1*0e209d39SAndroid Build Coastguard Worker // © 2016 and later: Unicode, Inc. and others.
2*0e209d39SAndroid Build Coastguard Worker // License & terms of use: http://www.unicode.org/copyright.html
3*0e209d39SAndroid Build Coastguard Worker /*
4*0e209d39SAndroid Build Coastguard Worker *******************************************************************************
5*0e209d39SAndroid Build Coastguard Worker *
6*0e209d39SAndroid Build Coastguard Worker * Copyright (C) 2002-2016, International Business Machines
7*0e209d39SAndroid Build Coastguard Worker * Corporation and others. All Rights Reserved.
8*0e209d39SAndroid Build Coastguard Worker *
9*0e209d39SAndroid Build Coastguard Worker *******************************************************************************
10*0e209d39SAndroid Build Coastguard Worker * file name: uprops.h
11*0e209d39SAndroid Build Coastguard Worker * encoding: UTF-8
12*0e209d39SAndroid Build Coastguard Worker * tab size: 8 (not used)
13*0e209d39SAndroid Build Coastguard Worker * indentation:4
14*0e209d39SAndroid Build Coastguard Worker *
15*0e209d39SAndroid Build Coastguard Worker * created on: 2002feb24
16*0e209d39SAndroid Build Coastguard Worker * created by: Markus W. Scherer
17*0e209d39SAndroid Build Coastguard Worker *
18*0e209d39SAndroid Build Coastguard Worker * Constants for mostly non-core Unicode character properties
19*0e209d39SAndroid Build Coastguard Worker * stored in uprops.icu.
20*0e209d39SAndroid Build Coastguard Worker */
21*0e209d39SAndroid Build Coastguard Worker
22*0e209d39SAndroid Build Coastguard Worker #ifndef __UPROPS_H__
23*0e209d39SAndroid Build Coastguard Worker #define __UPROPS_H__
24*0e209d39SAndroid Build Coastguard Worker
25*0e209d39SAndroid Build Coastguard Worker #include "unicode/utypes.h"
26*0e209d39SAndroid Build Coastguard Worker #include "unicode/uset.h"
27*0e209d39SAndroid Build Coastguard Worker #include "uset_imp.h"
28*0e209d39SAndroid Build Coastguard Worker #include "udataswp.h"
29*0e209d39SAndroid Build Coastguard Worker
30*0e209d39SAndroid Build Coastguard Worker /* indexes[] entries */
31*0e209d39SAndroid Build Coastguard Worker enum {
32*0e209d39SAndroid Build Coastguard Worker UPROPS_PROPS32_INDEX,
33*0e209d39SAndroid Build Coastguard Worker UPROPS_EXCEPTIONS_INDEX,
34*0e209d39SAndroid Build Coastguard Worker UPROPS_EXCEPTIONS_TOP_INDEX,
35*0e209d39SAndroid Build Coastguard Worker
36*0e209d39SAndroid Build Coastguard Worker UPROPS_ADDITIONAL_TRIE_INDEX,
37*0e209d39SAndroid Build Coastguard Worker UPROPS_ADDITIONAL_VECTORS_INDEX,
38*0e209d39SAndroid Build Coastguard Worker UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX,
39*0e209d39SAndroid Build Coastguard Worker
40*0e209d39SAndroid Build Coastguard Worker UPROPS_SCRIPT_EXTENSIONS_INDEX,
41*0e209d39SAndroid Build Coastguard Worker
42*0e209d39SAndroid Build Coastguard Worker UPROPS_RESERVED_INDEX_7,
43*0e209d39SAndroid Build Coastguard Worker UPROPS_RESERVED_INDEX_8,
44*0e209d39SAndroid Build Coastguard Worker
45*0e209d39SAndroid Build Coastguard Worker /* size of the data file (number of 32-bit units after the header) */
46*0e209d39SAndroid Build Coastguard Worker UPROPS_DATA_TOP_INDEX,
47*0e209d39SAndroid Build Coastguard Worker
48*0e209d39SAndroid Build Coastguard Worker /* maximum values for code values in vector word 0 */
49*0e209d39SAndroid Build Coastguard Worker UPROPS_MAX_VALUES_INDEX=10,
50*0e209d39SAndroid Build Coastguard Worker /* maximum values for code values in vector word 2 */
51*0e209d39SAndroid Build Coastguard Worker UPROPS_MAX_VALUES_2_INDEX,
52*0e209d39SAndroid Build Coastguard Worker
53*0e209d39SAndroid Build Coastguard Worker UPROPS_INDEX_COUNT=16
54*0e209d39SAndroid Build Coastguard Worker };
55*0e209d39SAndroid Build Coastguard Worker
56*0e209d39SAndroid Build Coastguard Worker /* definitions for the main properties words */
57*0e209d39SAndroid Build Coastguard Worker enum {
58*0e209d39SAndroid Build Coastguard Worker /* general category shift==0 0 (5 bits) */
59*0e209d39SAndroid Build Coastguard Worker /* reserved 5 (1 bit) */
60*0e209d39SAndroid Build Coastguard Worker UPROPS_NUMERIC_TYPE_VALUE_SHIFT=6 /* 6 (10 bits) */
61*0e209d39SAndroid Build Coastguard Worker };
62*0e209d39SAndroid Build Coastguard Worker
63*0e209d39SAndroid Build Coastguard Worker #define GET_CATEGORY(props) ((props)&0x1f)
