1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/uprops.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,445 @@
1.4 +/*
1.5 +*******************************************************************************
1.6 +*
1.7 +* Copyright (C) 2002-2012, International Business Machines
1.8 +* Corporation and others. All Rights Reserved.
1.9 +*
1.10 +*******************************************************************************
1.11 +* file name: uprops.h
1.12 +* encoding: US-ASCII
1.13 +* tab size: 8 (not used)
1.14 +* indentation:4
1.15 +*
1.16 +* created on: 2002feb24
1.17 +* created by: Markus W. Scherer
1.18 +*
1.19 +* Constants for mostly non-core Unicode character properties
1.20 +* stored in uprops.icu.
1.21 +*/
1.22 +
1.23 +#ifndef __UPROPS_H__
1.24 +#define __UPROPS_H__
1.25 +
1.26 +#include "unicode/utypes.h"
1.27 +#include "unicode/uset.h"
1.28 +#include "uset_imp.h"
1.29 +#include "udataswp.h"
1.30 +
1.31 +/* indexes[] entries */
1.32 +enum {
1.33 + UPROPS_PROPS32_INDEX,
1.34 + UPROPS_EXCEPTIONS_INDEX,
1.35 + UPROPS_EXCEPTIONS_TOP_INDEX,
1.36 +
1.37 + UPROPS_ADDITIONAL_TRIE_INDEX,
1.38 + UPROPS_ADDITIONAL_VECTORS_INDEX,
1.39 + UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX,
1.40 +
1.41 + UPROPS_SCRIPT_EXTENSIONS_INDEX,
1.42 +
1.43 + UPROPS_RESERVED_INDEX_7,
1.44 + UPROPS_RESERVED_INDEX_8,
1.45 +
1.46 + /* size of the data file (number of 32-bit units after the header) */
1.47 + UPROPS_DATA_TOP_INDEX,
1.48 +
1.49 + /* maximum values for code values in vector word 0 */
1.50 + UPROPS_MAX_VALUES_INDEX=10,
1.51 + /* maximum values for code values in vector word 2 */
1.52 + UPROPS_MAX_VALUES_2_INDEX,
1.53 +
1.54 + UPROPS_INDEX_COUNT=16
1.55 +};
1.56 +
1.57 +/* definitions for the main properties words */
1.58 +enum {
1.59 + /* general category shift==0 0 (5 bits) */
1.60 + /* reserved 5 (1 bit) */
1.61 + UPROPS_NUMERIC_TYPE_VALUE_SHIFT=6 /* 6 (10 bits) */
1.62 +};
1.63 +
1.64 +#define GET_CATEGORY(props) ((props)&0x1f)
1.65 +#define CAT_MASK(props) U_MASK(GET_CATEGORY(props))
1.66 +
1.67 +#define GET_NUMERIC_TYPE_VALUE(props) ((props)>>UPROPS_NUMERIC_TYPE_VALUE_SHIFT)
1.68 +
1.69 +/* constants for the storage form of numeric types and values */
1.70 +enum {
1.71 + /** No numeric value. */
1.72 + UPROPS_NTV_NONE=0,
1.73 + /** Decimal digits: nv=0..9 */
1.74 + UPROPS_NTV_DECIMAL_START=1,
1.75 + /** Other digits: nv=0..9 */
1.76 + UPROPS_NTV_DIGIT_START=11,
1.77 + /** Small integers: nv=0..154 */
1.78 + UPROPS_NTV_NUMERIC_START=21,
1.79 + /** Fractions: ((ntv>>4)-12) / ((ntv&0xf)+1) = -1..17 / 1..16 */
1.80 + UPROPS_NTV_FRACTION_START=0xb0,
1.81 + /**
