icuSources/common/uscript_props.cpp

   1 /*
   2 *******************************************************************************
   3 *   Copyright (C) 2013-2015, 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 #include "cmemory.h"
  21
  22 namespace {
  23
  24 // Script metadata (script properties).
  25 // See http://unicode.org/cldr/trac/browser/trunk/common/properties/scriptMetadata.txt
  26
  27 // 0 = NOT_ENCODED, no sample character, default false script properties.
  28 // Bits 20.. 0: sample character
  29
  30 // Bits 23..21: usage
  31 const int32_t UNKNOWN = 1 << 21;
  32 const int32_t EXCLUSION = 2 << 21;
  33 const int32_t LIMITED_USE = 3 << 21;
  34 const int32_t ASPIRATIONAL = 4 << 21;
  35 const int32_t RECOMMENDED = 5 << 21;
  36
  37 // Bits 31..24: Single-bit flags
  38 const int32_t RTL = 1 << 24;
  39 const int32_t LB_LETTERS = 1 << 25;
  40 const int32_t CASED = 1 << 26;
  41
  42 const int32_t SCRIPT_PROPS[] = {
  43     // Begin copy-paste output from
  44     // tools/trunk/unicode/py/parsescriptmetadata.py
  45     0x0040 | RECOMMENDED,  // Zyyy
  46     0x0308 | RECOMMENDED,  // Zinh
  47     0x0628 | RECOMMENDED | RTL,  // Arab
  48     0x0531 | RECOMMENDED | CASED,  // Armn
  49     0x0995 | RECOMMENDED,  // Beng
  50     0x3105 | RECOMMENDED | LB_LETTERS,  // Bopo
  51     0x13C4 | LIMITED_USE | CASED,  // Cher
  52     0x03E2 | EXCLUSION | CASED,  // Copt
  53     0x042F | RECOMMENDED | CASED,  // Cyrl
  54     0x10414 | EXCLUSION | CASED,  // Dsrt
  55     0x0905 | RECOMMENDED,  // Deva
  56     0x12A0 | RECOMMENDED,  // Ethi
  57     0x10D3 | RECOMMENDED,  // Geor
  58     0x10330 | EXCLUSION,  // Goth
  59     0x03A9 | RECOMMENDED | CASED,  // Grek
  60     0x0A95 | RECOMMENDED,  // Gujr
  61     0x0A15 | RECOMMENDED,  // Guru
  62     0x5B57 | RECOMMENDED | LB_LETTERS,  // Hani
  63     0xAC00 | RECOMMENDED,  // Hang
  64     0x05D0 | RECOMMENDED | RTL,  // Hebr
  65     0x304B | RECOMMENDED | LB_LETTERS,  // Hira
  66     0x0C95 | RECOMMENDED,  // Knda
  67     0x30AB | RECOMMENDED | LB_LETTERS,  // Kana
  68     0x1780 | RECOMMENDED | LB_LETTERS,  // Khmr
  69     0x0EA5 | RECOMMENDED | LB_LETTERS,  // Laoo
  70     0x004C | RECOMMENDED | CASED,  // Latn
  71     0x0D15 | RECOMMENDED,  // Mlym
  72     0x1826 | ASPIRATIONAL,  // Mong
  73     0x1000 | RECOMMENDED | LB_LETTERS,  // Mymr
  74     0x168F | EXCLUSION,  // Ogam
  75     0x10308 | EXCLUSION,  // Ital
  76     0x0B15 | RECOMMENDED,  // Orya
  77     0x16A0 | EXCLUSION,  // Runr
  78     0x0D85 | RECOMMENDED,  // Sinh
  79     0x0710 | LIMITED_USE | RTL,  // Syrc
  80     0x0B95 | RECOMMENDED,  // Taml
  81     0x0C15 | RECOMMENDED,  // Telu
  82     0x078C | RECOMMENDED | RTL,  // Thaa
  83     0x0E17 | RECOMMENDED | LB_LETTERS,  // Thai
  84     0x0F40 | RECOMMENDED,  // Tibt
  85     0x14C0 | ASPIRATIONAL,  // Cans
  86     0xA288 | ASPIRATIONAL | LB_LETTERS,  // Yiii
  87     0x1703 | EXCLUSION,  // Tglg
  88     0x1723 | EXCLUSION,  // Hano
  89     0x1743 | EXCLUSION,  // Buhd
  90     0x1763 | EXCLUSION,  // Tagb
