[apple/icu.git] / icuSources / common / uscript_props.cpp

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*   Copyright (C) 2013-2016, International Business Machines
*   Corporation and others.  All Rights Reserved.
*******************************************************************************
*   file name:  uscript_props.cpp
*   encoding:   UTF-8
*   tab size:   8 (not used)
*   indentation:4
*
*   created on: 2013feb16
*   created by: Markus W. Scherer
*/

#include "unicode/utypes.h"
#include "unicode/unistr.h"
#include "unicode/uscript.h"
#include "unicode/utf16.h"
#include "ustr_imp.h"
#include "cmemory.h"

namespace {

// Script metadata (script properties).
// See http://unicode.org/cldr/trac/browser/trunk/common/properties/scriptMetadata.txt

// 0 = NOT_ENCODED, no sample character, default false script properties.
// Bits 20.. 0: sample character

// Bits 23..21: usage
const int32_t UNKNOWN = 1 << 21;
const int32_t EXCLUSION = 2 << 21;
const int32_t LIMITED_USE = 3 << 21;
// st int32_t ASPIRATIONAL = 4 << 21; -- not used any more since Unicode 10
const int32_t RECOMMENDED = 5 << 21;

// Bits 31..24: Single-bit flags
const int32_t RTL = 1 << 24;
const int32_t LB_LETTERS = 1 << 25;
const int32_t CASED = 1 << 26;

