[apple/icu.git] / icuSources / common / uprops.h

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*
*   Copyright (C) 2002-2016, International Business Machines
*   Corporation and others.  All Rights Reserved.
*
*******************************************************************************
*   file name:  uprops.h
*   encoding:   UTF-8
*   tab size:   8 (not used)
*   indentation:4
*
*   created on: 2002feb24
*   created by: Markus W. Scherer
*
*   Constants for mostly non-core Unicode character properties
*   stored in uprops.icu.
*/

#ifndef __UPROPS_H__
#define __UPROPS_H__

#include "unicode/utypes.h"
#include "unicode/uset.h"
#include "uset_imp.h"
#include "udataswp.h"

/* indexes[] entries */
enum {
    UPROPS_PROPS32_INDEX,
    UPROPS_EXCEPTIONS_INDEX,
    UPROPS_EXCEPTIONS_TOP_INDEX,

    UPROPS_ADDITIONAL_TRIE_INDEX,
    UPROPS_ADDITIONAL_VECTORS_INDEX,
    UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX,

    UPROPS_SCRIPT_EXTENSIONS_INDEX,

    UPROPS_RESERVED_INDEX_7,
    UPROPS_RESERVED_INDEX_8,

    /* size of the data file (number of 32-bit units after the header) */
    UPROPS_DATA_TOP_INDEX,

    /* maximum values for code values in vector word 0 */
    UPROPS_MAX_VALUES_INDEX=10,
    /* maximum values for code values in vector word 2 */
    UPROPS_MAX_VALUES_2_INDEX,

    UPROPS_INDEX_COUNT=16
};

/* definitions for the main properties words */
enum {
    /* general category shift==0                                0 (5 bits) */
    /* reserved                                                 5 (1 bit) */
    UPROPS_NUMERIC_TYPE_VALUE_SHIFT=6                       /*  6 (10 bits) */
};

#define GET_CATEGORY(props) ((props)&0x1f)
#define CAT_MASK(props) U_MASK(GET_CATEGORY(props))

#define GET_NUMERIC_TYPE_VALUE(props) ((props)>>UPROPS_NUMERIC_TYPE_VALUE_SHIFT)

/* constants for the storage form of numeric types and values */
enum {
    /** No numeric value. */
    UPROPS_NTV_NONE=0,
    /** Decimal digits: nv=0..9 */
    UPROPS_NTV_DECIMAL_START=1,
    /** Other digits: nv=0..9 */
    UPROPS_NTV_DIGIT_START=11,
    /** Small integers: nv=0..154 */
    UPROPS_NTV_NUMERIC_START=21,
    /** Fractions: ((ntv>>4)-12) / ((ntv&0xf)+1) = -1..17 / 1..16 */
    UPROPS_NTV_FRACTION_START=0xb0,
    /**
     * Large integers:
     * ((ntv>>5)-14) * 10^((ntv&0x1f)+2) = (1..9)*(10^2..10^33)
     * (only one significant decimal digit)
     */
    UPROPS_NTV_LARGE_START=0x1e0,
    /**
     * Sexagesimal numbers:
     * ((ntv>>2)-0xbf) * 60^((ntv&3)+1) = (1..9)*(60^1..60^4)
     */
    UPROPS_NTV_BASE60_START=0x300,
    /**
     * Fraction-20 values:
     * frac20 = ntv-0x324 = 0..0x17 -> 1|3|5|7 / 20|40|80|160|320|640
     * numerator: num = 2*(frac20&3)+1
     * denominator: den = 20<<(frac20>>2)
     */
    UPROPS_NTV_FRACTION20_START=UPROPS_NTV_BASE60_START+36,  // 0x300+9*4=0x324
    /**
     * Fraction-32 values:
     * frac32 = ntv-0x34c = 0..15 -> 1|3|5|7 / 32|64|128|256
     * numerator: num = 2*(frac32&3)+1
     * denominator: den = 32<<(frac32>>2)
     */
    UPROPS_NTV_FRACTION32_START=UPROPS_NTV_FRACTION20_START+24,  // 0x324+6*4=0x34c
    /** No numeric value (yet). */
    UPROPS_NTV_RESERVED_START=UPROPS_NTV_FRACTION32_START+16,  // 0x34c+4*4=0x35c

    UPROPS_NTV_MAX_SMALL_INT=UPROPS_NTV_FRACTION_START-UPROPS_NTV_NUMERIC_START-1
};

#define UPROPS_NTV_GET_TYPE(ntv) \
    ((ntv==UPROPS_NTV_NONE) ? U_NT_NONE : \
    (ntv<UPROPS_NTV_DIGIT_START) ?  U_NT_DECIMAL : \
    (ntv<UPROPS_NTV_NUMERIC_START) ? U_NT_DIGIT : \
    U_NT_NUMERIC)

/* number of properties vector words */
#define UPROPS_VECTOR_WORDS     3

/*
 * Properties in vector word 0
 * Bits
 * 31..24   DerivedAge version major/minor one nibble each
 * 23..22   3..1: Bits 7..0 = Script_Extensions index
 *             3: Script value from Script_Extensions
 *             2: Script=Inherited
 *             1: Script=Common
 *             0: Script=bits 7..0
 * 21..20   reserved
 * 19..17   East Asian Width
 * 16.. 8   UBlockCode
 *  7.. 0   UScriptCode, or index to Script_Extensions
 */

/* derived age: one nibble each for major and minor version numbers */
#define UPROPS_AGE_MASK         0xff000000
#define UPROPS_AGE_SHIFT        24

