michael@0: /*
michael@0: *******************************************************************************
michael@0: *   Copyright (C) 2013, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *******************************************************************************
michael@0: *   file name:  uscript_props.cpp
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 2013feb16
michael@0: *   created by: Markus W. Scherer
michael@0: */
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/unistr.h"
michael@0: #include "unicode/uscript.h"
michael@0: #include "unicode/utf16.h"
michael@0: #include "ustr_imp.h"
michael@0: 
michael@0: #define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
michael@0: 
michael@0: namespace {
michael@0: 
michael@0: // Script metadata (script properties).
michael@0: // See http://unicode.org/cldr/trac/browser/trunk/common/properties/scriptMetadata.txt
michael@0: 
michael@0: // 0 = NOT_ENCODED, no sample character, default false script properties.
michael@0: // Bits 20.. 0: sample character
michael@0: 
michael@0: // Bits 23..21: usage
michael@0: const int32_t UNKNOWN = 1 << 21;
michael@0: const int32_t EXCLUSION = 2 << 21;
michael@0: const int32_t LIMITED_USE = 3 << 21;
michael@0: const int32_t ASPIRATIONAL = 4 << 21;
michael@0: const int32_t RECOMMENDED = 5 << 21;
michael@0: 
michael@0: // Bits 31..24: Single-bit flags
michael@0: const int32_t RTL = 1 << 24;
michael@0: const int32_t LB_LETTERS = 1 << 25;
michael@0: const int32_t CASED = 1 << 26;
michael@0: 
michael@0: const int32_t SCRIPT_PROPS[] = {
michael@0:     // Begin copy-paste output from
michael@0:     // tools/trunk/unicode/py/parsescriptmetadata.py
michael@0:     0x0040 | UNKNOWN,  // Zyyy
michael@0:     0x0308 | UNKNOWN,  // Zinh
michael@0:     0x0628 | RECOMMENDED | RTL,  // Arab
michael@0:     0x0531 | RECOMMENDED | CASED,  // Armn
michael@0:     0x0995 | RECOMMENDED,  // Beng
michael@0:     0x3105 | RECOMMENDED | LB_LETTERS,  // Bopo
michael@0:     0x13C4 | LIMITED_USE,  // Cher
michael@0:     0x03E2 | EXCLUSION | CASED,  // Copt
michael@0:     0x042F | RECOMMENDED | CASED,  // Cyrl
michael@0:     0x10414 | EXCLUSION | CASED,  // Dsrt
michael@0:     0x0905 | RECOMMENDED,  // Deva
michael@0:     0x12A0 | RECOMMENDED,  // Ethi
michael@0:     0x10D3 | RECOMMENDED,  // Geor
michael@0:     0x10330 | EXCLUSION,  // Goth
michael@0:     0x03A9 | RECOMMENDED | CASED,  // Grek
michael@0:     0x0A95 | RECOMMENDED,  // Gujr
michael@0:     0x0A15 | RECOMMENDED,  // Guru
michael@0:     0x5B57 | RECOMMENDED | LB_LETTERS,  // Hani
michael@0:     0xAC00 | RECOMMENDED,  // Hang
michael@0:     0x05D0 | RECOMMENDED | RTL,  // Hebr
michael@0:     0x304B | RECOMMENDED | LB_LETTERS,  // Hira
michael@0:     0x0C95 | RECOMMENDED,  // Knda
michael@0:     0x30AB | RECOMMENDED | LB_LETTERS,  // Kana
michael@0:     0x1780 | RECOMMENDED | LB_LETTERS,  // Khmr
michael@0:     0x0EA5 | RECOMMENDED | LB_LETTERS,  // Laoo
michael@0:     0x004C | RECOMMENDED | CASED,  // Latn
michael@0:     0x0D15 | RECOMMENDED,  // Mlym
michael@0:     0x1826 | ASPIRATIONAL,  // Mong
michael@0:     0x1000 | RECOMMENDED | LB_LETTERS,  // Mymr
michael@0:     0x168F | EXCLUSION,  // Ogam
michael@0:     0x10300 | EXCLUSION,  // Ital
michael@0:     0x0B15 | RECOMMENDED,  // Orya
