icuSources/common/ucase.h

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2004-2006, International Business Machines
   5 *   Corporation and others.  All Rights Reserved.
   6 *
   7 *******************************************************************************
   8 *   file name:  ucase.h
   9 *   encoding:   US-ASCII
  10 *   tab size:   8 (not used)
  11 *   indentation:4
  12 *
  13 *   created on: 2004aug30
  14 *   created by: Markus W. Scherer
  15 *
  16 *   Low-level Unicode character/string case mapping code.
  17 */
  18
  19 #ifndef __UCASE_H__
  20 #define __UCASE_H__
  21
  22 #include "unicode/utypes.h"
  23 #include "unicode/uset.h"
  24 #include "uset_imp.h"
  25 #include "udataswp.h"
  26
  27 U_CDECL_BEGIN
  28
  29 /* library API -------------------------------------------------------------- */
  30
  31 struct UCaseProps;
  32 typedef struct UCaseProps UCaseProps;
  33
  34 U_CAPI UCaseProps * U_EXPORT2
  35 ucase_open(UErrorCode *pErrorCode);
  36
  37 U_CAPI UCaseProps * U_EXPORT2
  38 ucase_openBinary(const uint8_t *bin, int32_t length, UErrorCode *pErrorCode);
  39
  40 U_CAPI void U_EXPORT2
  41 ucase_close(UCaseProps *csp);
  42
  43
  44 U_CAPI const UCaseProps * U_EXPORT2
  45 ucase_getSingleton(UErrorCode *pErrorCode);
  46
  47 /**
  48  * Get a singleton dummy object, one that works with no real data.
  49  * This can be used when the real data is not available.
  50  * Using the dummy can reduce checks for available data after an initial failure.
  51  */
  52 U_CAPI const UCaseProps * U_EXPORT2
  53 ucase_getDummy(UErrorCode *pErrorCode);
  54
  55
  56 U_CAPI int32_t U_EXPORT2
  57 ucase_swap(const UDataSwapper *ds,
  58            const void *inData, int32_t length, void *outData,
  59            UErrorCode *pErrorCode);
  60
  61 U_CAPI void U_EXPORT2
  62 ucase_addPropertyStarts(const UCaseProps *csp, const USetAdder *sa, UErrorCode *pErrorCode);
  63
  64 /**
  65  * Requires non-NULL locale ID but otherwise does the equivalent of
  66  * checking for language codes as if uloc_getLanguage() were called:
  67  * Accepts both 2- and 3-letter codes and accepts case variants.
  68  */
  69 U_CFUNC int32_t
  70 ucase_getCaseLocale(const char *locale, int32_t *locCache);
  71
  72 /**
  73  * Bit mask for getting just the options from a string compare options word
  74  * that are relevant for case-insensitive string comparison.
  75  * See uchar.h. Also include _STRNCMP_STYLE and U_COMPARE_CODE_POINT_ORDER.
  76  * @internal
  77  */
  78 #define _STRCASECMP_OPTIONS_MASK 0xffff
  79
  80 /**
  81  * Bit mask for getting just the options from a string compare options word
  82  * that are relevant for case folding (of a single string or code point).
  83  * See uchar.h.
  84  * @internal
  85  */
  86 #define _FOLD_CASE_OPTIONS_MASK 0xff
  87
  88 /* single-code point functions */
  89
  90 U_CAPI UChar32 U_EXPORT2
  91 ucase_tolower(const UCaseProps *csp, UChar32 c);
  92
  93 U_CAPI UChar32 U_EXPORT2
  94 ucase_toupper(const UCaseProps *csp, UChar32 c);
  95
  96 U_CAPI UChar32 U_EXPORT2
  97 ucase_totitle(const UCaseProps *csp, UChar32 c);
  98
  99 U_CAPI UChar32 U_EXPORT2
 100 ucase_fold(const UCaseProps *csp, UChar32 c, uint32_t options);
 101
 102 /**
 103  * Adds all simple case mappings and the full case folding for c to sa,
 104  * and also adds special case closure mappings.
