organicmaps/icu
13
Fork 0
mirror of https://github.com/unicode-org/icu.git synced 2025-04-06 14:05:32 +00:00
Code Issues 1 Projects Releases Wiki Activity Actions 8

ICU-12985 implement the binary Emoji_Component property for emoji 5

Browse source 
X-SVN-Rev: 40082
This commit is contained in:
Markus Scherer 2017-04-26 23:58:36 +00:00
parent 9e9bfda19c
commit 6ce7f348a3
17 changed files with 942 additions and 910 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1680
icu4c/source/common/propname_data.h
View file

File diff suppressed because it is too large Load diff

8
icu4c/source/common/uchar_props_data.h
View file

@ -3319,8 +3319,8 @@ static const uint32_t propsVectors[6372]={
0xe00011,0x11065700,0x4000008,0xe02c11,0x11065700,0x4000010,0x871411,0x11065700,0x4000010,0x1201411,0x11065700,0x4000010,0x1271011,0x11065700,0x4000020,0xe00011,
0x11065700,0x4000400,0xe00011,0x11065700,0x4000420,0xe00011,0x11065700,0x6800000,0xe01c11,0x11065700,0x6800040,0xe00011,0x11065700,0xc000010,0x80ac11,0x11065700,
0xc000010,0xb48011,0x11065719,0x7c00100,0xe20411,0x11065719,0x7c00100,0xe50411,0x11065719,0x7c00140,0xe20411,0x11065719,0x7c00140,0xe50411,0x11080100,0x6800000,
0x201c00,0x11080100,0x68000c0,0x11329800,0x11080100,0x24000000,0x200000,0x11080100,0x24000000,0x810000,0x11080100,0x24000000,0x1410000,0x11080100,0x24000000,0x1500000,
0x11080100,0x24000000,0x1600000,0x11080100,0x24000000,0x1b00000,0x11080100,0x24000000,0x2410000,0x11080100,0x24000000,0x10200000,0x11080100,0x24000006,0xd70000,0x11080100,
0x201c00,0x11080100,0x68000c0,0x19329800,0x11080100,0x24000000,0x200000,0x11080100,0x24000000,0x810000,0x11080100,0x24000000,0x1410000,0x11080100,0x24000000,0x1500000,
0x11080100,0x24000000,0x1600000,0x11080100,0x24000000,0x1b00000,0x11080100,0x24000000,0x2410000,0x11080100,0x24000000,0x18200000,0x11080100,0x24000006,0xd70000,0x11080100,
0x24000008,0x1713c00,0x11080100,0x24000008,0x1714000,0x11080100,0x24000010,0x1001400,0x11080100,0x24000010,0x1071000,0x11080100,0x24000010,0x1071400,0x11080100,0x24000020,
0x200000,0x11080100,0x24000020,0x400000,0x11080100,0x24000020,0x1600000,0x11080100,0x24000400,0x200000,0x11080100,0x24000420,0x200000,0x11080100,0x2c000010,0xb48000,
0x11080100,0x2c000010,0x100ac00,0x11080100,0x44000001,0x1a40000,0x11080119,0x7c00100,0x220400,0x11080119,0x7c00100,0x250400,0x11080119,0x7c001c0,0x220400,0x11080119,
@ -3506,7 +3506,7 @@ static const uint32_t propsVectors[6372]={
0x60001b27,0x7c00100,0x230400,0x60001f0b,0x2802000,0x962460,0x60002919,0x7c00100,0x22040e,0x60002a00,0x4000000,0x1600000,0x60003000,0x24000000,0x10200000,0x60003000,
