icuSources/common/utrie2.h

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 ******************************************************************************
   5 *
   6 *   Copyright (C) 2001-2014, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 ******************************************************************************
  10 *   file name:  utrie2.h
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 2008aug16 (starting from a copy of utrie.h)
  16 *   created by: Markus W. Scherer
  17 */
  18
  19 #ifndef __UTRIE2_H__
  20 #define __UTRIE2_H__
  21
  22 #include "unicode/utypes.h"
  23 #include "putilimp.h"
  24 #include "udataswp.h"
  25
  26 U_CDECL_BEGIN
  27
  28 struct UTrie;  /* forward declaration */
  29 #ifndef __UTRIE_H__
  30 typedef struct UTrie UTrie;
  31 #endif
  32
  33 /**
  34  * \file
  35  *
  36  * This is a common implementation of a Unicode trie.
  37  * It is a kind of compressed, serializable table of 16- or 32-bit values associated with
  38  * Unicode code points (0..0x10ffff). (A map from code points to integers.)
  39  *
  40  * This is the second common version of a Unicode trie (hence the name UTrie2).
  41  * Compared with UTrie version 1:
  42  * - Still splitting BMP code points 11:5 bits for index and data table lookups.
  43  * - Still separate data for lead surrogate code _units_ vs. code _points_,
  44  *   but the lead surrogate code unit values are not required any more
  45  *   for data lookup for supplementary code points.
  46  * - The "folding" mechanism is removed. In UTrie version 1, this somewhat
  47  *   hard-to-explain mechanism was meant to be used for optimized UTF-16
  48  *   processing, with application-specific encoding of indexing bits
  49  *   in the lead surrogate data for the associated supplementary code points.
  50  * - For the last single-value code point range (ending with U+10ffff),
  51  *   the starting code point ("highStart") and the value are stored.
  52  * - For supplementary code points U+10000..highStart-1 a three-table lookup
  53  *   (two index tables and one data table) is used. The first index
  54  *   is truncated, omitting both the BMP portion and the high range.
  55  * - There is a special small index for 2-byte UTF-8, and the initial data
  56  *   entries are designed for fast 1/2-byte UTF-8 lookup.
  57  */
  58
  59 /**
  60  * Trie structure.
  61  * Use only with public API macros and functions.
  62  */
  63 struct UTrie2;
  64 typedef struct UTrie2 UTrie2;
  65
  66 /* Public UTrie2 API functions: read-only access ---------------------------- */
  67
  68 /**
  69  * Selectors for the width of a UTrie2 data value.
  70  */
  71 enum UTrie2ValueBits {
  72     /** 16 bits per UTrie2 data value. */
  73     UTRIE2_16_VALUE_BITS,
  74     /** 32 bits per UTrie2 data value. */
  75     UTRIE2_32_VALUE_BITS,
  76     /** Number of selectors for the width of UTrie2 data values. */
  77     UTRIE2_COUNT_VALUE_BITS
  78 };
  79 typedef enum UTrie2ValueBits UTrie2ValueBits;
  80
  81 /**
  82  * Open a frozen trie from its serialized from, stored in 32-bit-aligned memory.
  83  * Inverse of utrie2_serialize().
  84  * The memory must remain valid and unchanged as long as the trie is used.
  85  * You must utrie2_close() the trie once you are done using it.
  86  *
  87  * @param valueBits selects the data entry size; results in an
  88  *                  U_INVALID_FORMAT_ERROR if it does not match the serialized form
  89  * @param data a pointer to 32-bit-aligned memory containing the serialized form of a UTrie2
  90  * @param length the number of bytes available at data;
  91  *               can be more than necessary
  92  * @param pActualLength receives the actual number of bytes at data taken up by the trie data;
  93  *                      can be NULL
  94  * @param pErrorCode an in/out ICU UErrorCode
  95  * @return the unserialized trie
  96  *
  97  * @see utrie2_open
  98  * @see utrie2_serialize
  99  */
 100 U_CAPI UTrie2 * U_EXPORT2
 101 utrie2_openFromSerialized(UTrie2ValueBits valueBits,
 102                           const void *data, int32_t length, int32_t *pActualLength,
 103                           UErrorCode *pErrorCode);
 104
 105 /**
 106  * Open a frozen, empty "dummy" trie.
 107  * A dummy trie is an empty trie, used when a real data trie cannot
 108  * be loaded. Equivalent to calling utrie2_open() and utrie2_freeze(),
 109  * but without internally creating and compacting/serializing the
 110  * builder data structure.
 111  *
 112  * The trie always returns the initialValue,
 113  * or the errorValue for out-of-range code points and illegal UTF-8.
 114  *
 115  * You must utrie2_close() the trie once you are done using it.
 116  *
 117  * @param valueBits selects the data entry size
 118  * @param initialValue the initial value that is set for all code points
 119  * @param errorValue the value for out-of-range code points and illegal UTF-8
 120  * @param pErrorCode an in/out ICU UErrorCode
 121  * @return the dummy trie
 122  *
 123  * @see utrie2_openFromSerialized
 124  * @see utrie2_open
 125  */
 126 U_CAPI UTrie2 * U_EXPORT2
 127 utrie2_openDummy(UTrie2ValueBits valueBits,
 128                  uint32_t initialValue, uint32_t errorValue,
 129                  UErrorCode *pErrorCode);
 130
 131 /**
 132  * Get a value from a code point as stored in the trie.
 133  * Easier to use than UTRIE2_GET16() and UTRIE2_GET32() but slower.
 134  * Easier to use because, unlike the macros, this function works on all UTrie2
 135  * objects, frozen or not, holding 16-bit or 32-bit data values.
 136  *
 137  * @param trie the trie
 138  * @param c the code point
 139  * @return the value
 140  */
 141 U_CAPI uint32_t U_EXPORT2
 142 utrie2_get32(const UTrie2 *trie, UChar32 c);
 143
 144 /* enumeration callback types */
 145
 146 /**
 147  * Callback from utrie2_enum(), extracts a uint32_t value from a
 148  * trie value. This value will be passed on to the UTrie2EnumRange function.
