third_party/harfbuzz-ng/src/hb-set-private.hh

   1 /*
   2  * Copyright © 2012  Google, Inc.
   3  *
   4  *  This is part of HarfBuzz, a text shaping library.
   5  *
   6  * Permission is hereby granted, without written agreement and without
   7  * license or royalty fees, to use, copy, modify, and distribute this
   8  * software and its documentation for any purpose, provided that the
   9  * above copyright notice and the following two paragraphs appear in
  10  * all copies of this software.
  11  *
  12  * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
  13  * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
  14  * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
  15  * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
  16  * DAMAGE.
  17  *
  18  * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
  19  * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
  20  * FITNESS FOR A PARTICULAR PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS
  21  * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
  22  * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
  23  *
  24  * Google Author(s): Behdad Esfahbod
  25  */
  26
  27 #ifndef HB_SET_PRIVATE_HH
  28 #define HB_SET_PRIVATE_HH
  29
  30 #include "hb-private.hh"
  31 #include "hb-object-private.hh"
  32
  33
  34 /*
  35  * The set digests here implement various "filters" that support
  36  * "approximate member query".  Conceptually these are like Bloom
  37  * Filter and Quotient Filter, however, much smaller, faster, and
  38  * designed to fit the requirements of our uses for glyph coverage
  39  * queries.  As a result, our filters have much higher.
  40  */
  41
  42 template <typename mask_t, unsigned int shift>
  43 struct hb_set_digest_lowest_bits_t
  44 {
  45   ASSERT_POD ();
  46
  47   static const unsigned int mask_bytes = sizeof (mask_t);
  48   static const unsigned int mask_bits = sizeof (mask_t) * 8;
  49   static const unsigned int num_bits = 0
  50                                      + (mask_bytes >= 1 ? 3 : 0)
  51                                      + (mask_bytes >= 2 ? 1 : 0)
  52                                      + (mask_bytes >= 4 ? 1 : 0)
  53                                      + (mask_bytes >= 8 ? 1 : 0)
  54                                      + (mask_bytes >= 16? 1 : 0)
  55                                      + 0;
  56
  57   ASSERT_STATIC (shift < sizeof (hb_codepoint_t) * 8);
  58   ASSERT_STATIC (shift + num_bits <= sizeof (hb_codepoint_t) * 8);
  59
  60   inline void init (void) {
  61     mask = 0;
  62   }
  63
  64   inline void add (hb_codepoint_t g) {
  65     mask |= mask_for (g);
  66   }
  67
  68   inline void add_range (hb_codepoint_t a, hb_codepoint_t b) {
  69     if ((b >> shift) - (a >> shift) >= mask_bits - 1)
  70       mask = (mask_t) -1;
  71     else {
  72       mask_t ma = mask_for (a);
  73       mask_t mb = mask_for (b);
  74       mask |= mb + (mb - ma) - (mb < ma);
  75     }
  76   }
  77
  78   inline bool may_have (hb_codepoint_t g) const {
  79     return !!(mask & mask_for (g));
  80   }
  81
  82   private:
  83
  84   static inline mask_t mask_for (hb_codepoint_t g) {
  85     return ((mask_t) 1) << ((g >> shift) & (mask_bits - 1));
  86   }
  87   mask_t mask;
  88 };
  89
  90 template <typename head_t, typename tail_t>
  91 struct hb_set_digest_combiner_t
  92 {
  93   ASSERT_POD ();
  94
  95   inline void init (void) {
  96     head.init ();
  97     tail.init ();
  98   }
  99
 100   inline void add (hb_codepoint_t g) {
 101     head.add (g);
 102     tail.add (g);
 103   }
 104
 105   inline void add_range (hb_codepoint_t a, hb_codepoint_t b) {
 106     head.add_range (a, b);
 107     tail.add_range (a, b);
 108   }
 109
 110   inline bool may_have (hb_codepoint_t g) const {
 111     return head.may_have (g) && tail.may_have (g);
 112   }
 113
 114   private:
 115   head_t head;
 116   tail_t tail;
 117 };
 118
 119
 120 /*
 121  * hb_set_digest_t
 122  *
 123  * This is a combination of digests that performs "best".
 124  * There is not much science to this: it's a result of intuition
 125  * and testing.