0x24000000,0x10e00000,0x60003700,0x24000000,0x200000,0x60003800,0x24000000,0x1710000,0x60005102,0x4000000,0x200000,0x60006108,0x7c00100,0x220400,0x60006108,0x7c00100,
0x250400,0x60006600,0x24000020,0x200000,0x60008301,0x2802000,0x962460,0x6000903c,0x2806000,0x962460,0x6000903c,0x4000000,0x400000,0x60009519,0x7c00100,0x220400,
0x60009519,0x7c00100,0x250400,0x6000a008,0x7c00100,0x220400,0x6000a008,0x7c00100,0x250400,0x6000c300,0x4000000,0x32703580,0x6000c654,0x2802000,0x962460,0x6000c654,
0x60009519,0x7c00100,0x250400,0x6000a008,0x7c00100,0x220400,0x6000a008,0x7c00100,0x250400,0x6000c300,0x4000000,0x3a703580,0x6000c654,0x2802000,0x962460,0x6000c654,
0x4000010,0x200000,0x6000c654,0x7c00100,0x230400,0x6000c73f,0x2802000,0x962460,0x6000c73f,0x2802100,0x962460,0x6000c73f,0x4000000,0x200000,0x6000c73f,0x6800100,
0x962540,0x6000c73f,0x6804000,0x962540,0x6000c73f,0x7c00100,0x230400,0x6000c80b,0x7c00100,0x230400,0x6000c941,0x2802100,0x962460,0x6000c941,0x2806000,0x962460,
0x6000c941,0x4000000,0x200000,0x6000c941,0x4000010,0x200000,0x6000c941,0x6800000,0x1329800,0x6000c941,0x6800100,0x962540,0x6000c941,0x7c00100,0x230400,0x6000c941,
@ -3575,7 +3575,7 @@ static const uint32_t propsVectors[6372]={
0x8000fda1,0xc000010,0x448000,0x8000fe9c,0x7c00100,0x230400,0x8000fe9c,0x7c00100,0x830400,0x8000fe9c,0x7c00100,0x1430400,0x8000ff06,0x7c00100,0x220400,0x80010165,
0x7c00100,0x230400,0x800102a2,0x4000000,0x200000,0x800102a2,0x7c00100,0x230400,0x800103a4,0x7c00100,0x230400,0x800103a4,0xc000010,0x448000,0x8001044c,0x4000000,
0x200000,0x8001044c,0x7c00100,0x220400,0x8001044c,0x7c00100,0x250400,0x80010670,0x2802000,0x962460,0x80010670,0x4000000,0x200000,0x80010670,0x4000010,0x400000,
0x80010670,0xc000010,0x448000,0x800a4711,0x7c40300,0xe30000,0x800acd00,0x4000000,0x30e00000,0x800acd00,0x4000000,0x72904de0,0x800ace00,0x4000000,0x30e00000,0x800acf00,
0x80010670,0xc000010,0x448000,0x800a4711,0x7c40300,0xe30000,0x800acd00,0x4000000,0x30e00000,0x800acd00,0x4000000,0x7a904de0,0x800ace00,0x4000000,0x30e00000,0x800acf00,
0x4000000,0x30e00000,0x800b0011,0x7c40300,0xe30000,0x800b0500,0x4000000,0x30e00000,0x800b0500,0x4000000,0xb28045a0,0x90001615,0x7c00100,0x230400,0x9000171a,0x4000000,
0x200000,0x9000171a,0x7c00100,0x230400,0x90003000,0x24000000,0x200000,0x90007f0e,0x4000000,0x200000,0x90008301,0x2802000,0x962460,0x90008e00,0x24000000,0x400000,
0x90009519,0x7c00100,0x250400,0x9000a16f,0x2802100,0x962460,0x9000d200,0,0x218960,0x9000d202,0x2802000,0x962460,0x9000d202,0x2802100,0x962460,0x9000d202,

