
1788          * in the configuration.  If blanks isn't set,
1789          * then allow any glyph to be blank
1790          */
1791         if (!blanks || FcBlanksIsMember (blanks, ucs4))
1792             return FcTrue;
1793         /* fall through ... */
1794     default:
1795         break;
1796     }
1797     return FcFalse;
1798 }
1799
1800 FcCharSet *
1801 FcFreeTypeCharSet (FT_Face face, FcBlanks *blanks)
1802 {
1803     FcChar32        page, off, max, ucs4;
1804 #ifdef CHECK
1805     FcChar32        font_max = 0;
1806 #endif
1807     FcCharSet       *fcs;
1808     FcCharLeaf      *leaf;
1809     const FcCharMap *map;
1810     int             o;
1811     int             i;
1812     FT_UInt         glyph;
1813
1814     fcs = FcCharSetCreate ();
1815     if (!fcs)
1816         goto bail0;
1817
1818     for (o = 0; o < NUM_DECODE; o++)
1819     {
1820         if (FT_Select_Charmap (face, fcFontDecoders[o].encoding) != 0)
1821             continue;
1822         map = fcFontDecoders[o].map;
1823         if (map)
1824         {
1825             /*
1826              * Non-Unicode tables are easy; there's a list of all possible
1827              * characters
1828              */
1829             for (i = 0; i < map->nent; i++)
1830             {
1831                 ucs4 = map->ent[i].bmp;
1832                 glyph = FT_Get_Char_Index (face, map->ent[i].encode);
1833                 if (glyph && FcFreeTypeCheckGlyph (face, ucs4, glyph, blanks))
1834                 {
1835                     leaf = FcCharSetFindLeafCreate (fcs, ucs4);
1836                     if (!leaf)
1837                         goto bail1;
1838                     leaf->map[(ucs4 & 0xff) >> 5] |= (1 << (ucs4 & 0x1f));
1839 #ifdef CHECK
1840                     if (ucs4 > font_max)
1841                         font_max = ucs4;
1842 #endif
1843                 }
1844             }
1845         }
1846         else
1847         {
1848             FT_UInt gindex;
1849
1850             max = fcFontDecoders[o].max;
1851             /*
1852              * Find the first encoded character in the font
1853              */
1854             if (FT_Get_Char_Index (face, 0))
1855             {
1856                 ucs4 = 0;
1857                 gindex = 1;
1858             }
1859             else
1860             {
1861                 ucs4 = FT_Get_Next_Char (face, 0, &gindex);
1862                 if (!ucs4)
1863                     gindex = 0;
1864             }
1865
1866             while (gindex)
1867             {
1868                 page = ucs4 >> 8;
1869                 leaf = 0;
1870                 while ((ucs4 >> 8) == page)
1871                 {
1872                     glyph = FT_Get_Char_Index (face, ucs4);
1873                     if (glyph && FcFreeTypeCheckGlyph (face, ucs4,
1874                                                        glyph, blanks))
1875                     {
1876                         if (!leaf)
1877                         {
1878                             leaf = FcCharSetFindLeafCreate (fcs, ucs4);
1879                             if (!leaf)
1880                                 goto bail1;
1881                         }
1882                         off = ucs4 & 0xff;
1883                         leaf->map[off >> 5] |= (1 << (off & 0x1f));
1884 #ifdef CHECK
1885                         if (ucs4 > font_max)
1886                             font_max = ucs4;
1887 #endif
1888                     }
1889                     ucs4++;
1890                 }
1891                 ucs4 = FT_Get_Next_Char (face, ucs4 - 1, &gindex);
1892                 if (!ucs4)
1893                     gindex = 0;
1894             }
1895 #ifdef CHECK
1896             for (ucs4 = 0; ucs4 < 0x10000; ucs4++)
1897             {
1898                 FcBool      FT_Has, FC_Has;
1899
1900                 FT_Has = FT_Get_Char_Index (face, ucs4) != 0;
1901                 FC_Has = FcCharSetHasChar (fcs, ucs4);
1902                 if (FT_Has != FC_Has)
1903                 {
1904                     printf ("0x%08x FT says %d FC says %d\n", ucs4, FT_Has, FC_Has);
1905                 }
1906             }
1907 #endif
1908         }
1909     }
1910 #ifdef CHECK
1911     printf ("%d glyphs %d encoded\n", (int) face->num_glyphs, FcCharSetCount (fcs));
1912     for (ucs4 = 0; ucs4 <= font_max; ucs4++)
1913     {
1914         FcBool  has_char = FcFreeTypeCharIndex (face, ucs4) != 0;
1915         FcBool  has_bit = FcCharSetHasChar (fcs, ucs4);
1916
1917         if (has_char && !has_bit)
1918             printf ("Bitmap missing char 0x%x\n", ucs4);
1919         else if (!has_char && has_bit)
1920             printf ("Bitmap extra char 0x%x\n", ucs4);
1921     }
1922 #endif
1923     return fcs;
1924 bail1:
1925     FcCharSetDestroy (fcs);
1926 bail0:
1927     return 0;
1928 }
1929