1 /*
2 **********************************************************************
3 *   Copyright (C) 1997-2014, International Business Machines
4 *   Corporation and others.  All Rights Reserved.
5 **********************************************************************
6 *
7 * File USCRIPT.C
8 *
9 * Modification History:
10 *
11 *   Date        Name        Description
12 *   07/06/2001    Ram         Creation.
13 ******************************************************************************
14 */
15 
16 #include "unicode/uchar.h"
17 #include "unicode/uscript.h"
18 #include "unicode/uloc.h"
19 #include "cmemory.h"
20 #include "cstring.h"
21 
22 static const UScriptCode JAPANESE[3] = { USCRIPT_KATAKANA, USCRIPT_HIRAGANA, USCRIPT_HAN };
23 static const UScriptCode KOREAN[2] = { USCRIPT_HANGUL, USCRIPT_HAN };
24 static const UScriptCode HAN_BOPO[2] = { USCRIPT_HAN, USCRIPT_BOPOMOFO };
25 
26 static int32_t
setCodes(const UScriptCode * src,int32_t length,UScriptCode * dest,int32_t capacity,UErrorCode * err)27 setCodes(const UScriptCode *src, int32_t length,
28          UScriptCode *dest, int32_t capacity, UErrorCode *err) {
29     int32_t i;
30     if(U_FAILURE(*err)) { return 0; }
31     if(length > capacity) {
32         *err = U_BUFFER_OVERFLOW_ERROR;
33         return length;
34     }
35     for(i = 0; i < length; ++i) {
36         dest[i] = src[i];
37     }
38     return length;
39 }
40 
41 static int32_t
setOneCode(UScriptCode script,UScriptCode * scripts,int32_t capacity,UErrorCode * err)42 setOneCode(UScriptCode script, UScriptCode *scripts, int32_t capacity, UErrorCode *err) {
43     if(U_FAILURE(*err)) { return 0; }
44     if(1 > capacity) {
45         *err = U_BUFFER_OVERFLOW_ERROR;
46         return 1;
47     }
48     scripts[0] = script;
49     return 1;
50 }
51 
52 static int32_t
getCodesFromLocale(const char * locale,UScriptCode * scripts,int32_t capacity,UErrorCode * err)53 getCodesFromLocale(const char *locale,
54                    UScriptCode *scripts, int32_t capacity, UErrorCode *err) {
55     UErrorCode internalErrorCode = U_ZERO_ERROR;
56     char lang[8];
57     char script[8];
58     int32_t scriptLength;
59     if(U_FAILURE(*err)) { return 0; }
60     // Multi-script languages, equivalent to the LocaleScript data
61     // that we used to load from locale resource bundles.
62     /*length = */ uloc_getLanguage(locale, lang, UPRV_LENGTHOF(lang), &internalErrorCode);
63     if(U_FAILURE(internalErrorCode) || internalErrorCode == U_STRING_NOT_TERMINATED_WARNING) {
64         return 0;
65     }
66     if(0 == uprv_strcmp(lang, "ja")) {
67         return setCodes(JAPANESE, UPRV_LENGTHOF(JAPANESE), scripts, capacity, err);
68     }
69     if(0 == uprv_strcmp(lang, "ko")) {
70         return setCodes(KOREAN, UPRV_LENGTHOF(KOREAN), scripts, capacity, err);
71     }
72     scriptLength = uloc_getScript(locale, script, UPRV_LENGTHOF(script), &internalErrorCode);
73     if(U_FAILURE(internalErrorCode) || internalErrorCode == U_STRING_NOT_TERMINATED_WARNING) {
74         return 0;
75     }
76     if(0 == uprv_strcmp(lang, "zh") && 0 == uprv_strcmp(script, "Hant")) {
77         return setCodes(HAN_BOPO, UPRV_LENGTHOF(HAN_BOPO), scripts, capacity, err);
78     }
79     // Explicit script code.
80     if(scriptLength != 0) {
81         UScriptCode scriptCode = (UScriptCode)u_getPropertyValueEnum(UCHAR_SCRIPT, script);
82         if(scriptCode != USCRIPT_INVALID_CODE) {
83             if(scriptCode == USCRIPT_SIMPLIFIED_HAN || scriptCode == USCRIPT_TRADITIONAL_HAN) {
84                 scriptCode = USCRIPT_HAN;
85             }
86             return setOneCode(scriptCode, scripts, capacity, err);
87         }
88     }
89     return 0;
90 }
91 
92 /* TODO: this is a bad API and should be deprecated, ticket #11141 */
93 U_CAPI int32_t  U_EXPORT2
uscript_getCode(const char * nameOrAbbrOrLocale,UScriptCode * fillIn,int32_t capacity,UErrorCode * err)94 uscript_getCode(const char* nameOrAbbrOrLocale,
95                 UScriptCode* fillIn,
96                 int32_t capacity,
97                 UErrorCode* err){
98     UBool triedCode;
99     char likely[ULOC_FULLNAME_CAPACITY];
100     UErrorCode internalErrorCode;
101     int32_t length;
102 
103     if(U_FAILURE(*err)) {
104         return 0;
105     }
106     if(nameOrAbbrOrLocale==NULL ||
107             (fillIn == NULL ? capacity != 0 : capacity < 0)) {
108         *err = U_ILLEGAL_ARGUMENT_ERROR;
109         return 0;
110     }
111 
112     triedCode = FALSE;
113     if(uprv_strchr(nameOrAbbrOrLocale, '-')==NULL && uprv_strchr(nameOrAbbrOrLocale, '_')==NULL ){
114         /* try long and abbreviated script names first */
115         UScriptCode code = (UScriptCode) u_getPropertyValueEnum(UCHAR_SCRIPT, nameOrAbbrOrLocale);
116         if(code!=USCRIPT_INVALID_CODE) {
117             return setOneCode(code, fillIn, capacity, err);
118         }
119         triedCode = TRUE;
120     }
121     internalErrorCode = U_ZERO_ERROR;
122     length = getCodesFromLocale(nameOrAbbrOrLocale, fillIn, capacity, err);
123     if(U_FAILURE(*err) || length != 0) {
124         return length;
125     }
126     (void)uloc_addLikelySubtags(nameOrAbbrOrLocale,
127                                 likely, UPRV_LENGTHOF(likely), &internalErrorCode);
128     if(U_SUCCESS(internalErrorCode) && internalErrorCode != U_STRING_NOT_TERMINATED_WARNING) {
129         length = getCodesFromLocale(likely, fillIn, capacity, err);
130         if(U_FAILURE(*err) || length != 0) {
131             return length;
132         }
133     }
134     if(!triedCode) {
135         /* still not found .. try long and abbreviated script names again */
136         UScriptCode code = (UScriptCode) u_getPropertyValueEnum(UCHAR_SCRIPT, nameOrAbbrOrLocale);
137         if(code!=USCRIPT_INVALID_CODE) {
138             return setOneCode(code, fillIn, capacity, err);
139         }
140     }
141     return 0;
142 }
143