 149  *
 150  * @param context an opaque pointer, as passed into utrie2_enum()
 151  * @param value a value from the trie
 152  * @return the value that is to be passed on to the UTrie2EnumRange function
 153  */
 154 typedef uint32_t U_CALLCONV
 155 UTrie2EnumValue(const void *context, uint32_t value);
 156
 157 /**
 158  * Callback from utrie2_enum(), is called for each contiguous range
 159  * of code points with the same value as retrieved from the trie and
 160  * transformed by the UTrie2EnumValue function.
 161  *
 162  * The callback function can stop the enumeration by returning FALSE.
 163  *
 164  * @param context an opaque pointer, as passed into utrie2_enum()
 165  * @param start the first code point in a contiguous range with value
 166  * @param end the last code point in a contiguous range with value (inclusive)
 167  * @param value the value that is set for all code points in [start..end]
 168  * @return FALSE to stop the enumeration
 169  */
 170 typedef UBool U_CALLCONV
 171 UTrie2EnumRange(const void *context, UChar32 start, UChar32 end, uint32_t value);
 172
 173 /**
 174  * Enumerate efficiently all values in a trie.
 175  * Do not modify the trie during the enumeration.
 176  *
 177  * For each entry in the trie, the value to be delivered is passed through
 178  * the UTrie2EnumValue function.
 179  * The value is unchanged if that function pointer is NULL.
 180  *
 181  * For each contiguous range of code points with a given (transformed) value,
 182  * the UTrie2EnumRange function is called.
 183  *
 184  * @param trie a pointer to the trie
 185  * @param enumValue a pointer to a function that may transform the trie entry value,
 186  *                  or NULL if the values from the trie are to be used directly
 187  * @param enumRange a pointer to a function that is called for each contiguous range
 188  *                  of code points with the same (transformed) value
 189  * @param context an opaque pointer that is passed on to the callback functions
 190  */
 191 U_CAPI void U_EXPORT2
 192 utrie2_enum(const UTrie2 *trie,
 193             UTrie2EnumValue *enumValue, UTrie2EnumRange *enumRange, const void *context);
 194
 195 /* Building a trie ---------------------------------------------------------- */
 196
 197 /**
 198  * Open an empty, writable trie. At build time, 32-bit data values are used.
 199  * utrie2_freeze() takes a valueBits parameter
 200  * which determines the data value width in the serialized and frozen forms.
 201  * You must utrie2_close() the trie once you are done using it.
 202  *
 203  * @param initialValue the initial value that is set for all code points
 204  * @param errorValue the value for out-of-range code points and illegal UTF-8
 205  * @param pErrorCode an in/out ICU UErrorCode
 206  * @return a pointer to the allocated and initialized new trie
 207  */
 208 U_CAPI UTrie2 * U_EXPORT2
 209 utrie2_open(uint32_t initialValue, uint32_t errorValue, UErrorCode *pErrorCode);
 210
 211 /**
 212  * Clone a trie.
 213  * You must utrie2_close() the clone once you are done using it.
 214  *
 215  * @param other the trie to clone
 216  * @param pErrorCode an in/out ICU UErrorCode
 217  * @return a pointer to the new trie clone
 218  */
 219 U_CAPI UTrie2 * U_EXPORT2
 220 utrie2_clone(const UTrie2 *other, UErrorCode *pErrorCode);
 221
 222 /**
 223  * Clone a trie. The clone will be mutable/writable even if the other trie
 224  * is frozen. (See utrie2_freeze().)
 225  * You must utrie2_close() the clone once you are done using it.
 226  *
 227  * @param other the trie to clone
 228  * @param pErrorCode an in/out ICU UErrorCode
 229  * @return a pointer to the new trie clone
 230  */
 231 U_CAPI UTrie2 * U_EXPORT2
 232 utrie2_cloneAsThawed(const UTrie2 *other, UErrorCode *pErrorCode);
 233
 234 /**
 235  * Close a trie and release associated memory.
 236  *
 237  * @param trie the trie
 238  */
 239 U_CAPI void U_EXPORT2
 240 utrie2_close(UTrie2 *trie);
 241
 242 /**
 243  * Set a value for a code point.
 244  *
 245  * @param trie the unfrozen trie
 246  * @param c the code point
 247  * @param value the value
 248  * @param pErrorCode an in/out ICU UErrorCode; among other possible error codes:
 249  * - U_NO_WRITE_PERMISSION if the trie is frozen
 250  */
 251 U_CAPI void U_EXPORT2
 252 utrie2_set32(UTrie2 *trie, UChar32 c, uint32_t value, UErrorCode *pErrorCode);
 253
 254 /**
 255  * Set a value in a range of code points [start..end].
 256  * All code points c with start<=c<=end will get the value if
 257  * overwrite is TRUE or if the old value is the initial value.
 258  *
 259  * @param trie the unfrozen trie
 260  * @param start the first code point to get the value
 261  * @param end the last code point to get the value (inclusive)
 262  * @param value the value
 263  * @param overwrite flag for whether old non-initial values are to be overwritten
 264  * @param pErrorCode an in/out ICU UErrorCode; among other possible error codes:
 265  * - U_NO_WRITE_PERMISSION if the trie is frozen
 266  */
 267 U_CAPI void U_EXPORT2
 268 utrie2_setRange32(UTrie2 *trie,
 269                   UChar32 start, UChar32 end,
 270                   uint32_t value, UBool overwrite,
 271                   UErrorCode *pErrorCode);
 272
 273 /**
 274  * Freeze a trie. Make it immutable (read-only) and compact it,
 275  * ready for serialization and for use with fast macros.
 276  * Functions to set values will fail after serializing.
 277  *
 278  * A trie can be frozen only once. If this function is called again with different
 279  * valueBits then it will set a U_ILLEGAL_ARGUMENT_ERROR.