  91     0x280E | UNKNOWN,  // Brai
  92     0x10800 | EXCLUSION | RTL,  // Cprt
  93     0x1900 | LIMITED_USE,  // Limb
  94     0x10000 | EXCLUSION,  // Linb
  95     0x10480 | EXCLUSION,  // Osma
  96     0x10450 | EXCLUSION,  // Shaw
  97     0x1950 | LIMITED_USE | LB_LETTERS,  // Tale
  98     0x10380 | EXCLUSION,  // Ugar
  99     0,
 100     0x1A00 | EXCLUSION,  // Bugi
 101     0x2C00 | EXCLUSION | CASED,  // Glag
 102     0x10A00 | EXCLUSION | RTL,  // Khar
 103     0xA800 | LIMITED_USE,  // Sylo
 104     0x1980 | LIMITED_USE | LB_LETTERS,  // Talu
 105     0x2D5E | ASPIRATIONAL,  // Tfng
 106     0x103A0 | EXCLUSION,  // Xpeo
 107     0x1B05 | LIMITED_USE,  // Bali
 108     0x1BC0 | LIMITED_USE,  // Batk
 109     0,
 110     0x11005 | EXCLUSION,  // Brah
 111     0xAA00 | LIMITED_USE,  // Cham
 112     0,
 113     0,
 114     0,
 115     0,
 116     0x13153 | EXCLUSION,  // Egyp
 117     0,
 118     0x5B57 | RECOMMENDED | LB_LETTERS,  // Hans
 119     0x5B57 | RECOMMENDED | LB_LETTERS,  // Hant
 120     0x16B1C | EXCLUSION,  // Hmng
 121     0x10CA1 | EXCLUSION | RTL | CASED,  // Hung
 122     0,
 123     0xA984 | LIMITED_USE,  // Java
 124     0xA90A | LIMITED_USE,  // Kali
 125     0,
 126     0,
 127     0x1C00 | LIMITED_USE,  // Lepc
 128     0x10647 | EXCLUSION,  // Lina
 129     0x0840 | LIMITED_USE | RTL,  // Mand
 130     0,
 131     0x10980 | EXCLUSION | RTL,  // Mero
 132     0x07D8 | LIMITED_USE | RTL,  // Nkoo
 133     0x10C00 | EXCLUSION | RTL,  // Orkh
 134     0x1036B | EXCLUSION,  // Perm
 135     0xA840 | EXCLUSION,  // Phag
 136     0x10900 | EXCLUSION | RTL,  // Phnx
 137     0x16F00 | ASPIRATIONAL,  // Plrd
 138     0,
 139     0,
 140     0,
 141     0,
 142     0,
 143     0,
 144     0xA549 | LIMITED_USE,  // Vaii
 145     0,
 146     0x12000 | EXCLUSION,  // Xsux
 147     0,
 148     0xFDD0 | UNKNOWN,  // Zzzz
 149     0x102B7 | EXCLUSION,  // Cari
 150     0x304B | RECOMMENDED | LB_LETTERS,  // Jpan
 151     0x1A20 | LIMITED_USE | LB_LETTERS,  // Lana
 152     0x10280 | EXCLUSION,  // Lyci
 153     0x10920 | EXCLUSION | RTL,  // Lydi
 154     0x1C5A | LIMITED_USE,  // Olck
 155     0xA930 | EXCLUSION,  // Rjng
 156     0xA882 | LIMITED_USE,  // Saur
 157     0x1D850 | EXCLUSION,  // Sgnw
 158     0x1B83 | LIMITED_USE,  // Sund
 159     0,
 160     0xABC0 | LIMITED_USE,  // Mtei
 161     0x10840 | EXCLUSION | RTL,  // Armi
 162     0x10B00 | EXCLUSION | RTL,  // Avst
 163     0x11103 | LIMITED_USE,  // Cakm
 164     0xAC00 | RECOMMENDED,  // Kore
 165     0x11083 | EXCLUSION,  // Kthi
 166     0x10AC1 | EXCLUSION | RTL,  // Mani
 167     0x10B60 | EXCLUSION | RTL,  // Phli
 168     0x10B8F | EXCLUSION | RTL,  // Phlp
 169     0,
 170     0x10B40 | EXCLUSION | RTL,  // Prti
 171     0x0800 | EXCLUSION | RTL,  // Samr
 172     0xAA80 | LIMITED_USE | LB_LETTERS,  // Tavt
 173     0,
 174     0,
 175     0xA6A0 | LIMITED_USE,  // Bamu
 176     0xA4E8 | LIMITED_USE,  // Lisu
 177     0,
 178     0x10A60 | EXCLUSION | RTL,  // Sarb
 179     0x16AE6 | EXCLUSION,  // Bass
 180     0x1BC20 | EXCLUSION,  // Dupl