9
icu4c/source/common/unicode/uchar.h
View file

@ -427,12 +427,19 @@ typedef enum UProperty {
* @stable ICU 57
*/
UCHAR_EMOJI_MODIFIER_BASE=60,
/**
* Binary property Emoji_Component.
* See http://www.unicode.org/reports/tr51/#Emoji_Properties
*
* @stable ICU 60
*/
UCHAR_EMOJI_COMPONENT=61,
#ifndef U_HIDE_DEPRECATED_API
/**
* One more than the last constant for binary Unicode properties.
* @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
*/
UCHAR_BINARY_LIMIT=61,
UCHAR_BINARY_LIMIT,
#endif // U_HIDE_DEPRECATED_API
/** Enumerated property Bidi_Class.

1
icu4c/source/common/uprops.cpp
View file

@ -276,6 +276,7 @@ static const BinaryProperty binProps[UCHAR_BINARY_LIMIT]={
{ 2, U_MASK(UPROPS_2_EMOJI_PRESENTATION), defaultContains },
{ 2, U_MASK(UPROPS_2_EMOJI_MODIFIER), defaultContains },
{ 2, U_MASK(UPROPS_2_EMOJI_MODIFIER_BASE), defaultContains },
{ 2, U_MASK(UPROPS_2_EMOJI_COMPONENT), defaultContains },
};
U_CAPI UBool U_EXPORT2

7
icu4c/source/common/uprops.h
View file

@ -196,8 +196,8 @@ enum {
/*
* Properties in vector word 2
* Bits
* 31..28 http://www.unicode.org/reports/tr51/#Emoji_Properties
* 27..26 reserved
* 31..27 http://www.unicode.org/reports/tr51/#Emoji_Properties
* 26 reserved
* 25..20 Line Break
* 19..15 Sentence Break
* 14..10 Word Break
@ -205,7 +205,8 @@ enum {
* 4.. 0 Decomposition Type
*/
enum {
UPROPS_2_EMOJI=28,
UPROPS_2_EMOJI_COMPONENT=27,
UPROPS_2_EMOJI,
UPROPS_2_EMOJI_PRESENTATION,
UPROPS_2_EMOJI_MODIFIER,
UPROPS_2_EMOJI_MODIFIER_BASE

BIN
icu4c/source/data/in/pnames.icu
View file

Binary file not shown.

BIN
icu4c/source/data/in/uprops.icu
View file

Binary file not shown.

4
icu4c/source/data/unidata/changes.txt
View file

@ -64,12 +64,14 @@ export LD_LIBRARY_PATH=$ICU_ROOT/dbg/icu4c/lib
- ticket:12985: Unicode 10
- ticket:13061: undo hacks from emoji 5.0 update
- ticket:13062: add Emoji_Component property
- ^/branches/markus/uni10
*** CLDR Trac
- cldrbug 10055: Unicode 10
- cldrbug 9882: Unicode 10 script metadata
- cldrbug 10219: numbering systems for Unicode 10
*** Unicode version numbers
- makedata.mak
@ -205,7 +207,7 @@ $ICU_SRC/tools/unicode/c/icudefs.txt:
# Location (--prefix) of where ICU was installed.
set(ICU_INST_DIR /usr/local/google/home/mscherer/svn.icu/trunk/inst/icu4c)
# Location of the ICU4C source tree.
set(ICU4C_SRC_DIR /usr/local/google/home/mscherer/svn.icu/trunk/src/icu4c)
set(ICU4C_SRC_DIR /usr/local/google/home/mscherer/svn.icu/uni10/src/icu4c)
$ICU_ROOT/dbg/tools/unicode/c$
cmake ../../../../src/tools/unicode/c

