icuSources/common/ucase.h

   1 /*
   2 *******************************************************************************
   3 *
   4 *   Copyright (C) 2004, 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 UCaseProps * U_EXPORT2
  45 ucase_getSingleton(UErrorCode *pErrorCode);
  46
  47
  48 U_CAPI int32_t U_EXPORT2
  49 ucase_swap(const UDataSwapper *ds,
  50            const void *inData, int32_t length, void *outData,
  51            UErrorCode *pErrorCode);
  52
  53 U_CAPI void U_EXPORT2
  54 ucase_addPropertyStarts(const UCaseProps *csp, USetAdder *sa, UErrorCode *pErrorCode);
  55
  56 /**
  57  * Bit mask for getting just the options from a string compare options word
  58  * that are relevant for case-insensitive string comparison.
  59  * See uchar.h. Also include _STRNCMP_STYLE and U_COMPARE_CODE_POINT_ORDER.
  60  * @internal
  61  */
  62 #define _STRCASECMP_OPTIONS_MASK 0xffff
  63
  64 /**
  65  * Bit mask for getting just the options from a string compare options word
  66  * that are relevant for case folding (of a single string or code point).
  67  * See uchar.h.
  68  * @internal
  69  */
  70 #define _FOLD_CASE_OPTIONS_MASK 0xff
  71
  72 /* single-code point functions */
  73
  74 U_CAPI UChar32 U_EXPORT2
  75 ucase_tolower(const UCaseProps *csp, UChar32 c);
  76
  77 U_CAPI UChar32 U_EXPORT2
  78 ucase_toupper(const UCaseProps *csp, UChar32 c);
  79
  80 U_CAPI UChar32 U_EXPORT2
  81 ucase_totitle(const UCaseProps *csp, UChar32 c);
  82
  83 U_CAPI UChar32 U_EXPORT2
  84 ucase_fold(UCaseProps *csp, UChar32 c, uint32_t options);
  85
  86 /** @return UCASE_NONE, UCASE_LOWER, UCASE_UPPER, UCASE_TITLE */
  87 U_CAPI int32_t U_EXPORT2
  88 ucase_getType(const UCaseProps *csp, UChar32 c);
  89
  90 /** @return same as ucase_getType(), or <0 if c is case-ignorable */
  91 U_CAPI int32_t U_EXPORT2
  92 ucase_getTypeOrIgnorable(const UCaseProps *csp, UChar32 c);
  93
  94 U_CAPI UBool U_EXPORT2
  95 ucase_isSoftDotted(const UCaseProps *csp, UChar32 c);
  96
  97 U_CAPI UBool U_EXPORT2
  98 ucase_isCaseSensitive(const UCaseProps *csp, UChar32 c);
  99
 100 /* string case mapping functions */
 101
 102 /**
 103  * Iterator function for string case mappings, which need to look at the
 104  * context (surrounding text) of a given character for conditional mappings.
 105  *
 106  * The iterator only needs to go backward or forward away from the
 107  * character in question. It does not use any indexes on this interface.
 108  * It does not support random access or an arbitrary change of
 109  * iteration direction.
 110  *
 111  * The direction parameter either starts
 112  *
 113  * @param context A pointer to the iterator's working data.
 114  * @param dir If <0 then start iterating backward from the character;
 115  *            if >0 then start iterating forward from the character;
 116  *            if 0 then continue iterating in the current direction.
 117  * @return Next code point, or <0 when the iteration is done.
 118  */
 119 typedef UChar32 U_CALLCONV
 120 UCaseContextIterator(void *context, int8_t dir);
 121
 122 /**
 123  * Sample struct which may be used by some implementations of
 124  * UCaseContextIterator.
