1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/ucase.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,409 @@
1.4 +/*
1.5 +*******************************************************************************
1.6 +*
1.7 +* Copyright (C) 2004-2012, International Business Machines
1.8 +* Corporation and others. All Rights Reserved.
1.9 +*
1.10 +*******************************************************************************
1.11 +* file name: ucase.h
1.12 +* encoding: US-ASCII
1.13 +* tab size: 8 (not used)
1.14 +* indentation:4
1.15 +*
1.16 +* created on: 2004aug30
1.17 +* created by: Markus W. Scherer
1.18 +*
1.19 +* Low-level Unicode character/string case mapping code.
1.20 +*/
1.21 +
1.22 +#ifndef __UCASE_H__
1.23 +#define __UCASE_H__
1.24 +
1.25 +#include "unicode/utypes.h"
1.26 +#include "unicode/uset.h"
1.27 +#include "putilimp.h"
1.28 +#include "uset_imp.h"
1.29 +#include "udataswp.h"
1.30 +
1.31 +#ifdef __cplusplus
1.32 +U_NAMESPACE_BEGIN
1.33 +
1.34 +class UnicodeString;
1.35 +
1.36 +U_NAMESPACE_END
1.37 +#endif
1.38 +
1.39 +/* library API -------------------------------------------------------------- */
1.40 +
1.41 +U_CDECL_BEGIN
1.42 +
1.43 +struct UCaseProps;
1.44 +typedef struct UCaseProps UCaseProps;
1.45 +
1.46 +U_CDECL_END
1.47 +
1.48 +U_CAPI const UCaseProps * U_EXPORT2
1.49 +ucase_getSingleton(void);
1.50 +
1.51 +U_CFUNC void U_EXPORT2
1.52 +ucase_addPropertyStarts(const UCaseProps *csp, const USetAdder *sa, UErrorCode *pErrorCode);
1.53 +
1.54 +/**
1.55 + * Requires non-NULL locale ID but otherwise does the equivalent of
1.56 + * checking for language codes as if uloc_getLanguage() were called:
1.57 + * Accepts both 2- and 3-letter codes and accepts case variants.
1.58 + */
1.59 +U_CFUNC int32_t
1.60 +ucase_getCaseLocale(const char *locale, int32_t *locCache);
1.61 +
1.62 +/* Casing locale types for ucase_getCaseLocale */
1.63 +enum {
1.64 + UCASE_LOC_UNKNOWN,
1.65 + UCASE_LOC_ROOT,
1.66 + UCASE_LOC_TURKISH,
1.67 + UCASE_LOC_LITHUANIAN,
1.68 + UCASE_LOC_DUTCH
1.69 +};
1.70 +
1.71 +/**
1.72 + * Bit mask for getting just the options from a string compare options word
1.73 + * that are relevant for case-insensitive string comparison.
1.74 + * See uchar.h. Also include _STRNCMP_STYLE and U_COMPARE_CODE_POINT_ORDER.
1.75 + * @internal
1.76 + */
1.77 +#define _STRCASECMP_OPTIONS_MASK 0xffff
1.78 +
1.79 +/**
1.80 + * Bit mask for getting just the options from a string compare options word
1.81 + * that are relevant for case folding (of a single string or code point).
1.82 + * See uchar.h.
1.83 + * @internal
1.84 + */
1.85 +#define _FOLD_CASE_OPTIONS_MASK 0xff
1.86 +
1.87 +/* single-code point functions */
1.88 +
1.89 +U_CAPI UChar32 U_EXPORT2
1.90 +ucase_tolower(const UCaseProps *csp, UChar32 c);
1.91 +
1.92 +U_CAPI UChar32 U_EXPORT2
1.93 +ucase_toupper(const UCaseProps *csp, UChar32 c);
1.94 +
1.95 +U_CAPI UChar32 U_EXPORT2
1.96 +ucase_totitle(const UCaseProps *csp, UChar32 c);
1.97 +
1.98 +U_CAPI UChar32 U_EXPORT2
1.99 +ucase_fold(const UCaseProps *csp, UChar32 c, uint32_t options);
1.100 +
1.101 +/**
1.102 + * Adds all simple case mappings and the full case folding for c to sa,
1.103 + * and also adds special case closure mappings.
