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

/*
*******************************************************************************
*   Copyright (C) 2013, International Business Machines
*   Corporation and others.  All Rights Reserved.
*******************************************************************************
*   file name:  uscript_props.cpp
*   encoding:   US-ASCII
*   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"

#define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))

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;
const int32_t ASPIRATIONAL = 4 << 21;
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 | UNKNOWN,  // Zyyy
    0x0308 | UNKNOWN,  // Zinh
    0x0628 | RECOMMENDED | RTL,  // Arab
    0x0531 | RECOMMENDED | CASED,  // Armn
    0x0995 | RECOMMENDED,  // Beng
    0x3105 | RECOMMENDED | LB_LETTERS,  // Bopo
    0x13C4 | LIMITED_USE,  // 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 | ASPIRATIONAL,  // 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 | ASPIRATIONAL,  // Cans
    0xA288 | ASPIRATIONAL | LB_LETTERS,  // Yiii
    0x1703 | EXCLUSION,  // Tglg
    0x1723 | EXCLUSION,  // Hano
    0x1743 | EXCLUSION,  // Buhd
    0x1763 | EXCLUSION,  // Tagb
    0x2800 | 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 | ASPIRATIONAL,  // Tfng
    0x103A0 | EXCLUSION,  // Xpeo
    0x1B05 | LIMITED_USE | LB_LETTERS,  // 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
    0,
    0,
    0,
    0xA984 | LIMITED_USE | LB_LETTERS,  // Java
    0xA90A | LIMITED_USE,  // Kali
    0,
    0,
    0x1C00 | LIMITED_USE,  // Lepc
    0,
    0x0840 | LIMITED_USE | RTL,  // Mand
    0,
    0x10980 | EXCLUSION | RTL,  // Mero
    0x07CA | LIMITED_USE | RTL,  // Nkoo
    0x10C00 | EXCLUSION | RTL,  // Orkh
    0,
    0xA840 | EXCLUSION,  // Phag
    0x10900 | EXCLUSION | RTL,  // Phnx
    0x16F00 | ASPIRATIONAL,  // 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
    0,
    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
    0,
    0x10B60 | EXCLUSION | RTL,  // Phli
    0,
    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
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0x109A0 | EXCLUSION | RTL,  // Merc
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    0x11183 | EXCLUSION,  // Shrd
    0x110D0 | EXCLUSION,  // Sora
    0x11680 | EXCLUSION,  // Takr
    0,
    0,
    0,
    0,
    0,
    0,
    0,
    // End copy-paste from parsescriptmetadata.py
};

int32_t getScriptProps(UScriptCode script) {
    if (0 <= script && script < 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
51004dcb A	1	/*
	2	*******************************************************************************
	3	* Copyright (C) 2013, International Business Machines
	4	* Corporation and others. All Rights Reserved.
	5	*******************************************************************************
	6	* file name: uscript_props.cpp
	7	* encoding: US-ASCII
	8	* tab size: 8 (not used)
	9	* indentation:4
	10	*
	11	* created on: 2013feb16
	12	* created by: Markus W. Scherer
	13	*/
	14
	15	#include "unicode/utypes.h"
	16	#include "unicode/unistr.h"
	17	#include "unicode/uscript.h"
	18	#include "unicode/utf16.h"
	19	#include "ustr_imp.h"
	20
	21	#define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
	22
	23	namespace {
	24
	25	// Script metadata (script properties).
	26	// See http://unicode.org/cldr/trac/browser/trunk/common/properties/scriptMetadata.txt
	27
	28	// 0 = NOT_ENCODED, no sample character, default false script properties.