87
icu4c/source/data/unidata/ppucd.txt
View file

@ -35,6 +35,7 @@ property;String;dm;Decomposition_Mapping
property;Enumerated;dt;Decomposition_Type
property;Enumerated;ea;East_Asian_Width
property;Binary;Emoji;Emoji
property;Binary;Emoji_Component;Emoji_Component
property;Binary;Emoji_Modifier;Emoji_Modifier
property;Binary;Emoji_Modifier_Base;Emoji_Modifier_Base
property;Binary;Emoji_Presentation;Emoji_Presentation
@ -1207,30 +1208,30 @@ cp;001F;bc=S;ea=N;GCB=CN;-Gr_Base;lb=CM;Name_Alias=control=INFORMATION SEPARATOR
cp;0020;bc=WS;gc=Zs;lb=SP;na=SPACE;Name_Alias=abbreviation=SP;Pat_WS;SB=SP;WSpace
cp;0021;bc=ON;gc=Po;lb=EX;na=EXCLAMATION MARK;Pat_Syn;SB=ST;STerm;Term
cp;0022;bc=ON;gc=Po;lb=QU;na=QUOTATION MARK;Pat_Syn;QMark;SB=CL;WB=DQ
cp;0023;bc=ET;Emoji;gc=Po;na=NUMBER SIGN;Pat_Syn
cp;0023;bc=ET;Emoji;Emoji_Component;gc=Po;na=NUMBER SIGN;Pat_Syn
cp;0024;bc=ET;gc=Sc;lb=PR;na=DOLLAR SIGN;Pat_Syn
cp;0025;bc=ET;gc=Po;lb=PO;na=PERCENT SIGN;Pat_Syn
cp;0026;bc=ON;gc=Po;na=AMPERSAND;Pat_Syn
cp;0027;bc=ON;CI;gc=Po;lb=QU;na=APOSTROPHE;Pat_Syn;QMark;SB=CL;WB=SQ
cp;0028;bc=ON;Bidi_M;bmg=0029;bpb=0029;bpt=o;gc=Ps;lb=OP;na=LEFT PARENTHESIS;Pat_Syn;SB=CL
cp;0029;bc=ON;Bidi_M;bmg=0028;bpb=0028;bpt=c;gc=Pe;lb=CP;na=RIGHT PARENTHESIS;Pat_Syn;SB=CL
cp;002A;bc=ON;Emoji;gc=Po;na=ASTERISK;Pat_Syn
cp;002A;bc=ON;Emoji;Emoji_Component;gc=Po;na=ASTERISK;Pat_Syn
cp;002B;bc=ES;gc=Sm;lb=PR;Math;na=PLUS SIGN;Pat_Syn
cp;002C;bc=CS;gc=Po;lb=IS;na=COMMA;Pat_Syn;SB=SC;Term;WB=MN
cp;002D;bc=ES;Dash;gc=Pd;Hyphen;InSC=Consonant_Placeholder;lb=HY;na=HYPHEN-MINUS;Pat_Syn;SB=SC
cp;002E;bc=CS;CI;gc=Po;lb=IS;na=FULL STOP;Pat_Syn;SB=AT;STerm;Term;WB=MB
cp;002F;bc=CS;gc=Po;lb=SY;na=SOLIDUS;Pat_Syn
# ASCII digits
cp;0030;AHex;bc=EN;Emoji;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT ZERO;nt=De;nv=0;SB=NU;WB=NU;XIDC
cp;0031;AHex;bc=EN;Emoji;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT ONE;nt=De;nv=1;SB=NU;WB=NU;XIDC
cp;0032;AHex;bc=EN;Emoji;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT TWO;nt=De;nv=2;SB=NU;WB=NU;XIDC
cp;0033;AHex;bc=EN;Emoji;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT THREE;nt=De;nv=3;SB=NU;WB=NU;XIDC
cp;0034;AHex;bc=EN;Emoji;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT FOUR;nt=De;nv=4;SB=NU;WB=NU;XIDC
cp;0035;AHex;bc=EN;Emoji;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT FIVE;nt=De;nv=5;SB=NU;WB=NU;XIDC
cp;0036;AHex;bc=EN;Emoji;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT SIX;nt=De;nv=6;SB=NU;WB=NU;XIDC
cp;0037;AHex;bc=EN;Emoji;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT SEVEN;nt=De;nv=7;SB=NU;WB=NU;XIDC
cp;0038;AHex;bc=EN;Emoji;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT EIGHT;nt=De;nv=8;SB=NU;WB=NU;XIDC
cp;0039;AHex;bc=EN;Emoji;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT NINE;nt=De;nv=9;SB=NU;WB=NU;XIDC
cp;0030;AHex;bc=EN;Emoji;Emoji_Component;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT ZERO;nt=De;nv=0;SB=NU;WB=NU;XIDC
cp;0031;AHex;bc=EN;Emoji;Emoji_Component;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT ONE;nt=De;nv=1;SB=NU;WB=NU;XIDC
cp;0032;AHex;bc=EN;Emoji;Emoji_Component;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT TWO;nt=De;nv=2;SB=NU;WB=NU;XIDC