/* Script_Extensions: mask includes Script */
#define UPROPS_SCRIPT_X_MASK    0x00c000ff
#define UPROPS_SCRIPT_X_SHIFT   22

#define UPROPS_EA_MASK          0x000e0000
#define UPROPS_EA_SHIFT         17

#define UPROPS_BLOCK_MASK       0x0001ff00
#define UPROPS_BLOCK_SHIFT      8

#define UPROPS_SCRIPT_MASK      0x000000ff

/* UPROPS_SCRIPT_X_WITH_COMMON must be the lowest value that involves Script_Extensions. */
#define UPROPS_SCRIPT_X_WITH_COMMON     0x400000
#define UPROPS_SCRIPT_X_WITH_INHERITED  0x800000
#define UPROPS_SCRIPT_X_WITH_OTHER      0xc00000

/*
 * Properties in vector word 1
 * Each bit encodes one binary property.
 * The following constants represent the bit number, use 1<<UPROPS_XYZ.
 * UPROPS_BINARY_1_TOP<=32!
 *
 * Keep this list of property enums in sync with
 * propListNames[] in icu/source/tools/genprops/props2.c!
 *
 * ICU 2.6/uprops format version 3.2 stores full properties instead of "Other_".
 */
enum {
    UPROPS_WHITE_SPACE,
    UPROPS_DASH,
    UPROPS_HYPHEN,
    UPROPS_QUOTATION_MARK,
    UPROPS_TERMINAL_PUNCTUATION,
    UPROPS_MATH,
    UPROPS_HEX_DIGIT,
    UPROPS_ASCII_HEX_DIGIT,
    UPROPS_ALPHABETIC,
    UPROPS_IDEOGRAPHIC,
    UPROPS_DIACRITIC,
    UPROPS_EXTENDER,
    UPROPS_NONCHARACTER_CODE_POINT,
    UPROPS_GRAPHEME_EXTEND,
    UPROPS_GRAPHEME_LINK,
    UPROPS_IDS_BINARY_OPERATOR,
    UPROPS_IDS_TRINARY_OPERATOR,
    UPROPS_RADICAL,
    UPROPS_UNIFIED_IDEOGRAPH,
    UPROPS_DEFAULT_IGNORABLE_CODE_POINT,
    UPROPS_DEPRECATED,
    UPROPS_LOGICAL_ORDER_EXCEPTION,
    UPROPS_XID_START,
    UPROPS_XID_CONTINUE,
    UPROPS_ID_START,                            /* ICU 2.6, uprops format version 3.2 */
    UPROPS_ID_CONTINUE,
    UPROPS_GRAPHEME_BASE,
    UPROPS_S_TERM,                              /* new in ICU 3.0 and Unicode 4.0.1 */
    UPROPS_VARIATION_SELECTOR,
    UPROPS_PATTERN_SYNTAX,                      /* new in ICU 3.4 and Unicode 4.1 */
    UPROPS_PATTERN_WHITE_SPACE,
    UPROPS_PREPENDED_CONCATENATION_MARK,        // new in ICU 60 and Unicode 10
    UPROPS_BINARY_1_TOP                         /* ==32 - full! */
};

/*
 * Properties in vector word 2
 * Bits
 * 31..26   http://www.unicode.org/reports/tr51/#Emoji_Properties
 * 25..20   Line Break
 * 19..15   Sentence Break
 * 14..10   Word Break
 *  9.. 5   Grapheme Cluster Break
 *  4.. 0   Decomposition Type
 */
enum {
    UPROPS_2_EXTENDED_PICTOGRAPHIC=26,
    UPROPS_2_EMOJI_COMPONENT,
    UPROPS_2_EMOJI,
    UPROPS_2_EMOJI_PRESENTATION,
    UPROPS_2_EMOJI_MODIFIER,
    UPROPS_2_EMOJI_MODIFIER_BASE
};

#define UPROPS_LB_MASK          0x03f00000
#define UPROPS_LB_SHIFT         20

#define UPROPS_SB_MASK          0x000f8000
#define UPROPS_SB_SHIFT         15

#define UPROPS_WB_MASK          0x00007c00
#define UPROPS_WB_SHIFT         10

#define UPROPS_GCB_MASK         0x000003e0
#define UPROPS_GCB_SHIFT        5

#define UPROPS_DT_MASK          0x0000001f

/**
 * Gets the main properties value for a code point.
 * Implemented in uchar.c for uprops.cpp.
 */
U_CFUNC uint32_t
u_getMainProperties(UChar32 c);

/**
 * Get a properties vector word for a code point.
 * Implemented in uchar.c for uprops.cpp.
 * @return 0 if no data or illegal argument
 */
U_CFUNC uint32_t
u_getUnicodeProperties(UChar32 c, int32_t column);

/**
 * Get the the maximum values for some enum/int properties.
 * Use the same column numbers as for u_getUnicodeProperties().
 * The returned value will contain maximum values stored in the same bit fields
 * as where the enum values are stored in the u_getUnicodeProperties()
 * return values for the same columns.
 *
 * Valid columns are those for properties words that contain enumerated values.
 * (ICU 2.6: columns 0 and 2)
 * For other column numbers, this function will return 0.
 *
 * @internal
 */
U_CFUNC int32_t
uprv_getMaxValues(int32_t column);

/**
 * Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM.
 * @internal
 */
U_CFUNC UBool
u_isalnumPOSIX(UChar32 c);

