icuSources/common/uscript_props.cpp

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 *   Copyright (C) 2013-2016, International Business Machines
   6 *   Corporation and others.  All Rights Reserved.
   7 *******************************************************************************
   8 *   file name:  uscript_props.cpp
   9 *   encoding:   UTF-8
  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"
  22 #include "cmemory.h"
  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;
  36 // st int32_t ASPIRATIONAL = 4 << 21; -- not used any more since Unicode 10
  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
  47     0x0040 | RECOMMENDED,  // Zyyy
  48     0x0308 | RECOMMENDED,  // Zinh
  49     0x0628 | RECOMMENDED | RTL,  // Arab
  50     0x0531 | RECOMMENDED | CASED,  // Armn
  51     0x0995 | RECOMMENDED,  // Beng
  52     0x3105 | RECOMMENDED | LB_LETTERS,  // Bopo
  53     0x13C4 | LIMITED_USE | CASED,  // Cher
  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
  74     0x1826 | EXCLUSION,  // Mong
  75     0x1000 | RECOMMENDED | LB_LETTERS,  // Mymr
  76     0x168F | EXCLUSION,  // Ogam
  77     0x10300 | EXCLUSION,  // Ital
  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
  87     0x14C0 | LIMITED_USE,  // Cans
  88     0xA288 | LIMITED_USE | LB_LETTERS,  // Yiii
  89     0x1703 | EXCLUSION,  // Tglg
  90     0x1723 | EXCLUSION,  // Hano
  91     0x1743 | EXCLUSION,  // Buhd
  92     0x1763 | EXCLUSION,  // Tagb
  93     0x280E | UNKNOWN,  // Brai
  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
 107     0x2D30 | LIMITED_USE,  // Tfng
 108     0x103A0 | EXCLUSION,  // Xpeo
 109     0x1B05 | LIMITED_USE,  // Bali
 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
 122     0x16B1C | EXCLUSION,  // Hmng
 123     0x10CA1 | EXCLUSION | RTL | CASED,  // Hung
 124     0,
 125     0xA984 | LIMITED_USE,  // Java
 126     0xA90A | LIMITED_USE,  // Kali
 127     0,
 128     0,
 129     0x1C00 | LIMITED_USE,  // Lepc
 130     0x10647 | EXCLUSION,  // Lina
 131     0x0840 | LIMITED_USE | RTL,  // Mand
 132     0,
 133     0x10980 | EXCLUSION | RTL,  // Mero
 134     0x07CA | LIMITED_USE | RTL,  // Nkoo
 135     0x10C00 | EXCLUSION | RTL,  // Orkh
 136     0x1036B | EXCLUSION,  // Perm
 137     0xA840 | EXCLUSION,  // Phag
 138     0x10900 | EXCLUSION | RTL,  // Phnx
 139     0x16F00 | LIMITED_USE,  // Plrd
 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
 151     0x102A0 | EXCLUSION,  // Cari
 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
 159     0x1D850 | EXCLUSION,  // Sgnw
 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
 168     0x10AD8 | EXCLUSION | RTL,  // Mani
 169     0x10B60 | EXCLUSION | RTL,  // Phli
 170     0x10B8F | EXCLUSION | RTL,  // Phlp
 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
 178     0xA4D0 | LIMITED_USE,  // Lisu
 179     0,
 180     0x10A60 | EXCLUSION | RTL,  // Sarb
 181     0x16AE6 | EXCLUSION,  // Bass
 182     0x1BC20 | EXCLUSION,  // Dupl
 183     0x10500 | EXCLUSION,  // Elba
 184     0x11315 | EXCLUSION,  // Gran
 185     0,
 186     0,
 187     0x1E802 | EXCLUSION | RTL,  // Mend
 188     0x109A0 | EXCLUSION | RTL,  // Merc
 189     0x10A95 | EXCLUSION | RTL,  // Narb
 190     0x10896 | EXCLUSION | RTL,  // Nbat
 191     0x10873 | EXCLUSION | RTL,  // Palm
 192     0x112BE | EXCLUSION,  // Sind
