ICU-2427 support property Hangul_Syllable_Type

X-SVN-Rev: 11272

icu4c/source/common/uchar.c

@@ -918,6 +918,9 @@ _enumPropertyStartsRange(const void *context, UChar32 start, UChar32 limit, uint
 
 U_CAPI void U_EXPORT2
 uchar_addPropertyStarts(USet *set) {
+    UChar32 c;
+    int32_t value, value2;
+
     if(!HAVE_DATA) {
         return;
     }
@@ -980,6 +983,38 @@ uchar_addPropertyStarts(USet *set) {
     uset_add(set, ZWNJ); /* range ZWNJ..ZWJ */
     uset_add(set, ZWJ+1);
 
+    /* add Jamo type boundaries for UCHAR_HANGUL_SYLLABLE_TYPE */
+    uset_add(set, 0x1100);
+    value=U_HST_LEADING_JAMO;
+    for(c=0x115a; c<=0x115f; ++c) {
+        value2=u_getIntPropertyValue(c, UCHAR_HANGUL_SYLLABLE_TYPE);
+        if(value!=value2) {
+            value=value2;
+            uset_add(set, c);
+        }
+    }
+
+    uset_add(set, 0x1160);
+    value=U_HST_VOWEL_JAMO;
+    for(c=0x11a3; c<=0x11a7; ++c) {
+        value2=u_getIntPropertyValue(c, UCHAR_HANGUL_SYLLABLE_TYPE);
+        if(value!=value2) {
+            value=value2;
+            uset_add(set, c);
+        }
+    }
+
+    uset_add(set, 0x11a8);
+    value=U_HST_TRAILING_JAMO;
+    for(c=0x11fa; c<=0x11ff; ++c) {
+        value2=u_getIntPropertyValue(c, UCHAR_HANGUL_SYLLABLE_TYPE);
+        if(value!=value2) {
+            value=value2;
+            uset_add(set, c);
+        }
+    }
+
+
     /*
      * Omit code points for u_charCellWidth() because
      * - it is deprecated and not a real Unicode property

icu4c/source/common/unicode/uchar.h

@@ -294,6 +294,9 @@ typedef enum UProperty {
     /** Enumerated property Script.
         Same as uscript_getScript, returns UScriptCode values. @draft ICU 2.2 */
     UCHAR_SCRIPT,
+    /** Enumerated property Hangul_Syllable_Type, new in Unicode 4.
+        Returns UHangulSyllableType values. @draft ICU 2.6 */
+    UCHAR_HANGUL_SYLLABLE_TYPE,
     /** One more than the last constant for enumerated/integer Unicode properties. @draft ICU 2.2 */
     UCHAR_INT_LIMIT,
 
@@ -1230,6 +1233,22 @@ typedef enum UNumericType {
     U_NT_COUNT
 } UNumericType;
 
+/**
+ * Hangul Syllable Type constants.
+ *
+ * @see UCHAR_HANGUL_SYLLABLE_TYPE
+ * @draft ICU 2.6
+ */
+typedef enum UHangulSyllableType {
+    U_HST_NOT_APPLICABLE,   /*[NA]*/ /*See note !!*/
+    U_HST_LEADING_JAMO,     /*[L]*/
+    U_HST_VOWEL_JAMO,       /*[V]*/
+    U_HST_TRAILING_JAMO,    /*[T]*/
+    U_HST_LV_SYLLABLE,      /*[LV]*/
+    U_HST_LVT_SYLLABLE,     /*[LVT]*/
+    U_HST_COUNT
+} UHangulSyllableType;
+
 /**
  * Check a binary Unicode property for a code point.
  *

icu4c/source/common/uprops.c

@@ -241,6 +241,35 @@ u_getIntPropertyValue(UChar32 c, UProperty which) {
         case UCHAR_SCRIPT:
             errorCode=U_ZERO_ERROR;
             return (int32_t)uscript_getScript(c, &errorCode);
+        case UCHAR_HANGUL_SYLLABLE_TYPE:
+            /* purely algorithmic; hardcode known characters, check for assigned new ones */
+            if(c<JAMO_L_BASE) {
+                /* NA */
+            } else if(c<=0x11ff) {
+                /* Jamo range */
+                if(c<=0x115f) {
+                    /* Jamo L range, HANGUL CHOSEONG ... */
+                    if(c==0x115f || c<=0x1159 || u_charType(c)==U_OTHER_LETTER) {
+                        return U_HST_LEADING_JAMO;
+                    }
+                } else if(c<=0x11a7) {
+                    /* Jamo V range, HANGUL JUNGSEONG ... */
+                    if(c<=0x11a2 || u_charType(c)==U_OTHER_LETTER) {
+                        return U_HST_VOWEL_JAMO;
+                    }
+                } else {
+                    /* Jamo T range */
+                    if(c<=0x11f9 || u_charType(c)==U_OTHER_LETTER) {
+                        return U_HST_TRAILING_JAMO;
+                    }
+                }
+            } else if((c-=HANGUL_BASE)<0) {
+                /* NA */
+            } else if(c<HANGUL_COUNT) {
+                /* Hangul syllable */
+                return c%JAMO_T_COUNT==0 ? U_HST_LV_SYLLABLE : U_HST_LVT_SYLLABLE;
+            }
+            return 0; /* NA */
         default:
             return 0; /* undefined */
         }
@@ -297,6 +326,8 @@ u_getIntPropertyMaxValue(UProperty which) {
         case UCHAR_SCRIPT:
             max=uprv_getMaxValues(0)&UPROPS_SCRIPT_MASK;
             return max!=0 ? max : (int32_t)USCRIPT_CODE_LIMIT-1;
+        case UCHAR_HANGUL_SYLLABLE_TYPE:
+            return (int32_t)U_HST_COUNT-1;
         default:
             return -1; /* undefined */
         }