 105  * c itself is not added.
 106  * For example, the mappings
 107  * - for s include long s
 108  * - for sharp s include ss
 109  * - for k include the Kelvin sign
 110  */
 111 U_CAPI void U_EXPORT2
 112 ucase_addCaseClosure(const UCaseProps *csp, UChar32 c, const USetAdder *sa);
 113
 114 /**
 115  * Maps the string to single code points and adds the associated case closure
 116  * mappings.
 117  * The string is mapped to code points if it is their full case folding string.
 118  * In other words, this performs a reverse full case folding and then
 119  * adds the case closure items of the resulting code points.
 120  * If the string is found and its closure applied, then
 121  * the string itself is added as well as part of its code points' closure.
 122  * It must be length>=0.
 123  *
 124  * @return TRUE if the string was found
 125  */
 126 U_CAPI UBool U_EXPORT2
 127 ucase_addStringCaseClosure(const UCaseProps *csp, const UChar *s, int32_t length, const USetAdder *sa);
 128
 129 /** @return UCASE_NONE, UCASE_LOWER, UCASE_UPPER, UCASE_TITLE */
 130 U_CAPI int32_t U_EXPORT2
 131 ucase_getType(const UCaseProps *csp, UChar32 c);
 132
 133 /** @return same as ucase_getType(), or <0 if c is case-ignorable */
 134 U_CAPI int32_t U_EXPORT2
 135 ucase_getTypeOrIgnorable(const UCaseProps *csp, UChar32 c);
 136
 137 U_CAPI UBool U_EXPORT2
 138 ucase_isSoftDotted(const UCaseProps *csp, UChar32 c);
 139
 140 U_CAPI UBool U_EXPORT2
 141 ucase_isCaseSensitive(const UCaseProps *csp, UChar32 c);
 142
 143 /* string case mapping functions */
 144
 145 /**
 146  * Iterator function for string case mappings, which need to look at the
 147  * context (surrounding text) of a given character for conditional mappings.
 148  *
 149  * The iterator only needs to go backward or forward away from the
 150  * character in question. It does not use any indexes on this interface.
 151  * It does not support random access or an arbitrary change of
 152  * iteration direction.
 153  *
 154  * The code point being case-mapped itself is never returned by
 155  * this iterator.
 156  *
 157  * @param context A pointer to the iterator's working data.
 158  * @param dir If <0 then start iterating backward from the character;
 159  *            if >0 then start iterating forward from the character;
 160  *            if 0 then continue iterating in the current direction.
 161  * @return Next code point, or <0 when the iteration is done.
 162  */
 163 typedef UChar32 U_CALLCONV
 164 UCaseContextIterator(void *context, int8_t dir);
 165
 166 /**
 167  * Sample struct which may be used by some implementations of
 168  * UCaseContextIterator.
 169  */
 170 struct UCaseContext {
 171     void *p;
 172     int32_t start, index, limit;
 173     int32_t cpStart, cpLimit;
 174     int8_t dir;
 175     int8_t b1, b2, b3;
 176 };
 177 typedef struct UCaseContext UCaseContext;
 178
 179 enum {
 180     /**
 181      * For string case mappings, a single character (a code point) is mapped
 182      * either to itself (in which case in-place mapping functions do nothing),
 183      * or to another single code point, or to a string.
 184      * Aside from the string contents, these are indicated with a single int32_t
 185      * value as follows:
 186      *
 187      * Mapping to self: Negative values (~self instead of -self to support U+0000)
 188      *
 189      * Mapping to another code point: Positive values >UCASE_MAX_STRING_LENGTH
 190      *
 191      * Mapping to a string: The string length (0..UCASE_MAX_STRING_LENGTH) is
 192      * returned. Note that the string result may indeed have zero length.