const int32_t SCRIPT_PROPS[] = {
    // Begin copy-paste output from
    // tools/trunk/unicode/py/parsescriptmetadata.py
    0x0040 | RECOMMENDED,  // Zyyy
    0x0308 | RECOMMENDED,  // Zinh
    0x0628 | RECOMMENDED | RTL,  // Arab
    0x0531 | RECOMMENDED | CASED,  // Armn
    0x0995 | RECOMMENDED,  // Beng
    0x3105 | RECOMMENDED | LB_LETTERS,  // Bopo
    0x13C4 | LIMITED_USE | CASED,  // Cher
    0x03E2 | EXCLUSION | CASED,  // Copt
    0x042F | RECOMMENDED | CASED,  // Cyrl
    0x10414 | EXCLUSION | CASED,  // Dsrt
    0x0905 | RECOMMENDED,  // Deva
    0x12A0 | RECOMMENDED,  // Ethi
    0x10D3 | RECOMMENDED,  // Geor
    0x10330 | EXCLUSION,  // Goth
    0x03A9 | RECOMMENDED | CASED,  // Grek
    0x0A95 | RECOMMENDED,  // Gujr
    0x0A15 | RECOMMENDED,  // Guru
    0x5B57 | RECOMMENDED | LB_LETTERS,  // Hani
    0xAC00 | RECOMMENDED,  // Hang
    0x05D0 | RECOMMENDED | RTL,  // Hebr
    0x304B | RECOMMENDED | LB_LETTERS,  // Hira
    0x0C95 | RECOMMENDED,  // Knda
    0x30AB | RECOMMENDED | LB_LETTERS,  // Kana
    0x1780 | RECOMMENDED | LB_LETTERS,  // Khmr
    0x0EA5 | RECOMMENDED | LB_LETTERS,  // Laoo
    0x004C | RECOMMENDED | CASED,  // Latn
    0x0D15 | RECOMMENDED,  // Mlym
    0x1826 | EXCLUSION,  // Mong
    0x1000 | RECOMMENDED | LB_LETTERS,  // Mymr
    0x168F | EXCLUSION,  // Ogam
    0x10300 | EXCLUSION,  // Ital
    0x0B15 | RECOMMENDED,  // Orya
    0x16A0 | EXCLUSION,  // Runr
    0x0D85 | RECOMMENDED,  // Sinh
    0x0710 | LIMITED_USE | RTL,  // Syrc
    0x0B95 | RECOMMENDED,  // Taml
    0x0C15 | RECOMMENDED,  // Telu
    0x078C | RECOMMENDED | RTL,  // Thaa
    0x0E17 | RECOMMENDED | LB_LETTERS,  // Thai
    0x0F40 | RECOMMENDED,  // Tibt
    0x14C0 | LIMITED_USE,  // Cans
    0xA288 | LIMITED_USE | LB_LETTERS,  // Yiii
    0x1703 | EXCLUSION,  // Tglg
    0x1723 | EXCLUSION,  // Hano
    0x1743 | EXCLUSION,  // Buhd
    0x1763 | EXCLUSION,  // Tagb
    0x280E | UNKNOWN,  // Brai
    0x10800 | EXCLUSION | RTL,  // Cprt
    0x1900 | LIMITED_USE,  // Limb
    0x10000 | EXCLUSION,  // Linb
    0x10480 | EXCLUSION,  // Osma
    0x10450 | EXCLUSION,  // Shaw
    0x1950 | LIMITED_USE | LB_LETTERS,  // Tale
    0x10380 | EXCLUSION,  // Ugar
    0,
    0x1A00 | EXCLUSION,  // Bugi
    0x2C00 | EXCLUSION | CASED,  // Glag
    0x10A00 | EXCLUSION | RTL,  // Khar
    0xA800 | LIMITED_USE,  // Sylo
    0x1980 | LIMITED_USE | LB_LETTERS,  // Talu
    0x2D30 | LIMITED_USE,  // Tfng
    0x103A0 | EXCLUSION,  // Xpeo
    0x1B05 | LIMITED_USE,  // Bali
    0x1BC0 | LIMITED_USE,  // Batk
    0,
    0x11005 | EXCLUSION,  // Brah
    0xAA00 | LIMITED_USE,  // Cham
    0,
    0,
    0,
    0,
    0x13153 | EXCLUSION,  // Egyp
    0,
    0x5B57 | RECOMMENDED | LB_LETTERS,  // Hans
    0x5B57 | RECOMMENDED | LB_LETTERS,  // Hant
    0x16B1C | EXCLUSION,  // Hmng
    0x10CA1 | EXCLUSION | RTL | CASED,  // Hung
    0,
    0xA984 | LIMITED_USE,  // Java
    0xA90A | LIMITED_USE,  // Kali
    0,
    0,
    0x1C00 | LIMITED_USE,  // Lepc
    0x10647 | EXCLUSION,  // Lina
    0x0840 | LIMITED_USE | RTL,  // Mand
    0,
    0x10980 | EXCLUSION | RTL,  // Mero
    0x07CA | LIMITED_USE | RTL,  // Nkoo
    0x10C00 | EXCLUSION | RTL,  // Orkh