/**
 * Checks if c is in
 * [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}]
 * with space=\p{Whitespace} and Control=Cc.
 * Implements UCHAR_POSIX_GRAPH.
 * @internal
 */
U_CFUNC UBool
u_isgraphPOSIX(UChar32 c);

/**
 * Checks if c is in \p{graph}\p{blank} - \p{cntrl}.
 * Implements UCHAR_POSIX_PRINT.
 * @internal
 */
U_CFUNC UBool
u_isprintPOSIX(UChar32 c);

/** Turn a bit index into a bit flag. @internal */
#define FLAG(n) ((uint32_t)1<<(n))

/** Flags for general categories in the order of UCharCategory. @internal */
#define _Cn     FLAG(U_GENERAL_OTHER_TYPES)
#define _Lu     FLAG(U_UPPERCASE_LETTER)
#define _Ll     FLAG(U_LOWERCASE_LETTER)
#define _Lt     FLAG(U_TITLECASE_LETTER)
#define _Lm     FLAG(U_MODIFIER_LETTER)
/* #define _Lo     FLAG(U_OTHER_LETTER) -- conflicts with MS Visual Studio 9.0 xiosbase */
#define _Mn     FLAG(U_NON_SPACING_MARK)
#define _Me     FLAG(U_ENCLOSING_MARK)
#define _Mc     FLAG(U_COMBINING_SPACING_MARK)
#define _Nd     FLAG(U_DECIMAL_DIGIT_NUMBER)
#define _Nl     FLAG(U_LETTER_NUMBER)
#define _No     FLAG(U_OTHER_NUMBER)
#define _Zs     FLAG(U_SPACE_SEPARATOR)
#define _Zl     FLAG(U_LINE_SEPARATOR)
#define _Zp     FLAG(U_PARAGRAPH_SEPARATOR)
#define _Cc     FLAG(U_CONTROL_CHAR)
#define _Cf     FLAG(U_FORMAT_CHAR)
#define _Co     FLAG(U_PRIVATE_USE_CHAR)
#define _Cs     FLAG(U_SURROGATE)
#define _Pd     FLAG(U_DASH_PUNCTUATION)
#define _Ps     FLAG(U_START_PUNCTUATION)
/* #define _Pe     FLAG(U_END_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 xlocnum */
/* #define _Pc     FLAG(U_CONNECTOR_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 streambuf */
#define _Po     FLAG(U_OTHER_PUNCTUATION)
#define _Sm     FLAG(U_MATH_SYMBOL)
#define _Sc     FLAG(U_CURRENCY_SYMBOL)
#define _Sk     FLAG(U_MODIFIER_SYMBOL)
#define _So     FLAG(U_OTHER_SYMBOL)
#define _Pi     FLAG(U_INITIAL_PUNCTUATION)
/* #define _Pf     FLAG(U_FINAL_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 streambuf */

/** Some code points. @internal */
enum {
    TAB     =0x0009,
    LF      =0x000a,
    FF      =0x000c,
    CR      =0x000d,
    U_A     =0x0041,
    U_F     =0x0046,
    U_Z     =0x005a,
    U_a     =0x0061,
    U_f     =0x0066,
    U_z     =0x007a,
    DEL     =0x007f,
    NL      =0x0085,
    NBSP    =0x00a0,
    CGJ     =0x034f,
    FIGURESP=0x2007,
    HAIRSP  =0x200a,
    ZWNJ    =0x200c,
    ZWJ     =0x200d,
    RLM     =0x200f,
    NNBSP   =0x202f,
    WJ      =0x2060,
    INHSWAP =0x206a,
    NOMDIG  =0x206f,
    U_FW_A  =0xff21,
    U_FW_F  =0xff26,
    U_FW_Z  =0xff3a,
    U_FW_a  =0xff41,
    U_FW_f  =0xff46,
    U_FW_z  =0xff5a,
    ZWNBSP  =0xfeff
};

/**
 * Get the maximum length of a (regular/1.0/extended) character name.
 * @return 0 if no character names available.
 */
U_CAPI int32_t U_EXPORT2
uprv_getMaxCharNameLength(void);

/**
 * Fills set with characters that are used in Unicode character names.
 * Includes all characters that are used in regular/Unicode 1.0/extended names.
 * Just empties the set if no character names are available.
 * @param sa USetAdder to receive characters.
 */
U_CAPI void U_EXPORT2
uprv_getCharNameCharacters(const USetAdder *sa);

/**
 * Constants for which data and implementation files provide which properties.
 * Used by UnicodeSet for service-specific property enumeration.
 * @internal
 */
enum UPropertySource {
    /** No source, not a supported property. */
    UPROPS_SRC_NONE,
    /** From uchar.c/uprops.icu main trie */
    UPROPS_SRC_CHAR,
    /** From uchar.c/uprops.icu properties vectors trie */
    UPROPS_SRC_PROPSVEC,
    /** From unames.c/unames.icu */
    UPROPS_SRC_NAMES,
    /** From ucase.c/ucase.icu */
    UPROPS_SRC_CASE,
    /** From ubidi_props.c/ubidi.icu */
    UPROPS_SRC_BIDI,
    /** From uchar.c/uprops.icu main trie as well as properties vectors trie */
    UPROPS_SRC_CHAR_AND_PROPSVEC,
    /** From ucase.c/ucase.icu as well as unorm.cpp/unorm.icu */
    UPROPS_SRC_CASE_AND_NORM,
    /** From normalizer2impl.cpp/nfc.nrm */
    UPROPS_SRC_NFC,
    /** From normalizer2impl.cpp/nfkc.nrm */
    UPROPS_SRC_NFKC,
    /** From normalizer2impl.cpp/nfkc_cf.nrm */
    UPROPS_SRC_NFKC_CF,
    /** From normalizer2impl.cpp/nfc.nrm canonical iterator data */
    UPROPS_SRC_NFC_CANON_ITER,
    // Text layout properties.
    UPROPS_SRC_INPC,
    UPROPS_SRC_INSC,
    UPROPS_SRC_VO,
    /** One more than the highest UPropertySource (UPROPS_SRC_) constant. */
    UPROPS_SRC_COUNT
};
typedef enum UPropertySource UPropertySource;

