ICU-551.24.tar.gz

[apple/icu.git] / icuSources / common / uscript.c

diff --git a/icuSources/common/uscript.c b/icuSources/common/uscript.c
index 546d5f88e6b7d9ffc5c533e613fb45e30d2c6587..3700099cd2a70e43f5322040e84ad5d4246c91b6 100644 (file)
--- a/icuSources/common/uscript.c
+++ b/icuSources/common/uscript.c
@@ -1,6 +1,6 @@
 /*
 **********************************************************************
 /*
 **********************************************************************

-*   Copyright (C) 1997-2004, International Business Machines
+*   Copyright (C) 1997-2014, International Business Machines

 *   Corporation and others.  All Rights Reserved.
 **********************************************************************
 *
 *   Corporation and others.  All Rights Reserved.
 **********************************************************************
 *


@@ -13,91 +13,130 @@
 ******************************************************************************
 */
 
 ******************************************************************************
 */
 

-#include "unicode/uscript.h"
-#include "unicode/ures.h"

 #include "unicode/uchar.h"
 #include "unicode/uchar.h"

-#include "unicode/putil.h"
-#include "uprops.h"
+#include "unicode/uscript.h"
+#include "unicode/uloc.h"

 #include "cmemory.h"
 #include "cstring.h"
 
 #include "cmemory.h"
 #include "cstring.h"
 

-static const char kLocaleScript[] = "LocaleScript";
+static const UScriptCode JAPANESE[3] = { USCRIPT_KATAKANA, USCRIPT_HIRAGANA, USCRIPT_HAN };
+static const UScriptCode KOREAN[2] = { USCRIPT_HANGUL, USCRIPT_HAN };
+static const UScriptCode HAN_BOPO[2] = { USCRIPT_HAN, USCRIPT_BOPOMOFO };
+
+static int32_t
+setCodes(const UScriptCode *src, int32_t length,
+         UScriptCode *dest, int32_t capacity, UErrorCode *err) {
+    int32_t i;
+    if(U_FAILURE(*err)) { return 0; }
+    if(length > capacity) {
+        *err = U_BUFFER_OVERFLOW_ERROR;
+        return length;
+    }
+    for(i = 0; i < length; ++i) {
+        dest[i] = src[i];
+    }
+    return length;
+}
+
+static int32_t
+setOneCode(UScriptCode script, UScriptCode *scripts, int32_t capacity, UErrorCode *err) {
+    if(U_FAILURE(*err)) { return 0; }
+    if(1 > capacity) {
+        *err = U_BUFFER_OVERFLOW_ERROR;
+        return 1;
+    }
+    scripts[0] = script;
+    return 1;
+}
+
+static int32_t
+getCodesFromLocale(const char *locale,
+                   UScriptCode *scripts, int32_t capacity, UErrorCode *err) {
+    UErrorCode internalErrorCode = U_ZERO_ERROR;
+    char lang[8];
+    char script[8];
+    int32_t scriptLength;
+    if(U_FAILURE(*err)) { return 0; }
+    // Multi-script languages, equivalent to the LocaleScript data
+    // that we used to load from locale resource bundles.
+    /*length = */ uloc_getLanguage(locale, lang, UPRV_LENGTHOF(lang), &internalErrorCode);
+    if(U_FAILURE(internalErrorCode) || internalErrorCode == U_STRING_NOT_TERMINATED_WARNING) {
+        return 0;
+    }
+    if(0 == uprv_strcmp(lang, "ja")) {
+        return setCodes(JAPANESE, UPRV_LENGTHOF(JAPANESE), scripts, capacity, err);
+    }
+    if(0 == uprv_strcmp(lang, "ko")) {
+        return setCodes(KOREAN, UPRV_LENGTHOF(KOREAN), scripts, capacity, err);
+    }
+    scriptLength = uloc_getScript(locale, script, UPRV_LENGTHOF(script), &internalErrorCode);
+    if(U_FAILURE(internalErrorCode) || internalErrorCode == U_STRING_NOT_TERMINATED_WARNING) {
+        return 0;
+    }
+    if(0 == uprv_strcmp(lang, "zh") && 0 == uprv_strcmp(script, "Hant")) {
+        return setCodes(HAN_BOPO, UPRV_LENGTHOF(HAN_BOPO), scripts, capacity, err);
+    }
+    // Explicit script code.
+    if(scriptLength != 0) {
+        UScriptCode scriptCode = (UScriptCode)u_getPropertyValueEnum(UCHAR_SCRIPT, script);
+        if(scriptCode != USCRIPT_INVALID_CODE) {
+            if(scriptCode == USCRIPT_SIMPLIFIED_HAN || scriptCode == USCRIPT_TRADITIONAL_HAN) {
+                scriptCode = USCRIPT_HAN;
+            }
+            return setOneCode(scriptCode, scripts, capacity, err);
+        }
+    }
+    return 0;
+}

 
 

+/* TODO: this is a bad API and should be deprecated, ticket #11141 */

 U_CAPI int32_t  U_EXPORT2
 uscript_getCode(const char* nameOrAbbrOrLocale,
                 UScriptCode* fillIn,
                 int32_t capacity,
                 UErrorCode* err){
 U_CAPI int32_t  U_EXPORT2
 uscript_getCode(const char* nameOrAbbrOrLocale,
                 UScriptCode* fillIn,
                 int32_t capacity,
                 UErrorCode* err){

