icuSources/common/uscript.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) 1997-2014, International Business Machines
   6 *   Corporation and others.  All Rights Reserved.
   7 **********************************************************************
   8 *
   9 * File USCRIPT.C
  10 *
  11 * Modification History:
  12 *
  13 *   Date        Name        Description
  14 *   07/06/2001    Ram         Creation.
  15 ******************************************************************************
  16 */
  17
  18 #include "unicode/uchar.h"
  19 #include "unicode/uscript.h"
  20 #include "unicode/uloc.h"
  21 #include "bytesinkutil.h"
  22 #include "charstr.h"
  23 #include "cmemory.h"
  24 #include "cstring.h"
  25 #include "ulocimp.h"
  26
  27 static const UScriptCode JAPANESE[3] = { USCRIPT_KATAKANA, USCRIPT_HIRAGANA, USCRIPT_HAN };
  28 static const UScriptCode KOREAN[2] = { USCRIPT_HANGUL, USCRIPT_HAN };
  29 static const UScriptCode HAN_BOPO[2] = { USCRIPT_HAN, USCRIPT_BOPOMOFO };
  30
  31 static int32_t
  32 setCodes(const UScriptCode *src, int32_t length,
  33          UScriptCode *dest, int32_t capacity, UErrorCode *err) {
  34     int32_t i;
  35     if(U_FAILURE(*err)) { return 0; }
  36     if(length > capacity) {
  37         *err = U_BUFFER_OVERFLOW_ERROR;
  38         return length;
  39     }
  40     for(i = 0; i < length; ++i) {
  41         dest[i] = src[i];
  42     }
  43     return length;
  44 }
  45
  46 static int32_t
  47 setOneCode(UScriptCode script, UScriptCode *scripts, int32_t capacity, UErrorCode *err) {
  48     if(U_FAILURE(*err)) { return 0; }
  49     if(1 > capacity) {
  50         *err = U_BUFFER_OVERFLOW_ERROR;
  51         return 1;
  52     }
  53     scripts[0] = script;
  54     return 1;
  55 }
  56
  57 static int32_t
  58 getCodesFromLocale(const char *locale,
  59                    UScriptCode *scripts, int32_t capacity, UErrorCode *err) {
  60     UErrorCode internalErrorCode = U_ZERO_ERROR;
  61     char lang[8];
  62     char script[8];
  63     int32_t scriptLength;
  64     if(U_FAILURE(*err)) { return 0; }
  65     // Multi-script languages, equivalent to the LocaleScript data
  66     // that we used to load from locale resource bundles.
  67     /*length = */ uloc_getLanguage(locale, lang, UPRV_LENGTHOF(lang), &internalErrorCode);
  68     if(U_FAILURE(internalErrorCode) || internalErrorCode == U_STRING_NOT_TERMINATED_WARNING) {
  69         return 0;
  70     }
  71     if(0 == uprv_strcmp(lang, "ja")) {
  72         return setCodes(JAPANESE, UPRV_LENGTHOF(JAPANESE), scripts, capacity, err);
  73     }
  74     if(0 == uprv_strcmp(lang, "ko")) {
  75         return setCodes(KOREAN, UPRV_LENGTHOF(KOREAN), scripts, capacity, err);
  76     }
  77     scriptLength = uloc_getScript(locale, script, UPRV_LENGTHOF(script), &internalErrorCode);
  78     if(U_FAILURE(internalErrorCode) || internalErrorCode == U_STRING_NOT_TERMINATED_WARNING) {
  79         return 0;
  80     }
  81     if(0 == uprv_strcmp(lang, "zh") && 0 == uprv_strcmp(script, "Hant")) {
  82         return setCodes(HAN_BOPO, UPRV_LENGTHOF(HAN_BOPO), scripts, capacity, err);
  83     }
  84     // Explicit script code.
  85     if(scriptLength != 0) {
  86         if (0 == uprv_strcmp(script, "Aran")) { // Apple <rdar://problem/47494884>
  87             uprv_strcpy(script, "Arab");
  88         }
  89         UScriptCode scriptCode = (UScriptCode)u_getPropertyValueEnum(UCHAR_SCRIPT, script);
  90         if(scriptCode != USCRIPT_INVALID_CODE) {
  91             if(scriptCode == USCRIPT_SIMPLIFIED_HAN || scriptCode == USCRIPT_TRADITIONAL_HAN) {
  92                 scriptCode = USCRIPT_HAN;
  93             }
  94             return setOneCode(scriptCode, scripts, capacity, err);
  95         }
  96     }
  97     return 0;
  98 }
  99
 100 /* TODO: this is a bad API and should be deprecated, ticket #11141 */
 101 U_CAPI int32_t  U_EXPORT2
 102 uscript_getCode(const char* nameOrAbbrOrLocale,
 103                 UScriptCode* fillIn,
 104                 int32_t capacity,
 105                 UErrorCode* err){
 106     UBool triedCode;
 107     UErrorCode internalErrorCode;
 108     int32_t length;
 109
 110     if(U_FAILURE(*err)) {
 111         return 0;
 112     }
 113     if(nameOrAbbrOrLocale==NULL ||
 114             (fillIn == NULL ? capacity != 0 : capacity < 0)) {
 115         *err = U_ILLEGAL_ARGUMENT_ERROR;
 116         return 0;
 117     }
 118     if (0 == uprv_strcmp(nameOrAbbrOrLocale, "Aran")) { // Apple <rdar://problem/47494884>
 119         nameOrAbbrOrLocale = "Arab";
 120     }
 121
 122     triedCode = FALSE;
 123     if(uprv_strchr(nameOrAbbrOrLocale, '-')==NULL && uprv_strchr(nameOrAbbrOrLocale, '_')==NULL ){
 124         /* try long and abbreviated script names first */
 125         UScriptCode code = (UScriptCode) u_getPropertyValueEnum(UCHAR_SCRIPT, nameOrAbbrOrLocale);
 126         if(code!=USCRIPT_INVALID_CODE) {
 127             return setOneCode(code, fillIn, capacity, err);
 128         }
 129         triedCode = TRUE;
 130     }
 131     internalErrorCode = U_ZERO_ERROR;
 132     length = getCodesFromLocale(nameOrAbbrOrLocale, fillIn, capacity, err);
 133     if(U_FAILURE(*err) || length != 0) {
 134         return length;
 135     }
 136     icu::CharString likely;
 137     {
 138         icu::CharStringByteSink sink(&likely);
 139         ulocimp_addLikelySubtags(nameOrAbbrOrLocale, sink, &internalErrorCode);
 140     }
 141     if(U_SUCCESS(internalErrorCode) && internalErrorCode != U_STRING_NOT_TERMINATED_WARNING) {
 142         length = getCodesFromLocale(likely.data(), fillIn, capacity, err);
 143         if(U_FAILURE(*err) || length != 0) {
 144             return length;
 145         }
 146     }
 147     if(!triedCode) {
 148         /* still not found .. try long and abbreviated script names again */
 149         UScriptCode code = (UScriptCode) u_getPropertyValueEnum(UCHAR_SCRIPT, nameOrAbbrOrLocale);
 150         if(code!=USCRIPT_INVALID_CODE) {
 151             return setOneCode(code, fillIn, capacity, err);
 152         }
 153     }
 154     return 0;
 155 }