icuSources/common/utrie.h

   1 /*
   2 ******************************************************************************
   3 *
   4 *   Copyright (C) 2001-2006, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 ******************************************************************************
   8 *   file name:  utrie.h
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 2001nov08
  14 *   created by: Markus W. Scherer
  15 */
  16
  17 #ifndef __UTRIE_H__
  18 #define __UTRIE_H__
  19
  20 #include "unicode/utypes.h"
  21 #include "udataswp.h"
  22
  23 U_CDECL_BEGIN
  24
  25 /**
  26  * \file
  27  *
  28  * This is a common implementation of a "folded" trie.
  29  * It is a kind of compressed, serializable table of 16- or 32-bit values associated with
  30  * Unicode code points (0..0x10ffff).
  31  *
  32  * This implementation is optimized for getting values while walking forward
  33  * through a UTF-16 string.
  34  * Therefore, the simplest and fastest access macros are the
  35  * _FROM_LEAD() and _FROM_OFFSET_TRAIL() macros.
  36  *
  37  * The _FROM_BMP() macros are a little more complicated; they get values
  38  * even for lead surrogate code _points_, while the _FROM_LEAD() macros
  39  * get special "folded" values for lead surrogate code _units_ if
  40  * there is relevant data associated with them.
  41  * From such a folded value, an offset needs to be extracted to supply
  42  * to the _FROM_OFFSET_TRAIL() macros.
  43  *
  44  * Most of the more complex (and more convenient) functions/macros call a callback function
  45  * to get that offset from the folded value for a lead surrogate unit.
  46  */
  47
  48 /**
  49  * Trie constants, defining shift widths, index array lengths, etc.
  50  */
  51 enum {
  52     /** Shift size for shifting right the input index. 1..9 */
  53     UTRIE_SHIFT=5,
  54
  55     /** Number of data values in a stage 2 (data array) block. 2, 4, 8, .., 0x200 */
  56     UTRIE_DATA_BLOCK_LENGTH=1<<UTRIE_SHIFT,
  57
  58     /** Mask for getting the lower bits from the input index. */
  59     UTRIE_MASK=UTRIE_DATA_BLOCK_LENGTH-1,
  60
  61     /**
  62      * Lead surrogate code points' index displacement in the index array.
  63      * 0x10000-0xd800=0x2800
  64      */
  65     UTRIE_LEAD_INDEX_DISP=0x2800>>UTRIE_SHIFT,
  66
  67     /**
  68      * Shift size for shifting left the index array values.
  69      * Increases possible data size with 16-bit index values at the cost
  70      * of compactability.
  71      * This requires blocks of stage 2 data to be aligned by UTRIE_DATA_GRANULARITY.
  72      * 0..UTRIE_SHIFT
  73      */
  74     UTRIE_INDEX_SHIFT=2,
  75
  76     /** The alignment size of a stage 2 data block. Also the granularity for compaction. */
  77     UTRIE_DATA_GRANULARITY=1<<UTRIE_INDEX_SHIFT,
  78
  79     /** Number of bits of a trail surrogate that are used in index table lookups. */
  80     UTRIE_SURROGATE_BLOCK_BITS=10-UTRIE_SHIFT,
  81
  82     /**
  83      * Number of index (stage 1) entries per lead surrogate.
  84      * Same as number of index entries for 1024 trail surrogates,
  85      * ==0x400>>UTRIE_SHIFT
  86      */
  87     UTRIE_SURROGATE_BLOCK_COUNT=(1<<UTRIE_SURROGATE_BLOCK_BITS),
  88
  89     /** Length of the BMP portion of the index (stage 1) array. */
  90     UTRIE_BMP_INDEX_LENGTH=0x10000>>UTRIE_SHIFT
  91 };
  92
  93 /**
  94  * Length of the index (stage 1) array before folding.
  95  * Maximum number of Unicode code points (0x110000) shifted right by UTRIE_SHIFT.
  96  */
  97 #define UTRIE_MAX_INDEX_LENGTH (0x110000>>UTRIE_SHIFT)
  98
  99 /**
 100  * Maximum length of the runtime data (stage 2) array.
 101  * Limited by 16-bit index values that are left-shifted by UTRIE_INDEX_SHIFT.
 102  */
 103 #define UTRIE_MAX_DATA_LENGTH (0x10000<<UTRIE_INDEX_SHIFT)
 104
 105 /**
 106  * Maximum length of the build-time data (stage 2) array.
 107  * The maximum length is 0x110000+UTRIE_DATA_BLOCK_LENGTH+0x400.
 108  * (Number of Unicode code points + one all-initial-value block +
 109  *  possible duplicate entries for 1024 lead surrogates.)