 280  *
 281  * @param trie the trie
 282  * @param valueBits selects the data entry size; if smaller than 32 bits, then
 283  *                  the values stored in the trie will be truncated
 284  * @param pErrorCode an in/out ICU UErrorCode; among other possible error codes:
 285  * - U_INDEX_OUTOFBOUNDS_ERROR if the compacted index or data arrays are too long
 286  *                             for serialization
 287  *                             (the trie will be immutable and usable,
 288  *                             but not frozen and not usable with the fast macros)
 289  *
 290  * @see utrie2_cloneAsThawed
 291  */
 292 U_CAPI void U_EXPORT2
 293 utrie2_freeze(UTrie2 *trie, UTrie2ValueBits valueBits, UErrorCode *pErrorCode);
 294
 295 /**
 296  * Test if the trie is frozen. (See utrie2_freeze().)
 297  *
 298  * @param trie the trie
 299  * @return TRUE if the trie is frozen, that is, immutable, ready for serialization
 300  *         and for use with fast macros
 301  */
 302 U_CAPI UBool U_EXPORT2
 303 utrie2_isFrozen(const UTrie2 *trie);
 304
 305 /**
 306  * Serialize a frozen trie into 32-bit aligned memory.
 307  * If the trie is not frozen, then the function returns with a U_ILLEGAL_ARGUMENT_ERROR.
 308  * A trie can be serialized multiple times.
 309  *
 310  * @param trie the frozen trie
 311  * @param data a pointer to 32-bit-aligned memory to be filled with the trie data,
 312  *             can be NULL if capacity==0
 313  * @param capacity the number of bytes available at data,
 314  *                 or 0 for preflighting
 315  * @param pErrorCode an in/out ICU UErrorCode; among other possible error codes:
 316  * - U_BUFFER_OVERFLOW_ERROR if the data storage block is too small for serialization
 317  * - U_ILLEGAL_ARGUMENT_ERROR if the trie is not frozen or the data and capacity
 318  *                            parameters are bad
 319  * @return the number of bytes written or needed for the trie
 320  *
 321  * @see utrie2_openFromSerialized()
 322  */
 323 U_CAPI int32_t U_EXPORT2
 324 utrie2_serialize(const UTrie2 *trie,
 325                  void *data, int32_t capacity,
 326                  UErrorCode *pErrorCode);
 327
 328 /* Public UTrie2 API: miscellaneous functions ------------------------------- */
 329
 330 /**
 331  * Get the UTrie version from 32-bit-aligned memory containing the serialized form
 332  * of either a UTrie (version 1) or a UTrie2 (version 2).
 333  *
 334  * @param data a pointer to 32-bit-aligned memory containing the serialized form
 335  *             of a UTrie, version 1 or 2
 336  * @param length the number of bytes available at data;
 337  *               can be more than necessary (see return value)
 338  * @param anyEndianOk If FALSE, only platform-endian serialized forms are recognized.
 339  *                    If TRUE, opposite-endian serialized forms are recognized as well.
 340  * @return the UTrie version of the serialized form, or 0 if it is not
 341  *         recognized as a serialized UTrie
 342  */
 343 U_CAPI int32_t U_EXPORT2
 344 utrie2_getVersion(const void *data, int32_t length, UBool anyEndianOk);
 345
 346 /**
 347  * Swap a serialized UTrie2.
 348  * @internal
 349  */
 350 U_CAPI int32_t U_EXPORT2
 351 utrie2_swap(const UDataSwapper *ds,
 352             const void *inData, int32_t length, void *outData,
 353             UErrorCode *pErrorCode);
 354
 355 /**
 356  * Swap a serialized UTrie or UTrie2.
 357  * @internal
 358  */
 359 U_CAPI int32_t U_EXPORT2
 360 utrie2_swapAnyVersion(const UDataSwapper *ds,
 361                       const void *inData, int32_t length, void *outData,
 362                       UErrorCode *pErrorCode);
 363
 364 /**
 365  * Build a UTrie2 (version 2) from a UTrie (version 1).
 366  * Enumerates all values in the UTrie and builds a UTrie2 with the same values.
 367  * The resulting UTrie2 will be frozen.
 368  *
 369  * @param trie1 the runtime UTrie structure to be enumerated
 370  * @param errorValue the value for out-of-range code points and illegal UTF-8
 371  * @param pErrorCode an in/out ICU UErrorCode
 372  * @return The frozen UTrie2 with the same values as the UTrie.
 373  */
 374 U_CAPI UTrie2 * U_EXPORT2
 375 utrie2_fromUTrie(const UTrie *trie1, uint32_t errorValue, UErrorCode *pErrorCode);
 376
 377 /* Public UTrie2 API macros ------------------------------------------------- */
 378
 379 /*
 380  * These macros provide fast data lookup from a frozen trie.
 381  * They will crash when used on an unfrozen trie.
 382  */
 383
 384 /**
 385  * Return a 16-bit trie value from a code point, with range checking.
 386  * Returns trie->errorValue if c is not in the range 0..U+10ffff.
 387  *
 388  * @param trie (const UTrie2 *, in) a frozen trie
 389  * @param c (UChar32, in) the input code point
 390  * @return (uint16_t) The code point's trie value.
 391  */
 392 #define UTRIE2_GET16(trie, c) _UTRIE2_GET((trie), index, (trie)->indexLength, (c))
 393
 394 /**
 395  * Return a 32-bit trie value from a code point, with range checking.
 396  * Returns trie->errorValue if c is not in the range 0..U+10ffff.
 397  *
 398  * @param trie (const UTrie2 *, in) a frozen trie
 399  * @param c (UChar32, in) the input code point
 400  * @return (uint32_t) The code point's trie value.
 401  */
 402 #define UTRIE2_GET32(trie, c) _UTRIE2_GET((trie), data32, 0, (c))
 403
 404 /**
 405  * UTF-16: Get the next code point (UChar32 c, out), post-increment src,
 406  * and get a 16-bit value from the trie.
 407  *
 408  * @param trie (const UTrie2 *, in) a frozen trie
 409  * @param src (const UChar *, in/out) the source text pointer
 410  * @param limit (const UChar *, in) the limit pointer for the text, or NULL if NUL-terminated
 411  * @param c (UChar32, out) variable for the code point
 412  * @param result (uint16_t, out) uint16_t variable for the trie lookup result
 413  */
 414 #define UTRIE2_U16_NEXT16(trie, src, limit, c, result) _UTRIE2_U16_NEXT(trie, index, src, limit, c, result)
 415
 416 /**
 417  * UTF-16: Get the next code point (UChar32 c, out), post-increment src,
 418  * and get a 32-bit value from the trie.
