+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*
-* Copyright (C) 2002-2006, International Business Machines
+* Copyright (C) 2002-2016, International Business Machines
* Corporation and others. All Rights Reserved.
*
*******************************************************************************
* file name: uprops.h
-* encoding: US-ASCII
+* encoding: UTF-8
* tab size: 8 (not used)
* indentation:4
*
UPROPS_ADDITIONAL_VECTORS_INDEX,
UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX,
- UPROPS_RESERVED_INDEX, /* 6 */
+ 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,
/* definitions for the main properties words */
enum {
/* general category shift==0 0 (5 bits) */
- UPROPS_NUMERIC_TYPE_SHIFT=5, /* 5 (3 bits) */
- UPROPS_NUMERIC_VALUE_SHIFT=8 /* 8 (8 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(props) (((props)>>UPROPS_NUMERIC_TYPE_SHIFT)&7)
-#define GET_NUMERIC_VALUE(props) (((props)>>UPROPS_NUMERIC_VALUE_SHIFT)&0xff)
+#define GET_NUMERIC_TYPE_VALUE(props) ((props)>>UPROPS_NUMERIC_TYPE_VALUE_SHIFT)
-/* internal numeric pseudo-types for special encodings of numeric values */
+/* constants for the storage form of numeric types and values */
enum {
- UPROPS_NT_FRACTION=4, /* ==U_NT_COUNT, must not change unless binary format version changes */
- UPROPS_NT_LARGE,
- UPROPS_NT_COUNT
+ /** 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
+ /** No numeric value (yet). */
+ UPROPS_NTV_RESERVED_START=UPROPS_NTV_FRACTION20_START+24, // 0x324+6*4=0x34c
+
+ UPROPS_NTV_MAX_SMALL_INT=UPROPS_NTV_FRACTION_START-UPROPS_NTV_NUMERIC_START-1
};
-/* encoding of fractional and large numbers */
-enum {
- UPROPS_MAX_SMALL_NUMBER=0xff,
-
- UPROPS_FRACTION_NUM_SHIFT=3, /* numerator: bits 7..3 */
- UPROPS_FRACTION_DEN_MASK=7, /* denominator: bits 2..0 */
-
- UPROPS_FRACTION_MAX_NUM=31,
- UPROPS_FRACTION_DEN_OFFSET=2, /* denominator values are 2..9 */
-
- UPROPS_FRACTION_MIN_DEN=UPROPS_FRACTION_DEN_OFFSET,
- UPROPS_FRACTION_MAX_DEN=UPROPS_FRACTION_MIN_DEN+UPROPS_FRACTION_DEN_MASK,
-
- UPROPS_LARGE_MANT_SHIFT=4, /* mantissa: bits 7..4 */
- UPROPS_LARGE_EXP_MASK=0xf, /* exponent: bits 3..0 */
- UPROPS_LARGE_EXP_OFFSET=2, /* regular exponents 2..17 */
- UPROPS_LARGE_EXP_OFFSET_EXTRA=18, /* extra large exponents 18..33 */
-
- UPROPS_LARGE_MIN_EXP=UPROPS_LARGE_EXP_OFFSET,
- UPROPS_LARGE_MAX_EXP=UPROPS_LARGE_MIN_EXP+UPROPS_LARGE_EXP_MASK,
- UPROPS_LARGE_MAX_EXP_EXTRA=UPROPS_LARGE_EXP_OFFSET_EXTRA+UPROPS_LARGE_EXP_MASK
-};
+#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..18 Line Break
- * 17..15 East Asian Width
- * 14.. 7 UBlockCode
- * 6.. 0 UScriptCode
+ * 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
-#define UPROPS_LB_MASK 0x00FC0000
-#define UPROPS_LB_SHIFT 18
+/* Script_Extensions: mask includes Script */
+#define UPROPS_SCRIPT_X_MASK 0x00c000ff
+#define UPROPS_SCRIPT_X_SHIFT 22
-#define UPROPS_EA_MASK 0x00038000
-#define UPROPS_EA_SHIFT 15
+#define UPROPS_EA_MASK 0x000e0000
+#define UPROPS_EA_SHIFT 17
-#define UPROPS_BLOCK_MASK 0x00007f80
-#define UPROPS_BLOCK_SHIFT 7
+#define UPROPS_BLOCK_MASK 0x0001ff00
+#define UPROPS_BLOCK_SHIFT 8
-#define UPROPS_SCRIPT_MASK 0x0000007f
+#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
*/
enum {
UPROPS_WHITE_SPACE,
- UPROPS_WAS_BIDI_CONTROL, /* reserved, was used in format version 3 */
- UPROPS_WAS_JOIN_CONTROL,
UPROPS_DASH,
UPROPS_HYPHEN,
UPROPS_QUOTATION_MARK,
UPROPS_IDEOGRAPHIC,
UPROPS_DIACRITIC,
UPROPS_EXTENDER,
- UPROPS_WAS_LOWERCASE, /* reserved, was used in format version 3 */
- UPROPS_WAS_UPPERCASE,
UPROPS_NONCHARACTER_CODE_POINT,
UPROPS_GRAPHEME_EXTEND,
UPROPS_GRAPHEME_LINK,
UPROPS_UNIFIED_IDEOGRAPH,
UPROPS_DEFAULT_IGNORABLE_CODE_POINT,
UPROPS_DEPRECATED,
- UPROPS_WAS_SOFT_DOTTED, /* reserved, was used in format version 3 */
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..24 More binary properties
- * 23..19 reserved
- * 18..14 Sentence Break
- * 13..10 Word Break
+ * 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
*/
-#define UPROPS_SB_MASK 0x0007c000
-#define UPROPS_SB_SHIFT 14
+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 0x00003c00
+#define UPROPS_WB_MASK 0x00007c00
#define UPROPS_WB_SHIFT 10
#define UPROPS_GCB_MASK 0x000003e0
#define UPROPS_DT_MASK 0x0000001f
-enum {
- UPROPS_V2_S_TERM=24, /* new in ICU 3.0 and Unicode 4.0.1 */
- UPROPS_V2_VARIATION_SELECTOR,
- UPROPS_V2_PATTERN_SYNTAX, /* new in ICU 3.4 and Unicode 4.1 */
- UPROPS_V2_PATTERN_WHITE_SPACE,
- UPROPS_V2_TOP /* must be <=32 */
-};
+/**
+ * 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.c.