1.104 + * c itself is not added.
1.105 + * For example, the mappings
1.106 + * - for s include long s
1.107 + * - for sharp s include ss
1.108 + * - for k include the Kelvin sign
1.109 + */
1.110 +U_CFUNC void U_EXPORT2
1.111 +ucase_addCaseClosure(const UCaseProps *csp, UChar32 c, const USetAdder *sa);
1.112 +
1.113 +/**
1.114 + * Maps the string to single code points and adds the associated case closure
1.115 + * mappings.
1.116 + * The string is mapped to code points if it is their full case folding string.
1.117 + * In other words, this performs a reverse full case folding and then
1.118 + * adds the case closure items of the resulting code points.
1.119 + * If the string is found and its closure applied, then
1.120 + * the string itself is added as well as part of its code points' closure.
1.121 + * It must be length>=0.
1.122 + *
1.123 + * @return TRUE if the string was found
1.124 + */
1.125 +U_CFUNC UBool U_EXPORT2
1.126 +ucase_addStringCaseClosure(const UCaseProps *csp, const UChar *s, int32_t length, const USetAdder *sa);
1.127 +
1.128 +#ifdef __cplusplus
1.129 +U_NAMESPACE_BEGIN
1.130 +
1.131 +/**
1.132 + * Iterator over characters with more than one code point in the full default Case_Folding.
1.133 + */
1.134 +class U_COMMON_API FullCaseFoldingIterator {
1.135 +public:
1.136 + /** Constructor. */
1.137 + FullCaseFoldingIterator();
1.138 + /**
1.139 + * Returns the next (cp, full) pair where "full" is cp's full default Case_Folding.
1.140 + * Returns a negative cp value at the end of the iteration.
1.141 + */
1.142 + UChar32 next(UnicodeString &full);
1.143 +private:
1.144 + FullCaseFoldingIterator(const FullCaseFoldingIterator &); // no copy
1.145 + FullCaseFoldingIterator &operator=(const FullCaseFoldingIterator &); // no assignment
1.146 +
1.147 + const UChar *unfold;
1.148 + int32_t unfoldRows;
1.149 + int32_t unfoldRowWidth;
1.150 + int32_t unfoldStringWidth;
1.151 + int32_t currentRow;
1.152 + int32_t rowCpIndex;
1.153 +};
1.154 +
1.155 +U_NAMESPACE_END
1.156 +#endif
1.157 +
1.158 +/** @return UCASE_NONE, UCASE_LOWER, UCASE_UPPER, UCASE_TITLE */
1.159 +U_CAPI int32_t U_EXPORT2
1.160 +ucase_getType(const UCaseProps *csp, UChar32 c);
1.161 +
1.162 +/** @return same as ucase_getType(), or <0 if c is case-ignorable */
1.163 +U_CAPI int32_t U_EXPORT2
1.164 +ucase_getTypeOrIgnorable(const UCaseProps *csp, UChar32 c);
1.165 +
1.166 +U_CAPI UBool U_EXPORT2
1.167 +ucase_isSoftDotted(const UCaseProps *csp, UChar32 c);
1.168 +
1.169 +U_CAPI UBool U_EXPORT2
1.170 +ucase_isCaseSensitive(const UCaseProps *csp, UChar32 c);
1.171 +
1.172 +/* string case mapping functions */
1.173 +
1.174 +U_CDECL_BEGIN
1.175 +
1.176 +/**
1.177 + * Iterator function for string case mappings, which need to look at the
1.178 + * context (surrounding text) of a given character for conditional mappings.