 419  *
 420  * @param trie (const UTrie2 *, in) a frozen trie
 421  * @param src (const UChar *, in/out) the source text pointer
 422  * @param limit (const UChar *, in) the limit pointer for the text, or NULL if NUL-terminated
 423  * @param c (UChar32, out) variable for the code point
 424  * @param result (uint32_t, out) uint32_t variable for the trie lookup result
 425  */
 426 #define UTRIE2_U16_NEXT32(trie, src, limit, c, result) _UTRIE2_U16_NEXT(trie, data32, src, limit, c, result)
 427
 428 /**
 429  * UTF-16: Get the previous code point (UChar32 c, out), pre-decrement src,
 430  * and get a 16-bit value from the trie.
 431  *
 432  * @param trie (const UTrie2 *, in) a frozen trie
 433  * @param start (const UChar *, in) the start pointer for the text
 434  * @param src (const UChar *, in/out) the source text pointer
 435  * @param c (UChar32, out) variable for the code point
 436  * @param result (uint16_t, out) uint16_t variable for the trie lookup result
 437  */
 438 #define UTRIE2_U16_PREV16(trie, start, src, c, result) _UTRIE2_U16_PREV(trie, index, start, src, c, result)
 439
 440 /**
 441  * UTF-16: Get the previous code point (UChar32 c, out), pre-decrement src,
 442  * and get a 32-bit value from the trie.
 443  *
 444  * @param trie (const UTrie2 *, in) a frozen trie
 445  * @param start (const UChar *, in) the start pointer for the text
 446  * @param src (const UChar *, in/out) the source text pointer
 447  * @param c (UChar32, out) variable for the code point
 448  * @param result (uint32_t, out) uint32_t variable for the trie lookup result
 449  */
 450 #define UTRIE2_U16_PREV32(trie, start, src, c, result) _UTRIE2_U16_PREV(trie, data32, start, src, c, result)
 451
 452 /**
 453  * UTF-8: Post-increment src and get a 16-bit value from the trie.
 454  *
 455  * @param trie (const UTrie2 *, in) a frozen trie
 456  * @param src (const char *, in/out) the source text pointer
 457  * @param limit (const char *, in) the limit pointer for the text (must not be NULL)
 458  * @param result (uint16_t, out) uint16_t variable for the trie lookup result
 459  */
 460 #define UTRIE2_U8_NEXT16(trie, src, limit, result)\
 461     _UTRIE2_U8_NEXT(trie, data16, index, src, limit, result)
 462
 463 /**
 464  * UTF-8: Post-increment src and get a 32-bit value from the trie.
 465  *
 466  * @param trie (const UTrie2 *, in) a frozen trie
 467  * @param src (const char *, in/out) the source text pointer
 468  * @param limit (const char *, in) the limit pointer for the text (must not be NULL)
 469  * @param result (uint16_t, out) uint32_t variable for the trie lookup result
 470  */
 471 #define UTRIE2_U8_NEXT32(trie, src, limit, result) \
 472     _UTRIE2_U8_NEXT(trie, data32, data32, src, limit, result)
 473
 474 /**
 475  * UTF-8: Pre-decrement src and get a 16-bit value from the trie.
 476  *
 477  * @param trie (const UTrie2 *, in) a frozen trie
 478  * @param start (const char *, in) the start pointer for the text
 479  * @param src (const char *, in/out) the source text pointer
 480  * @param result (uint16_t, out) uint16_t variable for the trie lookup result
 481  */
 482 #define UTRIE2_U8_PREV16(trie, start, src, result) \
 483     _UTRIE2_U8_PREV(trie, data16, index, start, src, result)
 484
 485 /**
 486  * UTF-8: Pre-decrement src and get a 32-bit value from the trie.
 487  *
 488  * @param trie (const UTrie2 *, in) a frozen trie
 489  * @param start (const char *, in) the start pointer for the text
 490  * @param src (const char *, in/out) the source text pointer
 491  * @param result (uint16_t, out) uint32_t variable for the trie lookup result
 492  */
 493 #define UTRIE2_U8_PREV32(trie, start, src, result) \
 494     _UTRIE2_U8_PREV(trie, data32, data32, start, src, result)
 495
 496 /* Public UTrie2 API: optimized UTF-16 access ------------------------------- */
 497
 498 /*
 499  * The following functions and macros are used for highly optimized UTF-16
 500  * text processing. The UTRIE2_U16_NEXTxy() macros do not depend on these.
 501  *
 502  * A UTrie2 stores separate values for lead surrogate code _units_ vs. code _points_.
 503  * UTF-16 text processing can be optimized by detecting surrogate pairs and
 504  * assembling supplementary code points only when there is non-trivial data
 505  * available.
 506  *
 507  * At build-time, use utrie2_enumForLeadSurrogate() to see if there
 508  * is non-trivial (non-initialValue) data for any of the supplementary
 509  * code points associated with a lead surrogate.
 510  * If so, then set a special (application-specific) value for the
 511  * lead surrogate code _unit_, with utrie2_set32ForLeadSurrogateCodeUnit().
 512  *
 513  * At runtime, use UTRIE2_GET16_FROM_U16_SINGLE_LEAD() or
 514  * UTRIE2_GET32_FROM_U16_SINGLE_LEAD() per code unit. If there is non-trivial
 515  * data and the code unit is a lead surrogate, then check if a trail surrogate
 516  * follows. If so, assemble the supplementary code point with
 517  * U16_GET_SUPPLEMENTARY() and look up its value with UTRIE2_GET16_FROM_SUPP()
 518  * or UTRIE2_GET32_FROM_SUPP(); otherwise reset the lead
 519  * surrogate's value or do a code point lookup for it.