64*0e209d39SAndroid Build Coastguard Worker #define CAT_MASK(props) U_MASK(GET_CATEGORY(props))
65*0e209d39SAndroid Build Coastguard Worker
66*0e209d39SAndroid Build Coastguard Worker #define GET_NUMERIC_TYPE_VALUE(props) ((props)>>UPROPS_NUMERIC_TYPE_VALUE_SHIFT)
67*0e209d39SAndroid Build Coastguard Worker
68*0e209d39SAndroid Build Coastguard Worker /* constants for the storage form of numeric types and values */
69*0e209d39SAndroid Build Coastguard Worker enum {
70*0e209d39SAndroid Build Coastguard Worker /** No numeric value. */
71*0e209d39SAndroid Build Coastguard Worker UPROPS_NTV_NONE=0,
72*0e209d39SAndroid Build Coastguard Worker /** Decimal digits: nv=0..9 */
73*0e209d39SAndroid Build Coastguard Worker UPROPS_NTV_DECIMAL_START=1,
74*0e209d39SAndroid Build Coastguard Worker /** Other digits: nv=0..9 */
75*0e209d39SAndroid Build Coastguard Worker UPROPS_NTV_DIGIT_START=11,
76*0e209d39SAndroid Build Coastguard Worker /** Small integers: nv=0..154 */
77*0e209d39SAndroid Build Coastguard Worker UPROPS_NTV_NUMERIC_START=21,
78*0e209d39SAndroid Build Coastguard Worker /** Fractions: ((ntv>>4)-12) / ((ntv&0xf)+1) = -1..17 / 1..16 */
79*0e209d39SAndroid Build Coastguard Worker UPROPS_NTV_FRACTION_START=0xb0,
80*0e209d39SAndroid Build Coastguard Worker /**
81*0e209d39SAndroid Build Coastguard Worker * Large integers:
82*0e209d39SAndroid Build Coastguard Worker * ((ntv>>5)-14) * 10^((ntv&0x1f)+2) = (1..9)*(10^2..10^33)
83*0e209d39SAndroid Build Coastguard Worker * (only one significant decimal digit)
84*0e209d39SAndroid Build Coastguard Worker */
85*0e209d39SAndroid Build Coastguard Worker UPROPS_NTV_LARGE_START=0x1e0,
86*0e209d39SAndroid Build Coastguard Worker /**
87*0e209d39SAndroid Build Coastguard Worker * Sexagesimal numbers:
88*0e209d39SAndroid Build Coastguard Worker * ((ntv>>2)-0xbf) * 60^((ntv&3)+1) = (1..9)*(60^1..60^4)
89*0e209d39SAndroid Build Coastguard Worker */
90*0e209d39SAndroid Build Coastguard Worker UPROPS_NTV_BASE60_START=0x300,
91*0e209d39SAndroid Build Coastguard Worker /**
92*0e209d39SAndroid Build Coastguard Worker * Fraction-20 values:
93*0e209d39SAndroid Build Coastguard Worker * frac20 = ntv-0x324 = 0..0x17 -> 1|3|5|7 / 20|40|80|160|320|640
94*0e209d39SAndroid Build Coastguard Worker * numerator: num = 2*(frac20&3)+1
95*0e209d39SAndroid Build Coastguard Worker * denominator: den = 20<<(frac20>>2)
96*0e209d39SAndroid Build Coastguard Worker */
97*0e209d39SAndroid Build Coastguard Worker UPROPS_NTV_FRACTION20_START=UPROPS_NTV_BASE60_START+36, // 0x300+9*4=0x324
98*0e209d39SAndroid Build Coastguard Worker /**
99*0e209d39SAndroid Build Coastguard Worker * Fraction-32 values:
100*0e209d39SAndroid Build Coastguard Worker * frac32 = ntv-0x34c = 0..15 -> 1|3|5|7 / 32|64|128|256
101*0e209d39SAndroid Build Coastguard Worker * numerator: num = 2*(frac32&3)+1
102*0e209d39SAndroid Build Coastguard Worker * denominator: den = 32<<(frac32>>2)
103*0e209d39SAndroid Build Coastguard Worker */
104*0e209d39SAndroid Build Coastguard Worker UPROPS_NTV_FRACTION32_START=UPROPS_NTV_FRACTION20_START+24, // 0x324+6*4=0x34c
105*0e209d39SAndroid Build Coastguard Worker /** No numeric value (yet). */
106*0e209d39SAndroid Build Coastguard Worker UPROPS_NTV_RESERVED_START=UPROPS_NTV_FRACTION32_START+16, // 0x34c+4*4=0x35c
107*0e209d39SAndroid Build Coastguard Worker
108*0e209d39SAndroid Build Coastguard Worker UPROPS_NTV_MAX_SMALL_INT=UPROPS_NTV_FRACTION_START-UPROPS_NTV_NUMERIC_START-1
109*0e209d39SAndroid Build Coastguard Worker };
110*0e209d39SAndroid Build Coastguard Worker
111*0e209d39SAndroid Build Coastguard Worker #define UPROPS_NTV_GET_TYPE(ntv) \
112*0e209d39SAndroid Build Coastguard Worker ((ntv==UPROPS_NTV_NONE) ? U_NT_NONE : \
113*0e209d39SAndroid Build Coastguard Worker (ntv<UPROPS_NTV_DIGIT_START) ? U_NT_DECIMAL : \
114*0e209d39SAndroid Build Coastguard Worker (ntv<UPROPS_NTV_NUMERIC_START) ? U_NT_DIGIT : \
115*0e209d39SAndroid Build Coastguard Worker U_NT_NUMERIC)
116*0e209d39SAndroid Build Coastguard Worker
117*0e209d39SAndroid Build Coastguard Worker /* number of properties vector words */
118*0e209d39SAndroid Build Coastguard Worker #define UPROPS_VECTOR_WORDS 3
119*0e209d39SAndroid Build Coastguard Worker
120*0e209d39SAndroid Build Coastguard Worker /*
121*0e209d39SAndroid Build Coastguard Worker * Properties in vector word 0
122*0e209d39SAndroid Build Coastguard Worker * Bits