1.82 + * Large integers:
1.83 + * ((ntv>>5)-14) * 10^((ntv&0x1f)+2) = (1..9)*(10^2..10^33)
1.84 + * (only one significant decimal digit)
1.85 + */
1.86 + UPROPS_NTV_LARGE_START=0x1e0,
1.87 + /**
1.88 + * Sexagesimal numbers:
1.89 + * ((ntv>>2)-0xbf) * 60^((ntv&3)+1) = (1..9)*(60^1..60^4)
1.90 + */
1.91 + UPROPS_NTV_BASE60_START=0x300,
1.92 + /** No numeric value (yet). */
1.93 + UPROPS_NTV_RESERVED_START=UPROPS_NTV_BASE60_START+36, /* 0x300+9*4=0x324 */
1.94 +
1.95 + UPROPS_NTV_MAX_SMALL_INT=UPROPS_NTV_FRACTION_START-UPROPS_NTV_NUMERIC_START-1
1.96 +};
1.97 +
1.98 +#define UPROPS_NTV_GET_TYPE(ntv) \
1.99 + ((ntv==UPROPS_NTV_NONE) ? U_NT_NONE : \
1.100 + (ntv<UPROPS_NTV_DIGIT_START) ? U_NT_DECIMAL : \
1.101 + (ntv<UPROPS_NTV_NUMERIC_START) ? U_NT_DIGIT : \
1.102 + U_NT_NUMERIC)
1.103 +
1.104 +/* number of properties vector words */
1.105 +#define UPROPS_VECTOR_WORDS 3
1.106 +
1.107 +/*
1.108 + * Properties in vector word 0
1.109 + * Bits
1.110 + * 31..24 DerivedAge version major/minor one nibble each
1.111 + * 23..22 3..1: Bits 7..0 = Script_Extensions index
1.112 + * 3: Script value from Script_Extensions
1.113 + * 2: Script=Inherited
1.114 + * 1: Script=Common
1.115 + * 0: Script=bits 7..0
1.116 + * 21..20 reserved
1.117 + * 19..17 East Asian Width
1.118 + * 16.. 8 UBlockCode
1.119 + * 7.. 0 UScriptCode, or index to Script_Extensions
1.120 + */
1.121 +
1.122 +/* derived age: one nibble each for major and minor version numbers */
1.123 +#define UPROPS_AGE_MASK 0xff000000
1.124 +#define UPROPS_AGE_SHIFT 24
1.125 +
1.126 +/* Script_Extensions: mask includes Script */
1.127 +#define UPROPS_SCRIPT_X_MASK 0x00c000ff
1.128 +#define UPROPS_SCRIPT_X_SHIFT 22
1.129 +
1.130 +#define UPROPS_EA_MASK 0x000e0000
1.131 +#define UPROPS_EA_SHIFT 17
1.132 +
1.133 +#define UPROPS_BLOCK_MASK 0x0001ff00
1.134 +#define UPROPS_BLOCK_SHIFT 8
1.135 +
1.136 +#define UPROPS_SCRIPT_MASK 0x000000ff
1.137 +
1.138 +/* UPROPS_SCRIPT_X_WITH_COMMON must be the lowest value that involves Script_Extensions. */
1.139 +#define UPROPS_SCRIPT_X_WITH_COMMON 0x400000
1.140 +#define UPROPS_SCRIPT_X_WITH_INHERITED 0x800000
1.141 +#define UPROPS_SCRIPT_X_WITH_OTHER 0xc00000
1.142 +
1.143 +/*
1.144 + * Properties in vector word 1
1.145 + * Each bit encodes one binary property.
1.146 + * The following constants represent the bit number, use 1<<UPROPS_XYZ.
1.147 + * UPROPS_BINARY_1_TOP<=32!
1.148 + *
1.149 + * Keep this list of property enums in sync with
1.150 + * propListNames[] in icu/source/tools/genprops/props2.c!
1.151 + *
1.152 + * ICU 2.6/uprops format version 3.2 stores full properties instead of "Other_".