	29	// Bits 20.. 0: sample character
	30
	31	// Bits 23..21: usage
	32	const int32_t UNKNOWN = 1 << 21;
	33	const int32_t EXCLUSION = 2 << 21;
	34	const int32_t LIMITED_USE = 3 << 21;
	35	const int32_t ASPIRATIONAL = 4 << 21;
	36	const int32_t RECOMMENDED = 5 << 21;
	37
	38	// Bits 31..24: Single-bit flags
	39	const int32_t RTL = 1 << 24;
	40	const int32_t LB_LETTERS = 1 << 25;
	41	const int32_t CASED = 1 << 26;
	42
	43	const int32_t SCRIPT_PROPS[] = {
	44	// Begin copy-paste output from
	45	// tools/trunk/unicode/py/parsescriptmetadata.py
	46	0x0040 \| UNKNOWN, // Zyyy
	47	0x0308 \| UNKNOWN, // Zinh
	48	0x0628 \| RECOMMENDED \| RTL, // Arab
	49	0x0531 \| RECOMMENDED \| CASED, // Armn
	50	0x0995 \| RECOMMENDED, // Beng
	51	0x3105 \| RECOMMENDED \| LB_LETTERS, // Bopo
	52	0x13C4 \| LIMITED_USE, // Cher
	53	0x03E2 \| EXCLUSION \| CASED, // Copt
	54	0x042F \| RECOMMENDED \| CASED, // Cyrl
	55	0x10414 \| EXCLUSION \| CASED, // Dsrt
	56	0x0905 \| RECOMMENDED, // Deva
	57	0x12A0 \| RECOMMENDED, // Ethi
	58	0x10D3 \| RECOMMENDED, // Geor
	59	0x10330 \| EXCLUSION, // Goth
	60	0x03A9 \| RECOMMENDED \| CASED, // Grek
	61	0x0A95 \| RECOMMENDED, // Gujr
	62	0x0A15 \| RECOMMENDED, // Guru
	63	0x5B57 \| RECOMMENDED \| LB_LETTERS, // Hani
	64	0xAC00 \| RECOMMENDED, // Hang
65	0x05D0 \| RECOMMENDED \| RTL, // Hebr
66	0x304B \| RECOMMENDED \| LB_LETTERS, // Hira
67	0x0C95 \| RECOMMENDED, // Knda
68	0x30AB \| RECOMMENDED \| LB_LETTERS, // Kana
69	0x1780 \| RECOMMENDED \| LB_LETTERS, // Khmr
70	0x0EA5 \| RECOMMENDED \| LB_LETTERS, // Laoo
71	0x004C \| RECOMMENDED \| CASED, // Latn
72	0x0D15 \| RECOMMENDED, // Mlym
73	0x1826 \| ASPIRATIONAL, // Mong
74	0x1000 \| RECOMMENDED \| LB_LETTERS, // Mymr
75	0x168F \| EXCLUSION, // Ogam
76	0x10300 \| EXCLUSION, // Ital
77	0x0B15 \| RECOMMENDED, // Orya
78	0x16A0 \| EXCLUSION, // Runr
79	0x0D85 \| RECOMMENDED, // Sinh
80	0x0710 \| LIMITED_USE \| RTL, // Syrc
81	0x0B95 \| RECOMMENDED, // Taml
82	0x0C15 \| RECOMMENDED, // Telu
83	0x078C \| RECOMMENDED \| RTL, // Thaa
84	0x0E17 \| RECOMMENDED \| LB_LETTERS, // Thai
85	0x0F40 \| RECOMMENDED, // Tibt
86	0x14C0 \| ASPIRATIONAL, // Cans
87	0xA288 \| ASPIRATIONAL \| LB_LETTERS, // Yiii
88	0x1703 \| EXCLUSION, // Tglg
89	0x1723 \| EXCLUSION, // Hano
90	0x1743 \| EXCLUSION, // Buhd
91	0x1763 \| EXCLUSION, // Tagb
92	0x2800 \| UNKNOWN, // Brai
93	0x10800 \| EXCLUSION \| RTL, // Cprt
94	0x1900 \| LIMITED_USE, // Limb
95	0x10000 \| EXCLUSION, // Linb
96	0x10480 \| EXCLUSION, // Osma
97	0x10450 \| EXCLUSION, // Shaw
98	0x1950 \| LIMITED_USE \| LB_LETTERS, // Tale
99	0x10380 \| EXCLUSION, // Ugar
100	0,
101	0x1A00 \| EXCLUSION, // Bugi
102	0x2C00 \| EXCLUSION \| CASED, // Glag
103	0x10A00 \| EXCLUSION \| RTL, // Khar
104	0xA800 \| LIMITED_USE, // Sylo
105	0x1980 \| LIMITED_USE \| LB_LETTERS, // Talu
106	0x2D30 \| ASPIRATIONAL, // Tfng
107	0x103A0 \| EXCLUSION, // Xpeo
108	0x1B05 \| LIMITED_USE \| LB_LETTERS, // Bali
109	0x1BC0 \| LIMITED_USE, // Batk
110	0,
111	0x11005 \| EXCLUSION, // Brah
112	0xAA00 \| LIMITED_USE, // Cham
113	0,
114	0,
115	0,
116	0,
117	0x13153 \| EXCLUSION, // Egyp
118	0,
119	0x5B57 \| RECOMMENDED \| LB_LETTERS, // Hans
120	0x5B57 \| RECOMMENDED \| LB_LETTERS, // Hant
121	0,
122	0,
123	0,
124	0xA984 \| LIMITED_USE \| LB_LETTERS, // Java
125	0xA90A \| LIMITED_USE, // Kali
126	0,
127	0,
128	0x1C00 \| LIMITED_USE, // Lepc
129	0,
130	0x0840 \| LIMITED_USE \| RTL, // Mand
131	0,
132	0x10980 \| EXCLUSION \| RTL, // Mero