123*0e209d39SAndroid Build Coastguard Worker * 31..24 DerivedAge version major/minor one nibble each
124*0e209d39SAndroid Build Coastguard Worker * 23..22 3..1: Bits 21..20 & 7..0 = Script_Extensions index
125*0e209d39SAndroid Build Coastguard Worker * 3: Script value from Script_Extensions
126*0e209d39SAndroid Build Coastguard Worker * 2: Script=Inherited
127*0e209d39SAndroid Build Coastguard Worker * 1: Script=Common
128*0e209d39SAndroid Build Coastguard Worker * 0: Script=bits 21..20 & 7..0
129*0e209d39SAndroid Build Coastguard Worker * 21..20 Bits 9..8 of the UScriptCode, or index to Script_Extensions
130*0e209d39SAndroid Build Coastguard Worker * 19..17 East Asian Width
131*0e209d39SAndroid Build Coastguard Worker * 16.. 8 UBlockCode
132*0e209d39SAndroid Build Coastguard Worker * 7.. 0 UScriptCode, or index to Script_Extensions
133*0e209d39SAndroid Build Coastguard Worker */
134*0e209d39SAndroid Build Coastguard Worker
135*0e209d39SAndroid Build Coastguard Worker /* derived age: one nibble each for major and minor version numbers */
136*0e209d39SAndroid Build Coastguard Worker #define UPROPS_AGE_MASK 0xff000000
137*0e209d39SAndroid Build Coastguard Worker #define UPROPS_AGE_SHIFT 24
138*0e209d39SAndroid Build Coastguard Worker
139*0e209d39SAndroid Build Coastguard Worker /* Script_Extensions: mask includes Script */
140*0e209d39SAndroid Build Coastguard Worker #define UPROPS_SCRIPT_X_MASK 0x00f000ff
141*0e209d39SAndroid Build Coastguard Worker #define UPROPS_SCRIPT_X_SHIFT 22
142*0e209d39SAndroid Build Coastguard Worker
143*0e209d39SAndroid Build Coastguard Worker // The UScriptCode or Script_Extensions index is split across two bit fields.
144*0e209d39SAndroid Build Coastguard Worker // (Starting with Unicode 13/ICU 66/2019 due to more varied Script_Extensions.)
145*0e209d39SAndroid Build Coastguard Worker // Shift the high bits right by 12 to assemble the full value.
146*0e209d39SAndroid Build Coastguard Worker #define UPROPS_SCRIPT_HIGH_MASK 0x00300000
147*0e209d39SAndroid Build Coastguard Worker #define UPROPS_SCRIPT_HIGH_SHIFT 12
148*0e209d39SAndroid Build Coastguard Worker #define UPROPS_MAX_SCRIPT 0x3ff
149*0e209d39SAndroid Build Coastguard Worker
150*0e209d39SAndroid Build Coastguard Worker #define UPROPS_EA_MASK 0x000e0000
151*0e209d39SAndroid Build Coastguard Worker #define UPROPS_EA_SHIFT 17
152*0e209d39SAndroid Build Coastguard Worker
153*0e209d39SAndroid Build Coastguard Worker #define UPROPS_BLOCK_MASK 0x0001ff00
154*0e209d39SAndroid Build Coastguard Worker #define UPROPS_BLOCK_SHIFT 8
155*0e209d39SAndroid Build Coastguard Worker
156*0e209d39SAndroid Build Coastguard Worker #define UPROPS_SCRIPT_LOW_MASK 0x000000ff
157*0e209d39SAndroid Build Coastguard Worker
158*0e209d39SAndroid Build Coastguard Worker /* UPROPS_SCRIPT_X_WITH_COMMON must be the lowest value that involves Script_Extensions. */
159*0e209d39SAndroid Build Coastguard Worker #define UPROPS_SCRIPT_X_WITH_COMMON 0x400000
160*0e209d39SAndroid Build Coastguard Worker #define UPROPS_SCRIPT_X_WITH_INHERITED 0x800000
161*0e209d39SAndroid Build Coastguard Worker #define UPROPS_SCRIPT_X_WITH_OTHER 0xc00000
162*0e209d39SAndroid Build Coastguard Worker
163*0e209d39SAndroid Build Coastguard Worker #ifdef __cplusplus
164*0e209d39SAndroid Build Coastguard Worker
165*0e209d39SAndroid Build Coastguard Worker namespace {
166*0e209d39SAndroid Build Coastguard Worker
uprops_mergeScriptCodeOrIndex(uint32_t scriptX)167*0e209d39SAndroid Build Coastguard Worker inline uint32_t uprops_mergeScriptCodeOrIndex(uint32_t scriptX) {
168*0e209d39SAndroid Build Coastguard Worker return
169*0e209d39SAndroid Build Coastguard Worker ((scriptX & UPROPS_SCRIPT_HIGH_MASK) >> UPROPS_SCRIPT_HIGH_SHIFT) |
170*0e209d39SAndroid Build Coastguard Worker (scriptX & UPROPS_SCRIPT_LOW_MASK);
171*0e209d39SAndroid Build Coastguard Worker }
172*0e209d39SAndroid Build Coastguard Worker
173*0e209d39SAndroid Build Coastguard Worker } // namespace
174*0e209d39SAndroid Build Coastguard Worker
175*0e209d39SAndroid Build Coastguard Worker #endif // __cplusplus
176*0e209d39SAndroid Build Coastguard Worker
177*0e209d39SAndroid Build Coastguard Worker /*
178*0e209d39SAndroid Build Coastguard Worker * Properties in vector word 1
179*0e209d39SAndroid Build Coastguard Worker * Each bit encodes one binary property.