cp;0033;AHex;bc=EN;Emoji;Emoji_Component;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT THREE;nt=De;nv=3;SB=NU;WB=NU;XIDC
cp;0034;AHex;bc=EN;Emoji;Emoji_Component;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT FOUR;nt=De;nv=4;SB=NU;WB=NU;XIDC
cp;0035;AHex;bc=EN;Emoji;Emoji_Component;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT FIVE;nt=De;nv=5;SB=NU;WB=NU;XIDC
cp;0036;AHex;bc=EN;Emoji;Emoji_Component;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT SIX;nt=De;nv=6;SB=NU;WB=NU;XIDC
cp;0037;AHex;bc=EN;Emoji;Emoji_Component;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT SEVEN;nt=De;nv=7;SB=NU;WB=NU;XIDC
cp;0038;AHex;bc=EN;Emoji;Emoji_Component;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT EIGHT;nt=De;nv=8;SB=NU;WB=NU;XIDC
cp;0039;AHex;bc=EN;Emoji;Emoji_Component;gc=Nd;Hex;IDC;InSC=Number;lb=NU;na=DIGIT NINE;nt=De;nv=9;SB=NU;WB=NU;XIDC
# ASCII punctuation and symbols
cp;003A;bc=CS;CI;gc=Po;lb=IS;na=COLON;Pat_Syn;SB=SC;Term;WB=ML
cp;003B;bc=ON;gc=Po;lb=IS;na=SEMICOLON;Pat_Syn;Term;WB=MN
@ -34106,32 +34107,32 @@ cp;1F1AB;age=9.0;na=SQUARED UHD
cp;1F1AC;age=9.0;na=SQUARED VOD
cp;1F1AD..1F1E5;age=NA;ea=N;gc=Cn;-Gr_Base;lb=ID;sc=Zzzz
# Regional indicator symbols
cp;1F1E6;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER A;RI;WB=RI
cp;1F1E7;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER B;RI;WB=RI
cp;1F1E8;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER C;RI;WB=RI
cp;1F1E9;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER D;RI;WB=RI
cp;1F1EA;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER E;RI;WB=RI
cp;1F1EB;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER F;RI;WB=RI
cp;1F1EC;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER G;RI;WB=RI
cp;1F1ED;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER H;RI;WB=RI
cp;1F1EE;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER I;RI;WB=RI
cp;1F1EF;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER J;RI;WB=RI
cp;1F1F0;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER K;RI;WB=RI
cp;1F1F1;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER L;RI;WB=RI
cp;1F1F2;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER M;RI;WB=RI
cp;1F1F3;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER N;RI;WB=RI
cp;1F1F4;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER O;RI;WB=RI
cp;1F1F5;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER P;RI;WB=RI
cp;1F1F6;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER Q;RI;WB=RI
cp;1F1F7;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER R;RI;WB=RI
cp;1F1F8;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER S;RI;WB=RI
cp;1F1F9;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER T;RI;WB=RI
cp;1F1FA;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER U;RI;WB=RI
cp;1F1FB;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER V;RI;WB=RI
cp;1F1FC;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER W;RI;WB=RI
cp;1F1FD;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER X;RI;WB=RI
cp;1F1FE;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER Y;RI;WB=RI
cp;1F1FF;ea=N;Emoji;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER Z;RI;WB=RI
cp;1F1E6;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER A;RI;WB=RI
cp;1F1E7;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER B;RI;WB=RI