 126  */
 127 typedef hb_set_digest_combiner_t
 128 <
 129   hb_set_digest_lowest_bits_t<unsigned long, 4>,
 130   hb_set_digest_combiner_t
 131   <
 132     hb_set_digest_lowest_bits_t<unsigned long, 0>,
 133     hb_set_digest_lowest_bits_t<unsigned long, 9>
 134   >
 135 > hb_set_digest_t;
 136
 137
 138
 139 /*
 140  * hb_set_t
 141  */
 142
 143
 144 /* TODO Make this faster and memmory efficient. */
 145
 146 struct hb_set_t
 147 {
 148   friend struct hb_frozen_set_t;
 149
 150   hb_object_header_t header;
 151   ASSERT_POD ();
 152   bool in_error;
 153
 154   inline void init (void) {
 155     hb_object_init (this);
 156     clear ();
 157   }
 158   inline void fini (void) {
 159   }
 160   inline void clear (void) {
 161     if (unlikely (hb_object_is_inert (this)))
 162       return;
 163     in_error = false;
 164     memset (elts, 0, sizeof elts);
 165   }
 166   inline bool is_empty (void) const {
 167     for (unsigned int i = 0; i < ARRAY_LENGTH (elts); i++)
 168       if (elts[i])
 169         return false;
 170     return true;
 171   }
 172   inline void add (hb_codepoint_t g)
 173   {
 174     if (unlikely (in_error)) return;
 175     if (unlikely (g == INVALID)) return;
 176     if (unlikely (g > MAX_G)) return;
 177     elt (g) |= mask (g);
 178   }
 179   inline void add_range (hb_codepoint_t a, hb_codepoint_t b)
 180   {
 181     if (unlikely (in_error)) return;
 182     /* TODO Speedup */
 183     for (unsigned int i = a; i < b + 1; i++)
 184       add (i);
 185   }
 186   inline void del (hb_codepoint_t g)
 187   {
 188     if (unlikely (in_error)) return;
 189     if (unlikely (g > MAX_G)) return;
 190     elt (g) &= ~mask (g);
 191   }
 192   inline void del_range (hb_codepoint_t a, hb_codepoint_t b)
 193   {
 194     if (unlikely (in_error)) return;
 195     /* TODO Speedup */
 196     for (unsigned int i = a; i < b + 1; i++)
 197       del (i);
 198   }
 199   inline bool has (hb_codepoint_t g) const
 200   {
 201     if (unlikely (g > MAX_G)) return false;
 202     return !!(elt (g) & mask (g));
 203   }
 204   inline bool intersects (hb_codepoint_t first,
 205                           hb_codepoint_t last) const
 206   {
 207     if (unlikely (first > MAX_G)) return false;
 208     if (unlikely (last  > MAX_G)) last = MAX_G;
 209     unsigned int end = last + 1;
 210     for (hb_codepoint_t i = first; i < end; i++)
 211       if (has (i))
 212         return true;
 213     return false;
 214   }
 215   inline bool is_equal (const hb_set_t *other) const
 216   {
 217     for (unsigned int i = 0; i < ELTS; i++)
 218       if (elts[i] != other->elts[i])
 219         return false;
 220     return true;
 221   }
 222   inline void set (const hb_set_t *other)
 223   {
 224     if (unlikely (in_error)) return;
 225     for (unsigned int i = 0; i < ELTS; i++)
 226       elts[i] = other->elts[i];
 227   }
 228   inline void union_ (const hb_set_t *other)
 229   {
 230     if (unlikely (in_error)) return;
 231     for (unsigned int i = 0; i < ELTS; i++)
 232       elts[i] |= other->elts[i];
 233   }
 234   inline void intersect (const hb_set_t *other)
 235   {
 236     if (unlikely (in_error)) return;
 237     for (unsigned int i = 0; i < ELTS; i++)
 238       elts[i] &= other->elts[i];
 239   }
 240   inline void subtract (const hb_set_t *other)
 241   {
 242     if (unlikely (in_error)) return;
 243     for (unsigned int i = 0; i < ELTS; i++)
 244       elts[i] &= ~other->elts[i];
 245   }
 246   inline void symmetric_difference (const hb_set_t *other)
 247   {
 248     if (unlikely (in_error)) return;
 249     for (unsigned int i = 0; i < ELTS; i++)
 250       elts[i] ^= other->elts[i];
 251   }
 252   inline void invert (void)
 253   {
 254     if (unlikely (in_error)) return;
 255     for (unsigned int i = 0; i < ELTS; i++)
 256       elts[i] = ~elts[i];
 257   }
 258   inline bool next (hb_codepoint_t *codepoint) const
 259   {
 260     if (unlikely (*codepoint == INVALID)) {
 261       hb_codepoint_t i = get_min ();