    0x1036B | EXCLUSION,  // Perm
    0xA840 | EXCLUSION,  // Phag
    0x10900 | EXCLUSION | RTL,  // Phnx
    0x16F00 | LIMITED_USE,  // Plrd
    0,
    0,
    0,
    0,
    0,
    0,
    0xA549 | LIMITED_USE,  // Vaii
    0,
    0x12000 | EXCLUSION,  // Xsux
    0,
    0xFDD0 | UNKNOWN,  // Zzzz
    0x102A0 | EXCLUSION,  // Cari
    0x304B | RECOMMENDED | LB_LETTERS,  // Jpan
    0x1A20 | LIMITED_USE | LB_LETTERS,  // Lana
    0x10280 | EXCLUSION,  // Lyci
    0x10920 | EXCLUSION | RTL,  // Lydi
    0x1C5A | LIMITED_USE,  // Olck
    0xA930 | EXCLUSION,  // Rjng
    0xA882 | LIMITED_USE,  // Saur
    0x1D850 | EXCLUSION,  // Sgnw
    0x1B83 | LIMITED_USE,  // Sund
    0,
    0xABC0 | LIMITED_USE,  // Mtei
    0x10840 | EXCLUSION | RTL,  // Armi
    0x10B00 | EXCLUSION | RTL,  // Avst
    0x11103 | LIMITED_USE,  // Cakm
    0xAC00 | RECOMMENDED,  // Kore
    0x11083 | EXCLUSION,  // Kthi
    0x10AD8 | EXCLUSION | RTL,  // Mani
    0x10B60 | EXCLUSION | RTL,  // Phli
    0x10B8F | EXCLUSION | RTL,  // Phlp
    0,
    0x10B40 | EXCLUSION | RTL,  // Prti
    0x0800 | EXCLUSION | RTL,  // Samr
    0xAA80 | LIMITED_USE | LB_LETTERS,  // Tavt
    0,
    0,
    0xA6A0 | LIMITED_USE,  // Bamu
    0xA4D0 | LIMITED_USE,  // Lisu
    0,
    0x10A60 | EXCLUSION | RTL,  // Sarb
    0x16AE6 | EXCLUSION,  // Bass
    0x1BC20 | EXCLUSION,  // Dupl
    0x10500 | EXCLUSION,  // Elba
    0x11315 | EXCLUSION,  // Gran
    0,
    0,
    0x1E802 | EXCLUSION | RTL,  // Mend
    0x109A0 | EXCLUSION | RTL,  // Merc
    0x10A95 | EXCLUSION | RTL,  // Narb
    0x10896 | EXCLUSION | RTL,  // Nbat
    0x10873 | EXCLUSION | RTL,  // Palm
    0x112BE | EXCLUSION,  // Sind
    0x118B4 | EXCLUSION | CASED,  // Wara
    0,
    0,
    0x16A4F | EXCLUSION,  // Mroo
    0x1B1C4 | EXCLUSION | LB_LETTERS,  // Nshu
    0x11183 | EXCLUSION,  // Shrd
    0x110D0 | EXCLUSION,  // Sora
    0x11680 | EXCLUSION,  // Takr
    0x18229 | EXCLUSION | LB_LETTERS,  // Tang
    0,
    0x14400 | EXCLUSION,  // Hluw
    0x11208 | EXCLUSION,  // Khoj
    0x11484 | EXCLUSION,  // Tirh
    0x10537 | EXCLUSION,  // Aghb
    0x11152 | EXCLUSION,  // Mahj
    0x11717 | EXCLUSION | LB_LETTERS,  // Ahom
    0x108F4 | EXCLUSION | RTL,  // Hatr
    0x1160E | EXCLUSION,  // Modi
    0x1128F | EXCLUSION,  // Mult
    0x11AC0 | EXCLUSION,  // Pauc
    0x1158E | EXCLUSION,  // Sidd
    0x1E909 | LIMITED_USE | RTL | CASED,  // Adlm
    0x11C0E | EXCLUSION,  // Bhks
    0x11C72 | EXCLUSION,  // Marc
    0x11412 | LIMITED_USE,  // Newa
    0x104B5 | LIMITED_USE | CASED,  // Osge
    0x5B57 | RECOMMENDED | LB_LETTERS,  // Hanb
    0x1112 | RECOMMENDED,  // Jamo
    0,
    0x11D10 | EXCLUSION,  // Gonm
    0x11A5C | EXCLUSION,  // Soyo
    0x11A0B | EXCLUSION,  // Zanb
    0x1180B | EXCLUSION,  // Dogr
    0x11D71 | LIMITED_USE,  // Gong
    0x11EE5 | EXCLUSION,  // Maka
    0x16E40 | EXCLUSION | CASED,  // Medf
    0x10D12 | LIMITED_USE | RTL,  // Rohg
    0x10F42 | EXCLUSION | RTL,  // Sogd
    0x10F19 | EXCLUSION | RTL,  // Sogo
    0x10FF1 | EXCLUSION | RTL,  // Elym
    0x1E108 | LIMITED_USE,  // Hmnp
    0x119CE | EXCLUSION,  // Nand
    0x1E2E1 | LIMITED_USE,  // Wcho
    // End copy-paste from parsescriptmetadata.py
};