 181     0x10500 | EXCLUSION,  // Elba
 182     0x11315 | EXCLUSION,  // Gran
 183     0,
 184     0,
 185     0x1E802 | EXCLUSION | RTL,  // Mend
 186     0x109A0 | EXCLUSION | RTL,  // Merc
 187     0x10A95 | EXCLUSION | RTL,  // Narb
 188     0x10896 | EXCLUSION | RTL,  // Nbat
 189     0x10873 | EXCLUSION | RTL,  // Palm
 190     0x112BE | EXCLUSION,  // Sind
 191     0x118B4 | EXCLUSION | CASED,  // Wara
 192     0,
 193     0,
 194     0x16A4F | EXCLUSION,  // Mroo
 195     0,
 196     0x11183 | EXCLUSION,  // Shrd
 197     0x110D0 | EXCLUSION,  // Sora
 198     0x11680 | EXCLUSION,  // Takr
 199     0,
 200     0,
 201     0x14400 | EXCLUSION,  // Hluw
 202     0x11208 | EXCLUSION,  // Khoj
 203     0x11484 | EXCLUSION,  // Tirh
 204     0x10537 | EXCLUSION,  // Aghb
 205     0x11152 | EXCLUSION,  // Mahj
 206     0x11717 | EXCLUSION | LB_LETTERS,  // Ahom
 207     0x108F4 | EXCLUSION | RTL,  // Hatr
 208     0x1160E | EXCLUSION,  // Modi
 209     0x1128F | EXCLUSION,  // Mult
 210     0x11AC0 | EXCLUSION,  // Pauc
 211     0x1158E | EXCLUSION,  // Sidd
 212     // End copy-paste from parsescriptmetadata.py
 213 };
 214
 215 int32_t getScriptProps(UScriptCode script) {
 216     if (0 <= script && script < UPRV_LENGTHOF(SCRIPT_PROPS)) {
 217         return SCRIPT_PROPS[script];
 218     } else {
 219         return 0;
 220     }
 221 }
 222
 223 }  // namespace
 224
 225 U_CAPI int32_t U_EXPORT2
 226 uscript_getSampleString(UScriptCode script, UChar *dest, int32_t capacity, UErrorCode *pErrorCode) {
 227     if(U_FAILURE(*pErrorCode)) { return 0; }
 228     if(capacity < 0 || (capacity > 0 && dest == NULL)) {
 229         *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
 230         return 0;
 231     }
 232     int32_t sampleChar = getScriptProps(script) & 0x1fffff;
 233     int32_t length;
 234     if(sampleChar == 0) {
 235         length = 0;
 236     } else {
 237         length = U16_LENGTH(sampleChar);
 238         if(length <= capacity) {
 239             int32_t i = 0;
 240             U16_APPEND_UNSAFE(dest, i, sampleChar);
 241         }
 242     }
 243     return u_terminateUChars(dest, capacity, length, pErrorCode);
 244 }
 245
 246 U_COMMON_API icu::UnicodeString U_EXPORT2
 247 uscript_getSampleUnicodeString(UScriptCode script) {
 248     icu::UnicodeString sample;
 249     int32_t sampleChar = getScriptProps(script) & 0x1fffff;
 250     if(sampleChar != 0) {
 251         sample.append(sampleChar);
 252     }
 253     return sample;
 254 }
 255
 256 U_CAPI UScriptUsage U_EXPORT2
 257 uscript_getUsage(UScriptCode script) {
 258     return (UScriptUsage)((getScriptProps(script) >> 21) & 7);
 259 }
 260
 261 U_CAPI UBool U_EXPORT2
 262 uscript_isRightToLeft(UScriptCode script) {
 263     return (getScriptProps(script) & RTL) != 0;
 264 }
 265
 266 U_CAPI UBool U_EXPORT2
 267 uscript_breaksBetweenLetters(UScriptCode script) {
 268     return (getScriptProps(script) & LB_LETTERS) != 0;
 269 }
 270
 271 U_CAPI UBool U_EXPORT2
 272 uscript_isCased(UScriptCode script) {
 273     return (getScriptProps(script) & CASED) != 0;
 274 }