cp;1F1E8;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER C;RI;WB=RI
cp;1F1E9;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER D;RI;WB=RI
cp;1F1EA;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER E;RI;WB=RI
cp;1F1EB;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER F;RI;WB=RI
cp;1F1EC;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER G;RI;WB=RI
cp;1F1ED;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER H;RI;WB=RI
cp;1F1EE;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER I;RI;WB=RI
cp;1F1EF;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER J;RI;WB=RI
cp;1F1F0;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER K;RI;WB=RI
cp;1F1F1;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER L;RI;WB=RI
cp;1F1F2;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER M;RI;WB=RI
cp;1F1F3;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER N;RI;WB=RI
cp;1F1F4;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER O;RI;WB=RI
cp;1F1F5;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER P;RI;WB=RI
cp;1F1F6;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER Q;RI;WB=RI
cp;1F1F7;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER R;RI;WB=RI
cp;1F1F8;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER S;RI;WB=RI
cp;1F1F9;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER T;RI;WB=RI
cp;1F1FA;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER U;RI;WB=RI
cp;1F1FB;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER V;RI;WB=RI
cp;1F1FC;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER W;RI;WB=RI
cp;1F1FD;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER X;RI;WB=RI
cp;1F1FE;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER Y;RI;WB=RI
cp;1F1FF;ea=N;Emoji;Emoji_Component;Emoji_Presentation;GCB=RI;lb=RI;na=REGIONAL INDICATOR SYMBOL LETTER Z;RI;WB=RI
block;1F200..1F2FF;age=5.2;blk=Enclosed_Ideographic_Sup;CWKCF;dt=Sqr;ea=W;gc=So;Gr_Base;lb=ID;NFKC_QC=N;NFKD_QC=N;sc=Zyyy
# 1F200..1F2FF Enclosed Ideographic Supplement
@ -34489,11 +34490,11 @@ cp;1F3F9;age=8.0;na=BOW AND ARROW
# Miscellaneous symbol
cp;1F3FA;age=8.0;na=AMPHORA
# Emoji modifiers
cp;1F3FB;age=8.0;CI;Emoji_Modifier;gc=Sk;GCB=EM;lb=EM;na=EMOJI MODIFIER FITZPATRICK TYPE-1-2;WB=EM
cp;1F3FC;age=8.0;CI;Emoji_Modifier;gc=Sk;GCB=EM;lb=EM;na=EMOJI MODIFIER FITZPATRICK TYPE-3;WB=EM
cp;1F3FD;age=8.0;CI;Emoji_Modifier;gc=Sk;GCB=EM;lb=EM;na=EMOJI MODIFIER FITZPATRICK TYPE-4;WB=EM
cp;1F3FE;age=8.0;CI;Emoji_Modifier;gc=Sk;GCB=EM;lb=EM;na=EMOJI MODIFIER FITZPATRICK TYPE-5;WB=EM
cp;1F3FF;age=8.0;CI;Emoji_Modifier;gc=Sk;GCB=EM;lb=EM;na=EMOJI MODIFIER FITZPATRICK TYPE-6;WB=EM
cp;1F3FB;age=8.0;CI;Emoji_Component;Emoji_Modifier;gc=Sk;GCB=EM;lb=EM;na=EMOJI MODIFIER FITZPATRICK TYPE-1-2;WB=EM
cp;1F3FC;age=8.0;CI;Emoji_Component;Emoji_Modifier;gc=Sk;GCB=EM;lb=EM;na=EMOJI MODIFIER FITZPATRICK TYPE-3;WB=EM
cp;1F3FD;age=8.0;CI;Emoji_Component;Emoji_Modifier;gc=Sk;GCB=EM;lb=EM;na=EMOJI MODIFIER FITZPATRICK TYPE-4;WB=EM
cp;1F3FE;age=8.0;CI;Emoji_Component;Emoji_Modifier;gc=Sk;GCB=EM;lb=EM;na=EMOJI MODIFIER FITZPATRICK TYPE-5;WB=EM
cp;1F3FF;age=8.0;CI;Emoji_Component;Emoji_Modifier;gc=Sk;GCB=EM;lb=EM;na=EMOJI MODIFIER FITZPATRICK TYPE-6;WB=EM
# Animal symbols
cp;1F400;na=RAT
cp;1F401;na=MOUSE