int32_t getScriptProps(UScriptCode script) {
    if (0 <= script && script < UPRV_LENGTHOF(SCRIPT_PROPS)) {
        return SCRIPT_PROPS[script];
    } else {
        return 0;
    }
}

}  // namespace

U_CAPI int32_t U_EXPORT2
uscript_getSampleString(UScriptCode script, UChar *dest, int32_t capacity, UErrorCode *pErrorCode) {
    if(U_FAILURE(*pErrorCode)) { return 0; }
    if(capacity < 0 || (capacity > 0 && dest == NULL)) {
        *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
        return 0;
    }
    int32_t sampleChar = getScriptProps(script) & 0x1fffff;
    int32_t length;
    if(sampleChar == 0) {
        length = 0;
    } else {
        length = U16_LENGTH(sampleChar);
        if(length <= capacity) {
            int32_t i = 0;
            U16_APPEND_UNSAFE(dest, i, sampleChar);
        }
    }
    return u_terminateUChars(dest, capacity, length, pErrorCode);
}

U_COMMON_API icu::UnicodeString U_EXPORT2
uscript_getSampleUnicodeString(UScriptCode script) {
    icu::UnicodeString sample;
    int32_t sampleChar = getScriptProps(script) & 0x1fffff;
    if(sampleChar != 0) {
        sample.append(sampleChar);
    }
    return sample;
}