 110  */
 111 #define UTRIE_MAX_BUILD_TIME_DATA_LENGTH (0x110000+UTRIE_DATA_BLOCK_LENGTH+0x400)
 112
 113 /**
 114  * Number of bytes for a dummy trie.
 115  * A dummy trie is an empty runtime trie, used when a real data trie cannot
 116  * be loaded.
 117  * The number of bytes works for Latin-1-linear tries with 32-bit data
 118  * (worst case).
 119  *
 120  * Calculation:
 121  *   BMP index + 1 index block for lead surrogate code points +
 122  *   Latin-1-linear array + 1 data block for lead surrogate code points
 123  *
 124  * Latin-1: if(UTRIE_SHIFT<=8) { 256 } else { included in first data block }
 125  *
 126  * @see utrie_unserializeDummy
 127  */
 128 #define UTRIE_DUMMY_SIZE ((UTRIE_BMP_INDEX_LENGTH+UTRIE_SURROGATE_BLOCK_COUNT)*2+(UTRIE_SHIFT<=8?256:UTRIE_DATA_BLOCK_LENGTH)*4+UTRIE_DATA_BLOCK_LENGTH*4)
 129
 130 /**
 131  * Runtime UTrie callback function.
 132  * Extract from a lead surrogate's data the
 133  * index array offset of the indexes for that lead surrogate.
 134  *
 135  * @param data data value for a surrogate from the trie, including the folding offset
 136  * @return offset>=UTRIE_BMP_INDEX_LENGTH, or 0 if there is no data for the lead surrogate
 137  */
 138 typedef int32_t U_CALLCONV
 139 UTrieGetFoldingOffset(uint32_t data);
 140
 141 /**
 142  * Run-time Trie structure.
 143  *
 144  * Either the data table is 16 bits wide and accessed via the index
 145  * pointer, with each index item increased by indexLength;
 146  * in this case, data32==NULL.
 147  *
 148  * Or the data table is 32 bits wide and accessed via the data32 pointer.
 149  */
 150 struct UTrie {
 151     const uint16_t *index;
 152     const uint32_t *data32; /* NULL if 16b data is used via index */
 153
 154     /**
 155      * This function is not used in _FROM_LEAD, _FROM_BMP, and _FROM_OFFSET_TRAIL macros.
 156      * If convenience macros like _GET16 or _NEXT32 are used, this function must be set.
 157      *
 158      * utrie_unserialize() sets a default function which simply returns
 159      * the lead surrogate's value itself - which is the inverse of the default
 160      * folding function used by utrie_serialize().
 161      *
 162      * @see UTrieGetFoldingOffset
 163      */
 164     UTrieGetFoldingOffset *getFoldingOffset;
 165
 166     int32_t indexLength, dataLength;
 167     uint32_t initialValue;
 168     UBool isLatin1Linear;
 169 };
 170
 171 typedef struct UTrie UTrie;
 172
 173 /** Internal trie getter from an offset (0 if c16 is a BMP/lead units) and a 16-bit unit */
 174 #define _UTRIE_GET_RAW(trie, data, offset, c16) \
 175     (trie)->data[ \
 176         ((int32_t)((trie)->index[(offset)+((c16)>>UTRIE_SHIFT)])<<UTRIE_INDEX_SHIFT)+ \
 177         ((c16)&UTRIE_MASK) \
 178     ]
 179
 180 /** Internal trie getter from a pair of surrogates */
 181 #define _UTRIE_GET_FROM_PAIR(trie, data, c, c2, result, resultType) { \
 182     int32_t __offset; \
 183 \
 184     /* get data for lead surrogate */ \
 185     (result)=_UTRIE_GET_RAW((trie), data, 0, (c)); \
 186     __offset=(trie)->getFoldingOffset(result); \
 187 \
 188     /* get the real data from the folded lead/trail units */ \
 189     if(__offset>0) { \
 190         (result)=_UTRIE_GET_RAW((trie), data, __offset, (c2)&0x3ff); \
 191     } else { \
 192         (result)=(resultType)((trie)->initialValue); \
 193     } \
 194 }
 195
 196 /** Internal trie getter from a BMP code point, treating a lead surrogate as a normal code point */
 197 #define _UTRIE_GET_FROM_BMP(trie, data, c16) \
 198     _UTRIE_GET_RAW(trie, data, 0xd800<=(c16) && (c16)<=0xdbff ? UTRIE_LEAD_INDEX_DISP : 0, c16);
 199
 200 /**
 201  * Internal trie getter from a code point.