2
icu4c/source/test/intltest/ucdtest.cpp
View file

@ -527,4 +527,6 @@ void UnicodeTest::TestEmojiProperties() {
u_hasBinaryProperty(0x1F3FF, UCHAR_EMOJI_MODIFIER));
assertTrue("happy person is Emoji_Modifier_Base",
u_hasBinaryProperty(0x1F64B, UCHAR_EMOJI_MODIFIER_BASE));
assertTrue("asterisk is Emoji_Component",
u_hasBinaryProperty(0x2A, UCHAR_EMOJI_COMPONENT));
}

6
icu4j/main/classes/core/src/com/ibm/icu/impl/UCharacterProperty.java
View file

@ -404,6 +404,7 @@ public final class UCharacterProperty
new BinaryProperty(2, 1<<PROPS_2_EMOJI_PRESENTATION),
new BinaryProperty(2, 1<<PROPS_2_EMOJI_MODIFIER),
new BinaryProperty(2, 1<<PROPS_2_EMOJI_MODIFIER_BASE),
new BinaryProperty(2, 1<<PROPS_2_EMOJI_COMPONENT),
};
public boolean hasBinaryProperty(int c, int which) {
@ -1155,14 +1156,15 @@ public final class UCharacterProperty
/*
* Properties in vector word 2
* Bits
* 31..28 http://www.unicode.org/reports/tr51/#Emoji_Properties
* 27..26 reserved
* 31..27 http://www.unicode.org/reports/tr51/#Emoji_Properties
* 26 reserved
* 25..20 Line Break
* 19..15 Sentence Break
* 14..10 Word Break
* 9.. 5 Grapheme Cluster Break
* 4.. 0 Decomposition Type
*/
private static final int PROPS_2_EMOJI_COMPONENT = 27;
private static final int PROPS_2_EMOJI = 28;
private static final int PROPS_2_EMOJI_PRESENTATION = 29;
private static final int PROPS_2_EMOJI_MODIFIER = 30;

21
icu4j/main/classes/core/src/com/ibm/icu/lang/UProperty.java
View file

@ -518,41 +518,44 @@ public interface UProperty
* Binary property Emoji.
* See http://www.unicode.org/reports/tr51/#Emoji_Properties
*
* @draft ICU 57
* @provisional This API might change or be removed in a future release.
* @stable ICU 57
*/
public static final int EMOJI=57;
/**
* Binary property Emoji_Presentation.
* See http://www.unicode.org/reports/tr51/#Emoji_Properties
*
* @draft ICU 57
* @provisional This API might change or be removed in a future release.
* @stable ICU 57
*/
public static final int EMOJI_PRESENTATION=58;
/**
* Binary property Emoji_Modifier.
* See http://www.unicode.org/reports/tr51/#Emoji_Properties
*
* @draft ICU 57
* @provisional This API might change or be removed in a future release.
* @stable ICU 57
*/
public static final int EMOJI_MODIFIER=59;
/**
* Binary property Emoji_Modifier_Base.
* See http://www.unicode.org/reports/tr51/#Emoji_Properties
*
* @draft ICU 57
* @provisional This API might change or be removed in a future release.
* @stable ICU 57
*/
public static final int EMOJI_MODIFIER_BASE=60;
/**
* Binary property Emoji_Component.
* See http://www.unicode.org/reports/tr51/#Emoji_Properties
*
* @stable ICU 60
*/
public static final int EMOJI_COMPONENT=61;
/**
* One more than the last constant for binary Unicode properties.
* @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
*/
@Deprecated
public static final int BINARY_LIMIT = 61;
public static final int BINARY_LIMIT = 62;
/**
* Enumerated property Bidi_Class.

4
icu4j/main/shared/data/icudata.jar
View file

@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:9938ffc38cb8a451fd020b244e96678109cdf153ab0344b535f9bbe16d6fdb0b
size 12152261
oid sha256:75e39c89d7b28489ff570eeeeb86d4b1398c46456a61f63372534f956c8b7f98
size 12152263

2
icu4j/main/tests/core/src/com/ibm/icu/dev/test/lang/UCharacterTest.java
View file

@ -2499,6 +2499,8 @@ public final class UCharacterTest extends TestFmwk
UCharacter.hasBinaryProperty(0x1F3FF, UProperty.EMOJI_MODIFIER));
assertTrue("happy person is Emoji_Modifier_Base",
UCharacter.hasBinaryProperty(0x1F64B, UProperty.EMOJI_MODIFIER_BASE));
assertTrue("asterisk is Emoji_Component",
UCharacter.hasBinaryProperty(0x2A, UProperty.EMOJI_COMPONENT));
}
@Test