/**
 * @see UPropertySource
 * @internal
 */
U_CFUNC UPropertySource U_EXPORT2
uprops_getSource(UProperty which);

/**
 * Enumerate uprops.icu's main data trie and add the
 * start of each range of same properties to the set.
 * @internal
 */
U_CFUNC void U_EXPORT2
uchar_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode);

/**
 * Enumerate uprops.icu's properties vectors trie and add the
 * start of each range of same properties to the set.
 * @internal
 */
U_CFUNC void U_EXPORT2
upropsvec_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode);

U_CFUNC void U_EXPORT2
uprops_addPropertyStarts(UPropertySource src, const USetAdder *sa, UErrorCode *pErrorCode);

/**
 * Return a set of characters for property enumeration.
 * For each two consecutive characters (start, limit) in the set,
 * all of the properties for start..limit-1 are all the same.
 *
 * @param sa USetAdder to receive result. Existing contents are lost.
 * @internal
 */
/*U_CFUNC void U_EXPORT2
uprv_getInclusions(const USetAdder *sa, UErrorCode *pErrorCode);
*/

/**
 * Swap the ICU Unicode character names file. See uchar.c.
 * @internal
 */
U_CAPI int32_t U_EXPORT2
uchar_swapNames(const UDataSwapper *ds,
                const void *inData, int32_t length, void *outData,
                UErrorCode *pErrorCode);

#ifdef __cplusplus

U_NAMESPACE_BEGIN

class UnicodeSet;

class CharacterProperties {
public:
    CharacterProperties() = delete;
    static const UnicodeSet *getInclusionsForProperty(UProperty prop, UErrorCode &errorCode);
};

// implemented in uniset_props.cpp
U_CFUNC UnicodeSet *
uniset_getUnicode32Instance(UErrorCode &errorCode);