 202  * Could be faster(?) but longer with
 203  *   if((c32)<=0xd7ff) { (result)=_UTRIE_GET_RAW(trie, data, 0, c32); }
 204  */
 205 #define _UTRIE_GET(trie, data, c32, result, resultType) \
 206     if((uint32_t)(c32)<=0xffff) { \
 207         /* BMP code points */ \
 208         (result)=_UTRIE_GET_FROM_BMP(trie, data, c32); \
 209     } else if((uint32_t)(c32)<=0x10ffff) { \
 210         /* supplementary code point */ \
 211         UChar __lead16=UTF16_LEAD(c32); \
 212         _UTRIE_GET_FROM_PAIR(trie, data, __lead16, c32, result, resultType); \
 213     } else { \
 214         /* out of range */ \
 215         (result)=(resultType)((trie)->initialValue); \
 216     }
 217
 218 /** Internal next-post-increment: get the next code point (c, c2) and its data */
 219 #define _UTRIE_NEXT(trie, data, src, limit, c, c2, result, resultType) { \
 220     (c)=*(src)++; \
 221     if(!UTF_IS_LEAD(c)) { \
 222         (c2)=0; \
 223         (result)=_UTRIE_GET_RAW((trie), data, 0, (c)); \
 224     } else if((src)!=(limit) && UTF_IS_TRAIL((c2)=*(src))) { \
 225         ++(src); \
 226         _UTRIE_GET_FROM_PAIR((trie), data, (c), (c2), (result), resultType); \
 227     } else { \
 228         /* unpaired lead surrogate code point */ \
 229         (c2)=0; \
 230         (result)=_UTRIE_GET_RAW((trie), data, UTRIE_LEAD_INDEX_DISP, (c)); \
 231     } \
 232 }
 233
 234 /** Internal previous: get the previous code point (c, c2) and its data */
 235 #define _UTRIE_PREVIOUS(trie, data, start, src, c, c2, result, resultType) { \
 236     (c)=*--(src); \
 237     if(!UTF_IS_SURROGATE(c)) { \
 238         (c2)=0; \
 239         (result)=_UTRIE_GET_RAW((trie), data, 0, (c)); \
 240     } else if(!UTF_IS_SURROGATE_FIRST(c)) { \
 241         /* trail surrogate */ \
 242         if((start)!=(src) && UTF_IS_LEAD((c2)=*((src)-1))) { \
 243             --(src); \
 244             (result)=(c); (c)=(c2); (c2)=(UChar)(result); /* swap c, c2 */ \
 245             _UTRIE_GET_FROM_PAIR((trie), data, (c), (c2), (result), resultType); \
 246         } else { \
 247             /* unpaired trail surrogate code point */ \
 248             (c2)=0; \
 249             (result)=_UTRIE_GET_RAW((trie), data, 0, (c)); \
 250         } \
 251     } else { \
 252         /* unpaired lead surrogate code point */ \
 253         (c2)=0; \
 254         (result)=_UTRIE_GET_RAW((trie), data, UTRIE_LEAD_INDEX_DISP, (c)); \
 255     } \
 256 }
 257
 258 /* Public UTrie API ---------------------------------------------------------*/
 259
 260 /**
 261  * Get a pointer to the contiguous part of the data array
 262  * for the Latin-1 range (U+0000..U+00ff).
 263  * Must be used only if the Latin-1 range is in fact linear
 264  * (trie->isLatin1Linear).
 265  *
 266  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 267  * @return (const uint16_t *) pointer to values for Latin-1 code points
 268  */
 269 #define UTRIE_GET16_LATIN1(trie) ((trie)->index+(trie)->indexLength+UTRIE_DATA_BLOCK_LENGTH)
 270
 271 /**
 272  * Get a pointer to the contiguous part of the data array
 273  * for the Latin-1 range (U+0000..U+00ff).
 274  * Must be used only if the Latin-1 range is in fact linear
 275  * (trie->isLatin1Linear).
 276  *
 277  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 278  * @return (const uint32_t *) pointer to values for Latin-1 code points
 279  */
 280 #define UTRIE_GET32_LATIN1(trie) ((trie)->data32+UTRIE_DATA_BLOCK_LENGTH)
 281
 282 /**
 283  * Get a 16-bit trie value from a BMP code point (UChar, <=U+ffff).
 284  * c16 may be a lead surrogate, which may have a value including a folding offset.
 285  *
 286  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 287  * @param c16 (UChar, in) the input BMP code point
 288  * @return (uint16_t) trie lookup result
 289  */
 290 #define UTRIE_GET16_FROM_LEAD(trie, c16) _UTRIE_GET_RAW(trie, index, 0, c16)
 291
 292 /**
 293  * Get a 32-bit trie value from a BMP code point (UChar, <=U+ffff).
 294  * c16 may be a lead surrogate, which may have a value including a folding offset.
