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 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 }