U_NAMESPACE_END

#endif

#endif
Commit	Line	Data
f3c0d7a5 A	1	// © 2016 and later: Unicode, Inc. and others.
f3c0d7a5 A	2	// License & terms of use: http://www.unicode.org/copyright.html
b75a7d8f A	3	/*
	4	*******************************************************************************
	5	*
2ca993e8	6	* Copyright (C) 2002-2016, International Business Machines
b75a7d8f A	7	* Corporation and others. All Rights Reserved.
	8	*
	9	*******************************************************************************
	10	* file name: uprops.h
f3c0d7a5	11	* encoding: UTF-8
b75a7d8f A	12	* tab size: 8 (not used)
	13	* indentation:4
	14	*
	15	* created on: 2002feb24
	16	* created by: Markus W. Scherer
	17	*
	18	* Constants for mostly non-core Unicode character properties
73c04bcf	19	* stored in uprops.icu.
b75a7d8f A	20	*/
	21
	22	#ifndef __UPROPS_H__
	23	#define __UPROPS_H__
	24
	25	#include "unicode/utypes.h"
	26	#include "unicode/uset.h"
374ca955	27	#include "uset_imp.h"
374ca955	28	#include "udataswp.h"
b75a7d8f A	29
	30	/* indexes[] entries */
	31	enum {
	32	UPROPS_PROPS32_INDEX,
	33	UPROPS_EXCEPTIONS_INDEX,
	34	UPROPS_EXCEPTIONS_TOP_INDEX,
	35
	36	UPROPS_ADDITIONAL_TRIE_INDEX,
	37	UPROPS_ADDITIONAL_VECTORS_INDEX,
	38	UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX,
	39
729e4ab9 A	40	UPROPS_SCRIPT_EXTENSIONS_INDEX,
	41
	42	UPROPS_RESERVED_INDEX_7,
	43	UPROPS_RESERVED_INDEX_8,
	44
	45	/* size of the data file (number of 32-bit units after the header) */
	46	UPROPS_DATA_TOP_INDEX,
b75a7d8f A	47
	48	/* maximum values for code values in vector word 0 */
	49	UPROPS_MAX_VALUES_INDEX=10,
	50	/* maximum values for code values in vector word 2 */
	51	UPROPS_MAX_VALUES_2_INDEX,
	52
	53	UPROPS_INDEX_COUNT=16
	54	};
	55
	56	/* definitions for the main properties words */
	57	enum {
	58	/* general category shift==0 0 (5 bits) */
729e4ab9 A	59	/* reserved 5 (1 bit) */
729e4ab9 A	60	UPROPS_NUMERIC_TYPE_VALUE_SHIFT=6 /* 6 (10 bits) */
b75a7d8f A	61	};
b75a7d8f A	62
b75a7d8f	63	#define GET_CATEGORY(props) ((props)&0x1f)
73c04bcf A	64	#define CAT_MASK(props) U_MASK(GET_CATEGORY(props))
73c04bcf A	65
729e4ab9	66	#define GET_NUMERIC_TYPE_VALUE(props) ((props)>>UPROPS_NUMERIC_TYPE_VALUE_SHIFT)
b75a7d8f	67
729e4ab9	68	/* constants for the storage form of numeric types and values */
73c04bcf	69	enum {
51004dcb	70	/** No numeric value. */
729e4ab9	71	UPROPS_NTV_NONE=0,
51004dcb	72	/** Decimal digits: nv=0..9 */
729e4ab9	73	UPROPS_NTV_DECIMAL_START=1,
51004dcb	74	/** Other digits: nv=0..9 */
729e4ab9	75	UPROPS_NTV_DIGIT_START=11,
51004dcb	76	/** Small integers: nv=0..154 */
729e4ab9	77	UPROPS_NTV_NUMERIC_START=21,
51004dcb	78	/** Fractions: ((ntv>>4)-12) / ((ntv&0xf)+1) = -1..17 / 1..16 */
729e4ab9	79	UPROPS_NTV_FRACTION_START=0xb0,
51004dcb A	80	/**
	81	* Large integers:
	82	* ((ntv>>5)-14) * 10^((ntv&0x1f)+2) = (1..9)*(10^2..10^33)
	83	* (only one significant decimal digit)
	84	*/
729e4ab9	85	UPROPS_NTV_LARGE_START=0x1e0,
51004dcb A	86	/**
	87	* Sexagesimal numbers:
	88	* ((ntv>>2)-0xbf) * 60^((ntv&3)+1) = (1..9)*(60^1..60^4)
	89	*/
	90	UPROPS_NTV_BASE60_START=0x300,
f3c0d7a5 A	91	/**
	92	* Fraction-20 values:
	93	* frac20 = ntv-0x324 = 0..0x17 -> 1\|3\|5\|7 / 20\|40\|80\|160\|320\|640
	94	* numerator: num = 2*(frac20&3)+1
	95	* denominator: den = 20<<(frac20>>2)
	96	*/
	97	UPROPS_NTV_FRACTION20_START=UPROPS_NTV_BASE60_START+36, // 0x300+9*4=0x324
3d1f044b A	98	/**
	99	* Fraction-32 values:
	100	* frac32 = ntv-0x34c = 0..15 -> 1\|3\|5\|7 / 32\|64\|128\|256
	101	* numerator: num = 2*(frac32&3)+1
	102	* denominator: den = 32<<(frac32>>2)
	103	*/
	104	UPROPS_NTV_FRACTION32_START=UPROPS_NTV_FRACTION20_START+24, // 0x324+6*4=0x34c
51004dcb	105	/** No numeric value (yet). */
3d1f044b	106	UPROPS_NTV_RESERVED_START=UPROPS_NTV_FRACTION32_START+16, // 0x34c+4*4=0x35c
729e4ab9 A	107
729e4ab9 A	108	UPROPS_NTV_MAX_SMALL_INT=UPROPS_NTV_FRACTION_START-UPROPS_NTV_NUMERIC_START-1
73c04bcf	109	};
b75a7d8f	110
729e4ab9 A	111	#define UPROPS_NTV_GET_TYPE(ntv) \
	112	((ntv==UPROPS_NTV_NONE) ? U_NT_NONE : \
	113	(ntv<UPROPS_NTV_DIGIT_START) ? U_NT_DECIMAL : \
	114	(ntv<UPROPS_NTV_NUMERIC_START) ? U_NT_DIGIT : \
	115	U_NT_NUMERIC)
b75a7d8f A	116