1.153 + */
1.154 +enum {
1.155 + UPROPS_WHITE_SPACE,
1.156 + UPROPS_DASH,
1.157 + UPROPS_HYPHEN,
1.158 + UPROPS_QUOTATION_MARK,
1.159 + UPROPS_TERMINAL_PUNCTUATION,
1.160 + UPROPS_MATH,
1.161 + UPROPS_HEX_DIGIT,
1.162 + UPROPS_ASCII_HEX_DIGIT,
1.163 + UPROPS_ALPHABETIC,
1.164 + UPROPS_IDEOGRAPHIC,
1.165 + UPROPS_DIACRITIC,
1.166 + UPROPS_EXTENDER,
1.167 + UPROPS_NONCHARACTER_CODE_POINT,
1.168 + UPROPS_GRAPHEME_EXTEND,
1.169 + UPROPS_GRAPHEME_LINK,
1.170 + UPROPS_IDS_BINARY_OPERATOR,
1.171 + UPROPS_IDS_TRINARY_OPERATOR,
1.172 + UPROPS_RADICAL,
1.173 + UPROPS_UNIFIED_IDEOGRAPH,
1.174 + UPROPS_DEFAULT_IGNORABLE_CODE_POINT,
1.175 + UPROPS_DEPRECATED,
1.176 + UPROPS_LOGICAL_ORDER_EXCEPTION,
1.177 + UPROPS_XID_START,
1.178 + UPROPS_XID_CONTINUE,
1.179 + UPROPS_ID_START, /* ICU 2.6, uprops format version 3.2 */
1.180 + UPROPS_ID_CONTINUE,
1.181 + UPROPS_GRAPHEME_BASE,
1.182 + UPROPS_S_TERM, /* new in ICU 3.0 and Unicode 4.0.1 */
1.183 + UPROPS_VARIATION_SELECTOR,
1.184 + UPROPS_PATTERN_SYNTAX, /* new in ICU 3.4 and Unicode 4.1 */
1.185 + UPROPS_PATTERN_WHITE_SPACE,
1.186 + UPROPS_RESERVED, /* reserved & unused */
1.187 + UPROPS_BINARY_1_TOP /* ==32 - full! */
1.188 +};
1.189 +
1.190 +/*
1.191 + * Properties in vector word 2
1.192 + * Bits
1.193 + * 31..26 reserved
1.194 + * 25..20 Line Break
1.195 + * 19..15 Sentence Break
1.196 + * 14..10 Word Break
1.197 + * 9.. 5 Grapheme Cluster Break
1.198 + * 4.. 0 Decomposition Type
1.199 + */
1.200 +#define UPROPS_LB_MASK 0x03f00000
1.201 +#define UPROPS_LB_SHIFT 20
1.202 +
1.203 +#define UPROPS_SB_MASK 0x000f8000
1.204 +#define UPROPS_SB_SHIFT 15
1.205 +
1.206 +#define UPROPS_WB_MASK 0x00007c00
1.207 +#define UPROPS_WB_SHIFT 10
1.208 +
1.209 +#define UPROPS_GCB_MASK 0x000003e0
1.210 +#define UPROPS_GCB_SHIFT 5
1.211 +
1.212 +#define UPROPS_DT_MASK 0x0000001f
1.213 +
1.214 +/**
1.215 + * Gets the main properties value for a code point.
1.216 + * Implemented in uchar.c for uprops.cpp.
1.217 + */
1.218 +U_CFUNC uint32_t
1.219 +u_getMainProperties(UChar32 c);
1.220 +
1.221 +/**
1.222 + * Get a properties vector word for a code point.
1.223 + * Implemented in uchar.c for uprops.cpp.
1.224 + * @return 0 if no data or illegal argument
1.225 + */
1.226 +U_CFUNC uint32_t
1.227 +u_getUnicodeProperties(UChar32 c, int32_t column);
1.228 +
1.229 +/**
1.230 + * Get the the maximum values for some enum/int properties.
1.231 + * Use the same column numbers as for u_getUnicodeProperties().
1.232 + * The returned value will contain maximum values stored in the same bit fields
1.233 + * as where the enum values are stored in the u_getUnicodeProperties()
1.234 + * return values for the same columns.
1.235 + *
1.236 + * Valid columns are those for properties words that contain enumerated values.
1.237 + * (ICU 2.6: columns 0 and 2)
1.238 + * For other column numbers, this function will return 0.
1.239 + *
1.240 + * @internal
1.241 + */
1.242 +U_CFUNC int32_t
1.243 +uprv_getMaxValues(int32_t column);
1.244 +
1.245 +/**
1.246 + * Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM.
1.247 + * @internal
1.248 + */
1.249 +U_CFUNC UBool
1.250 +u_isalnumPOSIX(UChar32 c);
1.251 +
1.252 +/**
1.253 + * Checks if c is in
1.254 + * [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}]
1.255 + * with space=\p{Whitespace} and Control=Cc.
1.256 + * Implements UCHAR_POSIX_GRAPH.
1.257 + * @internal
1.258 + */
1.259 +U_CFUNC UBool
1.260 +u_isgraphPOSIX(UChar32 c);
1.261 +
1.262 +/**
1.263 + * Checks if c is in \p{graph}\p{blank} - \p{cntrl}.
1.264 + * Implements UCHAR_POSIX_PRINT.
