icuSources/common/ucase.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) 2004-2012, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 *******************************************************************************
  10 *   file name:  ucase.h
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 2004aug30
  16 *   created by: Markus W. Scherer
  17 *
  18 *   Low-level Unicode character/string case mapping code.
  19 */
  20
  21 #ifndef __UCASE_H__
  22 #define __UCASE_H__
  23
  24 #include "unicode/utypes.h"
  25 #include "unicode/uset.h"
  26 #include "putilimp.h"
  27 #include "uset_imp.h"
  28 #include "udataswp.h"
  29
  30 #ifdef __cplusplus
  31 U_NAMESPACE_BEGIN
  32
  33 class UnicodeString;
  34
  35 U_NAMESPACE_END
  36 #endif
  37
  38 /* library API -------------------------------------------------------------- */
  39
  40 U_CFUNC void U_EXPORT2
  41 ucase_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode);
  42
  43 /**
  44  * Requires non-NULL locale ID but otherwise does the equivalent of
  45  * checking for language codes as if uloc_getLanguage() were called:
  46  * Accepts both 2- and 3-letter codes and accepts case variants.
  47  */
  48 U_CFUNC int32_t
  49 ucase_getCaseLocale(const char *locale);
  50
  51 /* Casing locale types for ucase_getCaseLocale */
  52 enum {
  53     UCASE_LOC_UNKNOWN,
  54     UCASE_LOC_ROOT,
  55     UCASE_LOC_TURKISH,
  56     UCASE_LOC_LITHUANIAN,
  57     UCASE_LOC_GREEK,
  58     UCASE_LOC_DUTCH
  59 };
  60
  61 /**
  62  * Bit mask for getting just the options from a string compare options word
  63  * that are relevant for case-insensitive string comparison.
  64  * See uchar.h. Also include _STRNCMP_STYLE and U_COMPARE_CODE_POINT_ORDER.
  65  * @internal
  66  */
  67 #define _STRCASECMP_OPTIONS_MASK 0xffff
  68
  69 /**
  70  * Bit mask for getting just the options from a string compare options word
  71  * that are relevant for case folding (of a single string or code point).
  72  * See uchar.h.
  73  * @internal
  74  */
  75 #define _FOLD_CASE_OPTIONS_MASK 0xff
  76
  77 /* single-code point functions */
  78
  79 U_CAPI UChar32 U_EXPORT2
  80 ucase_tolower(UChar32 c);
  81
  82 U_CAPI UChar32 U_EXPORT2
  83 ucase_toupper(UChar32 c);
  84
  85 U_CAPI UChar32 U_EXPORT2
  86 ucase_totitle(UChar32 c);
  87
  88 U_CAPI UChar32 U_EXPORT2
  89 ucase_fold(UChar32 c, uint32_t options);
  90
  91 /**
  92  * Adds all simple case mappings and the full case folding for c to sa,
  93  * and also adds special case closure mappings.
  94  * c itself is not added.
  95  * For example, the mappings
  96  * - for s include long s
  97  * - for sharp s include ss
  98  * - for k include the Kelvin sign
  99  */
 100 U_CFUNC void U_EXPORT2
 101 ucase_addCaseClosure(UChar32 c, const USetAdder *sa);
 102
 103 /**
 104  * Maps the string to single code points and adds the associated case closure
 105  * mappings.
 106  * The string is mapped to code points if it is their full case folding string.
 107  * In other words, this performs a reverse full case folding and then
 108  * adds the case closure items of the resulting code points.
 109  * If the string is found and its closure applied, then
 110  * the string itself is added as well as part of its code points' closure.
 111  * It must be length>=0.
 112  *
 113  * @return TRUE if the string was found
 114  */
 115 U_CFUNC UBool U_EXPORT2
 116 ucase_addStringCaseClosure(const UChar *s, int32_t length, const USetAdder *sa);
 117
 118 #ifdef __cplusplus
 119 U_NAMESPACE_BEGIN
 120
 121 /**
 122  * Iterator over characters with more than one code point in the full default Case_Folding.