U_CAPI UScriptUsage U_EXPORT2
uscript_getUsage(UScriptCode script) {
    return (UScriptUsage)((getScriptProps(script) >> 21) & 7);
}

U_CAPI UBool U_EXPORT2
uscript_isRightToLeft(UScriptCode script) {
    return (getScriptProps(script) & RTL) != 0;
}

U_CAPI UBool U_EXPORT2
uscript_breaksBetweenLetters(UScriptCode script) {
    return (getScriptProps(script) & LB_LETTERS) != 0;
}

U_CAPI UBool U_EXPORT2
uscript_isCased(UScriptCode script) {
    return (getScriptProps(script) & CASED) != 0;
}
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
51004dcb A	3	/*
51004dcb A	4	*******************************************************************************
f3c0d7a5	5	* Copyright (C) 2013-2016, International Business Machines
51004dcb A	6	* Corporation and others. All Rights Reserved.
	7	*******************************************************************************
	8	* file name: uscript_props.cpp
f3c0d7a5	9	* encoding: UTF-8
51004dcb A	10	* tab size: 8 (not used)
	11	* indentation:4
	12	*
	13	* created on: 2013feb16
	14	* created by: Markus W. Scherer
	15	*/
	16
	17	#include "unicode/utypes.h"
	18	#include "unicode/unistr.h"
	19	#include "unicode/uscript.h"
	20	#include "unicode/utf16.h"
	21	#include "ustr_imp.h"
b331163b	22	#include "cmemory.h"
51004dcb A	23
	24	namespace {
	25
	26	// Script metadata (script properties).
	27	// See http://unicode.org/cldr/trac/browser/trunk/common/properties/scriptMetadata.txt
	28
	29	// 0 = NOT_ENCODED, no sample character, default false script properties.
	30	// Bits 20.. 0: sample character
	31
	32	// Bits 23..21: usage
	33	const int32_t UNKNOWN = 1 << 21;
	34	const int32_t EXCLUSION = 2 << 21;
	35	const int32_t LIMITED_USE = 3 << 21;
6be67b06	36	// st int32_t ASPIRATIONAL = 4 << 21; -- not used any more since Unicode 10
51004dcb A	37	const int32_t RECOMMENDED = 5 << 21;
	38
	39	// Bits 31..24: Single-bit flags
	40	const int32_t RTL = 1 << 24;
	41	const int32_t LB_LETTERS = 1 << 25;
	42	const int32_t CASED = 1 << 26;
	43
	44	const int32_t SCRIPT_PROPS[] = {
	45	// Begin copy-paste output from
	46	// tools/trunk/unicode/py/parsescriptmetadata.py
b331163b	47	0x0040 \| RECOMMENDED, // Zyyy
2ca993e8	48	0x0308 \| RECOMMENDED, // Zinh
51004dcb A	49	0x0628 \| RECOMMENDED \| RTL, // Arab
	50	0x0531 \| RECOMMENDED \| CASED, // Armn
	51	0x0995 \| RECOMMENDED, // Beng
	52	0x3105 \| RECOMMENDED \| LB_LETTERS, // Bopo
2ca993e8	53	0x13C4 \| LIMITED_USE \| CASED, // Cher
51004dcb A	54	0x03E2 \| EXCLUSION \| CASED, // Copt
	55	0x042F \| RECOMMENDED \| CASED, // Cyrl
	56	0x10414 \| EXCLUSION \| CASED, // Dsrt
	57	0x0905 \| RECOMMENDED, // Deva
	58	0x12A0 \| RECOMMENDED, // Ethi
	59	0x10D3 \| RECOMMENDED, // Geor
	60	0x10330 \| EXCLUSION, // Goth
	61	0x03A9 \| RECOMMENDED \| CASED, // Grek
	62	0x0A95 \| RECOMMENDED, // Gujr
	63	0x0A15 \| RECOMMENDED, // Guru
	64	0x5B57 \| RECOMMENDED \| LB_LETTERS, // Hani
	65	0xAC00 \| RECOMMENDED, // Hang
	66	0x05D0 \| RECOMMENDED \| RTL, // Hebr
	67	0x304B \| RECOMMENDED \| LB_LETTERS, // Hira
	68	0x0C95 \| RECOMMENDED, // Knda
	69	0x30AB \| RECOMMENDED \| LB_LETTERS, // Kana
	70	0x1780 \| RECOMMENDED \| LB_LETTERS, // Khmr
	71	0x0EA5 \| RECOMMENDED \| LB_LETTERS, // Laoo
	72	0x004C \| RECOMMENDED \| CASED, // Latn
	73	0x0D15 \| RECOMMENDED, // Mlym
0f5d89e8	74	0x1826 \| EXCLUSION, // Mong
51004dcb A	75	0x1000 \| RECOMMENDED \| LB_LETTERS, // Mymr
51004dcb A	76	0x168F \| EXCLUSION, // Ogam
f3c0d7a5	77	0x10300 \| EXCLUSION, // Ital
51004dcb A	78	0x0B15 \| RECOMMENDED, // Orya
	79	0x16A0 \| EXCLUSION, // Runr
	80	0x0D85 \| RECOMMENDED, // Sinh
	81	0x0710 \| LIMITED_USE \| RTL, // Syrc
	82	0x0B95 \| RECOMMENDED, // Taml
	83	0x0C15 \| RECOMMENDED, // Telu