 125  */
 126 struct UCaseContext {
 127     void *p;
 128     int32_t start, index, limit;
 129     int32_t cpStart, cpLimit;
 130     int8_t dir;
 131     int8_t b1, b2, b3;
 132 };
 133 typedef struct UCaseContext UCaseContext;
 134
 135 enum {
 136     /**
 137      * For string case mappings, a single character (a code point) is mapped
 138      * either to itself (in which case in-place mapping functions do nothing),
 139      * or to another single code point, or to a string.
 140      * Aside from the string contents, these are indicated with a single int32_t
 141      * value as follows:
 142      *
 143      * Mapping to self: Negative values (~self instead of -self to support U+0000)
 144      *
 145      * Mapping to another code point: Positive values >UCASE_MAX_STRING_LENGTH
 146      *
 147      * Mapping to a string: The string length (0..UCASE_MAX_STRING_LENGTH) is
 148      * returned. Note that the string result may indeed have zero length.
 149      */
 150     UCASE_MAX_STRING_LENGTH=0x1f
 151 };
 152
 153 /**
 154  * Get the full lowercase mapping for c.
 155  *
 156  * @param csp Case mapping properties.
 157  * @param c Character to be mapped.
 158  * @param iter Character iterator, used for context-sensitive mappings.
 159  *             See UCaseContextIterator for details.
 160  *             If iter==NULL then a context-independent result is returned.
 161  * @param context Pointer to be passed into iter.
 162  * @param pString If the mapping result is a string, then the pointer is
 163  *                written to *pString.
 164  * @param locale Locale ID for locale-dependent mappings.
 165  * @param locCache Initialize to 0; may be used to cache the result of parsing
 166  *                 the locale ID for subsequent calls.
 167  *                 Can be NULL.
 168  * @return Output code point or string length, see UCASE_MAX_STRING_LENGTH.
 169  *
 170  * @see UCaseContextIterator
 171  * @see UCASE_MAX_STRING_LENGTH
 172  * @internal
 173  */
 174 U_CAPI int32_t U_EXPORT2
 175 ucase_toFullLower(const UCaseProps *csp, UChar32 c,
 176                   UCaseContextIterator *iter, void *context,
 177                   const UChar **pString,
 178                   const char *locale, int32_t *locCache);
 179
 180 U_CAPI int32_t U_EXPORT2
 181 ucase_toFullUpper(const UCaseProps *csp, UChar32 c,
 182                   UCaseContextIterator *iter, void *context,
 183                   const UChar **pString,
 184                   const char *locale, int32_t *locCache);
 185
 186 U_CAPI int32_t U_EXPORT2
 187 ucase_toFullTitle(const UCaseProps *csp, UChar32 c,
 188                   UCaseContextIterator *iter, void *context,
 189                   const UChar **pString,
 190                   const char *locale, int32_t *locCache);
 191
 192 U_CAPI int32_t U_EXPORT2
 193 ucase_toFullFolding(const UCaseProps *csp, UChar32 c,
 194                     const UChar **pString,
 195                     uint32_t options);
 196
 197 /* file definitions --------------------------------------------------------- */
 198
 199 #define UCASE_DATA_NAME "ucase"
 200 #define UCASE_DATA_TYPE "icu"
 201
 202 /* format "cAsE" */
 203 #define UCASE_FMT_0 0x63
 204 #define UCASE_FMT_1 0x41