	117	/* number of properties vector words */
	118	#define UPROPS_VECTOR_WORDS 3
	119
	120	/*
	121	* Properties in vector word 0
	122	* Bits
	123	* 31..24 DerivedAge version major/minor one nibble each
729e4ab9 A	124	* 23..22 3..1: Bits 7..0 = Script_Extensions index
	125	* 3: Script value from Script_Extensions
	126	* 2: Script=Inherited
	127	* 1: Script=Common
	128	* 0: Script=bits 7..0
	129	* 21..20 reserved
46f4442e A	130	* 19..17 East Asian Width
46f4442e A	131	* 16.. 8 UBlockCode
729e4ab9	132	* 7.. 0 UScriptCode, or index to Script_Extensions
b75a7d8f A	133	*/
	134
	135	/* derived age: one nibble each for major and minor version numbers */
	136	#define UPROPS_AGE_MASK 0xff000000
	137	#define UPROPS_AGE_SHIFT 24
	138
729e4ab9 A	139	/* Script_Extensions: mask includes Script */
	140	#define UPROPS_SCRIPT_X_MASK 0x00c000ff
	141	#define UPROPS_SCRIPT_X_SHIFT 22
	142
46f4442e A	143	#define UPROPS_EA_MASK 0x000e0000
46f4442e A	144	#define UPROPS_EA_SHIFT 17
b75a7d8f	145
46f4442e A	146	#define UPROPS_BLOCK_MASK 0x0001ff00
46f4442e A	147	#define UPROPS_BLOCK_SHIFT 8
b75a7d8f	148
46f4442e	149	#define UPROPS_SCRIPT_MASK 0x000000ff
b75a7d8f	150
729e4ab9 A	151	/* UPROPS_SCRIPT_X_WITH_COMMON must be the lowest value that involves Script_Extensions. */
	152	#define UPROPS_SCRIPT_X_WITH_COMMON 0x400000
	153	#define UPROPS_SCRIPT_X_WITH_INHERITED 0x800000
	154	#define UPROPS_SCRIPT_X_WITH_OTHER 0xc00000
	155
b75a7d8f A	156	/*
	157	* Properties in vector word 1
	158	* Each bit encodes one binary property.
	159	* The following constants represent the bit number, use 1<<UPROPS_XYZ.
	160	* UPROPS_BINARY_1_TOP<=32!
	161	*
	162	* Keep this list of property enums in sync with
	163	* propListNames[] in icu/source/tools/genprops/props2.c!
	164	*
	165	* ICU 2.6/uprops format version 3.2 stores full properties instead of "Other_".
	166	*/
	167	enum {
	168	UPROPS_WHITE_SPACE,
b75a7d8f A	169	UPROPS_DASH,
	170	UPROPS_HYPHEN,
	171	UPROPS_QUOTATION_MARK,
	172	UPROPS_TERMINAL_PUNCTUATION,
	173	UPROPS_MATH,
	174	UPROPS_HEX_DIGIT,
	175	UPROPS_ASCII_HEX_DIGIT,
	176	UPROPS_ALPHABETIC,
	177	UPROPS_IDEOGRAPHIC,
	178	UPROPS_DIACRITIC,
	179	UPROPS_EXTENDER,
b75a7d8f A	180	UPROPS_NONCHARACTER_CODE_POINT,
	181	UPROPS_GRAPHEME_EXTEND,
	182	UPROPS_GRAPHEME_LINK,
	183	UPROPS_IDS_BINARY_OPERATOR,
	184	UPROPS_IDS_TRINARY_OPERATOR,
	185	UPROPS_RADICAL,
	186	UPROPS_UNIFIED_IDEOGRAPH,
	187	UPROPS_DEFAULT_IGNORABLE_CODE_POINT,
	188	UPROPS_DEPRECATED,
b75a7d8f A	189	UPROPS_LOGICAL_ORDER_EXCEPTION,
	190	UPROPS_XID_START,
	191	UPROPS_XID_CONTINUE,
	192	UPROPS_ID_START, /* ICU 2.6, uprops format version 3.2 */
	193	UPROPS_ID_CONTINUE,
	194	UPROPS_GRAPHEME_BASE,
46f4442e A	195	UPROPS_S_TERM, /* new in ICU 3.0 and Unicode 4.0.1 */
	196	UPROPS_VARIATION_SELECTOR,
	197	UPROPS_PATTERN_SYNTAX, /* new in ICU 3.4 and Unicode 4.1 */
	198	UPROPS_PATTERN_WHITE_SPACE,
6be67b06	199	UPROPS_PREPENDED_CONCATENATION_MARK, // new in ICU 60 and Unicode 10
b75a7d8f A	200	UPROPS_BINARY_1_TOP /* ==32 - full! */
	201	};
	202
	203	/*
	204	* Properties in vector word 2
	205	* Bits
0f5d89e8	206	* 31..26 http://www.unicode.org/reports/tr51/#Emoji_Properties
46f4442e A	207	* 25..20 Line Break
	208	* 19..15 Sentence Break
	209	* 14..10 Word Break
73c04bcf	210	* 9.. 5 Grapheme Cluster Break
b75a7d8f A	211	* 4.. 0 Decomposition Type
b75a7d8f A	212	*/
2ca993e8	213	enum {
0f5d89e8 A	214	UPROPS_2_EXTENDED_PICTOGRAPHIC=26,
0f5d89e8 A	215	UPROPS_2_EMOJI_COMPONENT,
6be67b06	216	UPROPS_2_EMOJI,
2ca993e8 A	217	UPROPS_2_EMOJI_PRESENTATION,
	218	UPROPS_2_EMOJI_MODIFIER,
	219	UPROPS_2_EMOJI_MODIFIER_BASE
	220	};
	221
46f4442e A	222	#define UPROPS_LB_MASK 0x03f00000
46f4442e A	223	#define UPROPS_LB_SHIFT 20
46f4442e A	224
	225	#define UPROPS_SB_MASK 0x000f8000
	226	#define UPROPS_SB_SHIFT 15
73c04bcf	227
46f4442e	228	#define UPROPS_WB_MASK 0x00007c00
73c04bcf	229	#define UPROPS_WB_SHIFT 10
b75a7d8f	230
73c04bcf A	231	#define UPROPS_GCB_MASK 0x000003e0
73c04bcf A	232	#define UPROPS_GCB_SHIFT 5
b75a7d8f A	233
	234	#define UPROPS_DT_MASK 0x0000001f
	235
4388f060 A	236	/**
	237	* Gets the main properties value for a code point.
	238	* Implemented in uchar.c for uprops.cpp.
	239	*/
	240	U_CFUNC uint32_t
	241	u_getMainProperties(UChar32 c);