15
tools/unicode/c/genprops/corepropsbuilder.cpp
View file

@ -1,3 +1,5 @@
// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*
@ -45,7 +47,7 @@ the udata API for loading ICU data. Especially, a UDataInfo structure
precedes the actual data. It contains platform properties values and the
file format version.
The following is a description of format version 7.3 .
The following is a description of format version 7.4 .
Data contents:
@ -258,6 +260,12 @@ http://www.unicode.org/reports/tr51/#Emoji_Properties
ICU 58 adds fraction-20 numeric values for new Unicode 9 Malayalam fraction characters.
--- Changes in format version 7.4 ---
ICU 60 adds the Emoji_Component property to vector word 2, for emoji 5.
http://bugs.icu-project.org/trac/ticket/13062
http://www.unicode.org/reports/tr51/#Emoji_Properties
----------------------------------------------------------------------------- */
U_NAMESPACE_USE
@ -273,8 +281,8 @@ static UDataInfo dataInfo={
0,
{ 0x55, 0x50, 0x72, 0x6f }, /* dataFormat="UPro" */
{ 7, 3, 0, 0 }, /* formatVersion */
{ 8, 0, 0, 0 } /* dataVersion */
{ 7, 4, 0, 0 }, /* formatVersion */
{ 10, 0, 0, 0 } /* dataVersion */
};
class CorePropsBuilder : public PropsBuilder {
@ -568,6 +576,7 @@ propToBinaries[]={
{ UCHAR_EMOJI_PRESENTATION, 2, UPROPS_2_EMOJI_PRESENTATION },
{ UCHAR_EMOJI_MODIFIER, 2, UPROPS_2_EMOJI_MODIFIER },
{ UCHAR_EMOJI_MODIFIER_BASE, 2, UPROPS_2_EMOJI_MODIFIER_BASE },
{ UCHAR_EMOJI_COMPONENT, 2, UPROPS_2_EMOJI_COMPONENT },
};
struct PropToEnum {

3
tools/unicode/c/genprops/pnames_data.h
View file

@ -1055,7 +1055,7 @@ static const Value VALUES_gcm[38] = {
Value((int32_t)U_GC_ZS_MASK, "Zs Space_Separator"),
};
static const Property PROPERTIES[100] = {
static const Property PROPERTIES[101] = {
Property(UCHAR_ALPHABETIC, "Alpha Alphabetic"),
Property(UCHAR_ASCII_HEX_DIGIT, "AHex ASCII_Hex_Digit"),
Property(UCHAR_BIDI_CONTROL, "Bidi_C Bidi_Control"),
@ -1117,6 +1117,7 @@ static const Property PROPERTIES[100] = {
Property(UCHAR_EMOJI_PRESENTATION, "Emoji_Presentation Emoji_Presentation"),
Property(UCHAR_EMOJI_MODIFIER, "Emoji_Modifier Emoji_Modifier"),
Property(UCHAR_EMOJI_MODIFIER_BASE, "Emoji_Modifier_Base Emoji_Modifier_Base"),
Property(UCHAR_EMOJI_COMPONENT, "Emoji_Component Emoji_Component"),
Property(UCHAR_BIDI_CLASS, "bc Bidi_Class", VALUES_bc, 23),
Property(UCHAR_BLOCK, "blk Block", VALUES_blk, 281),
Property(UCHAR_CANONICAL_COMBINING_CLASS, "ccc Canonical_Combining_Class", VALUES_ccc, 57),

3
tools/unicode/py/preparseucd.py
View file

@ -76,8 +76,6 @@ _ignored_properties = set((
"XO_NFD",
"XO_NFKC",
"XO_NFKD",
# TODO: ICU 60 should support this after UTS #51 Emoji 5.0 is released.
"Emoji_Component",
# ICU does not (yet) support Vertical_Orientation.
"vo",
# ICU does not use Unihan properties.
@ -528,6 +526,7 @@ def ParsePropertyAliases(in_file):
AddBinaryProperty("Emoji_Presentation", "Emoji_Presentation")
AddBinaryProperty("Emoji_Modifier", "Emoji_Modifier")
AddBinaryProperty("Emoji_Modifier_Base", "Emoji_Modifier_Base")
AddBinaryProperty("Emoji_Component", "Emoji_Component")
# C/POSIX character classes that do not have Unicode property [value] aliases.
# See uchar.h.
AddPOSIXBinaryProperty("alnum")