1.179 + *
1.180 + * The iterator only needs to go backward or forward away from the
1.181 + * character in question. It does not use any indexes on this interface.
1.182 + * It does not support random access or an arbitrary change of
1.183 + * iteration direction.
1.184 + *
1.185 + * The code point being case-mapped itself is never returned by
1.186 + * this iterator.
1.187 + *
1.188 + * @param context A pointer to the iterator's working data.
1.189 + * @param dir If <0 then start iterating backward from the character;
1.190 + * if >0 then start iterating forward from the character;
1.191 + * if 0 then continue iterating in the current direction.
1.192 + * @return Next code point, or <0 when the iteration is done.
1.193 + */
1.194 +typedef UChar32 U_CALLCONV
1.195 +UCaseContextIterator(void *context, int8_t dir);
1.196 +
1.197 +/**
1.198 + * Sample struct which may be used by some implementations of
1.199 + * UCaseContextIterator.
1.200 + */
1.201 +struct UCaseContext {
1.202 + void *p;
1.203 + int32_t start, index, limit;
1.204 + int32_t cpStart, cpLimit;
1.205 + int8_t dir;
1.206 + int8_t b1, b2, b3;
1.207 +};
1.208 +typedef struct UCaseContext UCaseContext;
1.209 +
1.210 +U_CDECL_END
1.211 +
1.212 +#define UCASECONTEXT_INITIALIZER { NULL, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
1.213 +
1.214 +enum {
1.215 + /**
1.216 + * For string case mappings, a single character (a code point) is mapped
1.217 + * either to itself (in which case in-place mapping functions do nothing),
1.218 + * or to another single code point, or to a string.
1.219 + * Aside from the string contents, these are indicated with a single int32_t
1.220 + * value as follows:
1.221 + *
1.222 + * Mapping to self: Negative values (~self instead of -self to support U+0000)
1.223 + *
1.224 + * Mapping to another code point: Positive values >UCASE_MAX_STRING_LENGTH
1.225 + *
1.226 + * Mapping to a string: The string length (0..UCASE_MAX_STRING_LENGTH) is
1.227 + * returned. Note that the string result may indeed have zero length.
1.228 + */
1.229 + UCASE_MAX_STRING_LENGTH=0x1f
1.230 +};
1.231 +
1.232 +/**
1.233 + * Get the full lowercase mapping for c.
1.234 + *
1.235 + * @param csp Case mapping properties.
1.236 + * @param c Character to be mapped.
1.237 + * @param iter Character iterator, used for context-sensitive mappings.
1.238 + * See UCaseContextIterator for details.
1.239 + * If iter==NULL then a context-independent result is returned.
1.240 + * @param context Pointer to be passed into iter.
1.241 + * @param pString If the mapping result is a string, then the pointer is
1.242 + * written to *pString.
1.243 + * @param locale Locale ID for locale-dependent mappings.
1.244 + * @param locCache Initialize to 0; may be used to cache the result of parsing
1.245 + * the locale ID for subsequent calls.
1.246 + * Can be NULL.
1.247 + * @return Output code point or string length, see UCASE_MAX_STRING_LENGTH.