 295  *
 296  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 297  * @param c16 (UChar, in) the input BMP code point
 298  * @return (uint32_t) trie lookup result
 299  */
 300 #define UTRIE_GET32_FROM_LEAD(trie, c16) _UTRIE_GET_RAW(trie, data32, 0, c16)
 301
 302 /**
 303  * Get a 16-bit trie value from a BMP code point (UChar, <=U+ffff).
 304  * Even lead surrogate code points are treated as normal code points,
 305  * with unfolded values that may differ from _FROM_LEAD() macro results for them.
 306  *
 307  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 308  * @param c16 (UChar, in) the input BMP code point
 309  * @return (uint16_t) trie lookup result
 310  */
 311 #define UTRIE_GET16_FROM_BMP(trie, c16) _UTRIE_GET_FROM_BMP(trie, index, c16)
 312
 313 /**
 314  * Get a 32-bit trie value from a BMP code point (UChar, <=U+ffff).
 315  * Even lead surrogate code points are treated as normal code points,
 316  * with unfolded values that may differ from _FROM_LEAD() macro results for them.
 317  *
 318  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 319  * @param c16 (UChar, in) the input BMP code point
 320  * @return (uint32_t) trie lookup result
 321  */
 322 #define UTRIE_GET32_FROM_BMP(trie, c16) _UTRIE_GET_FROM_BMP(trie, data32, c16)
 323
 324 /**
 325  * Get a 16-bit trie value from a code point.
 326  * Even lead surrogate code points are treated as normal code points,
 327  * with unfolded values that may differ from _FROM_LEAD() macro results for them.
 328  *
 329  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 330  * @param c32 (UChar32, in) the input code point
 331  * @param result (uint16_t, out) uint16_t variable for the trie lookup result
 332  */
 333 #define UTRIE_GET16(trie, c32, result) _UTRIE_GET(trie, index, c32, result, uint16_t)
 334
 335 /**
 336  * Get a 32-bit trie value from a code point.
 337  * Even lead surrogate code points are treated as normal code points,
 338  * with unfolded values that may differ from _FROM_LEAD() macro results for them.
 339  *
 340  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 341  * @param c32 (UChar32, in) the input code point
 342  * @param result (uint32_t, out) uint32_t variable for the trie lookup result
 343  */
 344 #define UTRIE_GET32(trie, c32, result) _UTRIE_GET(trie, data32, c32, result, uint32_t)
 345
 346 /**
 347  * Get the next code point (c, c2), post-increment src,
 348  * and get a 16-bit value from the trie.
 349  *
 350  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 351  * @param src (const UChar *, in/out) the source text pointer
 352  * @param limit (const UChar *, in) the limit pointer for the text, or NULL
 353  * @param c (UChar, out) variable for the BMP or lead code unit
 354  * @param c2 (UChar, out) variable for 0 or the trail code unit
 355  * @param result (uint16_t, out) uint16_t variable for the trie lookup result
 356  */
 357 #define UTRIE_NEXT16(trie, src, limit, c, c2, result) _UTRIE_NEXT(trie, index, src, limit, c, c2, result, uint16_t)
 358
 359 /**
 360  * Get the next code point (c, c2), post-increment src,
 361  * and get a 32-bit value from the trie.
 362  *
 363  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 364  * @param src (const UChar *, in/out) the source text pointer
 365  * @param limit (const UChar *, in) the limit pointer for the text, or NULL
 366  * @param c (UChar, out) variable for the BMP or lead code unit
 367  * @param c2 (UChar, out) variable for 0 or the trail code unit
 368  * @param result (uint32_t, out) uint32_t variable for the trie lookup result
 369  */
 370 #define UTRIE_NEXT32(trie, src, limit, c, c2, result) _UTRIE_NEXT(trie, data32, src, limit, c, c2, result, uint32_t)
 371
 372 /**
 373  * Get the previous code point (c, c2), pre-decrement src,
 374  * and get a 16-bit value from the trie.
 375  *
 376  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 377  * @param start (const UChar *, in) the start pointer for the text, or NULL
 378  * @param src (const UChar *, in/out) the source text pointer
 379  * @param c (UChar, out) variable for the BMP or lead code unit
 380  * @param c2 (UChar, out) variable for 0 or the trail code unit
 381  * @param result (uint16_t, out) uint16_t variable for the trie lookup result
 382  */
 383 #define UTRIE_PREVIOUS16(trie, start, src, c, c2, result) _UTRIE_PREVIOUS(trie, index, start, src, c, c2, result, uint16_t)
 384
 385 /**
 386  * Get the previous code point (c, c2), pre-decrement src,
 387  * and get a 32-bit value from the trie.