- * column==-1 gets the 32-bit main properties word instead.
+ * Implemented in uchar.c for uprops.cpp.
* @return 0 if no data or illegal argument
*/
U_CFUNC uint32_t
U_CFUNC int32_t
uprv_getMaxValues(int32_t column);
-/**
- * Get the Hangul Syllable Type for c.
- * @internal
- */
-U_CFUNC UHangulSyllableType
-uchar_getHST(UChar32 c);
-
/**
* Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM.
* @internal
#define _Ll FLAG(U_LOWERCASE_LETTER)
#define _Lt FLAG(U_TITLECASE_LETTER)
#define _Lm FLAG(U_MODIFIER_LETTER)
-#define _Lo FLAG(U_OTHER_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 _Cs FLAG(U_SURROGATE)
#define _Pd FLAG(U_DASH_PUNCTUATION)
#define _Ps FLAG(U_START_PUNCTUATION)
-#define _Pe FLAG(U_END_PUNCTUATION)
-#define _Pc FLAG(U_CONNECTOR_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)
+/* #define _Pf FLAG(U_FINAL_PUNCTUATION) -- conflicts with MS Visual Studio 9.0 streambuf */
/** Some code points. @internal */
enum {
UPROPS_SRC_CHAR,
/** From uchar.c/uprops.icu properties vectors trie */
UPROPS_SRC_PROPSVEC,
- /** Hangul_Syllable_Type, from uchar.c/uprops.icu */
- UPROPS_SRC_HST,
/** From unames.c/unames.icu */
UPROPS_SRC_NAMES,
- /** From unorm.cpp/unorm.icu */
- UPROPS_SRC_NORM,
/** 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,
/** One more than the highest UPropertySource (UPROPS_SRC_) constant. */
UPROPS_SRC_COUNT
};
* @see UPropertySource
* @internal
*/
-U_CAPI UPropertySource U_EXPORT2
+U_CFUNC UPropertySource U_EXPORT2
uprops_getSource(UProperty which);
/**
* start of each range of same properties to the set.
* @internal
*/
-U_CAPI void U_EXPORT2
+U_CFUNC void U_EXPORT2
uchar_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode);
/**
* start of each range of same properties to the set.
* @internal
*/
-U_CAPI void U_EXPORT2
+U_CFUNC void U_EXPORT2
upropsvec_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode);
-/**
- * Same as uchar_addPropertyStarts() but only for Hangul_Syllable_Type.
- * @internal
- */
-U_CAPI void U_EXPORT2
-uhst_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode);
-
/**
* Return a set of characters for property enumeration.
* For each two consecutive characters (start, limit) in the set,
* @param sa USetAdder to receive result. Existing contents are lost.
* @internal
*/
-U_CAPI void U_EXPORT2
+/*U_CFUNC void U_EXPORT2
uprv_getInclusions(const USetAdder *sa, UErrorCode *pErrorCode);
-
-/**
- * Swap the ICU Unicode properties file. See uchar.c.
- * @internal
- */
-U_CAPI int32_t U_EXPORT2
-uprops_swap(const UDataSwapper *ds,
- const void *inData, int32_t length, void *outData,
- UErrorCode *pErrorCode);
+*/
/**
* Swap the ICU Unicode character names file. See uchar.c.
const void *inData, int32_t length, void *outData,
UErrorCode *pErrorCode);
+#ifdef __cplusplus
+
+U_NAMESPACE_BEGIN
+
+class UnicodeSet;
+
+// implemented in uniset_props.cpp
+U_CFUNC UnicodeSet *
+uniset_getUnicode32Instance(UErrorCode &errorCode);
+
+U_NAMESPACE_END
+
+#endif
+
#endif
projects / apple / icu.git / blobdiff