	84	0x078C \| RECOMMENDED \| RTL, // Thaa
	85	0x0E17 \| RECOMMENDED \| LB_LETTERS, // Thai
	86	0x0F40 \| RECOMMENDED, // Tibt
6be67b06 A	87	0x14C0 \| LIMITED_USE, // Cans
6be67b06 A	88	0xA288 \| LIMITED_USE \| LB_LETTERS, // Yiii
51004dcb A	89	0x1703 \| EXCLUSION, // Tglg
	90	0x1723 \| EXCLUSION, // Hano
	91	0x1743 \| EXCLUSION, // Buhd
	92	0x1763 \| EXCLUSION, // Tagb
2ca993e8	93	0x280E \| UNKNOWN, // Brai
51004dcb A	94	0x10800 \| EXCLUSION \| RTL, // Cprt
	95	0x1900 \| LIMITED_USE, // Limb
	96	0x10000 \| EXCLUSION, // Linb
	97	0x10480 \| EXCLUSION, // Osma
	98	0x10450 \| EXCLUSION, // Shaw
	99	0x1950 \| LIMITED_USE \| LB_LETTERS, // Tale
	100	0x10380 \| EXCLUSION, // Ugar
	101	0,
	102	0x1A00 \| EXCLUSION, // Bugi
	103	0x2C00 \| EXCLUSION \| CASED, // Glag
	104	0x10A00 \| EXCLUSION \| RTL, // Khar
	105	0xA800 \| LIMITED_USE, // Sylo
	106	0x1980 \| LIMITED_USE \| LB_LETTERS, // Talu
6be67b06	107	0x2D30 \| LIMITED_USE, // Tfng
51004dcb	108	0x103A0 \| EXCLUSION, // Xpeo
b331163b	109	0x1B05 \| LIMITED_USE, // Bali
51004dcb A	110	0x1BC0 \| LIMITED_USE, // Batk
	111	0,
	112	0x11005 \| EXCLUSION, // Brah
	113	0xAA00 \| LIMITED_USE, // Cham
	114	0,
	115	0,
	116	0,
	117	0,
	118	0x13153 \| EXCLUSION, // Egyp
	119	0,
	120	0x5B57 \| RECOMMENDED \| LB_LETTERS, // Hans
	121	0x5B57 \| RECOMMENDED \| LB_LETTERS, // Hant
b331163b	122	0x16B1C \| EXCLUSION, // Hmng
2ca993e8	123	0x10CA1 \| EXCLUSION \| RTL \| CASED, // Hung
51004dcb	124	0,
b331163b	125	0xA984 \| LIMITED_USE, // Java
51004dcb A	126	0xA90A \| LIMITED_USE, // Kali
	127	0,
	128	0,
	129	0x1C00 \| LIMITED_USE, // Lepc
b331163b	130	0x10647 \| EXCLUSION, // Lina
51004dcb A	131	0x0840 \| LIMITED_USE \| RTL, // Mand
	132	0,
	133	0x10980 \| EXCLUSION \| RTL, // Mero
f3c0d7a5	134	0x07CA \| LIMITED_USE \| RTL, // Nkoo
51004dcb	135	0x10C00 \| EXCLUSION \| RTL, // Orkh
b331163b	136	0x1036B \| EXCLUSION, // Perm
51004dcb A	137	0xA840 \| EXCLUSION, // Phag
51004dcb A	138	0x10900 \| EXCLUSION \| RTL, // Phnx
6be67b06	139	0x16F00 \| LIMITED_USE, // Plrd
51004dcb A	140	0,
	141	0,
	142	0,
	143	0,
	144	0,
	145	0,
	146	0xA549 \| LIMITED_USE, // Vaii
	147	0,
	148	0x12000 \| EXCLUSION, // Xsux
	149	0,
	150	0xFDD0 \| UNKNOWN, // Zzzz
f3c0d7a5	151	0x102A0 \| EXCLUSION, // Cari
51004dcb A	152	0x304B \| RECOMMENDED \| LB_LETTERS, // Jpan
	153	0x1A20 \| LIMITED_USE \| LB_LETTERS, // Lana
	154	0x10280 \| EXCLUSION, // Lyci
	155	0x10920 \| EXCLUSION \| RTL, // Lydi
	156	0x1C5A \| LIMITED_USE, // Olck
	157	0xA930 \| EXCLUSION, // Rjng
	158	0xA882 \| LIMITED_USE, // Saur
2ca993e8	159	0x1D850 \| EXCLUSION, // Sgnw
51004dcb A	160	0x1B83 \| LIMITED_USE, // Sund
	161	0,
	162	0xABC0 \| LIMITED_USE, // Mtei
	163	0x10840 \| EXCLUSION \| RTL, // Armi
	164	0x10B00 \| EXCLUSION \| RTL, // Avst
	165	0x11103 \| LIMITED_USE, // Cakm
	166	0xAC00 \| RECOMMENDED, // Kore
	167	0x11083 \| EXCLUSION, // Kthi
f3c0d7a5	168	0x10AD8 \| EXCLUSION \| RTL, // Mani
51004dcb	169	0x10B60 \| EXCLUSION \| RTL, // Phli
b331163b	170	0x10B8F \| EXCLUSION \| RTL, // Phlp
51004dcb A	171	0,
	172	0x10B40 \| EXCLUSION \| RTL, // Prti
	173	0x0800 \| EXCLUSION \| RTL, // Samr
	174	0xAA80 \| LIMITED_USE \| LB_LETTERS, // Tavt
	175	0,
	176	0,
	177	0xA6A0 \| LIMITED_USE, // Bamu
f3c0d7a5	178	0xA4D0 \| LIMITED_USE, // Lisu
51004dcb A	179	0,
51004dcb A	180	0x10A60 \| EXCLUSION \| RTL, // Sarb
b331163b A	181	0x16AE6 \| EXCLUSION, // Bass
	182	0x1BC20 \| EXCLUSION, // Dupl
	183	0x10500 \| EXCLUSION, // Elba
	184	0x11315 \| EXCLUSION, // Gran
51004dcb A	185	0,
51004dcb A	186	0,
b331163b	187	0x1E802 \| EXCLUSION \| RTL, // Mend
51004dcb	188	0x109A0 \| EXCLUSION \| RTL, // Merc
b331163b A	189	0x10A95 \| EXCLUSION \| RTL, // Narb
	190	0x10896 \| EXCLUSION \| RTL, // Nbat