180*0e209d39SAndroid Build Coastguard Worker * The following constants represent the bit number, use 1<<UPROPS_XYZ.
181*0e209d39SAndroid Build Coastguard Worker * UPROPS_BINARY_1_TOP<=32!
182*0e209d39SAndroid Build Coastguard Worker *
183*0e209d39SAndroid Build Coastguard Worker * Keep this list of property enums in sync with
184*0e209d39SAndroid Build Coastguard Worker * propListNames[] in icu/source/tools/genprops/props2.c!
185*0e209d39SAndroid Build Coastguard Worker *
186*0e209d39SAndroid Build Coastguard Worker * ICU 2.6/uprops format version 3.2 stores full properties instead of "Other_".
187*0e209d39SAndroid Build Coastguard Worker */
188*0e209d39SAndroid Build Coastguard Worker enum {
189*0e209d39SAndroid Build Coastguard Worker UPROPS_WHITE_SPACE,
190*0e209d39SAndroid Build Coastguard Worker UPROPS_DASH,
191*0e209d39SAndroid Build Coastguard Worker UPROPS_HYPHEN,
192*0e209d39SAndroid Build Coastguard Worker UPROPS_QUOTATION_MARK,
193*0e209d39SAndroid Build Coastguard Worker UPROPS_TERMINAL_PUNCTUATION,
194*0e209d39SAndroid Build Coastguard Worker UPROPS_MATH,
195*0e209d39SAndroid Build Coastguard Worker UPROPS_HEX_DIGIT,
196*0e209d39SAndroid Build Coastguard Worker UPROPS_ASCII_HEX_DIGIT,
197*0e209d39SAndroid Build Coastguard Worker UPROPS_ALPHABETIC,
198*0e209d39SAndroid Build Coastguard Worker UPROPS_IDEOGRAPHIC,
199*0e209d39SAndroid Build Coastguard Worker UPROPS_DIACRITIC,
200*0e209d39SAndroid Build Coastguard Worker UPROPS_EXTENDER,
201*0e209d39SAndroid Build Coastguard Worker UPROPS_NONCHARACTER_CODE_POINT,
202*0e209d39SAndroid Build Coastguard Worker UPROPS_GRAPHEME_EXTEND,
203*0e209d39SAndroid Build Coastguard Worker UPROPS_GRAPHEME_LINK,
204*0e209d39SAndroid Build Coastguard Worker UPROPS_IDS_BINARY_OPERATOR,
205*0e209d39SAndroid Build Coastguard Worker UPROPS_IDS_TRINARY_OPERATOR,
206*0e209d39SAndroid Build Coastguard Worker UPROPS_RADICAL,
207*0e209d39SAndroid Build Coastguard Worker UPROPS_UNIFIED_IDEOGRAPH,
208*0e209d39SAndroid Build Coastguard Worker UPROPS_DEFAULT_IGNORABLE_CODE_POINT,
209*0e209d39SAndroid Build Coastguard Worker UPROPS_DEPRECATED,
210*0e209d39SAndroid Build Coastguard Worker UPROPS_LOGICAL_ORDER_EXCEPTION,
211*0e209d39SAndroid Build Coastguard Worker UPROPS_XID_START,
212*0e209d39SAndroid Build Coastguard Worker UPROPS_XID_CONTINUE,
213*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_START, /* ICU 2.6, uprops format version 3.2 */
214*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_CONTINUE,
215*0e209d39SAndroid Build Coastguard Worker UPROPS_GRAPHEME_BASE,
216*0e209d39SAndroid Build Coastguard Worker UPROPS_S_TERM, /* new in ICU 3.0 and Unicode 4.0.1 */
217*0e209d39SAndroid Build Coastguard Worker UPROPS_VARIATION_SELECTOR,
218*0e209d39SAndroid Build Coastguard Worker UPROPS_PATTERN_SYNTAX, /* new in ICU 3.4 and Unicode 4.1 */
219*0e209d39SAndroid Build Coastguard Worker UPROPS_PATTERN_WHITE_SPACE,
220*0e209d39SAndroid Build Coastguard Worker UPROPS_PREPENDED_CONCATENATION_MARK, // new in ICU 60 and Unicode 10
221*0e209d39SAndroid Build Coastguard Worker UPROPS_BINARY_1_TOP /* ==32 - full! */
222*0e209d39SAndroid Build Coastguard Worker };
223*0e209d39SAndroid Build Coastguard Worker
224*0e209d39SAndroid Build Coastguard Worker /*
225*0e209d39SAndroid Build Coastguard Worker * Properties in vector word 2
226*0e209d39SAndroid Build Coastguard Worker * Bits
227*0e209d39SAndroid Build Coastguard Worker * 31..26 ICU 75: Identifier_Type bit set
228*0e209d39SAndroid Build Coastguard Worker * ICU 70..74: unused
229*0e209d39SAndroid Build Coastguard Worker * ICU 57..69: emoji properties; moved to uemoji.icu in ICU 70
230*0e209d39SAndroid Build Coastguard Worker * 25..20 Line Break
231*0e209d39SAndroid Build Coastguard Worker * 19..15 Sentence Break
232*0e209d39SAndroid Build Coastguard Worker * 14..10 Word Break
233*0e209d39SAndroid Build Coastguard Worker * 9.. 5 Grapheme Cluster Break
234*0e209d39SAndroid Build Coastguard Worker * 4.. 0 Decomposition Type
235*0e209d39SAndroid Build Coastguard Worker */
236*0e209d39SAndroid Build Coastguard Worker
237*0e209d39SAndroid Build Coastguard Worker #ifdef __cplusplus
238*0e209d39SAndroid Build Coastguard Worker
239*0e209d39SAndroid Build Coastguard Worker // https://www.unicode.org/reports/tr39/#Identifier_Status_and_Type
240*0e209d39SAndroid Build Coastguard Worker // The Identifier_Type maps each code point to a *set* of one or more values.