 262       if (i != INVALID) {
 263         *codepoint = i;
 264         return true;
 265       } else {
 266         *codepoint = INVALID;
 267         return false;
 268       }
 269     }
 270     for (hb_codepoint_t i = *codepoint + 1; i < MAX_G + 1; i++)
 271       if (has (i)) {
 272         *codepoint = i;
 273         return true;
 274       }
 275     *codepoint = INVALID;
 276     return false;
 277   }
 278   inline bool next_range (hb_codepoint_t *first, hb_codepoint_t *last) const
 279   {
 280     hb_codepoint_t i;
 281
 282     i = *last;
 283     if (!next (&i))
 284     {
 285       *last = *first = INVALID;
 286       return false;
 287     }
 288
 289     *last = *first = i;
 290     while (next (&i) && i == *last + 1)
 291       (*last)++;
 292
 293     return true;
 294   }
 295
 296   inline unsigned int get_population (void) const
 297   {
 298     unsigned int count = 0;
 299     for (unsigned int i = 0; i < ELTS; i++)
 300       count += _hb_popcount32 (elts[i]);
 301     return count;
 302   }
 303   inline hb_codepoint_t get_min (void) const
 304   {
 305     for (unsigned int i = 0; i < ELTS; i++)
 306       if (elts[i])
 307         for (unsigned int j = 0; j < BITS; j++)
 308           if (elts[i] & (1 << j))
 309             return i * BITS + j;
 310     return INVALID;
 311   }
 312   inline hb_codepoint_t get_max (void) const
 313   {
 314     for (unsigned int i = ELTS; i; i--)
 315       if (elts[i - 1])
 316         for (unsigned int j = BITS; j; j--)
 317           if (elts[i - 1] & (1 << (j - 1)))
 318             return (i - 1) * BITS + (j - 1);
 319     return INVALID;
 320   }
 321
 322   typedef uint32_t elt_t;
 323   static const unsigned int MAX_G = 65536 - 1; /* XXX Fix this... */
 324   static const unsigned int SHIFT = 5;
 325   static const unsigned int BITS = (1 << SHIFT);
 326   static const unsigned int MASK = BITS - 1;
 327   static const unsigned int ELTS = (MAX_G + 1 + (BITS - 1)) / BITS;
 328   static  const hb_codepoint_t INVALID = HB_SET_VALUE_INVALID;
 329
 330   elt_t &elt (hb_codepoint_t g) { return elts[g >> SHIFT]; }
 331   elt_t const &elt (hb_codepoint_t g) const { return elts[g >> SHIFT]; }
 332   elt_t mask (hb_codepoint_t g) const { return elt_t (1) << (g & MASK); }
 333
 334   elt_t elts[ELTS]; /* XXX 8kb */
 335
 336   ASSERT_STATIC (sizeof (elt_t) * 8 == BITS);
 337   ASSERT_STATIC (sizeof (elt_t) * 8 * ELTS > MAX_G);
 338 };
 339
 340 struct hb_frozen_set_t
 341 {
 342   static const unsigned int SHIFT = hb_set_t::SHIFT;
 343   static const unsigned int BITS = hb_set_t::BITS;
 344   static const unsigned int MASK = hb_set_t::MASK;
 345   typedef hb_set_t::elt_t elt_t;
 346
 347   inline void init (const hb_set_t &set)
 348   {
 349     start = count = 0;
 350     elts = NULL;
 351
 352     unsigned int max = set.get_max ();
 353     if (max == set.INVALID)
 354       return;
 355     unsigned int min = set.get_min ();
 356     const elt_t &min_elt = set.elt (min);
 357     const elt_t &max_elt = set.elt (max);
 358
 359     start = min & ~MASK;
 360     count = max - start + 1;
 361     unsigned int num_elts = (count + BITS - 1) / BITS;
 362     unsigned int elts_size = num_elts * sizeof (elt_t);
 363     elts = (elt_t *) malloc (elts_size);
 364     if (unlikely (!elts))
 365     {
 366       start = count = 0;
 367       return;
 368     }
 369     memcpy (elts, &min_elt, elts_size);
 370   }
 371
 372   inline void fini (void)
 373   {
 374     if (elts)
 375       free (elts);
 376   }
 377
 378   inline bool has (hb_codepoint_t g) const
 379   {
 380     /* hb_codepoint_t is unsigned. */
 381     g -= start;
 382     if (unlikely (g > count)) return false;
 383     return !!(elt (g) & mask (g));
 384   }
 385
 386   elt_t const &elt (hb_codepoint_t g) const { return elts[g >> SHIFT]; }
 387   elt_t mask (hb_codepoint_t g) const { return elt_t (1) << (g & MASK); }
 388
 389   private:
 390   hb_codepoint_t start, count;
 391   elt_t *elts;
 392 };
 393
 394
 395 #endif /* HB_SET_PRIVATE_HH */