 388  *
 389  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 390  * @param start (const UChar *, in) the start pointer for the text, or NULL
 391  * @param src (const UChar *, in/out) the source text pointer
 392  * @param c (UChar, out) variable for the BMP or lead code unit
 393  * @param c2 (UChar, out) variable for 0 or the trail code unit
 394  * @param result (uint32_t, out) uint32_t variable for the trie lookup result
 395  */
 396 #define UTRIE_PREVIOUS32(trie, start, src, c, c2, result) _UTRIE_PREVIOUS(trie, data32, start, src, c, c2, result, uint32_t)
 397
 398 /**
 399  * Get a 16-bit trie value from a pair of surrogates.
 400  *
 401  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 402  * @param c (UChar, in) a lead surrogate
 403  * @param c2 (UChar, in) a trail surrogate
 404  * @param result (uint16_t, out) uint16_t variable for the trie lookup result
 405  */
 406 #define UTRIE_GET16_FROM_PAIR(trie, c, c2, result) _UTRIE_GET_FROM_PAIR(trie, index, c, c2, result, uint16_t)
 407
 408 /**
 409  * Get a 32-bit trie value from a pair of surrogates.
 410  *
 411  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 412  * @param c (UChar, in) a lead surrogate
 413  * @param c2 (UChar, in) a trail surrogate
 414  * @param result (uint32_t, out) uint32_t variable for the trie lookup result
 415  */
 416 #define UTRIE_GET32_FROM_PAIR(trie, c, c2, result) _UTRIE_GET_FROM_PAIR(trie, data32, c, c2, result, uint32_t)
 417
 418 /**
 419  * Get a 16-bit trie value from a folding offset (from the value of a lead surrogate)
 420  * and a trail surrogate.
 421  *
 422  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 423  * @param offset (int32_t, in) the folding offset from the value of a lead surrogate
 424  * @param c2 (UChar, in) a trail surrogate (only the 10 low bits are significant)
 425  * @return (uint16_t) trie lookup result
 426  */
 427 #define UTRIE_GET16_FROM_OFFSET_TRAIL(trie, offset, c2) _UTRIE_GET_RAW(trie, index, offset, (c2)&0x3ff)
 428
 429 /**
 430  * Get a 32-bit trie value from a folding offset (from the value of a lead surrogate)
 431  * and a trail surrogate.
 432  *
 433  * @param trie (const UTrie *, in) a pointer to the runtime trie structure
 434  * @param offset (int32_t, in) the folding offset from the value of a lead surrogate
 435  * @param c2 (UChar, in) a trail surrogate (only the 10 low bits are significant)
 436  * @return (uint32_t) trie lookup result
 437  */
 438 #define UTRIE_GET32_FROM_OFFSET_TRAIL(trie, offset, c2) _UTRIE_GET_RAW(trie, data32, offset, (c2)&0x3ff)
 439
 440 /* enumeration callback types */
 441
 442 /**
 443  * Callback from utrie_enum(), extracts a uint32_t value from a
 444  * trie value. This value will be passed on to the UTrieEnumRange function.
 445  *
 446  * @param context an opaque pointer, as passed into utrie_enum()
 447  * @param value a value from the trie
 448  * @return the value that is to be passed on to the UTrieEnumRange function
 449  */
 450 typedef uint32_t U_CALLCONV
 451 UTrieEnumValue(const void *context, uint32_t value);
 452
 453 /**
 454  * Callback from utrie_enum(), is called for each contiguous range
 455  * of code points with the same value as retrieved from the trie and
 456  * transformed by the UTrieEnumValue function.
 457  *
 458  * The callback function can stop the enumeration by returning FALSE.
 459  *
 460  * @param context an opaque pointer, as passed into utrie_enum()
 461  * @param start the first code point in a contiguous range with value
 462  * @param limit one past the last code point in a contiguous range with value
 463  * @param value the value that is set for all code points in [start..limit[
 464  * @return FALSE to stop the enumeration
 465  */
 466 typedef UBool U_CALLCONV
 467 UTrieEnumRange(const void *context, UChar32 start, UChar32 limit, uint32_t value);
 468
 469 /**
 470  * Enumerate efficiently all values in a trie.
 471  * For each entry in the trie, the value to be delivered is passed through
 472  * the UTrieEnumValue function.
 473  * The value is unchanged if that function pointer is NULL.
 474  *
 475  * For each contiguous range of code points with a given value,
 476  * the UTrieEnumRange function is called.
 477  *
 478  * @param trie a pointer to the runtime trie structure
 479  * @param enumValue a pointer to a function that may transform the trie entry value,
 480  *                  or NULL if the values from the trie are to be used directly
 481  * @param enumRange a pointer to a function that is called for each contiguous range
 482  *                  of code points with the same value
 483  * @param context an opaque pointer that is passed on to the callback functions
 484  */
 485 U_CAPI void U_EXPORT2
 486 utrie_enum(const UTrie *trie,
 487            UTrieEnumValue *enumValue, UTrieEnumRange *enumRange, const void *context);
 488
 489 /**
 490  * Unserialize a trie from 32-bit-aligned memory.