241*0e209d39SAndroid Build Coastguard Worker // Some can be combined with others, some can only occur alone.
242*0e209d39SAndroid Build Coastguard Worker // Exclusion & Limited_Use are combinable bits, but cannot occur together.
243*0e209d39SAndroid Build Coastguard Worker // We use this forbidden combination for enumerated values.
244*0e209d39SAndroid Build Coastguard Worker // We use 6 bits for all possible combinations.
245*0e209d39SAndroid Build Coastguard Worker // If more combinable values are added, then we need to use more bits.
246*0e209d39SAndroid Build Coastguard Worker //
247*0e209d39SAndroid Build Coastguard Worker // We do not store separate data for Identifier_Status:
248*0e209d39SAndroid Build Coastguard Worker // We can derive that from the encoded Identifier_Type via a simple range check.
249*0e209d39SAndroid Build Coastguard Worker
250*0e209d39SAndroid Build Coastguard Worker inline constexpr uint32_t UPROPS_2_ID_TYPE_MASK = 0xfc000000;
251*0e209d39SAndroid Build Coastguard Worker inline constexpr int32_t UPROPS_2_ID_TYPE_SHIFT = 26;
252*0e209d39SAndroid Build Coastguard Worker
253*0e209d39SAndroid Build Coastguard Worker enum {
254*0e209d39SAndroid Build Coastguard Worker // A high bit for use in idTypeToEncoded[] but not used in the data
255*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_BIT = 0x80,
256*0e209d39SAndroid Build Coastguard Worker
257*0e209d39SAndroid Build Coastguard Worker // Combinable bits
258*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_EXCLUSION = 0x20,
259*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_LIMITED_USE = 0x10,
260*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_UNCOMMON_USE = 8,
261*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_TECHNICAL = 4,
262*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_OBSOLETE = 2,
263*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_NOT_XID = 1,
264*0e209d39SAndroid Build Coastguard Worker
265*0e209d39SAndroid Build Coastguard Worker // Exclusive values
266*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_NOT_CHARACTER = 0,
267*0e209d39SAndroid Build Coastguard Worker
268*0e209d39SAndroid Build Coastguard Worker // Forbidden bit combination used for enumerating other exclusive values
269*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_FORBIDDEN = UPROPS_ID_TYPE_EXCLUSION | UPROPS_ID_TYPE_LIMITED_USE, // 0x30
270*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_DEPRECATED = UPROPS_ID_TYPE_FORBIDDEN, // 0x30
271*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_DEFAULT_IGNORABLE, // 0x31
272*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_NOT_NFKC, // 0x32
273*0e209d39SAndroid Build Coastguard Worker
274*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_ALLOWED_MIN = UPROPS_ID_TYPE_FORBIDDEN + 0xc, // 0x3c
275*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_INCLUSION = UPROPS_ID_TYPE_FORBIDDEN + 0xe, // 0x3e
276*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_RECOMMENDED = UPROPS_ID_TYPE_FORBIDDEN + 0xf, // 0x3f
277*0e209d39SAndroid Build Coastguard Worker };
278*0e209d39SAndroid Build Coastguard Worker
279*0e209d39SAndroid Build Coastguard Worker /**
280*0e209d39SAndroid Build Coastguard Worker * Maps UIdentifierType to encoded bits.
281*0e209d39SAndroid Build Coastguard Worker * When UPROPS_ID_TYPE_BIT is set, then use "&" to test whether the value bit is set.
282*0e209d39SAndroid Build Coastguard Worker * When UPROPS_ID_TYPE_BIT is not set, then compare ("==") the array value with the data value.
283*0e209d39SAndroid Build Coastguard Worker */
284*0e209d39SAndroid Build Coastguard Worker inline constexpr uint8_t uprops_idTypeToEncoded[] = {
285*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_NOT_CHARACTER,
286*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_DEPRECATED,
287*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_DEFAULT_IGNORABLE,
288*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_NOT_NFKC,
289*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_BIT | UPROPS_ID_TYPE_NOT_XID,
290*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_BIT | UPROPS_ID_TYPE_EXCLUSION,
291*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_BIT | UPROPS_ID_TYPE_OBSOLETE,
292*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_BIT | UPROPS_ID_TYPE_TECHNICAL,
293*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_BIT | UPROPS_ID_TYPE_UNCOMMON_USE,
294*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_BIT | UPROPS_ID_TYPE_LIMITED_USE,
295*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_INCLUSION,
296*0e209d39SAndroid Build Coastguard Worker UPROPS_ID_TYPE_RECOMMENDED
297*0e209d39SAndroid Build Coastguard Worker };
298*0e209d39SAndroid Build Coastguard Worker
299*0e209d39SAndroid Build Coastguard Worker #endif // __cplusplus
300*0e209d39SAndroid Build Coastguard Worker
301*0e209d39SAndroid Build Coastguard Worker #define UPROPS_LB_MASK 0x03f00000
302*0e209d39SAndroid Build Coastguard Worker #define UPROPS_LB_SHIFT 20
303*0e209d39SAndroid Build Coastguard Worker
304*0e209d39SAndroid Build Coastguard Worker #define UPROPS_SB_MASK 0x000f8000
305*0e209d39SAndroid Build Coastguard Worker #define UPROPS_SB_SHIFT 15
306*0e209d39SAndroid Build Coastguard Worker
307*0e209d39SAndroid Build Coastguard Worker #define UPROPS_WB_MASK 0x00007c00
308*0e209d39SAndroid Build Coastguard Worker #define UPROPS_WB_SHIFT 10
309*0e209d39SAndroid Build Coastguard Worker
310*0e209d39SAndroid Build Coastguard Worker #define UPROPS_GCB_MASK 0x000003e0
311*0e209d39SAndroid Build Coastguard Worker #define UPROPS_GCB_SHIFT 5
312*0e209d39SAndroid Build Coastguard Worker
313*0e209d39SAndroid Build Coastguard Worker #define UPROPS_DT_MASK 0x0000001f
314*0e209d39SAndroid Build Coastguard Worker
315*0e209d39SAndroid Build Coastguard Worker /**
316*0e209d39SAndroid Build Coastguard Worker * Gets the main properties value for a code point.