 123  */
 124 class U_COMMON_API FullCaseFoldingIterator {
 125 public:
 126     /** Constructor. */
 127     FullCaseFoldingIterator();
 128     /**
 129      * Returns the next (cp, full) pair where "full" is cp's full default Case_Folding.
 130      * Returns a negative cp value at the end of the iteration.
 131      */
 132     UChar32 next(UnicodeString &full);
 133 private:
 134     FullCaseFoldingIterator(const FullCaseFoldingIterator &);  // no copy
 135     FullCaseFoldingIterator &operator=(const FullCaseFoldingIterator &);  // no assignment
 136
 137     const UChar *unfold;
 138     int32_t unfoldRows;
 139     int32_t unfoldRowWidth;
 140     int32_t unfoldStringWidth;
 141     int32_t currentRow;
 142     int32_t rowCpIndex;
 143 };
 144
 145 U_NAMESPACE_END
 146 #endif
 147
 148 /** @return UCASE_NONE, UCASE_LOWER, UCASE_UPPER, UCASE_TITLE */
 149 U_CAPI int32_t U_EXPORT2
 150 ucase_getType(UChar32 c);
 151
 152 /** @return like ucase_getType() but also sets UCASE_IGNORABLE if c is case-ignorable */
 153 U_CAPI int32_t U_EXPORT2
 154 ucase_getTypeOrIgnorable(UChar32 c);
 155
 156 U_CAPI UBool U_EXPORT2
 157 ucase_isSoftDotted(UChar32 c);
 158
 159 U_CAPI UBool U_EXPORT2
 160 ucase_isCaseSensitive(UChar32 c);
 161
 162 /* string case mapping functions */
 163
 164 U_CDECL_BEGIN
 165
 166 /**
 167  * Iterator function for string case mappings, which need to look at the
 168  * context (surrounding text) of a given character for conditional mappings.
 169  *
 170  * The iterator only needs to go backward or forward away from the
 171  * character in question. It does not use any indexes on this interface.
 172  * It does not support random access or an arbitrary change of
 173  * iteration direction.
 174  *
 175  * The code point being case-mapped itself is never returned by
 176  * this iterator.
 177  *
 178  * @param context A pointer to the iterator's working data.
 179  * @param dir If <0 then start iterating backward from the character;
 180  *            if >0 then start iterating forward from the character;
 181  *            if 0 then continue iterating in the current direction.
 182  * @return Next code point, or <0 when the iteration is done.
 183  */
 184 typedef UChar32 U_CALLCONV
 185 UCaseContextIterator(void *context, int8_t dir);
 186
 187 /**
 188  * Sample struct which may be used by some implementations of
 189  * UCaseContextIterator.
 190  */
 191 struct UCaseContext {
 192     void *p;
 193     int32_t start, index, limit;
 194     int32_t cpStart, cpLimit;
 195     int8_t dir;
 196     int8_t b1, b2, b3;
 197 };
 198 typedef struct UCaseContext UCaseContext;
 199
 200 U_CDECL_END
 201
 202 #define UCASECONTEXT_INITIALIZER { NULL,  0, 0, 0,  0, 0,  0,  0, 0, 0 }
 203
 204 enum {
 205     /**
 206      * For string case mappings, a single character (a code point) is mapped
 207      * either to itself (in which case in-place mapping functions do nothing),
 208      * or to another single code point, or to a string.
 209      * Aside from the string contents, these are indicated with a single int32_t
 210      * value as follows:
 211      *
 212      * Mapping to self: Negative values (~self instead of -self to support U+0000)
 213      *
 214      * Mapping to another code point: Positive values >UCASE_MAX_STRING_LENGTH
 215      *
 216      * Mapping to a string: The string length (0..UCASE_MAX_STRING_LENGTH) is
 217      * returned. Note that the string result may indeed have zero length.
 218      */
 219     UCASE_MAX_STRING_LENGTH=0x1f
 220 };
 221
 222 /**
 223  * Get the full lowercase mapping for c.
 224  *
 225  * @param csp Case mapping properties.