 193     0x118B4 | EXCLUSION | CASED,  // Wara
 194     0,
 195     0,
 196     0x16A4F | EXCLUSION,  // Mroo
 197     0x1B1C4 | EXCLUSION | LB_LETTERS,  // Nshu
 198     0x11183 | EXCLUSION,  // Shrd
 199     0x110D0 | EXCLUSION,  // Sora
 200     0x11680 | EXCLUSION,  // Takr
 201     0x18229 | EXCLUSION | LB_LETTERS,  // Tang
 202     0,
 203     0x14400 | EXCLUSION,  // Hluw
 204     0x11208 | EXCLUSION,  // Khoj
 205     0x11484 | EXCLUSION,  // Tirh
 206     0x10537 | EXCLUSION,  // Aghb
 207     0x11152 | EXCLUSION,  // Mahj
 208     0x11717 | EXCLUSION | LB_LETTERS,  // Ahom
 209     0x108F4 | EXCLUSION | RTL,  // Hatr
 210     0x1160E | EXCLUSION,  // Modi
 211     0x1128F | EXCLUSION,  // Mult
 212     0x11AC0 | EXCLUSION,  // Pauc
 213     0x1158E | EXCLUSION,  // Sidd
 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,
 222     0x11D10 | EXCLUSION,  // Gonm
 223     0x11A5C | EXCLUSION,  // Soyo
 224     0x11A0B | EXCLUSION,  // Zanb
 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
 232     0x10FF1 | EXCLUSION | RTL,  // Elym
 233     0x1E108 | LIMITED_USE,  // Hmnp
 234     0x119CE | EXCLUSION,  // Nand
 235     0x1E2E1 | LIMITED_USE,  // Wcho
 236     0x10FBF | EXCLUSION | RTL,  // Chrs
 237     0x1190C | EXCLUSION,  // Diak
 238     0x18C65 | EXCLUSION | LB_LETTERS,  // Kits
 239     0x10E88 | EXCLUSION | RTL,  // Yezi
 240     // End copy-paste from parsescriptmetadata.py
 241 };
 242
 243 int32_t getScriptProps(UScriptCode script) {
 244     if (0 <= script && script < UPRV_LENGTHOF(SCRIPT_PROPS)) {
 245         return SCRIPT_PROPS[script];
 246     } else {
 247         return 0;
 248     }
 249 }
 250
 251 }  // namespace
 252
 253 U_CAPI int32_t U_EXPORT2
 254 uscript_getSampleString(UScriptCode script, UChar *dest, int32_t capacity, UErrorCode *pErrorCode) {
 255     if(U_FAILURE(*pErrorCode)) { return 0; }
 256     if(capacity < 0 || (capacity > 0 && dest == NULL)) {
 257         *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
 258         return 0;
 259     }
 260     int32_t sampleChar = getScriptProps(script) & 0x1fffff;
 261     int32_t length;
 262     if(sampleChar == 0) {
 263         length = 0;
 264     } else {
 265         length = U16_LENGTH(sampleChar);
 266         if(length <= capacity) {
 267             int32_t i = 0;
 268             U16_APPEND_UNSAFE(dest, i, sampleChar);
 269         }
 270     }
 271     return u_terminateUChars(dest, capacity, length, pErrorCode);
 272 }
 273
 274 U_COMMON_API icu::UnicodeString U_EXPORT2
 275 uscript_getSampleUnicodeString(UScriptCode script) {
 276     icu::UnicodeString sample;
 277     int32_t sampleChar = getScriptProps(script) & 0x1fffff;
 278     if(sampleChar != 0) {
 279         sample.append(sampleChar);
 280     }
 281     return sample;
 282 }
 283
 284 U_CAPI UScriptUsage U_EXPORT2
 285 uscript_getUsage(UScriptCode script) {
 286     return (UScriptUsage)((getScriptProps(script) >> 21) & 7);
 287 }
 288
 289 U_CAPI UBool U_EXPORT2
 290 uscript_isRightToLeft(UScriptCode script) {
 291     return (getScriptProps(script) & RTL) != 0;
 292 }
 293
 294 U_CAPI UBool U_EXPORT2
 295 uscript_breaksBetweenLetters(UScriptCode script) {
 296     return (getScriptProps(script) & LB_LETTERS) != 0;
 297 }
 298
 299 U_CAPI UBool U_EXPORT2
 300 uscript_isCased(UScriptCode script) {
 301     return (getScriptProps(script) & CASED) != 0;
 302 }