	191	0x10873 \| EXCLUSION \| RTL, // Palm
	192	0x112BE \| EXCLUSION, // Sind
	193	0x118B4 \| EXCLUSION \| CASED, // Wara
51004dcb A	194	0,
51004dcb A	195	0,
b331163b	196	0x16A4F \| EXCLUSION, // Mroo
6be67b06	197	0x1B1C4 \| EXCLUSION \| LB_LETTERS, // Nshu
51004dcb A	198	0x11183 \| EXCLUSION, // Shrd
	199	0x110D0 \| EXCLUSION, // Sora
	200	0x11680 \| EXCLUSION, // Takr
f3c0d7a5	201	0x18229 \| EXCLUSION \| LB_LETTERS, // Tang
51004dcb	202	0,
2ca993e8	203	0x14400 \| EXCLUSION, // Hluw
b331163b A	204	0x11208 \| EXCLUSION, // Khoj
	205	0x11484 \| EXCLUSION, // Tirh
	206	0x10537 \| EXCLUSION, // Aghb
	207	0x11152 \| EXCLUSION, // Mahj
2ca993e8 A	208	0x11717 \| EXCLUSION \| LB_LETTERS, // Ahom
2ca993e8 A	209	0x108F4 \| EXCLUSION \| RTL, // Hatr
b331163b	210	0x1160E \| EXCLUSION, // Modi
2ca993e8	211	0x1128F \| EXCLUSION, // Mult
b331163b A	212	0x11AC0 \| EXCLUSION, // Pauc
b331163b A	213	0x1158E \| EXCLUSION, // Sidd
f3c0d7a5 A	214	0x1E909 \| LIMITED_USE \| RTL \| CASED, // Adlm
	215	0x11C0E \| EXCLUSION, // Bhks
	216	0x11C72 \| EXCLUSION, // Marc
	217	0x11412 \| LIMITED_USE, // Newa
	218	0x104B5 \| LIMITED_USE \| CASED, // Osge
	219	0x5B57 \| RECOMMENDED \| LB_LETTERS, // Hanb
	220	0x1112 \| RECOMMENDED, // Jamo
	221	0,
6be67b06 A	222	0x11D10 \| EXCLUSION, // Gonm
	223	0x11A5C \| EXCLUSION, // Soyo
	224	0x11A0B \| EXCLUSION, // Zanb
0f5d89e8 A	225	0x1180B \| EXCLUSION, // Dogr
	226	0x11D71 \| LIMITED_USE, // Gong
	227	0x11EE5 \| EXCLUSION, // Maka
	228	0x16E40 \| EXCLUSION \| CASED, // Medf
	229	0x10D12 \| LIMITED_USE \| RTL, // Rohg
	230	0x10F42 \| EXCLUSION \| RTL, // Sogd
	231	0x10F19 \| EXCLUSION \| RTL, // Sogo
3d1f044b A	232	0x10FF1 \| EXCLUSION \| RTL, // Elym
	233	0x1E108 \| LIMITED_USE, // Hmnp
	234	0x119CE \| EXCLUSION, // Nand
	235	0x1E2E1 \| LIMITED_USE, // Wcho
51004dcb A	236	// End copy-paste from parsescriptmetadata.py
	237	};
	238
	239	int32_t getScriptProps(UScriptCode script) {
b331163b	240	if (0 <= script && script < UPRV_LENGTHOF(SCRIPT_PROPS)) {
51004dcb A	241	return SCRIPT_PROPS[script];
	242	} else {
	243	return 0;
	244	}
	245	}
	246
	247	} // namespace
	248
	249	U_CAPI int32_t U_EXPORT2
	250	uscript_getSampleString(UScriptCode script, UChar dest, int32_t capacity, UErrorCode pErrorCode) {
	251	if(U_FAILURE(*pErrorCode)) { return 0; }
	252	if(capacity < 0 \|\| (capacity > 0 && dest == NULL)) {
	253	*pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
	254	return 0;
	255	}
	256	int32_t sampleChar = getScriptProps(script) & 0x1fffff;
	257	int32_t length;
	258	if(sampleChar == 0) {
	259	length = 0;
	260	} else {
	261	length = U16_LENGTH(sampleChar);
	262	if(length <= capacity) {
	263	int32_t i = 0;
	264	U16_APPEND_UNSAFE(dest, i, sampleChar);
	265	}
	266	}
	267	return u_terminateUChars(dest, capacity, length, pErrorCode);
	268	}
	269
	270	U_COMMON_API icu::UnicodeString U_EXPORT2
	271	uscript_getSampleUnicodeString(UScriptCode script) {
	272	icu::UnicodeString sample;
	273	int32_t sampleChar = getScriptProps(script) & 0x1fffff;
	274	if(sampleChar != 0) {
	275	sample.append(sampleChar);
	276	}
	277	return sample;
	278	}
	279
	280	U_CAPI UScriptUsage U_EXPORT2
	281	uscript_getUsage(UScriptCode script) {
	282	return (UScriptUsage)((getScriptProps(script) >> 21) & 7);
	283	}
	284
	285	U_CAPI UBool U_EXPORT2
	286	uscript_isRightToLeft(UScriptCode script) {
	287	return (getScriptProps(script) & RTL) != 0;
	288	}
	289
	290	U_CAPI UBool U_EXPORT2
	291	uscript_breaksBetweenLetters(UScriptCode script) {
	292	return (getScriptProps(script) & LB_LETTERS) != 0;
	293	}
	294
	295	U_CAPI UBool U_EXPORT2
	296	uscript_isCased(UScriptCode script) {
	297	return (getScriptProps(script) & CASED) != 0;
	298	}