 226  * @param c Character to be mapped.
 227  * @param iter Character iterator, used for context-sensitive mappings.
 228  *             See UCaseContextIterator for details.
 229  *             If iter==NULL then a context-independent result is returned.
 230  * @param context Pointer to be passed into iter.
 231  * @param pString If the mapping result is a string, then the pointer is
 232  *                written to *pString.
 233  * @param caseLocale Case locale value from ucase_getCaseLocale().
 234  * @return Output code point or string length, see UCASE_MAX_STRING_LENGTH.
 235  *
 236  * @see UCaseContextIterator
 237  * @see UCASE_MAX_STRING_LENGTH
 238  * @internal
 239  */
 240 U_CAPI int32_t U_EXPORT2
 241 ucase_toFullLower(UChar32 c,
 242                   UCaseContextIterator *iter, void *context,
 243                   const UChar **pString,
 244                   int32_t caseLocale);
 245
 246 U_CAPI int32_t U_EXPORT2
 247 ucase_toFullUpper(UChar32 c,
 248                   UCaseContextIterator *iter, void *context,
 249                   const UChar **pString,
 250                   int32_t caseLocale);
 251
 252 U_CAPI int32_t U_EXPORT2
 253 ucase_toFullTitle(UChar32 c,
 254                   UCaseContextIterator *iter, void *context,
 255                   const UChar **pString,
 256                   int32_t caseLocale);
 257
 258 U_CAPI int32_t U_EXPORT2
 259 ucase_toFullFolding(UChar32 c,
 260                     const UChar **pString,
 261                     uint32_t options);
 262
 263 U_CFUNC int32_t U_EXPORT2
 264 ucase_hasBinaryProperty(UChar32 c, UProperty which);
 265
 266
 267 U_CDECL_BEGIN
 268
 269 /**
 270  * @internal
 271  */
 272 typedef int32_t U_CALLCONV
 273 UCaseMapFull(UChar32 c,
 274              UCaseContextIterator *iter, void *context,
 275              const UChar **pString,
 276              int32_t caseLocale);
 277
 278 U_CDECL_END
 279
 280 /* file definitions --------------------------------------------------------- */
 281
 282 #define UCASE_DATA_NAME "ucase"
 283 #define UCASE_DATA_TYPE "icu"
 284
 285 /* format "cAsE" */
 286 #define UCASE_FMT_0 0x63
 287 #define UCASE_FMT_1 0x41
 288 #define UCASE_FMT_2 0x53
 289 #define UCASE_FMT_3 0x45
 290
 291 /* indexes into indexes[] */
 292 enum {
 293     UCASE_IX_INDEX_TOP,
 294     UCASE_IX_LENGTH,
 295     UCASE_IX_TRIE_SIZE,
 296     UCASE_IX_EXC_LENGTH,
 297     UCASE_IX_UNFOLD_LENGTH,
 298
 299     UCASE_IX_MAX_FULL_LENGTH=15,
 300     UCASE_IX_TOP=16
 301 };
 302
 303 /* definitions for 16-bit case properties word ------------------------------ */
 304
 305 /* 2-bit constants for types of cased characters */
 306 #define UCASE_TYPE_MASK     3
 307 enum {
 308     UCASE_NONE,
 309     UCASE_LOWER,
 310     UCASE_UPPER,
 311     UCASE_TITLE
 312 };
 313
 314 #define UCASE_GET_TYPE(props) ((props)&UCASE_TYPE_MASK)
 315 #define UCASE_GET_TYPE_AND_IGNORABLE(props) ((props)&7)
 316
 317 #define UCASE_IGNORABLE         4
 318 #define UCASE_SENSITIVE         8