 520  *
 521  * If there is only trivial data for lead and trail surrogates, then processing
 522  * can often skip them. For example, in normalization or case mapping
 523  * all characters that do not have any mappings are simply copied as is.
 524  */
 525
 526 /**
 527  * Get a value from a lead surrogate code unit as stored in the trie.
 528  *
 529  * @param trie the trie
 530  * @param c the code unit (U+D800..U+DBFF)
 531  * @return the value
 532  */
 533 U_CAPI uint32_t U_EXPORT2
 534 utrie2_get32FromLeadSurrogateCodeUnit(const UTrie2 *trie, UChar32 c);
 535
 536 /**
 537  * Enumerate the trie values for the 1024=0x400 code points
 538  * corresponding to a given lead surrogate.
 539  * For example, for the lead surrogate U+D87E it will enumerate the values
 540  * for [U+2F800..U+2FC00[.
 541  * Used by data builder code that sets special lead surrogate code unit values
 542  * for optimized UTF-16 string processing.
 543  *
 544  * Do not modify the trie during the enumeration.
 545  *
 546  * Except for the limited code point range, this functions just like utrie2_enum():
 547  * For each entry in the trie, the value to be delivered is passed through
 548  * the UTrie2EnumValue function.
 549  * The value is unchanged if that function pointer is NULL.
 550  *
 551  * For each contiguous range of code points with a given (transformed) value,
 552  * the UTrie2EnumRange function is called.
 553  *
 554  * @param trie a pointer to the trie
 555  * @param enumValue a pointer to a function that may transform the trie entry value,
 556  *                  or NULL if the values from the trie are to be used directly
 557  * @param enumRange a pointer to a function that is called for each contiguous range
 558  *                  of code points with the same (transformed) value
 559  * @param context an opaque pointer that is passed on to the callback functions
 560  */
 561 U_CAPI void U_EXPORT2
 562 utrie2_enumForLeadSurrogate(const UTrie2 *trie, UChar32 lead,
 563                             UTrie2EnumValue *enumValue, UTrie2EnumRange *enumRange,
 564                             const void *context);
 565
 566 /**
 567  * Set a value for a lead surrogate code unit.
 568  *
 569  * @param trie the unfrozen trie
 570  * @param lead the lead surrogate code unit (U+D800..U+DBFF)
 571  * @param value the value
 572  * @param pErrorCode an in/out ICU UErrorCode; among other possible error codes:
 573  * - U_NO_WRITE_PERMISSION if the trie is frozen
 574  */
 575 U_CAPI void U_EXPORT2
 576 utrie2_set32ForLeadSurrogateCodeUnit(UTrie2 *trie,
 577                                      UChar32 lead, uint32_t value,
 578                                      UErrorCode *pErrorCode);
 579
 580 /**
 581  * Return a 16-bit trie value from a UTF-16 single/lead code unit (<=U+ffff).
 582  * Same as UTRIE2_GET16() if c is a BMP code point except for lead surrogates,
 583  * but smaller and faster.
 584  *
 585  * @param trie (const UTrie2 *, in) a frozen trie
 586  * @param c (UChar32, in) the input code unit, must be 0<=c<=U+ffff
 587  * @return (uint16_t) The code unit's trie value.
 588  */
 589 #define UTRIE2_GET16_FROM_U16_SINGLE_LEAD(trie, c) _UTRIE2_GET_FROM_U16_SINGLE_LEAD((trie), index, c)
 590
 591 /**
 592  * Return a 32-bit trie value from a UTF-16 single/lead code unit (<=U+ffff).
 593  * Same as UTRIE2_GET32() if c is a BMP code point except for lead surrogates,
 594  * but smaller and faster.
 595  *
 596  * @param trie (const UTrie2 *, in) a frozen trie
 597  * @param c (UChar32, in) the input code unit, must be 0<=c<=U+ffff
 598  * @return (uint32_t) The code unit's trie value.
 599  */
 600 #define UTRIE2_GET32_FROM_U16_SINGLE_LEAD(trie, c) _UTRIE2_GET_FROM_U16_SINGLE_LEAD((trie), data32, c)
 601
 602 /**
 603  * Return a 16-bit trie value from a supplementary code point (U+10000..U+10ffff).
 604  *
 605  * @param trie (const UTrie2 *, in) a frozen trie
 606  * @param c (UChar32, in) the input code point, must be U+10000<=c<=U+10ffff
 607  * @return (uint16_t) The code point's trie value.
 608  */
 609 #define UTRIE2_GET16_FROM_SUPP(trie, c) _UTRIE2_GET_FROM_SUPP((trie), index, c)
 610
 611 /**
 612  * Return a 32-bit trie value from a supplementary code point (U+10000..U+10ffff).
 613  *
 614  * @param trie (const UTrie2 *, in) a frozen trie
 615  * @param c (UChar32, in) the input code point, must be U+10000<=c<=U+10ffff
 616  * @return (uint32_t) The code point's trie value.
 617  */
 618 #define UTRIE2_GET32_FROM_SUPP(trie, c) _UTRIE2_GET_FROM_SUPP((trie), data32, c)
 619
 620 U_CDECL_END
 621
 622 /* C++ convenience wrappers ------------------------------------------------- */
 623
 624 #ifdef __cplusplus
 625
 626 #include "unicode/utf.h"
 627 #include "mutex.h"
 628
 629 U_NAMESPACE_BEGIN
 630
 631 // Use the Forward/Backward subclasses below.
 632 class UTrie2StringIterator : public UMemory {
 633 public:
 634     UTrie2StringIterator(const UTrie2 *t, const UChar *p) :
 635         trie(t), codePointStart(p), codePointLimit(p), codePoint(U_SENTINEL) {}
 636
 637     const UTrie2 *trie;
 638     const UChar *codePointStart, *codePointLimit;
 639     UChar32 codePoint;
 640 };
 641
 642 class BackwardUTrie2StringIterator : public UTrie2StringIterator {
 643 public:
 644     BackwardUTrie2StringIterator(const UTrie2 *t, const UChar *s, const UChar *p) :
 645         UTrie2StringIterator(t, p), start(s) {}
 646
 647     uint16_t previous16();
 648
 649     const UChar *start;
 650 };
 651
 652 class ForwardUTrie2StringIterator : public UTrie2StringIterator {
 653 public:
 654     // Iteration limit l can be NULL.
