From 132489e8477c55d3626ba88a53d74f67989c13a3 Mon Sep 17 00:00:00 2001
From: Markus Scherer <markus.icu@gmail.com>
Date: Mon, 20 Aug 2007 21:27:51 +0000
Subject: [PATCH] ICU-5609 add Hangul type boundaries for
 UCHAR_HANGUL_SYLLABLE_TYPE; fix UCharacter.getPropertyValueName() for lccc
 and tccc; replace test hacks and time bombs

X-SVN-Rev: 22429
---
 .../icu/dev/test/translit/UnicodeSetTest.java | 57 +++++++++++++------
 .../com/ibm/icu/impl/UCharacterProperty.java  |  9 +++
 icu4j/src/com/ibm/icu/lang/UCharacter.java    |  4 +-
 3 files changed, 53 insertions(+), 17 deletions(-)

diff --git a/icu4j/src/com/ibm/icu/dev/test/translit/UnicodeSetTest.java b/icu4j/src/com/ibm/icu/dev/test/translit/UnicodeSetTest.java
index 440cae1dae..5289cdeefa 100755
--- a/icu4j/src/com/ibm/icu/dev/test/translit/UnicodeSetTest.java
+++ b/icu4j/src/com/ibm/icu/dev/test/translit/UnicodeSetTest.java
@@ -44,13 +44,41 @@ public class UnicodeSetTest extends TestFmwk {
   public static void main(String[] args) throws Exception {
     new UnicodeSetTest().run(args);
   }
-  
-  
+
+  private static final boolean isCccValue(int ccc) {
+    switch (ccc) {
+    case 0:
+    case 1:
+    case 7:
+    case 8:
+    case 9:
+    case 200:
+    case 202:
+    case 216:
+    case 218:
+    case 220:
+    case 222:
+    case 224:
+    case 226:
+    case 228:
+    case 230:
+    case 232:
+    case 233:
+    case 234:
+    case 240:
+      return true;
+    default:
+      return false;
+    }
+  }
+
   public void TestPropertyAccess() {
     // test to see that all of the names work
     for (int propNum = UProperty.BINARY_START; propNum < UProperty.INT_LIMIT; ++propNum) {
+      if (propNum >= UProperty.BINARY_LIMIT && propNum < UProperty.INT_START) { // skip the gap
+        propNum = UProperty.INT_START;
+      }
       for (int nameChoice = UProperty.NameChoice.SHORT; nameChoice <= UProperty.NameChoice.LONG; ++nameChoice) {
-        if (propNum >= UProperty.BINARY_LIMIT && propNum < UProperty.INT_START) continue; // skip the gap
         String propName;
         try {
           propName = UCharacter.getPropertyName(propNum, nameChoice);
@@ -72,18 +100,18 @@ public class UnicodeSetTest extends TestFmwk {
             valueName = UCharacter.getPropertyValueName(propNum, valueNum, nameChoice);
             if (valueName == null) {
               if (nameChoice == UProperty.NameChoice.SHORT) continue; // allow non-existent short names
-              throw new NullPointerException();
+              if ((propNum == UProperty.CANONICAL_COMBINING_CLASS ||
+                   propNum == UProperty.LEAD_CANONICAL_COMBINING_CLASS ||
+                   propNum == UProperty.TRAIL_CANONICAL_COMBINING_CLASS) &&
+                  !isCccValue(valueNum)) {
+                // Only a few of the canonical combining classes have names.
+                // Otherwise they are just integer values.
+                continue;
+              } else {
+                throw new NullPointerException();
+              }
             }
           } catch (RuntimeException e1) {
-            // HACK for now
-            
-            if (propNum == 4098 && e1 instanceof NullPointerException) {
-              if (skipIfBeforeICU(3,8,0)) continue;
-            }
-            if (propNum == 4112 || propNum == 4113) {
-              if (skipIfBeforeICU(3,8,0)) continue;
-            }
-            
             errln("Can't get property value name for: "
                 + "Property (" + propNum + "): " + propName + ", " 
                 + "Value (" + valueNum + ") "
@@ -110,9 +138,6 @@ public class UnicodeSetTest extends TestFmwk {
             }
           }
           if (collectedErrors.size() != 0) {
-            if (propNum == 4107 && valueNum == 0 && skipIfBeforeICU(3,8,0)) {
-              continue; 
-            }
             errln("Property Value Differs: " 
                 + "Property (" + propNum + "): " + propName + ", " 
                 + "Value (" + valueNum + "): " + valueName + ", "
diff --git a/icu4j/src/com/ibm/icu/impl/UCharacterProperty.java b/icu4j/src/com/ibm/icu/impl/UCharacterProperty.java
index 0a650ff136..e9e225619f 100644
--- a/icu4j/src/com/ibm/icu/impl/UCharacterProperty.java
+++ b/icu4j/src/com/ibm/icu/impl/UCharacterProperty.java
@@ -21,6 +21,8 @@ import com.ibm.icu.text.UTF16;
 import com.ibm.icu.util.RangeValueIterator;
 import com.ibm.icu.util.VersionInfo;
 
+import com.ibm.icu.impl.NormalizerImpl;
+
 /**
 * <p>Internal class used for Unicode character property database.</p>
 * <p>This classes store binary data read from uprops.icu.
@@ -955,6 +957,13 @@ public final class UCharacterProperty
                 set.add(c);
             }
         }
+
+        /* Add Hangul type boundaries for UCHAR_HANGUL_SYLLABLE_TYPE. */
+        for(c=NormalizerImpl.HANGUL_BASE; c<(NormalizerImpl.HANGUL_BASE+NormalizerImpl.HANGUL_COUNT); c+=NormalizerImpl.JAMO_T_COUNT) {
+            set.add(c);
+            set.add(c+1);
+        }
+        set.add(c);
     }
 
     public UnicodeSet addPropertyStarts(UnicodeSet set) {
diff --git a/icu4j/src/com/ibm/icu/lang/UCharacter.java b/icu4j/src/com/ibm/icu/lang/UCharacter.java
index c1ea006f49..0376687b26 100755
--- a/icu4j/src/com/ibm/icu/lang/UCharacter.java
+++ b/icu4j/src/com/ibm/icu/lang/UCharacter.java
@@ -3978,7 +3978,9 @@ public final class UCharacter implements ECharacterCategory, ECharacterDirection
                                               int value,
                                               int nameChoice) 
     {
-        if (property == UProperty.CANONICAL_COMBINING_CLASS 
+        if ((property == UProperty.CANONICAL_COMBINING_CLASS
+             || property == UProperty.LEAD_CANONICAL_COMBINING_CLASS
+             || property == UProperty.TRAIL_CANONICAL_COMBINING_CLASS)
             && value >= UCharacter.getIntPropertyMinValue(
                               UProperty.CANONICAL_COMBINING_CLASS)
             && value <= UCharacter.getIntPropertyMaxValue(