 319 #define UCASE_EXCEPTION         0x10
 320
 321 #define UCASE_DOT_MASK      0x60
 322 enum {
 323     UCASE_NO_DOT=0,         /* normal characters with cc=0 */
 324     UCASE_SOFT_DOTTED=0x20, /* soft-dotted characters with cc=0 */
 325     UCASE_ABOVE=0x40,       /* "above" accents with cc=230 */
 326     UCASE_OTHER_ACCENT=0x60 /* other accent character (0<cc!=230) */
 327 };
 328
 329 /* no exception: bits 15..7 are a 9-bit signed case mapping delta */
 330 #define UCASE_DELTA_SHIFT   7
 331 #define UCASE_DELTA_MASK    0xff80
 332 #define UCASE_MAX_DELTA     0xff
 333 #define UCASE_MIN_DELTA     (-UCASE_MAX_DELTA-1)
 334
 335 #if U_SIGNED_RIGHT_SHIFT_IS_ARITHMETIC
 336 #   define UCASE_GET_DELTA(props) ((int16_t)(props)>>UCASE_DELTA_SHIFT)
 337 #else
 338 #   define UCASE_GET_DELTA(props) (int16_t)(((props)&0x8000) ? (((props)>>UCASE_DELTA_SHIFT)|0xfe00) : ((uint16_t)(props)>>UCASE_DELTA_SHIFT))
 339 #endif
 340
 341 /* exception: bits 15..5 are an unsigned 11-bit index into the exceptions array */
 342 #define UCASE_EXC_SHIFT     5
 343 #define UCASE_EXC_MASK      0xffe0
 344 #define UCASE_MAX_EXCEPTIONS ((UCASE_EXC_MASK>>UCASE_EXC_SHIFT)+1)
 345
 346 /* definitions for 16-bit main exceptions word ------------------------------ */
 347
 348 /* first 8 bits indicate values in optional slots */
 349 enum {
 350     UCASE_EXC_LOWER,
 351     UCASE_EXC_FOLD,
 352     UCASE_EXC_UPPER,
 353     UCASE_EXC_TITLE,
 354     UCASE_EXC_4,            /* reserved */
 355     UCASE_EXC_5,            /* reserved */
 356     UCASE_EXC_CLOSURE,
 357     UCASE_EXC_FULL_MAPPINGS,
 358     UCASE_EXC_ALL_SLOTS     /* one past the last slot */
 359 };
 360
 361 /* each slot is 2 uint16_t instead of 1 */
 362 #define UCASE_EXC_DOUBLE_SLOTS      0x100
 363
 364 /* reserved: exception bits 11..9 */
 365
 366 /* UCASE_EXC_DOT_MASK=UCASE_DOT_MASK<<UCASE_EXC_DOT_SHIFT */
 367 #define UCASE_EXC_DOT_SHIFT     7
 368
 369 /* normally stored in the main word, but pushed out for larger exception indexes */
 370 #define UCASE_EXC_DOT_MASK      0x3000
 371 enum {
 372     UCASE_EXC_NO_DOT=0,
 373     UCASE_EXC_SOFT_DOTTED=0x1000,
 374     UCASE_EXC_ABOVE=0x2000,         /* "above" accents with cc=230 */
 375     UCASE_EXC_OTHER_ACCENT=0x3000   /* other character (0<cc!=230) */
 376 };
 377
 378 /* complex/conditional mappings */
 379 #define UCASE_EXC_CONDITIONAL_SPECIAL   0x4000
 380 #define UCASE_EXC_CONDITIONAL_FOLD      0x8000
 381
 382 /* definitions for lengths word for full case mappings */
 383 #define UCASE_FULL_LOWER    0xf
 384 #define UCASE_FULL_FOLDING  0xf0
 385 #define UCASE_FULL_UPPER    0xf00
 386 #define UCASE_FULL_TITLE    0xf000
 387
 388 /* maximum lengths */
 389 #define UCASE_FULL_MAPPINGS_MAX_LENGTH (4*0xf)
 390 #define UCASE_CLOSURE_MAX_LENGTH 0xf
 391
 392 /* constants for reverse case folding ("unfold") data */
 393 enum {
 394     UCASE_UNFOLD_ROWS,
 395     UCASE_UNFOLD_ROW_WIDTH,
 396     UCASE_UNFOLD_STRING_WIDTH
 397 };
 398
 399 #endif