 491  * Inverse of utrie_serialize().
 492  * Fills the UTrie runtime trie structure with the settings for the trie data.
 493  *
 494  * @param trie a pointer to the runtime trie structure
 495  * @param data a pointer to 32-bit-aligned memory containing trie data
 496  * @param length the number of bytes available at data
 497  * @param pErrorCode an in/out ICU UErrorCode
 498  * @return the number of bytes at data taken up by the trie data
 499  */
 500 U_CAPI int32_t U_EXPORT2
 501 utrie_unserialize(UTrie *trie, const void *data, int32_t length, UErrorCode *pErrorCode);
 502
 503 /**
 504  * "Unserialize" a dummy trie.
 505  * A dummy trie is an empty runtime trie, used when a real data trie cannot
 506  * be loaded.
 507  *
 508  * The input memory is filled so that the trie always returns the initialValue,
 509  * or the leadUnitValue for lead surrogate code points.
 510  * The Latin-1 part is always set up to be linear.
 511  *
 512  * @param trie a pointer to the runtime trie structure
 513  * @param data a pointer to 32-bit-aligned memory to be filled with the dummy trie data
 514  * @param length the number of bytes available at data (recommended to use UTRIE_DUMMY_SIZE)
 515  * @param initialValue the initial value that is set for all code points
 516  * @param leadUnitValue the value for lead surrogate code _units_ that do not
 517  *                      have associated supplementary data
 518  * @param pErrorCode an in/out ICU UErrorCode
 519  *
 520  * @see UTRIE_DUMMY_SIZE
 521  * @see utrie_open
 522  */
 523 U_CAPI int32_t U_EXPORT2
 524 utrie_unserializeDummy(UTrie *trie,
 525                        void *data, int32_t length,
 526                        uint32_t initialValue, uint32_t leadUnitValue,
 527                        UBool make16BitTrie,
 528                        UErrorCode *pErrorCode);
 529
 530 /**
 531  * Default implementation for UTrie.getFoldingOffset, set automatically by
 532  * utrie_unserialize().
 533  * Simply returns the lead surrogate's value itself - which is the inverse
 534  * of the default folding function used by utrie_serialize().
 535  * Exported for static const UTrie structures.
 536  *
 537  * @see UTrieGetFoldingOffset
 538  */
 539 U_CAPI int32_t U_EXPORT2
 540 utrie_defaultGetFoldingOffset(uint32_t data);
 541
 542 /* Building a trie ----------------------------------------------------------*/
 543
 544 /**
 545  * Build-time trie structure.
 546  * Opaque definition, here only to make fillIn parameters possible
 547  * for utrie_open() and utrie_clone().
 548  */
 549 struct UNewTrie {
 550     /**
 551      * Index values at build-time are 32 bits wide for easier processing.
 552      * Bit 31 is set if the data block is used by multiple index values (from utrie_setRange()).
 553      */
 554     int32_t index[UTRIE_MAX_INDEX_LENGTH];
 555     uint32_t *data;
 556
 557     uint32_t leadUnitValue;
 558     int32_t indexLength, dataCapacity, dataLength;
 559     UBool isAllocated, isDataAllocated;
 560     UBool isLatin1Linear, isCompacted;
 561
 562     /**
 563      * Map of adjusted indexes, used in utrie_compact().
 564      * Maps from original indexes to new ones.
 565      */
 566     int32_t map[UTRIE_MAX_BUILD_TIME_DATA_LENGTH>>UTRIE_SHIFT];
 567 };
 568
 569 typedef struct UNewTrie UNewTrie;
 570
 571 /**
 572  * Build-time trie callback function, used with utrie_serialize().
 573  * This function calculates a lead surrogate's value including a folding offset
 574  * from the 1024 supplementary code points [start..start+1024[ .
 575  * It is U+10000 <= start <= U+10fc00 and (start&0x3ff)==0.
 576  *
 577  * The folding offset is provided by the caller.
 578  * It is offset=UTRIE_BMP_INDEX_LENGTH+n*UTRIE_SURROGATE_BLOCK_COUNT with n=0..1023.
 579  * Instead of the offset itself, n can be stored in 10 bits -
 580  * or fewer if it can be assumed that few lead surrogates have associated data.
 581  *
 582  * The returned value must be
 583  * - not zero if and only if there is relevant data
 584  *   for the corresponding 1024 supplementary code points
 585  * - such that UTrie.getFoldingOffset(UNewTrieGetFoldedValue(..., offset))==offset
 586  *
 587  * @return a folded value, or 0 if there is no relevant data for the lead surrogate.