	242
b75a7d8f A	243	/**
b75a7d8f A	244	* Get a properties vector word for a code point.
4388f060	245	* Implemented in uchar.c for uprops.cpp.
b75a7d8f A	246	* @return 0 if no data or illegal argument
	247	*/
	248	U_CFUNC uint32_t
	249	u_getUnicodeProperties(UChar32 c, int32_t column);
	250
	251	/**
	252	* Get the the maximum values for some enum/int properties.
	253	* Use the same column numbers as for u_getUnicodeProperties().
	254	* The returned value will contain maximum values stored in the same bit fields
	255	* as where the enum values are stored in the u_getUnicodeProperties()
	256	* return values for the same columns.
	257	*
	258	* Valid columns are those for properties words that contain enumerated values.
	259	* (ICU 2.6: columns 0 and 2)
	260	* For other column numbers, this function will return 0.
	261	*
	262	* @internal
	263	*/
	264	U_CFUNC int32_t
	265	uprv_getMaxValues(int32_t column);
	266
73c04bcf A	267	/**
	268	* Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM.
	269	* @internal
	270	*/
	271	U_CFUNC UBool
	272	u_isalnumPOSIX(UChar32 c);
	273
	274	/**
	275	* Checks if c is in
	276	* [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}]
	277	* with space=\p{Whitespace} and Control=Cc.
	278	* Implements UCHAR_POSIX_GRAPH.
	279	* @internal
	280	*/
	281	U_CFUNC UBool
	282	u_isgraphPOSIX(UChar32 c);
	283
	284	/**
	285	* Checks if c is in \p{graph}\p{blank} - \p{cntrl}.
	286	* Implements UCHAR_POSIX_PRINT.
	287	* @internal
	288	*/
	289	U_CFUNC UBool
	290	u_isprintPOSIX(UChar32 c);
	291
b75a7d8f A	292	/** Turn a bit index into a bit flag. @internal */
	293	#define FLAG(n) ((uint32_t)1<<(n))
	294
	295	/** Flags for general categories in the order of UCharCategory. @internal */
	296	#define _Cn FLAG(U_GENERAL_OTHER_TYPES)
	297	#define _Lu FLAG(U_UPPERCASE_LETTER)
	298	#define _Ll FLAG(U_LOWERCASE_LETTER)
	299	#define _Lt FLAG(U_TITLECASE_LETTER)
	300	#define _Lm FLAG(U_MODIFIER_LETTER)
729e4ab9	301	/* #define _Lo FLAG(U_OTHER_LETTER) -- conflicts with MS Visual Studio 9.0 xiosbase */
b75a7d8f A	302	#define _Mn FLAG(U_NON_SPACING_MARK)
	303	#define _Me FLAG(U_ENCLOSING_MARK)
	304	#define _Mc FLAG(U_COMBINING_SPACING_MARK)
	305	#define _Nd FLAG(U_DECIMAL_DIGIT_NUMBER)
	306	#define _Nl FLAG(U_LETTER_NUMBER)
	307	#define _No FLAG(U_OTHER_NUMBER)
	308	#define _Zs FLAG(U_SPACE_SEPARATOR)
	309	#define _Zl FLAG(U_LINE_SEPARATOR)
	310	#define _Zp FLAG(U_PARAGRAPH_SEPARATOR)
	311	#define _Cc FLAG(U_CONTROL_CHAR)
	312	#define _Cf FLAG(U_FORMAT_CHAR)
	313	#define _Co FLAG(U_PRIVATE_USE_CHAR)
	314	#define _Cs FLAG(U_SURROGATE)
	315	#define _Pd FLAG(U_DASH_PUNCTUATION)
	316	#define _Ps FLAG(U_START_PUNCTUATION)
729e4ab9 A	317	/* #define _Pe FLAG(U_END_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 xlocnum */
729e4ab9 A	318	/* #define _Pc FLAG(U_CONNECTOR_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 streambuf */
b75a7d8f A	319	#define _Po FLAG(U_OTHER_PUNCTUATION)
	320	#define _Sm FLAG(U_MATH_SYMBOL)
	321	#define _Sc FLAG(U_CURRENCY_SYMBOL)
	322	#define _Sk FLAG(U_MODIFIER_SYMBOL)
	323	#define _So FLAG(U_OTHER_SYMBOL)
	324	#define _Pi FLAG(U_INITIAL_PUNCTUATION)
729e4ab9	325	/* #define _Pf FLAG(U_FINAL_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 streambuf */
b75a7d8f A	326
	327	/** Some code points. @internal */
	328	enum {
	329	TAB =0x0009,
	330	LF =0x000a,
	331	FF =0x000c,
	332	CR =0x000d,
	333	U_A =0x0041,
73c04bcf	334	U_F =0x0046,
b75a7d8f A	335	U_Z =0x005a,
b75a7d8f A	336	U_a =0x0061,
73c04bcf	337	U_f =0x0066,
b75a7d8f A	338	U_z =0x007a,
	339	DEL =0x007f,
	340	NL =0x0085,
	341	NBSP =0x00a0,
	342	CGJ =0x034f,
	343	FIGURESP=0x2007,
	344	HAIRSP =0x200a,
	345	ZWNJ =0x200c,
	346	ZWJ =0x200d,
	347	RLM =0x200f,
	348	NNBSP =0x202f,
	349	WJ =0x2060,
	350	INHSWAP =0x206a,
	351	NOMDIG =0x206f,
73c04bcf A	352	U_FW_A =0xff21,
	353	U_FW_F =0xff26,
	354	U_FW_Z =0xff3a,
	355	U_FW_a =0xff41,
	356	U_FW_f =0xff46,
	357	U_FW_z =0xff5a,
b75a7d8f A	358	ZWNBSP =0xfeff
	359	};
	360
b75a7d8f A	361	/**
	362	* Get the maximum length of a (regular/1.0/extended) character name.
	363	* @return 0 if no character names available.
	364	*/
	365	U_CAPI int32_t U_EXPORT2
	366	uprv_getMaxCharNameLength(void);