michael@0:     0x16A0 | EXCLUSION,  // Runr
michael@0:     0x0D85 | RECOMMENDED,  // Sinh
michael@0:     0x0710 | LIMITED_USE | RTL,  // Syrc
michael@0:     0x0B95 | RECOMMENDED,  // Taml
michael@0:     0x0C15 | RECOMMENDED,  // Telu
michael@0:     0x078C | RECOMMENDED | RTL,  // Thaa
michael@0:     0x0E17 | RECOMMENDED | LB_LETTERS,  // Thai
michael@0:     0x0F40 | RECOMMENDED,  // Tibt
michael@0:     0x14C0 | ASPIRATIONAL,  // Cans
michael@0:     0xA288 | ASPIRATIONAL | LB_LETTERS,  // Yiii
michael@0:     0x1703 | EXCLUSION,  // Tglg
michael@0:     0x1723 | EXCLUSION,  // Hano
michael@0:     0x1743 | EXCLUSION,  // Buhd
michael@0:     0x1763 | EXCLUSION,  // Tagb
michael@0:     0x2800 | UNKNOWN,  // Brai
michael@0:     0x10800 | EXCLUSION | RTL,  // Cprt
michael@0:     0x1900 | LIMITED_USE,  // Limb
michael@0:     0x10000 | EXCLUSION,  // Linb
michael@0:     0x10480 | EXCLUSION,  // Osma
michael@0:     0x10450 | EXCLUSION,  // Shaw
michael@0:     0x1950 | LIMITED_USE | LB_LETTERS,  // Tale
michael@0:     0x10380 | EXCLUSION,  // Ugar
michael@0:     0,
michael@0:     0x1A00 | EXCLUSION,  // Bugi
michael@0:     0x2C00 | EXCLUSION | CASED,  // Glag
michael@0:     0x10A00 | EXCLUSION | RTL,  // Khar
michael@0:     0xA800 | LIMITED_USE,  // Sylo
michael@0:     0x1980 | LIMITED_USE | LB_LETTERS,  // Talu
michael@0:     0x2D30 | ASPIRATIONAL,  // Tfng
michael@0:     0x103A0 | EXCLUSION,  // Xpeo
michael@0:     0x1B05 | LIMITED_USE | LB_LETTERS,  // Bali
michael@0:     0x1BC0 | LIMITED_USE,  // Batk
michael@0:     0,
michael@0:     0x11005 | EXCLUSION,  // Brah
michael@0:     0xAA00 | LIMITED_USE,  // Cham
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0x13153 | EXCLUSION,  // Egyp
michael@0:     0,
michael@0:     0x5B57 | RECOMMENDED | LB_LETTERS,  // Hans
michael@0:     0x5B57 | RECOMMENDED | LB_LETTERS,  // Hant
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0xA984 | LIMITED_USE | LB_LETTERS,  // Java
michael@0:     0xA90A | LIMITED_USE,  // Kali
michael@0:     0,
michael@0:     0,
michael@0:     0x1C00 | LIMITED_USE,  // Lepc
michael@0:     0,
michael@0:     0x0840 | LIMITED_USE | RTL,  // Mand
michael@0:     0,
michael@0:     0x10980 | EXCLUSION | RTL,  // Mero
michael@0:     0x07CA | LIMITED_USE | RTL,  // Nkoo
michael@0:     0x10C00 | EXCLUSION | RTL,  // Orkh
michael@0:     0,
michael@0:     0xA840 | EXCLUSION,  // Phag
michael@0:     0x10900 | EXCLUSION | RTL,  // Phnx
michael@0:     0x16F00 | ASPIRATIONAL,  // Plrd
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0xA549 | LIMITED_USE,  // Vaii
michael@0:     0,
michael@0:     0x12000 | EXCLUSION,  // Xsux
michael@0:     0,
michael@0:     0xFDD0 | UNKNOWN,  // Zzzz
michael@0:     0x102A0 | EXCLUSION,  // Cari
michael@0:     0x304B | RECOMMENDED | LB_LETTERS,  // Jpan
michael@0:     0x1A20 | LIMITED_USE | LB_LETTERS,  // Lana
michael@0:     0x10280 | EXCLUSION,  // Lyci
michael@0:     0x10920 | EXCLUSION | RTL,  // Lydi
michael@0:     0x1C5A | LIMITED_USE,  // Olck
michael@0:     0xA930 | EXCLUSION,  // Rjng
michael@0:     0xA882 | LIMITED_USE,  // Saur
michael@0:     0,
michael@0:     0x1B83 | LIMITED_USE,  // Sund
michael@0:     0,
michael@0:     0xABC0 | LIMITED_USE,  // Mtei
michael@0:     0x10840 | EXCLUSION | RTL,  // Armi
michael@0:     0x10B00 | EXCLUSION | RTL,  // Avst