 655     // In that case, the caller must detect c==0 and stop.
 656     ForwardUTrie2StringIterator(const UTrie2 *t, const UChar *p, const UChar *l) :
 657         UTrie2StringIterator(t, p), limit(l) {}
 658
 659     uint16_t next16();
 660
 661     const UChar *limit;
 662 };
 663
 664 U_NAMESPACE_END
 665
 666 #endif
 667
 668 /* Internal definitions ----------------------------------------------------- */
 669
 670 U_CDECL_BEGIN
 671
 672 /** Build-time trie structure. */
 673 struct UNewTrie2;
 674 typedef struct UNewTrie2 UNewTrie2;
 675
 676 /*
 677  * Trie structure definition.
 678  *
 679  * Either the data table is 16 bits wide and accessed via the index
 680  * pointer, with each index item increased by indexLength;
 681  * in this case, data32==NULL, and data16 is used for direct ASCII access.
 682  *
 683  * Or the data table is 32 bits wide and accessed via the data32 pointer.
 684  */
 685 struct UTrie2 {
 686     /* protected: used by macros and functions for reading values */
 687     const uint16_t *index;
 688     const uint16_t *data16;     /* for fast UTF-8 ASCII access, if 16b data */
 689     const uint32_t *data32;     /* NULL if 16b data is used via index */
 690
 691     int32_t indexLength, dataLength;
 692     uint16_t index2NullOffset;  /* 0xffff if there is no dedicated index-2 null block */
 693     uint16_t dataNullOffset;
 694     uint32_t initialValue;
 695     /** Value returned for out-of-range code points and illegal UTF-8. */
 696     uint32_t errorValue;
 697
 698     /* Start of the last range which ends at U+10ffff, and its value. */
 699     UChar32 highStart;
 700     int32_t highValueIndex;
 701
 702     /* private: used by builder and unserialization functions */
 703     void *memory;           /* serialized bytes; NULL if not frozen yet */
 704     int32_t length;         /* number of serialized bytes at memory; 0 if not frozen yet */
 705     UBool isMemoryOwned;    /* TRUE if the trie owns the memory */
 706     UBool padding1;
 707     int16_t padding2;
 708     UNewTrie2 *newTrie;     /* builder object; NULL when frozen */
 709 };
 710
 711 /**
 712  * Trie constants, defining shift widths, index array lengths, etc.
 713  *
 714  * These are needed for the runtime macros but users can treat these as
 715  * implementation details and skip to the actual public API further below.
 716  */
 717 enum {
 718     /** Shift size for getting the index-1 table offset. */
 719     UTRIE2_SHIFT_1=6+5,
 720
 721     /** Shift size for getting the index-2 table offset. */
 722     UTRIE2_SHIFT_2=5,
 723
 724     /**
 725      * Difference between the two shift sizes,
 726      * for getting an index-1 offset from an index-2 offset. 6=11-5
 727      */
 728     UTRIE2_SHIFT_1_2=UTRIE2_SHIFT_1-UTRIE2_SHIFT_2,
 729
 730     /**
 731      * Number of index-1 entries for the BMP. 32=0x20
 732      * This part of the index-1 table is omitted from the serialized form.
 733      */
 734     UTRIE2_OMITTED_BMP_INDEX_1_LENGTH=0x10000>>UTRIE2_SHIFT_1,
 735
 736     /** Number of code points per index-1 table entry. 2048=0x800 */
 737     UTRIE2_CP_PER_INDEX_1_ENTRY=1<<UTRIE2_SHIFT_1,
 738
 739     /** Number of entries in an index-2 block. 64=0x40 */
 740     UTRIE2_INDEX_2_BLOCK_LENGTH=1<<UTRIE2_SHIFT_1_2,
 741
 742     /** Mask for getting the lower bits for the in-index-2-block offset. */
 743     UTRIE2_INDEX_2_MASK=UTRIE2_INDEX_2_BLOCK_LENGTH-1,
 744
 745     /** Number of entries in a data block. 32=0x20 */
 746     UTRIE2_DATA_BLOCK_LENGTH=1<<UTRIE2_SHIFT_2,
 747
 748     /** Mask for getting the lower bits for the in-data-block offset. */
 749     UTRIE2_DATA_MASK=UTRIE2_DATA_BLOCK_LENGTH-1,
 750
 751     /**
 752      * Shift size for shifting left the index array values.
 753      * Increases possible data size with 16-bit index values at the cost
 754      * of compactability.
 755      * This requires data blocks to be aligned by UTRIE2_DATA_GRANULARITY.
 756      */
 757     UTRIE2_INDEX_SHIFT=2,
 758
 759     /** The alignment size of a data block. Also the granularity for compaction. */
 760     UTRIE2_DATA_GRANULARITY=1<<UTRIE2_INDEX_SHIFT,
 761
 762     /* Fixed layout of the first part of the index array. ------------------- */
 763
 764     /**
 765      * The BMP part of the index-2 table is fixed and linear and starts at offset 0.
 766      * Length=2048=0x800=0x10000>>UTRIE2_SHIFT_2.
 767      */
 768     UTRIE2_INDEX_2_OFFSET=0,
 769
 770     /**
 771      * The part of the index-2 table for U+D800..U+DBFF stores values for
 772      * lead surrogate code _units_ not code _points_.
 773      * Values for lead surrogate code _points_ are indexed with this portion of the table.
 774      * Length=32=0x20=0x400>>UTRIE2_SHIFT_2. (There are 1024=0x400 lead surrogates.)
 775      */
 776     UTRIE2_LSCP_INDEX_2_OFFSET=0x10000>>UTRIE2_SHIFT_2,
 777     UTRIE2_LSCP_INDEX_2_LENGTH=0x400>>UTRIE2_SHIFT_2,
 778
 779     /** Count the lengths of both BMP pieces. 2080=0x820 */
 780     UTRIE2_INDEX_2_BMP_LENGTH=UTRIE2_LSCP_INDEX_2_OFFSET+UTRIE2_LSCP_INDEX_2_LENGTH,
 781
 782     /**
 783      * The 2-byte UTF-8 version of the index-2 table follows at offset 2080=0x820.