 193      */
 194     UCASE_MAX_STRING_LENGTH=0x1f
 195 };
 196
 197 /**
 198  * Get the full lowercase mapping for c.
 199  *
 200  * @param csp Case mapping properties.
 201  * @param c Character to be mapped.
 202  * @param iter Character iterator, used for context-sensitive mappings.
 203  *             See UCaseContextIterator for details.
 204  *             If iter==NULL then a context-independent result is returned.
 205  * @param context Pointer to be passed into iter.
 206  * @param pString If the mapping result is a string, then the pointer is
 207  *                written to *pString.
 208  * @param locale Locale ID for locale-dependent mappings.
 209  * @param locCache Initialize to 0; may be used to cache the result of parsing
 210  *                 the locale ID for subsequent calls.
 211  *                 Can be NULL.
 212  * @return Output code point or string length, see UCASE_MAX_STRING_LENGTH.
 213  *
 214  * @see UCaseContextIterator
 215  * @see UCASE_MAX_STRING_LENGTH
 216  * @internal
 217  */
 218 U_CAPI int32_t U_EXPORT2
 219 ucase_toFullLower(const UCaseProps *csp, UChar32 c,
 220                   UCaseContextIterator *iter, void *context,
 221                   const UChar **pString,
 222                   const char *locale, int32_t *locCache);
 223
 224 U_CAPI int32_t U_EXPORT2
 225 ucase_toFullUpper(const UCaseProps *csp, UChar32 c,
 226                   UCaseContextIterator *iter, void *context,
 227                   const UChar **pString,
 228                   const char *locale, int32_t *locCache);
 229
 230 U_CAPI int32_t U_EXPORT2
 231 ucase_toFullTitle(const UCaseProps *csp, UChar32 c,
 232                   UCaseContextIterator *iter, void *context,
 233                   const UChar **pString,
 234                   const char *locale, int32_t *locCache);
 235
 236 U_CAPI int32_t U_EXPORT2
 237 ucase_toFullFolding(const UCaseProps *csp, UChar32 c,
 238                     const UChar **pString,
 239                     uint32_t options);
 240
 241 U_CFUNC int32_t U_EXPORT2
 242 ucase_hasBinaryProperty(UChar32 c, UProperty which);
 243
 244 /* file definitions --------------------------------------------------------- */
 245
 246 #define UCASE_DATA_NAME "ucase"
 247 #define UCASE_DATA_TYPE "icu"
 248
 249 /* format "cAsE" */
 250 #define UCASE_FMT_0 0x63
 251 #define UCASE_FMT_1 0x41
 252 #define UCASE_FMT_2 0x53
 253 #define UCASE_FMT_3 0x45
 254
 255 /* indexes into indexes[] */
 256 enum {
 257     UCASE_IX_INDEX_TOP,
 258     UCASE_IX_LENGTH,
 259     UCASE_IX_TRIE_SIZE,
 260     UCASE_IX_EXC_LENGTH,
 261     UCASE_IX_UNFOLD_LENGTH,
 262
 263     UCASE_IX_MAX_FULL_LENGTH=15,
 264     UCASE_IX_TOP=16
 265 };
 266
 267 /* definitions for 16-bit case properties word ------------------------------ */
 268
 269 /* 2-bit constants for types of cased characters */
 270 #define UCASE_TYPE_MASK     3
 271 enum {
 272     UCASE_NONE,
 273     UCASE_LOWER,
 274     UCASE_UPPER,
 275     UCASE_TITLE