+    UBool triedCode;
+    char likely[ULOC_FULLNAME_CAPACITY];
+    UErrorCode internalErrorCode;
+    int32_t length;

 
 

-    UScriptCode code = USCRIPT_INVALID_CODE;
-    int32_t numFilled=0;
-    int32_t len=0;
-    /* check arguments */
-    if(err==NULL ||U_FAILURE(*err)){
-        return numFilled;
+    if(U_FAILURE(*err)) {
+        return 0;

     }
     }

-    if(nameOrAbbrOrLocale==NULL || fillIn == NULL || capacity<0){
+    if(nameOrAbbrOrLocale==NULL ||
+            (fillIn == NULL ? capacity != 0 : capacity < 0)) {

         *err = U_ILLEGAL_ARGUMENT_ERROR;
         *err = U_ILLEGAL_ARGUMENT_ERROR;

-        return numFilled;
+        return 0;

     }
 
     }
 

-    /* try long and abbreviated script names first */
-    code = (UScriptCode) u_getPropertyValueEnum(UCHAR_SCRIPT, nameOrAbbrOrLocale);
-
-    /* we still haven't found it try locale */
-    if(code==(UScriptCode)UCHAR_INVALID_CODE){
-        /* Do not propagate error codes from just not finding a locale bundle. */
-        UErrorCode localErrorCode = U_ZERO_ERROR;
-        UResourceBundle* resB = ures_open(NULL,nameOrAbbrOrLocale,&localErrorCode);
-        if(U_SUCCESS(localErrorCode)&& localErrorCode != U_USING_DEFAULT_WARNING){
-            UResourceBundle* resD = ures_getByKey(resB,kLocaleScript,NULL,&localErrorCode);
-            if(U_SUCCESS(localErrorCode) ){
-                len =0;
-                while(ures_hasNext(resD)){
-                    const UChar* name = ures_getNextString(resD,&len,NULL,&localErrorCode);
-                    if(U_SUCCESS(localErrorCode)){
-                        char cName[50] = {'\0'};
-                        u_UCharsToChars(name,cName,len);
-                        code = (UScriptCode) u_getPropertyValueEnum(UCHAR_SCRIPT, cName);
-                        /* got the script code now fill in the buffer */
-                        if(numFilled<=capacity){ 
-                            *(fillIn)++=code;
-                            numFilled++;
-                        }else{
-                            ures_close(resD);
-                            ures_close(resB);
-                            *err=U_BUFFER_OVERFLOW_ERROR;
-                            return len;
-                        }
-                    }
-                }
-            }
-            ures_close(resD);
-        
+    triedCode = FALSE;
+    if(uprv_strchr(nameOrAbbrOrLocale, '-')==NULL && uprv_strchr(nameOrAbbrOrLocale, '_')==NULL ){
+        /* try long and abbreviated script names first */
+        UScriptCode code = (UScriptCode) u_getPropertyValueEnum(UCHAR_SCRIPT, nameOrAbbrOrLocale);
+        if(code!=USCRIPT_INVALID_CODE) {
+            return setOneCode(code, fillIn, capacity, err);

         }
         }

-        ures_close(resB);
-    }else{
-        /* we found it */
-        if(numFilled<=capacity){ 
-            *(fillIn)++=code;
-            numFilled++;
-        }else{
-            *err=U_BUFFER_OVERFLOW_ERROR;
-            return len;
+        triedCode = TRUE;
+    }
+    internalErrorCode = U_ZERO_ERROR;
+    length = getCodesFromLocale(nameOrAbbrOrLocale, fillIn, capacity, err);
+    if(U_FAILURE(*err) || length != 0) {
+        return length;
+    }
+    (void)uloc_addLikelySubtags(nameOrAbbrOrLocale,
+                                likely, UPRV_LENGTHOF(likely), &internalErrorCode);
+    if(U_SUCCESS(internalErrorCode) && internalErrorCode != U_STRING_NOT_TERMINATED_WARNING) {
+        length = getCodesFromLocale(likely, fillIn, capacity, err);
+        if(U_FAILURE(*err) || length != 0) {
+            return length;

         }
     }
         }
     }

-    return numFilled;
-}
-
-U_CAPI const char*  U_EXPORT2
-uscript_getName(UScriptCode scriptCode){
-    return u_getPropertyValueName(UCHAR_SCRIPT, scriptCode,
-                                  U_LONG_PROPERTY_NAME);
-}
-
-U_CAPI const char*  U_EXPORT2
-uscript_getShortName(UScriptCode scriptCode){
-    return u_getPropertyValueName(UCHAR_SCRIPT, scriptCode,
-                                  U_SHORT_PROPERTY_NAME);
+    if(!triedCode) {
+        /* still not found .. try long and abbreviated script names again */
+        UScriptCode code = (UScriptCode) u_getPropertyValueEnum(UCHAR_SCRIPT, nameOrAbbrOrLocale);
+        if(code!=USCRIPT_INVALID_CODE) {
+            return setOneCode(code, fillIn, capacity, err);
+        }
+    }
+    return 0;

 }
 }

-