317*0e209d39SAndroid Build Coastguard Worker * Implemented in uchar.c for uprops.cpp.
318*0e209d39SAndroid Build Coastguard Worker */
319*0e209d39SAndroid Build Coastguard Worker U_CFUNC uint32_t
320*0e209d39SAndroid Build Coastguard Worker u_getMainProperties(UChar32 c);
321*0e209d39SAndroid Build Coastguard Worker
322*0e209d39SAndroid Build Coastguard Worker /**
323*0e209d39SAndroid Build Coastguard Worker * Get a properties vector word for a code point.
324*0e209d39SAndroid Build Coastguard Worker * Implemented in uchar.c for uprops.cpp.
325*0e209d39SAndroid Build Coastguard Worker * @return 0 if no data or illegal argument
326*0e209d39SAndroid Build Coastguard Worker */
327*0e209d39SAndroid Build Coastguard Worker U_CFUNC uint32_t
328*0e209d39SAndroid Build Coastguard Worker u_getUnicodeProperties(UChar32 c, int32_t column);
329*0e209d39SAndroid Build Coastguard Worker
330*0e209d39SAndroid Build Coastguard Worker /**
331*0e209d39SAndroid Build Coastguard Worker * Get the the maximum values for some enum/int properties.
332*0e209d39SAndroid Build Coastguard Worker * Use the same column numbers as for u_getUnicodeProperties().
333*0e209d39SAndroid Build Coastguard Worker * The returned value will contain maximum values stored in the same bit fields
334*0e209d39SAndroid Build Coastguard Worker * as where the enum values are stored in the u_getUnicodeProperties()
335*0e209d39SAndroid Build Coastguard Worker * return values for the same columns.
336*0e209d39SAndroid Build Coastguard Worker *
337*0e209d39SAndroid Build Coastguard Worker * Valid columns are those for properties words that contain enumerated values.
338*0e209d39SAndroid Build Coastguard Worker * (ICU 2.6: columns 0 and 2)
339*0e209d39SAndroid Build Coastguard Worker * For other column numbers, this function will return 0.
340*0e209d39SAndroid Build Coastguard Worker *
341*0e209d39SAndroid Build Coastguard Worker * @internal
342*0e209d39SAndroid Build Coastguard Worker */
343*0e209d39SAndroid Build Coastguard Worker U_CFUNC int32_t
344*0e209d39SAndroid Build Coastguard Worker uprv_getMaxValues(int32_t column);
345*0e209d39SAndroid Build Coastguard Worker
346*0e209d39SAndroid Build Coastguard Worker /**
347*0e209d39SAndroid Build Coastguard Worker * Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM.
348*0e209d39SAndroid Build Coastguard Worker * @internal
349*0e209d39SAndroid Build Coastguard Worker */
350*0e209d39SAndroid Build Coastguard Worker U_CFUNC UBool
351*0e209d39SAndroid Build Coastguard Worker u_isalnumPOSIX(UChar32 c);
352*0e209d39SAndroid Build Coastguard Worker
353*0e209d39SAndroid Build Coastguard Worker /**
354*0e209d39SAndroid Build Coastguard Worker * Checks if c is in
355*0e209d39SAndroid Build Coastguard Worker * [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}]
356*0e209d39SAndroid Build Coastguard Worker * with space=\p{Whitespace} and Control=Cc.
357*0e209d39SAndroid Build Coastguard Worker * Implements UCHAR_POSIX_GRAPH.
358*0e209d39SAndroid Build Coastguard Worker * @internal
359*0e209d39SAndroid Build Coastguard Worker */
360*0e209d39SAndroid Build Coastguard Worker U_CFUNC UBool
361*0e209d39SAndroid Build Coastguard Worker u_isgraphPOSIX(UChar32 c);
362*0e209d39SAndroid Build Coastguard Worker
363*0e209d39SAndroid Build Coastguard Worker /**
364*0e209d39SAndroid Build Coastguard Worker * Checks if c is in \p{graph}\p{blank} - \p{cntrl}.
365*0e209d39SAndroid Build Coastguard Worker * Implements UCHAR_POSIX_PRINT.