1.265 + * @internal
1.266 + */
1.267 +U_CFUNC UBool
1.268 +u_isprintPOSIX(UChar32 c);
1.269 +
1.270 +/** Turn a bit index into a bit flag. @internal */
1.271 +#define FLAG(n) ((uint32_t)1<<(n))
1.272 +
1.273 +/** Flags for general categories in the order of UCharCategory. @internal */
1.274 +#define _Cn FLAG(U_GENERAL_OTHER_TYPES)
1.275 +#define _Lu FLAG(U_UPPERCASE_LETTER)
1.276 +#define _Ll FLAG(U_LOWERCASE_LETTER)
1.277 +#define _Lt FLAG(U_TITLECASE_LETTER)
1.278 +#define _Lm FLAG(U_MODIFIER_LETTER)
1.279 +/* #define _Lo FLAG(U_OTHER_LETTER) -- conflicts with MS Visual Studio 9.0 xiosbase */
1.280 +#define _Mn FLAG(U_NON_SPACING_MARK)
1.281 +#define _Me FLAG(U_ENCLOSING_MARK)
1.282 +#define _Mc FLAG(U_COMBINING_SPACING_MARK)
1.283 +#define _Nd FLAG(U_DECIMAL_DIGIT_NUMBER)
1.284 +#define _Nl FLAG(U_LETTER_NUMBER)
1.285 +#define _No FLAG(U_OTHER_NUMBER)
1.286 +#define _Zs FLAG(U_SPACE_SEPARATOR)
1.287 +#define _Zl FLAG(U_LINE_SEPARATOR)
1.288 +#define _Zp FLAG(U_PARAGRAPH_SEPARATOR)
1.289 +#define _Cc FLAG(U_CONTROL_CHAR)
1.290 +#define _Cf FLAG(U_FORMAT_CHAR)
1.291 +#define _Co FLAG(U_PRIVATE_USE_CHAR)
1.292 +#define _Cs FLAG(U_SURROGATE)
1.293 +#define _Pd FLAG(U_DASH_PUNCTUATION)
1.294 +#define _Ps FLAG(U_START_PUNCTUATION)
1.295 +/* #define _Pe FLAG(U_END_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 xlocnum */
1.296 +/* #define _Pc FLAG(U_CONNECTOR_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 streambuf */
1.297 +#define _Po FLAG(U_OTHER_PUNCTUATION)
1.298 +#define _Sm FLAG(U_MATH_SYMBOL)
1.299 +#define _Sc FLAG(U_CURRENCY_SYMBOL)
1.300 +#define _Sk FLAG(U_MODIFIER_SYMBOL)
1.301 +#define _So FLAG(U_OTHER_SYMBOL)
1.302 +#define _Pi FLAG(U_INITIAL_PUNCTUATION)
1.303 +/* #define _Pf FLAG(U_FINAL_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 streambuf */
1.304 +
1.305 +/** Some code points. @internal */
1.306 +enum {
1.307 + TAB =0x0009,
1.308 + LF =0x000a,
1.309 + FF =0x000c,
1.310 + CR =0x000d,
1.311 + U_A =0x0041,
1.312 + U_F =0x0046,
1.313 + U_Z =0x005a,
1.314 + U_a =0x0061,
1.315 + U_f =0x0066,
1.316 + U_z =0x007a,
1.317 + DEL =0x007f,
1.318 + NL =0x0085,
1.319 + NBSP =0x00a0,
1.320 + CGJ =0x034f,
1.321 + FIGURESP=0x2007,
1.322 + HAIRSP =0x200a,
1.323 + ZWNJ =0x200c,
1.324 + ZWJ =0x200d,
1.325 + RLM =0x200f,
1.326 + NNBSP =0x202f,
1.327 + WJ =0x2060,
1.328 + INHSWAP =0x206a,
1.329 + NOMDIG =0x206f,
1.330 + U_FW_A =0xff21,
1.331 + U_FW_F =0xff26,
1.332 + U_FW_Z =0xff3a,
1.333 + U_FW_a =0xff41,
1.334 + U_FW_f =0xff46,
1.335 + U_FW_z =0xff5a,
1.336 + ZWNBSP =0xfeff
1.337 +};
1.338 +
1.339 +/**
1.340 + * Get the maximum length of a (regular/1.0/extended) character name.
1.341 + * @return 0 if no character names available.
1.342 + */
1.343 +U_CAPI int32_t U_EXPORT2
1.344 +uprv_getMaxCharNameLength(void);
1.345 +
1.346 +/**
1.347 + * Fills set with characters that are used in Unicode character names.
1.348 + * Includes all characters that are used in regular/Unicode 1.0/extended names.
1.349 + * Just empties the set if no character names are available.