 588  */
 589 typedef uint32_t U_CALLCONV
 590 UNewTrieGetFoldedValue(UNewTrie *trie, UChar32 start, int32_t offset);
 591
 592 /**
 593  * Open a build-time trie structure.
 594  * The size of the build-time data array is specified to avoid allocating a large
 595  * array in all cases. The array itself can also be passed in.
 596  *
 597  * Although the trie is never fully expanded to a linear array, especially when
 598  * utrie_setRange32() is used, the data array could be large during build time.
 599  * The maximum length is
 600  * UTRIE_MAX_BUILD_TIME_DATA_LENGTH=0x110000+UTRIE_DATA_BLOCK_LENGTH+0x400.
 601  * (Number of Unicode code points + one all-initial-value block +
 602  *  possible duplicate entries for 1024 lead surrogates.)
 603  * (UTRIE_DATA_BLOCK_LENGTH<=0x200 in all cases.)
 604  *
 605  * @param fillIn a pointer to a UNewTrie structure to be initialized (will not be released), or
 606  *               NULL if one is to be allocated
 607  * @param aliasData a pointer to a data array to be used (will not be released), or
 608  *                  NULL if one is to be allocated
 609  * @param maxDataLength the capacity of aliasData (if not NULL) or
 610  *                      the length of the data array to be allocated
 611  * @param initialValue the initial value that is set for all code points
 612  * @param leadUnitValue the value for lead surrogate code _units_ that do not
 613  *                      have associated supplementary data
 614  * @param latin1Linear a flag indicating whether the Latin-1 range is to be allocated and
 615  *                     kept in a linear, contiguous part of the data array
 616  * @return a pointer to the initialized fillIn or the allocated and initialized new UNewTrie
 617  */
 618 U_CAPI UNewTrie * U_EXPORT2
 619 utrie_open(UNewTrie *fillIn,
 620            uint32_t *aliasData, int32_t maxDataLength,
 621            uint32_t initialValue, uint32_t leadUnitValue,
 622            UBool latin1Linear);
 623
 624 /**
 625  * Clone a build-time trie structure with all entries.
 626  *
 627  * @param fillIn like in utrie_open()
 628  * @param other the build-time trie structure to clone
 629  * @param aliasData like in utrie_open(),
 630  *                  used if aliasDataLength>=(capacity of other's data array)
 631  * @param aliasDataLength the length of aliasData
 632  * @return a pointer to the initialized fillIn or the allocated and initialized new UNewTrie
 633  */
 634 U_CAPI UNewTrie * U_EXPORT2
 635 utrie_clone(UNewTrie *fillIn, const UNewTrie *other, uint32_t *aliasData, int32_t aliasDataLength);
 636
 637 /**
 638  * Close a build-time trie structure, and release memory
 639  * that was allocated by utrie_open() or utrie_clone().
 640  *
 641  * @param trie the build-time trie
 642  */
 643 U_CAPI void U_EXPORT2
 644 utrie_close(UNewTrie *trie);
 645
 646 /**
 647  * Get the data array of a build-time trie.
 648  * The data may be modified, but entries that are equal before
 649  * must still be equal after modification.
 650  *
 651  * @param trie the build-time trie
 652  * @param pLength (out) a pointer to a variable that receives the number
 653  *                of entries in the data array
 654  * @return the data array
 655  */
 656 U_CAPI uint32_t * U_EXPORT2
 657 utrie_getData(UNewTrie *trie, int32_t *pLength);
 658
 659 /**
 660  * Set a value for a code point.
 661  *
 662  * @param trie the build-time trie
 663  * @param c the code point
 664  * @param value the value
 665  * @return FALSE if a failure occurred (illegal argument or data array overrun)
 666  */
 667 U_CAPI UBool U_EXPORT2
 668 utrie_set32(UNewTrie *trie, UChar32 c, uint32_t value);
 669
 670 /**
 671  * Get a value from a code point as stored in the build-time trie.
 672  *
 673  * @param trie the build-time trie
 674  * @param c the code point
 675  * @param pInBlockZero if not NULL, then *pInBlockZero is set to TRUE
 676  *                     iff the value is retrieved from block 0;
 677  *                     block 0 is the all-initial-value initial block
 678  * @return the value
 679  */
 680 U_CAPI uint32_t U_EXPORT2
 681 utrie_get32(UNewTrie *trie, UChar32 c, UBool *pInBlockZero);
 682
 683 /**
 684  * Set a value in a range of code points [start..limit[.