133	0x07CA \| LIMITED_USE \| RTL, // Nkoo
134	0x10C00 \| EXCLUSION \| RTL, // Orkh
135	0,
136	0xA840 \| EXCLUSION, // Phag
137	0x10900 \| EXCLUSION \| RTL, // Phnx
138	0x16F00 \| ASPIRATIONAL, // Plrd
139	0,
140	0,
141	0,
142	0,
143	0,
144	0,
145	0xA549 \| LIMITED_USE, // Vaii
146	0,
147	0x12000 \| EXCLUSION, // Xsux
148	0,
149	0xFDD0 \| UNKNOWN, // Zzzz
150	0x102A0 \| EXCLUSION, // Cari
151	0x304B \| RECOMMENDED \| LB_LETTERS, // Jpan
152	0x1A20 \| LIMITED_USE \| LB_LETTERS, // Lana
153	0x10280 \| EXCLUSION, // Lyci
154	0x10920 \| EXCLUSION \| RTL, // Lydi
155	0x1C5A \| LIMITED_USE, // Olck
156	0xA930 \| EXCLUSION, // Rjng
157	0xA882 \| LIMITED_USE, // Saur
158	0,
159	0x1B83 \| LIMITED_USE, // Sund
160	0,
161	0xABC0 \| LIMITED_USE, // Mtei
162	0x10840 \| EXCLUSION \| RTL, // Armi
163	0x10B00 \| EXCLUSION \| RTL, // Avst
164	0x11103 \| LIMITED_USE, // Cakm
165	0xAC00 \| RECOMMENDED, // Kore
166	0x11083 \| EXCLUSION, // Kthi
167	0,
168	0x10B60 \| EXCLUSION \| RTL, // Phli
169	0,
170	0,
171	0x10B40 \| EXCLUSION \| RTL, // Prti
172	0x0800 \| EXCLUSION \| RTL, // Samr
173	0xAA80 \| LIMITED_USE \| LB_LETTERS, // Tavt
174	0,
175	0,
176	0xA6A0 \| LIMITED_USE, // Bamu
177	0xA4D0 \| LIMITED_USE, // Lisu
178	0,
179	0x10A60 \| EXCLUSION \| RTL, // Sarb
180	0,
181	0,
182	0,
183	0,
184	0,
185	0,
186	0,
187	0x109A0 \| EXCLUSION \| RTL, // Merc
188	0,
189	0,
190	0,
191	0,
192	0,
193	0,
194	0,
195	0,
196	0,
197	0x11183 \| EXCLUSION, // Shrd
198	0x110D0 \| EXCLUSION, // Sora
199	0x11680 \| EXCLUSION, // Takr
200	0,
201	0,
202	0,
203	0,
204	0,
57a6839d A	205	0,
57a6839d A	206	0,
51004dcb A	207	// End copy-paste from parsescriptmetadata.py
	208	};
	209
	210	int32_t getScriptProps(UScriptCode script) {
	211	if (0 <= script && script < LENGTHOF(SCRIPT_PROPS)) {
	212	return SCRIPT_PROPS[script];
	213	} else {
	214	return 0;
	215	}
	216	}
	217
	218	} // namespace
	219
	220	U_CAPI int32_t U_EXPORT2
	221	uscript_getSampleString(UScriptCode script, UChar dest, int32_t capacity, UErrorCode pErrorCode) {
	222	if(U_FAILURE(*pErrorCode)) { return 0; }
	223	if(capacity < 0 \|\| (capacity > 0 && dest == NULL)) {
	224	*pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
	225	return 0;
	226	}
	227	int32_t sampleChar = getScriptProps(script) & 0x1fffff;
	228	int32_t length;
	229	if(sampleChar == 0) {
	230	length = 0;
	231	} else {
	232	length = U16_LENGTH(sampleChar);
	233	if(length <= capacity) {
	234	int32_t i = 0;
	235	U16_APPEND_UNSAFE(dest, i, sampleChar);
	236	}
	237	}
	238	return u_terminateUChars(dest, capacity, length, pErrorCode);
	239	}
	240
	241	U_COMMON_API icu::UnicodeString U_EXPORT2
	242	uscript_getSampleUnicodeString(UScriptCode script) {
	243	icu::UnicodeString sample;
	244	int32_t sampleChar = getScriptProps(script) & 0x1fffff;
	245	if(sampleChar != 0) {
	246	sample.append(sampleChar);
	247	}
	248	return sample;
	249	}
	250
	251	U_CAPI UScriptUsage U_EXPORT2
	252	uscript_getUsage(UScriptCode script) {
	253	return (UScriptUsage)((getScriptProps(script) >> 21) & 7);
	254	}
	255
	256	U_CAPI UBool U_EXPORT2
	257	uscript_isRightToLeft(UScriptCode script) {
	258	return (getScriptProps(script) & RTL) != 0;
	259	}
	260
	261	U_CAPI UBool U_EXPORT2
	262	uscript_breaksBetweenLetters(UScriptCode script) {
	263	return (getScriptProps(script) & LB_LETTERS) != 0;
	264	}
	265
	266	U_CAPI UBool U_EXPORT2
	267	uscript_isCased(UScriptCode script) {
	268	return (getScriptProps(script) & CASED) != 0;
	269	}