 205 #define UCASE_FMT_2 0x53
 206 #define UCASE_FMT_3 0x45
 207
 208 /* indexes into indexes[] */
 209 enum {
 210     UCASE_IX_INDEX_TOP,
 211     UCASE_IX_LENGTH,
 212     UCASE_IX_TRIE_SIZE,
 213     UCASE_IX_EXC_LENGTH,
 214
 215     UCASE_IX_MAX_FULL_LENGTH=15,
 216     UCASE_IX_TOP=16
 217 };
 218
 219 /* definitions for 16-bit case properties word ------------------------------ */
 220
 221 /* 2-bit constants for types of cased characters */
 222 #define UCASE_TYPE_MASK     3
 223 enum {
 224     UCASE_NONE,
 225     UCASE_LOWER,
 226     UCASE_UPPER,
 227     UCASE_TITLE
 228 };
 229
 230 #define UCASE_SENSITIVE     4
 231 #define UCASE_EXCEPTION     8
 232
 233 #define UCASE_DOT_MASK      0x30
 234 enum {
 235     UCASE_NO_DOT=0,         /* normal characters with cc=0 */
 236     UCASE_SOFT_DOTTED=0x10, /* soft-dotted characters with cc=0 */
 237     UCASE_ABOVE=0x20,       /* "above" accents with cc=230 */
 238     UCASE_OTHER_ACCENT=0x30 /* other accent character (0<cc!=230) */
 239 };
 240
 241 /* no exception: bits 15..6 are a 10-bit signed case mapping delta */
 242 #define UCASE_DELTA_SHIFT   6
 243 #define UCASE_DELTA_MASK    0xffc0
 244 #define UCASE_MAX_DELTA     0x1ff
 245 #define UCASE_MIN_DELTA     (-UCASE_MAX_DELTA-1)
 246
 247 #define UCASE_GET_DELTA(props) ((int16_t)(props)>>UCASE_DELTA_SHIFT)
 248
 249 /* case-ignorable uses one of the delta bits, see gencase/store.c */
 250 #define UCASE_CASE_IGNORABLE 0x40
 251
 252 /* exception: bits 15..4 are an unsigned 12-bit index into the exceptions array */
 253 #define UCASE_EXC_SHIFT     4
 254 #define UCASE_EXC_MASK      0xfff0
 255 #define UCASE_MAX_EXCEPTIONS 0x1000
 256
 257 /* definitions for 16-bit main exceptions word ------------------------------ */
 258
 259 /* first 8 bits indicate values in optional slots */
 260 enum {
 261     UCASE_EXC_LOWER,
 262     UCASE_EXC_FOLD,
 263     UCASE_EXC_UPPER,
 264     UCASE_EXC_TITLE,
 265     UCASE_EXC_4,            /* reserved */
 266     UCASE_EXC_5,            /* reserved */
 267     UCASE_EXC_6,            /* reserved */
 268     UCASE_EXC_FULL_MAPPINGS,
 269     UCASE_EXC_ALL_SLOTS     /* one past the last slot */
 270 };
 271
 272 /* each slot is 2 uint16_t instead of 1 */
 273 #define UCASE_EXC_DOUBLE_SLOTS      0x100
 274
 275 /* reserved: exception bits 11..9 */
 276
 277 /* UCASE_EXC_DOT_MASK=UCASE_DOT_MASK<<UCASE_EXC_DOT_SHIFT */
 278 #define UCASE_EXC_DOT_SHIFT     8
 279
 280 /* normally stored in the main word, but pushed out for larger exception indexes */
 281 #define UCASE_EXC_DOT_MASK      0x3000
 282 enum {
 283     UCASE_EXC_NO_DOT=0,
 284     UCASE_EXC_SOFT_DOTTED=0x1000,
 285     UCASE_EXC_ABOVE=0x2000,         /* "above" accents with cc=230 */
 286     UCASE_EXC_OTHER_ACCENT=0x3000   /* other character (0<cc!=230) */
 287 };
 288
 289 /* complex/conditional mappings */
 290 #define UCASE_EXC_CONDITIONAL_SPECIAL   0x4000
 291 #define UCASE_EXC_CONDITIONAL_FOLD      0x8000
 292
 293 /* definitions for lengths word for full case mappings */
 294 #define UCASE_FULL_LOWER    0xf
 295 #define UCASE_FULL_FOLDING  0xf0
 296 #define UCASE_FULL_UPPER    0xf00
 297 #define UCASE_FULL_TITLE    0xf000
 298
 299 U_CDECL_END
 300
 301 #endif