1.248 + *
1.249 + * @see UCaseContextIterator
1.250 + * @see UCASE_MAX_STRING_LENGTH
1.251 + * @internal
1.252 + */
1.253 +U_CAPI int32_t U_EXPORT2
1.254 +ucase_toFullLower(const UCaseProps *csp, UChar32 c,
1.255 + UCaseContextIterator *iter, void *context,
1.256 + const UChar **pString,
1.257 + const char *locale, int32_t *locCache);
1.258 +
1.259 +U_CAPI int32_t U_EXPORT2
1.260 +ucase_toFullUpper(const UCaseProps *csp, UChar32 c,
1.261 + UCaseContextIterator *iter, void *context,
1.262 + const UChar **pString,
1.263 + const char *locale, int32_t *locCache);
1.264 +
1.265 +U_CAPI int32_t U_EXPORT2
1.266 +ucase_toFullTitle(const UCaseProps *csp, UChar32 c,
1.267 + UCaseContextIterator *iter, void *context,
1.268 + const UChar **pString,
1.269 + const char *locale, int32_t *locCache);
1.270 +
1.271 +U_CAPI int32_t U_EXPORT2
1.272 +ucase_toFullFolding(const UCaseProps *csp, UChar32 c,
1.273 + const UChar **pString,
1.274 + uint32_t options);
1.275 +
1.276 +U_CFUNC int32_t U_EXPORT2
1.277 +ucase_hasBinaryProperty(UChar32 c, UProperty which);
1.278 +
1.279 +
1.280 +U_CDECL_BEGIN
1.281 +
1.282 +/**
1.283 + * @internal
1.284 + */
1.285 +typedef int32_t U_CALLCONV
1.286 +UCaseMapFull(const UCaseProps *csp, UChar32 c,
1.287 + UCaseContextIterator *iter, void *context,
1.288 + const UChar **pString,
1.289 + const char *locale, int32_t *locCache);
1.290 +
1.291 +U_CDECL_END
1.292 +
1.293 +/* file definitions --------------------------------------------------------- */
1.294 +
1.295 +#define UCASE_DATA_NAME "ucase"
1.296 +#define UCASE_DATA_TYPE "icu"
1.297 +
1.298 +/* format "cAsE" */
1.299 +#define UCASE_FMT_0 0x63
1.300 +#define UCASE_FMT_1 0x41
1.301 +#define UCASE_FMT_2 0x53
1.302 +#define UCASE_FMT_3 0x45
1.303 +
1.304 +/* indexes into indexes[] */
1.305 +enum {
1.306 + UCASE_IX_INDEX_TOP,
1.307 + UCASE_IX_LENGTH,
1.308 + UCASE_IX_TRIE_SIZE,
1.309 + UCASE_IX_EXC_LENGTH,
1.310 + UCASE_IX_UNFOLD_LENGTH,
1.311 +
1.312 + UCASE_IX_MAX_FULL_LENGTH=15,
1.313 + UCASE_IX_TOP=16
1.314 +};
1.315 +
1.316 +/* definitions for 16-bit case properties word ------------------------------ */
1.317 +
1.318 +/* 2-bit constants for types of cased characters */
1.319 +#define UCASE_TYPE_MASK 3
1.320 +enum {
1.321 + UCASE_NONE,
1.322 + UCASE_LOWER,
1.323 + UCASE_UPPER,
1.324 + UCASE_TITLE
1.325 +};
1.326 +
1.327 +#define UCASE_GET_TYPE(props) ((props)&UCASE_TYPE_MASK)
1.328 +#define UCASE_GET_TYPE_AND_IGNORABLE(props) ((props)&7)
1.329 +
1.330 +#define UCASE_IGNORABLE 4
1.331 +#define UCASE_SENSITIVE 8
1.332 +#define UCASE_EXCEPTION 0x10
1.333 +
1.334 +#define UCASE_DOT_MASK 0x60
1.335 +enum {
1.336 + UCASE_NO_DOT=0, /* normal characters with cc=0 */
1.337 + UCASE_SOFT_DOTTED=0x20, /* soft-dotted characters with cc=0 */