366*0e209d39SAndroid Build Coastguard Worker * @internal
367*0e209d39SAndroid Build Coastguard Worker */
368*0e209d39SAndroid Build Coastguard Worker U_CFUNC UBool
369*0e209d39SAndroid Build Coastguard Worker u_isprintPOSIX(UChar32 c);
370*0e209d39SAndroid Build Coastguard Worker
371*0e209d39SAndroid Build Coastguard Worker /** Some code points. @internal */
372*0e209d39SAndroid Build Coastguard Worker enum {
373*0e209d39SAndroid Build Coastguard Worker TAB =0x0009,
374*0e209d39SAndroid Build Coastguard Worker LF =0x000a,
375*0e209d39SAndroid Build Coastguard Worker FF =0x000c,
376*0e209d39SAndroid Build Coastguard Worker CR =0x000d,
377*0e209d39SAndroid Build Coastguard Worker NBSP =0x00a0,
378*0e209d39SAndroid Build Coastguard Worker CGJ =0x034f,
379*0e209d39SAndroid Build Coastguard Worker FIGURESP=0x2007,
380*0e209d39SAndroid Build Coastguard Worker HAIRSP =0x200a,
381*0e209d39SAndroid Build Coastguard Worker ZWNJ =0x200c,
382*0e209d39SAndroid Build Coastguard Worker ZWJ =0x200d,
383*0e209d39SAndroid Build Coastguard Worker RLM =0x200f,
384*0e209d39SAndroid Build Coastguard Worker NNBSP =0x202f,
385*0e209d39SAndroid Build Coastguard Worker ZWNBSP =0xfeff
386*0e209d39SAndroid Build Coastguard Worker };
387*0e209d39SAndroid Build Coastguard Worker
388*0e209d39SAndroid Build Coastguard Worker /**
389*0e209d39SAndroid Build Coastguard Worker * Get the maximum length of a (regular/1.0/extended) character name.
390*0e209d39SAndroid Build Coastguard Worker * @return 0 if no character names available.
391*0e209d39SAndroid Build Coastguard Worker */
392*0e209d39SAndroid Build Coastguard Worker U_CAPI int32_t U_EXPORT2
393*0e209d39SAndroid Build Coastguard Worker uprv_getMaxCharNameLength(void);
394*0e209d39SAndroid Build Coastguard Worker
395*0e209d39SAndroid Build Coastguard Worker /**
396*0e209d39SAndroid Build Coastguard Worker * Fills set with characters that are used in Unicode character names.
397*0e209d39SAndroid Build Coastguard Worker * Includes all characters that are used in regular/Unicode 1.0/extended names.
398*0e209d39SAndroid Build Coastguard Worker * Just empties the set if no character names are available.
399*0e209d39SAndroid Build Coastguard Worker * @param sa USetAdder to receive characters.
400*0e209d39SAndroid Build Coastguard Worker */
401*0e209d39SAndroid Build Coastguard Worker U_CAPI void U_EXPORT2
402*0e209d39SAndroid Build Coastguard Worker uprv_getCharNameCharacters(const USetAdder *sa);
403*0e209d39SAndroid Build Coastguard Worker
404*0e209d39SAndroid Build Coastguard Worker /**
405*0e209d39SAndroid Build Coastguard Worker * Constants for which data and implementation files provide which properties.
406*0e209d39SAndroid Build Coastguard Worker * Used by UnicodeSet for service-specific property enumeration.
407*0e209d39SAndroid Build Coastguard Worker * @internal
408*0e209d39SAndroid Build Coastguard Worker */
409*0e209d39SAndroid Build Coastguard Worker enum UPropertySource {
410*0e209d39SAndroid Build Coastguard Worker /** No source, not a supported property. */
411*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_NONE,
412*0e209d39SAndroid Build Coastguard Worker /** From uchar.c/uprops.icu main trie */
413*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_CHAR,
414*0e209d39SAndroid Build Coastguard Worker /** From uchar.c/uprops.icu properties vectors trie */
415*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_PROPSVEC,
416*0e209d39SAndroid Build Coastguard Worker /** From unames.c/unames.icu */
417*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_NAMES,
418*0e209d39SAndroid Build Coastguard Worker /** From ucase.c/ucase.icu */
419*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_CASE,
420*0e209d39SAndroid Build Coastguard Worker /** From ubidi_props.c/ubidi.icu */
421*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_BIDI,
422*0e209d39SAndroid Build Coastguard Worker /** From uchar.c/uprops.icu main trie as well as properties vectors trie */
423*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_CHAR_AND_PROPSVEC,
424*0e209d39SAndroid Build Coastguard Worker /** From ucase.c/ucase.icu as well as unorm.cpp/unorm.icu */
425*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_CASE_AND_NORM,
426*0e209d39SAndroid Build Coastguard Worker /** From normalizer2impl.cpp/nfc.nrm */
427*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_NFC,
428*0e209d39SAndroid Build Coastguard Worker /** From normalizer2impl.cpp/nfkc.nrm */
429*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_NFKC,
430*0e209d39SAndroid Build Coastguard Worker /** From normalizer2impl.cpp/nfkc_cf.nrm */
431*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_NFKC_CF,
432*0e209d39SAndroid Build Coastguard Worker /** From normalizer2impl.cpp/nfc.nrm canonical iterator data */
433*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_NFC_CANON_ITER,
434*0e209d39SAndroid Build Coastguard Worker // Text layout properties.