 784      * Length 32=0x20 for lead bytes C0..DF, regardless of UTRIE2_SHIFT_2.
 785      */
 786     UTRIE2_UTF8_2B_INDEX_2_OFFSET=UTRIE2_INDEX_2_BMP_LENGTH,
 787     UTRIE2_UTF8_2B_INDEX_2_LENGTH=0x800>>6,  /* U+0800 is the first code point after 2-byte UTF-8 */
 788
 789     /**
 790      * The index-1 table, only used for supplementary code points, at offset 2112=0x840.
 791      * Variable length, for code points up to highStart, where the last single-value range starts.
 792      * Maximum length 512=0x200=0x100000>>UTRIE2_SHIFT_1.
 793      * (For 0x100000 supplementary code points U+10000..U+10ffff.)
 794      *
 795      * The part of the index-2 table for supplementary code points starts
 796      * after this index-1 table.
 797      *
 798      * Both the index-1 table and the following part of the index-2 table
 799      * are omitted completely if there is only BMP data.
 800      */
 801     UTRIE2_INDEX_1_OFFSET=UTRIE2_UTF8_2B_INDEX_2_OFFSET+UTRIE2_UTF8_2B_INDEX_2_LENGTH,
 802     UTRIE2_MAX_INDEX_1_LENGTH=0x100000>>UTRIE2_SHIFT_1,
 803
 804     /*
 805      * Fixed layout of the first part of the data array. -----------------------
 806      * Starts with 4 blocks (128=0x80 entries) for ASCII.
 807      */
 808
 809     /**
 810      * The illegal-UTF-8 data block follows the ASCII block, at offset 128=0x80.
 811      * Used with linear access for single bytes 0..0xbf for simple error handling.
 812      * Length 64=0x40, not UTRIE2_DATA_BLOCK_LENGTH.
 813      */
 814     UTRIE2_BAD_UTF8_DATA_OFFSET=0x80,
 815
 816     /** The start of non-linear-ASCII data blocks, at offset 192=0xc0. */
 817     UTRIE2_DATA_START_OFFSET=0xc0
 818 };
 819
 820 /* Internal functions and macros -------------------------------------------- */
 821
 822 /**
 823  * Internal function for part of the UTRIE2_U8_NEXTxx() macro implementations.
 824  * Do not call directly.
 825  * @internal
 826  */
 827 U_INTERNAL int32_t U_EXPORT2
 828 utrie2_internalU8NextIndex(const UTrie2 *trie, UChar32 c,
 829                            const uint8_t *src, const uint8_t *limit);
 830
 831 /**
 832  * Internal function for part of the UTRIE2_U8_PREVxx() macro implementations.
 833  * Do not call directly.
 834  * @internal
 835  */
 836 U_INTERNAL int32_t U_EXPORT2
 837 utrie2_internalU8PrevIndex(const UTrie2 *trie, UChar32 c,
 838                            const uint8_t *start, const uint8_t *src);
 839
 840
 841 /** Internal low-level trie getter. Returns a data index. */
 842 #define _UTRIE2_INDEX_RAW(offset, trieIndex, c) \
 843     (((int32_t)((trieIndex)[(offset)+((c)>>UTRIE2_SHIFT_2)]) \
 844     <<UTRIE2_INDEX_SHIFT)+ \
 845     ((c)&UTRIE2_DATA_MASK))
 846
 847 /** Internal trie getter from a UTF-16 single/lead code unit. Returns the data index. */
 848 #define _UTRIE2_INDEX_FROM_U16_SINGLE_LEAD(trieIndex, c) _UTRIE2_INDEX_RAW(0, trieIndex, c)
 849
 850 /** Internal trie getter from a lead surrogate code point (D800..DBFF). Returns the data index. */
 851 #define _UTRIE2_INDEX_FROM_LSCP(trieIndex, c) \
 852     _UTRIE2_INDEX_RAW(UTRIE2_LSCP_INDEX_2_OFFSET-(0xd800>>UTRIE2_SHIFT_2), trieIndex, c)
 853
 854 /** Internal trie getter from a BMP code point. Returns the data index. */
 855 #define _UTRIE2_INDEX_FROM_BMP(trieIndex, c) \
 856     _UTRIE2_INDEX_RAW(U_IS_LEAD(c) ? UTRIE2_LSCP_INDEX_2_OFFSET-(0xd800>>UTRIE2_SHIFT_2) : 0, \
 857                       trieIndex, c)
 858
 859 /** Internal trie getter from a supplementary code point below highStart. Returns the data index. */
 860 #define _UTRIE2_INDEX_FROM_SUPP(trieIndex, c) \
 861     (((int32_t)((trieIndex)[ \
 862         (trieIndex)[(UTRIE2_INDEX_1_OFFSET-UTRIE2_OMITTED_BMP_INDEX_1_LENGTH)+ \
 863                       ((c)>>UTRIE2_SHIFT_1)]+ \
 864         (((c)>>UTRIE2_SHIFT_2)&UTRIE2_INDEX_2_MASK)]) \
 865     <<UTRIE2_INDEX_SHIFT)+ \
 866     ((c)&UTRIE2_DATA_MASK))
 867
 868 /**
 869  * Internal trie getter from a code point, with checking that c is in 0..10FFFF.
 870  * Returns the data index.