	367
b75a7d8f A	368	/**
	369	* Fills set with characters that are used in Unicode character names.
	370	* Includes all characters that are used in regular/Unicode 1.0/extended names.
	371	* Just empties the set if no character names are available.
374ca955	372	* @param sa USetAdder to receive characters.
b75a7d8f A	373	*/
b75a7d8f A	374	U_CAPI void U_EXPORT2
73c04bcf	375	uprv_getCharNameCharacters(const USetAdder *sa);
b75a7d8f	376
374ca955 A	377	/**
	378	* Constants for which data and implementation files provide which properties.
	379	* Used by UnicodeSet for service-specific property enumeration.
	380	* @internal
	381	*/
	382	enum UPropertySource {
	383	/** No source, not a supported property. */
	384	UPROPS_SRC_NONE,
73c04bcf	385	/** From uchar.c/uprops.icu main trie */
374ca955	386	UPROPS_SRC_CHAR,
73c04bcf A	387	/** From uchar.c/uprops.icu properties vectors trie */
73c04bcf A	388	UPROPS_SRC_PROPSVEC,
374ca955 A	389	/** From unames.c/unames.icu */
374ca955 A	390	UPROPS_SRC_NAMES,
374ca955 A	391	/** From ucase.c/ucase.icu */
374ca955 A	392	UPROPS_SRC_CASE,
73c04bcf	393	/** From ubidi_props.c/ubidi.icu */
374ca955	394	UPROPS_SRC_BIDI,
73c04bcf A	395	/** From uchar.c/uprops.icu main trie as well as properties vectors trie */
73c04bcf A	396	UPROPS_SRC_CHAR_AND_PROPSVEC,
729e4ab9 A	397	/** From ucase.c/ucase.icu as well as unorm.cpp/unorm.icu */
	398	UPROPS_SRC_CASE_AND_NORM,
	399	/** From normalizer2impl.cpp/nfc.nrm */
	400	UPROPS_SRC_NFC,
	401	/** From normalizer2impl.cpp/nfkc.nrm */
	402	UPROPS_SRC_NFKC,
	403	/** From normalizer2impl.cpp/nfkc_cf.nrm */
	404	UPROPS_SRC_NFKC_CF,
	405	/** From normalizer2impl.cpp/nfc.nrm canonical iterator data */
	406	UPROPS_SRC_NFC_CANON_ITER,
3d1f044b A	407	// Text layout properties.
	408	UPROPS_SRC_INPC,
	409	UPROPS_SRC_INSC,
	410	UPROPS_SRC_VO,
73c04bcf	411	/** One more than the highest UPropertySource (UPROPS_SRC_) constant. */
374ca955 A	412	UPROPS_SRC_COUNT
	413	};
	414	typedef enum UPropertySource UPropertySource;
	415
	416	/**
	417	* @see UPropertySource
	418	* @internal
	419	*/
46f4442e	420	U_CFUNC UPropertySource U_EXPORT2
374ca955 A	421	uprops_getSource(UProperty which);
374ca955 A	422
b75a7d8f	423	/**
73c04bcf A	424	* Enumerate uprops.icu's main data trie and add the
	425	* start of each range of same properties to the set.
	426	* @internal
	427	*/
46f4442e	428	U_CFUNC void U_EXPORT2
73c04bcf A	429	uchar_addPropertyStarts(const USetAdder sa, UErrorCode pErrorCode);
	430
	431	/**
	432	* Enumerate uprops.icu's properties vectors trie and add the
b75a7d8f A	433	* start of each range of same properties to the set.
	434	* @internal
	435	*/
46f4442e	436	U_CFUNC void U_EXPORT2
73c04bcf	437	upropsvec_addPropertyStarts(const USetAdder sa, UErrorCode pErrorCode);
374ca955	438
3d1f044b A	439	U_CFUNC void U_EXPORT2
	440	uprops_addPropertyStarts(UPropertySource src, const USetAdder sa, UErrorCode pErrorCode);
	441
b75a7d8f A	442	/**
	443	* Return a set of characters for property enumeration.
	444	* For each two consecutive characters (start, limit) in the set,
	445	* all of the properties for start..limit-1 are all the same.
	446	*
374ca955	447	* @param sa USetAdder to receive result. Existing contents are lost.
b75a7d8f A	448	* @internal
b75a7d8f A	449	*/
46f4442e	450	/*U_CFUNC void U_EXPORT2
73c04bcf	451	uprv_getInclusions(const USetAdder sa, UErrorCode pErrorCode);
46f4442e	452	*/
374ca955	453
374ca955 A	454	/**
	455	* Swap the ICU Unicode character names file. See uchar.c.
	456	* @internal
	457	*/
	458	U_CAPI int32_t U_EXPORT2
	459	uchar_swapNames(const UDataSwapper *ds,
	460	const void inData, int32_t length, void outData,
	461	UErrorCode *pErrorCode);
b75a7d8f	462
4388f060	463	#ifdef __cplusplus
729e4ab9 A	464
	465	U_NAMESPACE_BEGIN
	466
	467	class UnicodeSet;
	468
3d1f044b A	469	class CharacterProperties {
	470	public:
	471	CharacterProperties() = delete;
	472	static const UnicodeSet *getInclusionsForProperty(UProperty prop, UErrorCode &errorCode);
	473	};
	474
729e4ab9 A	475	// implemented in uniset_props.cpp
	476	U_CFUNC UnicodeSet *
	477	uniset_getUnicode32Instance(UErrorCode &errorCode);
	478
	479	U_NAMESPACE_END
	480
	481	#endif
	482
b75a7d8f	483	#endif