1.338 + UCASE_ABOVE=0x40, /* "above" accents with cc=230 */
1.339 + UCASE_OTHER_ACCENT=0x60 /* other accent character (0<cc!=230) */
1.340 +};
1.341 +
1.342 +/* no exception: bits 15..7 are a 9-bit signed case mapping delta */
1.343 +#define UCASE_DELTA_SHIFT 7
1.344 +#define UCASE_DELTA_MASK 0xff80
1.345 +#define UCASE_MAX_DELTA 0xff
1.346 +#define UCASE_MIN_DELTA (-UCASE_MAX_DELTA-1)
1.347 +
1.348 +#if U_SIGNED_RIGHT_SHIFT_IS_ARITHMETIC
1.349 +# define UCASE_GET_DELTA(props) ((int16_t)(props)>>UCASE_DELTA_SHIFT)
1.350 +#else
1.351 +# define UCASE_GET_DELTA(props) (int16_t)(((props)&0x8000) ? (((props)>>UCASE_DELTA_SHIFT)|0xfe00) : ((uint16_t)(props)>>UCASE_DELTA_SHIFT))
1.352 +#endif
1.353 +
1.354 +/* exception: bits 15..5 are an unsigned 11-bit index into the exceptions array */
1.355 +#define UCASE_EXC_SHIFT 5
1.356 +#define UCASE_EXC_MASK 0xffe0
1.357 +#define UCASE_MAX_EXCEPTIONS ((UCASE_EXC_MASK>>UCASE_EXC_SHIFT)+1)
1.358 +
1.359 +/* definitions for 16-bit main exceptions word ------------------------------ */
1.360 +
1.361 +/* first 8 bits indicate values in optional slots */
1.362 +enum {
1.363 + UCASE_EXC_LOWER,
1.364 + UCASE_EXC_FOLD,
1.365 + UCASE_EXC_UPPER,
1.366 + UCASE_EXC_TITLE,
1.367 + UCASE_EXC_4, /* reserved */
1.368 + UCASE_EXC_5, /* reserved */
1.369 + UCASE_EXC_CLOSURE,
1.370 + UCASE_EXC_FULL_MAPPINGS,
1.371 + UCASE_EXC_ALL_SLOTS /* one past the last slot */
1.372 +};
1.373 +
1.374 +/* each slot is 2 uint16_t instead of 1 */
1.375 +#define UCASE_EXC_DOUBLE_SLOTS 0x100
1.376 +
1.377 +/* reserved: exception bits 11..9 */
1.378 +
1.379 +/* UCASE_EXC_DOT_MASK=UCASE_DOT_MASK<<UCASE_EXC_DOT_SHIFT */
1.380 +#define UCASE_EXC_DOT_SHIFT 7
1.381 +
1.382 +/* normally stored in the main word, but pushed out for larger exception indexes */
1.383 +#define UCASE_EXC_DOT_MASK 0x3000
1.384 +enum {
1.385 + UCASE_EXC_NO_DOT=0,
1.386 + UCASE_EXC_SOFT_DOTTED=0x1000,
1.387 + UCASE_EXC_ABOVE=0x2000, /* "above" accents with cc=230 */
1.388 + UCASE_EXC_OTHER_ACCENT=0x3000 /* other character (0<cc!=230) */
1.389 +};
1.390 +
1.391 +/* complex/conditional mappings */
1.392 +#define UCASE_EXC_CONDITIONAL_SPECIAL 0x4000
1.393 +#define UCASE_EXC_CONDITIONAL_FOLD 0x8000
1.394 +
1.395 +/* definitions for lengths word for full case mappings */
1.396 +#define UCASE_FULL_LOWER 0xf
1.397 +#define UCASE_FULL_FOLDING 0xf0
1.398 +#define UCASE_FULL_UPPER 0xf00
1.399 +#define UCASE_FULL_TITLE 0xf000
1.400 +
1.401 +/* maximum lengths */
1.402 +#define UCASE_FULL_MAPPINGS_MAX_LENGTH (4*0xf)
1.403 +#define UCASE_CLOSURE_MAX_LENGTH 0xf
1.404 +
1.405 +/* constants for reverse case folding ("unfold") data */
1.406 +enum {
1.407 + UCASE_UNFOLD_ROWS,
1.408 + UCASE_UNFOLD_ROW_WIDTH,
1.409 + UCASE_UNFOLD_STRING_WIDTH
1.410 +};
1.411 +
1.412 +#endif