 871  */
 872 #define _UTRIE2_INDEX_FROM_CP(trie, asciiOffset, c) \
 873     ((uint32_t)(c)<0xd800 ? \
 874         _UTRIE2_INDEX_RAW(0, (trie)->index, c) : \
 875         (uint32_t)(c)<=0xffff ? \
 876             _UTRIE2_INDEX_RAW( \
 877                 (c)<=0xdbff ? UTRIE2_LSCP_INDEX_2_OFFSET-(0xd800>>UTRIE2_SHIFT_2) : 0, \
 878                 (trie)->index, c) : \
 879             (uint32_t)(c)>0x10ffff ? \
 880                 (asciiOffset)+UTRIE2_BAD_UTF8_DATA_OFFSET : \
 881                 (c)>=(trie)->highStart ? \
 882                     (trie)->highValueIndex : \
 883                     _UTRIE2_INDEX_FROM_SUPP((trie)->index, c))
 884
 885 /** Internal trie getter from a UTF-16 single/lead code unit. Returns the data. */
 886 #define _UTRIE2_GET_FROM_U16_SINGLE_LEAD(trie, data, c) \
 887     (trie)->data[_UTRIE2_INDEX_FROM_U16_SINGLE_LEAD((trie)->index, c)]
 888
 889 /** Internal trie getter from a supplementary code point. Returns the data. */
 890 #define _UTRIE2_GET_FROM_SUPP(trie, data, c) \
 891     (trie)->data[(c)>=(trie)->highStart ? (trie)->highValueIndex : \
 892                  _UTRIE2_INDEX_FROM_SUPP((trie)->index, c)]
 893
 894 /**
 895  * Internal trie getter from a code point, with checking that c is in 0..10FFFF.
 896  * Returns the data.
 897  */
 898 #define _UTRIE2_GET(trie, data, asciiOffset, c) \
 899     (trie)->data[_UTRIE2_INDEX_FROM_CP(trie, asciiOffset, c)]
 900
 901 /** Internal next-post-increment: get the next code point (c) and its data. */
 902 #define _UTRIE2_U16_NEXT(trie, data, src, limit, c, result) { \
 903     { \
 904         uint16_t __c2; \
 905         (c)=*(src)++; \
 906         if(!U16_IS_LEAD(c)) { \
 907             (result)=_UTRIE2_GET_FROM_U16_SINGLE_LEAD(trie, data, c); \
 908         } else if((src)==(limit) || !U16_IS_TRAIL(__c2=*(src))) { \
 909             (result)=(trie)->data[_UTRIE2_INDEX_FROM_LSCP((trie)->index, c)]; \
 910         } else { \
 911             ++(src); \
 912             (c)=U16_GET_SUPPLEMENTARY((c), __c2); \
 913             (result)=_UTRIE2_GET_FROM_SUPP((trie), data, (c)); \
 914         } \
 915     } \
 916 }
 917
 918 /** Internal pre-decrement-previous: get the previous code point (c) and its data */
 919 #define _UTRIE2_U16_PREV(trie, data, start, src, c, result) { \
 920     { \
 921         uint16_t __c2; \
 922         (c)=*--(src); \
 923         if(!U16_IS_TRAIL(c) || (src)==(start) || !U16_IS_LEAD(__c2=*((src)-1))) { \
 924             (result)=(trie)->data[_UTRIE2_INDEX_FROM_BMP((trie)->index, c)]; \
 925         } else { \
 926             --(src); \
 927             (c)=U16_GET_SUPPLEMENTARY(__c2, (c)); \
 928             (result)=_UTRIE2_GET_FROM_SUPP((trie), data, (c)); \
 929         } \
 930     } \
 931 }
 932
 933 /** Internal UTF-8 next-post-increment: get the next code point's data. */
 934 #define _UTRIE2_U8_NEXT(trie, ascii, data, src, limit, result) { \
 935     uint8_t __lead=(uint8_t)*(src)++; \
 936     if(__lead<0xc0) { \
 937         (result)=(trie)->ascii[__lead]; \
 938     } else { \
 939         uint8_t __t1, __t2; \
 940         if( /* handle U+0000..U+07FF inline */ \
 941             __lead<0xe0 && (src)<(limit) && \
 942             (__t1=(uint8_t)(*(src)-0x80))<=0x3f \
 943         ) { \
 944             ++(src); \
 945             (result)=(trie)->data[ \
 946                 (trie)->index[(UTRIE2_UTF8_2B_INDEX_2_OFFSET-0xc0)+__lead]+ \
 947                 __t1]; \
 948         } else if( /* handle U+0000..U+CFFF inline */ \
 949             __lead<0xed && ((src)+1)<(limit) && \
 950             (__t1=(uint8_t)(*(src)-0x80))<=0x3f && (__lead>0xe0 || __t1>=0x20) && \
 951             (__t2=(uint8_t)(*((src)+1)-0x80))<= 0x3f \
 952         ) { \
 953             (src)+=2; \
 954             (result)=(trie)->data[ \
 955                 ((int32_t)((trie)->index[((__lead-0xe0)<<(12-UTRIE2_SHIFT_2))+ \
 956                                          (__t1<<(6-UTRIE2_SHIFT_2))+(__t2>>UTRIE2_SHIFT_2)]) \
 957                 <<UTRIE2_INDEX_SHIFT)+ \
 958                 (__t2&UTRIE2_DATA_MASK)]; \
 959         } else { \
 960             int32_t __index=utrie2_internalU8NextIndex((trie), __lead, (const uint8_t *)(src), \
 961                                                                        (const uint8_t *)(limit)); \
 962             (src)+=__index&7; \
 963             (result)=(trie)->data[__index>>3]; \
 964         } \
 965     } \
 966 }
 967
 968 /** Internal UTF-8 pre-decrement-previous: get the previous code point's data. */
 969 #define _UTRIE2_U8_PREV(trie, ascii, data, start, src, result) { \
 970     uint8_t __b=(uint8_t)*--(src); \
 971     if(__b<0x80) { \
 972         (result)=(trie)->ascii[__b]; \
 973     } else { \
 974         int32_t __index=utrie2_internalU8PrevIndex((trie), __b, (const uint8_t *)(start), \
 975                                                                 (const uint8_t *)(src)); \
 976         (src)-=__index&7; \
 977         (result)=(trie)->data[__index>>3]; \
 978     } \
 979 }
 980
 981 U_CDECL_END
 982
 983 /**
 984  * Work around MSVC 2003 optimization bugs.
 985  */
 986 #if defined (U_HAVE_MSVC_2003_OR_EARLIER)
 987 #pragma optimize("", off)
 988 #endif
 989
 990 #endif