 276 };
 277
 278 #define UCASE_GET_TYPE(props) ((props)&UCASE_TYPE_MASK)
 279
 280 #define UCASE_SENSITIVE     4
 281 #define UCASE_EXCEPTION     8
 282
 283 #define UCASE_DOT_MASK      0x30
 284 enum {
 285     UCASE_NO_DOT=0,         /* normal characters with cc=0 */
 286     UCASE_SOFT_DOTTED=0x10, /* soft-dotted characters with cc=0 */
 287     UCASE_ABOVE=0x20,       /* "above" accents with cc=230 */
 288     UCASE_OTHER_ACCENT=0x30 /* other accent character (0<cc!=230) */
 289 };
 290
 291 /* no exception: bits 15..6 are a 10-bit signed case mapping delta */
 292 #define UCASE_DELTA_SHIFT   6
 293 #define UCASE_DELTA_MASK    0xffc0
 294 #define UCASE_MAX_DELTA     0x1ff
 295 #define UCASE_MIN_DELTA     (-UCASE_MAX_DELTA-1)
 296
 297 #define UCASE_GET_DELTA(props) ((int16_t)(props)>>UCASE_DELTA_SHIFT)
 298
 299 /* case-ignorable uses one of the delta bits, see gencase/store.c */
 300 #define UCASE_CASE_IGNORABLE 0x40
 301
 302 /* exception: bits 15..4 are an unsigned 12-bit index into the exceptions array */
 303 #define UCASE_EXC_SHIFT     4
 304 #define UCASE_EXC_MASK      0xfff0
 305 #define UCASE_MAX_EXCEPTIONS 0x1000
 306
 307 /* definitions for 16-bit main exceptions word ------------------------------ */
 308
 309 /* first 8 bits indicate values in optional slots */
 310 enum {
 311     UCASE_EXC_LOWER,
 312     UCASE_EXC_FOLD,
 313     UCASE_EXC_UPPER,
 314     UCASE_EXC_TITLE,
 315     UCASE_EXC_4,            /* reserved */
 316     UCASE_EXC_5,            /* reserved */
 317     UCASE_EXC_CLOSURE,
 318     UCASE_EXC_FULL_MAPPINGS,
 319     UCASE_EXC_ALL_SLOTS     /* one past the last slot */
 320 };
 321
 322 /* each slot is 2 uint16_t instead of 1 */
 323 #define UCASE_EXC_DOUBLE_SLOTS      0x100
 324
 325 /* reserved: exception bits 11..9 */
 326
 327 /* UCASE_EXC_DOT_MASK=UCASE_DOT_MASK<<UCASE_EXC_DOT_SHIFT */
 328 #define UCASE_EXC_DOT_SHIFT     8
 329
 330 /* normally stored in the main word, but pushed out for larger exception indexes */
 331 #define UCASE_EXC_DOT_MASK      0x3000
 332 enum {
 333     UCASE_EXC_NO_DOT=0,
 334     UCASE_EXC_SOFT_DOTTED=0x1000,
 335     UCASE_EXC_ABOVE=0x2000,         /* "above" accents with cc=230 */
 336     UCASE_EXC_OTHER_ACCENT=0x3000   /* other character (0<cc!=230) */
 337 };
 338
 339 /* complex/conditional mappings */
 340 #define UCASE_EXC_CONDITIONAL_SPECIAL   0x4000
 341 #define UCASE_EXC_CONDITIONAL_FOLD      0x8000
 342
 343 /* definitions for lengths word for full case mappings */
 344 #define UCASE_FULL_LOWER    0xf
 345 #define UCASE_FULL_FOLDING  0xf0
 346 #define UCASE_FULL_UPPER    0xf00
 347 #define UCASE_FULL_TITLE    0xf000
 348
 349 /* maximum lengths */
 350 #define UCASE_FULL_MAPPINGS_MAX_LENGTH (4*0xf)
 351 #define UCASE_CLOSURE_MAX_LENGTH 0xf
 352
 353 /* constants for reverse case folding ("unfold") data */
 354 enum {
 355     UCASE_UNFOLD_ROWS,
 356     UCASE_UNFOLD_ROW_WIDTH,
 357     UCASE_UNFOLD_STRING_WIDTH
 358 };
 359
 360 U_CDECL_END
 361
 362 #endif