435*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_INPC,
436*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_INSC,
437*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_VO,
438*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_EMOJI,
439*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_IDSU,
440*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_ID_COMPAT_MATH,
441*0e209d39SAndroid Build Coastguard Worker /** One more than the highest UPropertySource (UPROPS_SRC_) constant. */
442*0e209d39SAndroid Build Coastguard Worker UPROPS_SRC_COUNT
443*0e209d39SAndroid Build Coastguard Worker };
444*0e209d39SAndroid Build Coastguard Worker typedef enum UPropertySource UPropertySource;
445*0e209d39SAndroid Build Coastguard Worker
446*0e209d39SAndroid Build Coastguard Worker /**
447*0e209d39SAndroid Build Coastguard Worker * @see UPropertySource
448*0e209d39SAndroid Build Coastguard Worker * @internal
449*0e209d39SAndroid Build Coastguard Worker */
450*0e209d39SAndroid Build Coastguard Worker U_CFUNC UPropertySource U_EXPORT2
451*0e209d39SAndroid Build Coastguard Worker uprops_getSource(UProperty which);
452*0e209d39SAndroid Build Coastguard Worker
453*0e209d39SAndroid Build Coastguard Worker /**
454*0e209d39SAndroid Build Coastguard Worker * Enumerate uprops.icu's main data trie and add the
455*0e209d39SAndroid Build Coastguard Worker * start of each range of same properties to the set.
456*0e209d39SAndroid Build Coastguard Worker * @internal
457*0e209d39SAndroid Build Coastguard Worker */
458*0e209d39SAndroid Build Coastguard Worker U_CFUNC void U_EXPORT2
459*0e209d39SAndroid Build Coastguard Worker uchar_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode);
460*0e209d39SAndroid Build Coastguard Worker
461*0e209d39SAndroid Build Coastguard Worker /**
462*0e209d39SAndroid Build Coastguard Worker * Enumerate uprops.icu's properties vectors trie and add the
463*0e209d39SAndroid Build Coastguard Worker * start of each range of same properties to the set.
464*0e209d39SAndroid Build Coastguard Worker * @internal
465*0e209d39SAndroid Build Coastguard Worker */
466*0e209d39SAndroid Build Coastguard Worker U_CFUNC void U_EXPORT2
467*0e209d39SAndroid Build Coastguard Worker upropsvec_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode);
468*0e209d39SAndroid Build Coastguard Worker
469*0e209d39SAndroid Build Coastguard Worker U_CFUNC void U_EXPORT2
470*0e209d39SAndroid Build Coastguard Worker uprops_addPropertyStarts(UPropertySource src, const USetAdder *sa, UErrorCode *pErrorCode);
471*0e209d39SAndroid Build Coastguard Worker
472*0e209d39SAndroid Build Coastguard Worker /**
473*0e209d39SAndroid Build Coastguard Worker * Return a set of characters for property enumeration.
474*0e209d39SAndroid Build Coastguard Worker * For each two consecutive characters (start, limit) in the set,
475*0e209d39SAndroid Build Coastguard Worker * all of the properties for start..limit-1 are all the same.
476*0e209d39SAndroid Build Coastguard Worker *
477*0e209d39SAndroid Build Coastguard Worker * @param sa USetAdder to receive result. Existing contents are lost.
478*0e209d39SAndroid Build Coastguard Worker * @internal
479*0e209d39SAndroid Build Coastguard Worker */
480*0e209d39SAndroid Build Coastguard Worker /*U_CFUNC void U_EXPORT2
481*0e209d39SAndroid Build Coastguard Worker uprv_getInclusions(const USetAdder *sa, UErrorCode *pErrorCode);
482*0e209d39SAndroid Build Coastguard Worker */
483*0e209d39SAndroid Build Coastguard Worker
484*0e209d39SAndroid Build Coastguard Worker /**
485*0e209d39SAndroid Build Coastguard Worker * Swap the ICU Unicode character names file. See uchar.c.
486*0e209d39SAndroid Build Coastguard Worker * @internal
487*0e209d39SAndroid Build Coastguard Worker */
488*0e209d39SAndroid Build Coastguard Worker U_CAPI int32_t U_EXPORT2
489*0e209d39SAndroid Build Coastguard Worker uchar_swapNames(const UDataSwapper *ds,
490*0e209d39SAndroid Build Coastguard Worker const void *inData, int32_t length, void *outData,
491*0e209d39SAndroid Build Coastguard Worker UErrorCode *pErrorCode);
492*0e209d39SAndroid Build Coastguard Worker
493*0e209d39SAndroid Build Coastguard Worker #ifdef __cplusplus
494*0e209d39SAndroid Build Coastguard Worker
495*0e209d39SAndroid Build Coastguard Worker U_NAMESPACE_BEGIN
496*0e209d39SAndroid Build Coastguard Worker
497*0e209d39SAndroid Build Coastguard Worker class UnicodeSet;
498*0e209d39SAndroid Build Coastguard Worker
499*0e209d39SAndroid Build Coastguard Worker class CharacterProperties {
500*0e209d39SAndroid Build Coastguard Worker public:
501*0e209d39SAndroid Build Coastguard Worker CharacterProperties() = delete;
502*0e209d39SAndroid Build Coastguard Worker static const UnicodeSet *getInclusionsForProperty(UProperty prop, UErrorCode &errorCode);
503*0e209d39SAndroid Build Coastguard Worker static const UnicodeSet *getBinaryPropertySet(UProperty property, UErrorCode &errorCode);
504*0e209d39SAndroid Build Coastguard Worker };
505*0e209d39SAndroid Build Coastguard Worker
506*0e209d39SAndroid Build Coastguard Worker // implemented in uniset_props.cpp
507*0e209d39SAndroid Build Coastguard Worker U_CFUNC UnicodeSet *
508*0e209d39SAndroid Build Coastguard Worker uniset_getUnicode32Instance(UErrorCode &errorCode);
509*0e209d39SAndroid Build Coastguard Worker
510*0e209d39SAndroid Build Coastguard Worker U_NAMESPACE_END
511*0e209d39SAndroid Build Coastguard Worker
512*0e209d39SAndroid Build Coastguard Worker #endif
513*0e209d39SAndroid Build Coastguard Worker
514*0e209d39SAndroid Build Coastguard Worker #endif
515