 685  * All code points c with start<=c<limit will get the value if
 686  * overwrite is TRUE or if the old value is 0.
 687  *
 688  * @param trie the build-time trie
 689  * @param start the first code point to get the value
 690  * @param limit one past the last code point to get the value
 691  * @param value the value
 692  * @param overwrite flag for whether old non-initial values are to be overwritten
 693  * @return FALSE if a failure occurred (illegal argument or data array overrun)
 694  */
 695 U_CAPI UBool U_EXPORT2
 696 utrie_setRange32(UNewTrie *trie, UChar32 start, UChar32 limit, uint32_t value, UBool overwrite);
 697
 698 /**
 699  * Compact the build-time trie after all values are set, and then
 700  * serialize it into 32-bit aligned memory.
 701  *
 702  * After this, the trie can only be serizalized again and/or closed;
 703  * no further values can be added.
 704  *
 705  * @see utrie_unserialize()
 706  *
 707  * @param trie the build-time trie
 708  * @param data a pointer to 32-bit-aligned memory for the trie data
 709  * @param capacity the number of bytes available at data
 710  * @param getFoldedValue a callback function that calculates the value for
 711  *                       a lead surrogate from all of its supplementary code points
 712  *                       and the folding offset;
 713  *                       if NULL, then a default function is used which returns just
 714  *                       the input offset when there are any non-initial-value entries
 715  * @param reduceTo16Bits flag for whether the values are to be reduced to a
 716  *                       width of 16 bits for serialization and runtime
 717  * @param pErrorCode a UErrorCode argument; among other possible error codes:
 718  * - U_BUFFER_OVERFLOW_ERROR if the data storage block is too small for serialization
 719  * - U_MEMORY_ALLOCATION_ERROR if the trie data array is too small
 720  * - U_INDEX_OUTOFBOUNDS_ERROR if the index or data arrays are too long after compaction for serialization
 721  *
 722  * @return the number of bytes written for the trie
 723  */
 724 U_CAPI int32_t U_EXPORT2
 725 utrie_serialize(UNewTrie *trie, void *data, int32_t capacity,
 726                 UNewTrieGetFoldedValue *getFoldedValue,
 727                 UBool reduceTo16Bits,
 728                 UErrorCode *pErrorCode);
 729
 730 /**
 731  * Swap a serialized UTrie.
 732  * @internal
 733  */
 734 U_CAPI int32_t U_EXPORT2
 735 utrie_swap(const UDataSwapper *ds,
 736            const void *inData, int32_t length, void *outData,
 737            UErrorCode *pErrorCode);
 738
 739 /* serialization ------------------------------------------------------------ */
 740
 741 /**
 742  * Trie data structure in serialized form:
 743  *
 744  * UTrieHeader header;
 745  * uint16_t index[header.indexLength];
 746  * uint16_t data[header.dataLength];
 747  * @internal
 748  */
 749 typedef struct UTrieHeader {
 750     /** "Trie" in big-endian US-ASCII (0x54726965) */
 751     uint32_t signature;
 752
 753     /**
 754      * options bit field:
 755      *     9    1=Latin-1 data is stored linearly at data+UTRIE_DATA_BLOCK_LENGTH
 756      *     8    0=16-bit data, 1=32-bit data
 757      *  7..4    UTRIE_INDEX_SHIFT   // 0..UTRIE_SHIFT
 758      *  3..0    UTRIE_SHIFT         // 1..9
 759      */
 760     uint32_t options;
 761
 762     /** indexLength is a multiple of UTRIE_SURROGATE_BLOCK_COUNT */
 763     int32_t indexLength;
 764
 765     /** dataLength>=UTRIE_DATA_BLOCK_LENGTH */
 766     int32_t dataLength;
 767 } UTrieHeader;
 768
 769 /**
 770  * Constants for use with UTrieHeader.options.
 771  * @internal
 772  */
 773 enum {
 774     /** Mask to get the UTRIE_SHIFT value from options. */
 775     UTRIE_OPTIONS_SHIFT_MASK=0xf,
 776
 777     /** Shift options right this much to get the UTRIE_INDEX_SHIFT value. */
 778     UTRIE_OPTIONS_INDEX_SHIFT=4,
 779
 780     /** If set, then the data (stage 2) array is 32 bits wide. */
 781     UTRIE_OPTIONS_DATA_IS_32_BIT=0x100,
 782
 783     /**
 784      * If set, then Latin-1 data (for U+0000..U+00ff) is stored in the data (stage 2) array
 785      * as a simple, linear array at data+UTRIE_DATA_BLOCK_LENGTH.
 786      */
 787     UTRIE_OPTIONS_LATIN1_IS_LINEAR=0x200
 788 };
 789
 790 U_CDECL_END
 791
 792 #endif