michael@0:     0x11103 | LIMITED_USE,  // Cakm
michael@0:     0xAC00 | RECOMMENDED,  // Kore
michael@0:     0x11083 | EXCLUSION,  // Kthi
michael@0:     0,
michael@0:     0x10B60 | EXCLUSION | RTL,  // Phli
michael@0:     0,
michael@0:     0,
michael@0:     0x10B40 | EXCLUSION | RTL,  // Prti
michael@0:     0x0800 | EXCLUSION | RTL,  // Samr
michael@0:     0xAA80 | LIMITED_USE | LB_LETTERS,  // Tavt
michael@0:     0,
michael@0:     0,
michael@0:     0xA6A0 | LIMITED_USE,  // Bamu
michael@0:     0xA4D0 | LIMITED_USE,  // Lisu
michael@0:     0,
michael@0:     0x10A60 | EXCLUSION | RTL,  // Sarb
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0x109A0 | EXCLUSION | RTL,  // Merc
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0x11183 | EXCLUSION,  // Shrd
michael@0:     0x110D0 | EXCLUSION,  // Sora
michael@0:     0x11680 | EXCLUSION,  // Takr
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     0,
michael@0:     // End copy-paste from parsescriptmetadata.py
michael@0: };
michael@0: 
michael@0: int32_t getScriptProps(UScriptCode script) {
michael@0:     if (0 <= script && script < LENGTHOF(SCRIPT_PROPS)) {
michael@0:         return SCRIPT_PROPS[script];
michael@0:     } else {
michael@0:         return 0;
michael@0:     }
michael@0: }
michael@0: 
michael@0: }  // namespace
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: uscript_getSampleString(UScriptCode script, UChar *dest, int32_t capacity, UErrorCode *pErrorCode) {
michael@0:     if(U_FAILURE(*pErrorCode)) { return 0; }
michael@0:     if(capacity < 0 || (capacity > 0 && dest == NULL)) {
michael@0:         *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return 0;
michael@0:     }
michael@0:     int32_t sampleChar = getScriptProps(script) & 0x1fffff;
michael@0:     int32_t length;
michael@0:     if(sampleChar == 0) {
michael@0:         length = 0;
michael@0:     } else {
michael@0:         length = U16_LENGTH(sampleChar);
michael@0:         if(length <= capacity) {
michael@0:             int32_t i = 0;
michael@0:             U16_APPEND_UNSAFE(dest, i, sampleChar);
michael@0:         }
michael@0:     }
michael@0:     return u_terminateUChars(dest, capacity, length, pErrorCode);
michael@0: }
michael@0: 
michael@0: U_COMMON_API icu::UnicodeString U_EXPORT2
michael@0: uscript_getSampleUnicodeString(UScriptCode script) {
michael@0:     icu::UnicodeString sample;
michael@0:     int32_t sampleChar = getScriptProps(script) & 0x1fffff;
michael@0:     if(sampleChar != 0) {
michael@0:         sample.append(sampleChar);
michael@0:     }
michael@0:     return sample;
michael@0: }
michael@0: 
michael@0: U_CAPI UScriptUsage U_EXPORT2
michael@0: uscript_getUsage(UScriptCode script) {
michael@0:     return (UScriptUsage)((getScriptProps(script) >> 21) & 7);
michael@0: }
michael@0: 
michael@0: U_CAPI UBool U_EXPORT2
michael@0: uscript_isRightToLeft(UScriptCode script) {
michael@0:     return (getScriptProps(script) & RTL) != 0;
michael@0: }
michael@0: 
michael@0: U_CAPI UBool U_EXPORT2
michael@0: uscript_breaksBetweenLetters(UScriptCode script) {
michael@0:     return (getScriptProps(script) & LB_LETTERS) != 0;
michael@0: }
michael@0: 
michael@0: U_CAPI UBool U_EXPORT2
michael@0: uscript_isCased(UScriptCode script) {
michael@0:     return (getScriptProps(script) & CASED) != 0;
michael@0: }