1.350 + * @param sa USetAdder to receive characters.
1.351 + */
1.352 +U_CAPI void U_EXPORT2
1.353 +uprv_getCharNameCharacters(const USetAdder *sa);
1.354 +
1.355 +/**
1.356 + * Constants for which data and implementation files provide which properties.
1.357 + * Used by UnicodeSet for service-specific property enumeration.
1.358 + * @internal
1.359 + */
1.360 +enum UPropertySource {
1.361 + /** No source, not a supported property. */
1.362 + UPROPS_SRC_NONE,
1.363 + /** From uchar.c/uprops.icu main trie */
1.364 + UPROPS_SRC_CHAR,
1.365 + /** From uchar.c/uprops.icu properties vectors trie */
1.366 + UPROPS_SRC_PROPSVEC,
1.367 + /** From unames.c/unames.icu */
1.368 + UPROPS_SRC_NAMES,
1.369 + /** From ucase.c/ucase.icu */
1.370 + UPROPS_SRC_CASE,
1.371 + /** From ubidi_props.c/ubidi.icu */
1.372 + UPROPS_SRC_BIDI,
1.373 + /** From uchar.c/uprops.icu main trie as well as properties vectors trie */
1.374 + UPROPS_SRC_CHAR_AND_PROPSVEC,
1.375 + /** From ucase.c/ucase.icu as well as unorm.cpp/unorm.icu */
1.376 + UPROPS_SRC_CASE_AND_NORM,
1.377 + /** From normalizer2impl.cpp/nfc.nrm */
1.378 + UPROPS_SRC_NFC,
1.379 + /** From normalizer2impl.cpp/nfkc.nrm */
1.380 + UPROPS_SRC_NFKC,
1.381 + /** From normalizer2impl.cpp/nfkc_cf.nrm */
1.382 + UPROPS_SRC_NFKC_CF,
1.383 + /** From normalizer2impl.cpp/nfc.nrm canonical iterator data */
1.384 + UPROPS_SRC_NFC_CANON_ITER,
1.385 + /** One more than the highest UPropertySource (UPROPS_SRC_) constant. */
1.386 + UPROPS_SRC_COUNT
1.387 +};
1.388 +typedef enum UPropertySource UPropertySource;
1.389 +
1.390 +/**
1.391 + * @see UPropertySource
1.392 + * @internal
1.393 + */
1.394 +U_CFUNC UPropertySource U_EXPORT2
1.395 +uprops_getSource(UProperty which);
1.396 +
1.397 +/**
1.398 + * Enumerate uprops.icu's main data trie and add the
1.399 + * start of each range of same properties to the set.
1.400 + * @internal
1.401 + */
1.402 +U_CFUNC void U_EXPORT2
1.403 +uchar_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode);
1.404 +
1.405 +/**
1.406 + * Enumerate uprops.icu's properties vectors trie and add the
1.407 + * start of each range of same properties to the set.
1.408 + * @internal
1.409 + */
1.410 +U_CFUNC void U_EXPORT2
1.411 +upropsvec_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode);
1.412 +
1.413 +/**
1.414 + * Return a set of characters for property enumeration.
1.415 + * For each two consecutive characters (start, limit) in the set,
1.416 + * all of the properties for start..limit-1 are all the same.
1.417 + *
1.418 + * @param sa USetAdder to receive result. Existing contents are lost.
1.419 + * @internal
1.420 + */
1.421 +/*U_CFUNC void U_EXPORT2
1.422 +uprv_getInclusions(const USetAdder *sa, UErrorCode *pErrorCode);
1.423 +*/
1.424 +
1.425 +/**
1.426 + * Swap the ICU Unicode character names file. See uchar.c.
1.427 + * @internal
1.428 + */
1.429 +U_CAPI int32_t U_EXPORT2
1.430 +uchar_swapNames(const UDataSwapper *ds,
1.431 + const void *inData, int32_t length, void *outData,
1.432 + UErrorCode *pErrorCode);
1.433 +
1.434 +#ifdef __cplusplus
1.435 +
1.436 +U_NAMESPACE_BEGIN
1.437 +
1.438 +class UnicodeSet;
1.439 +
1.440 +// implemented in uniset_props.cpp
1.441 +U_CFUNC UnicodeSet *
1.442 +uniset_getUnicode32Instance(UErrorCode &errorCode);
1.443 +
1.444 +U_NAMESPACE_END
1.445 +
1.446 +#endif
1.447 +
1.448 +#endif
