Update UnicodeJsps to U13.0 and E13.1 (#44)

unicode-org · Jul 17, 2021 · 84a9500 · 84a9500
1 parent fdc77be
commit 84a9500
Show file tree

Hide file tree

Showing 22 changed files with 45,118 additions and 46 deletions.
diff --git a/UnicodeJsps/src/main/java/org/unicode/jsp/UnicodeJsp.java b/UnicodeJsps/src/main/java/org/unicode/jsp/UnicodeJsp.java
@@ -350,7 +350,7 @@ public static String getIdentifier(String script) {
 
     static final String VERSIONS = "Version 3.9; "
             + "ICU version: " + VersionInfo.ICU_VERSION.getVersionString(2, 2) + "; "
-            + "Unicode version: " + UCharacter.getUnicodeVersion().getVersionString(2, 2) + "; "
+            + "Unicode/Emoji version: " + UCharacter.getUnicodeVersion().getVersionString(2, 2) + "; "
             + (CachedProps.IS_BETA ? "Unicodeβ version: " + CachedProps.CACHED_PROPS.version.getVersionString(2, 2) + "; " : "");
 
     public static String getVersions() {

diff --git a/UnicodeJsps/src/main/java/org/unicode/jsp/UnicodeProperty.java b/UnicodeJsps/src/main/java/org/unicode/jsp/UnicodeProperty.java
@@ -69,7 +69,7 @@ public abstract class UnicodeProperty extends UnicodeLabel {
     .add(SAMPLE_PRIVATE_USE_AREA_A)
     .add(SAMPLE_PRIVATE_USE_AREA_B)
     .freeze();
-    
+
     public static final UnicodeSet STUFF_TO_TEST_WITH_UNASSIGNED = new UnicodeSet(STUFF_TO_TEST).addAll(UNASSIGNED).freeze();
 
     public static boolean DEBUG = false;
@@ -90,7 +90,7 @@ public abstract class UnicodeProperty extends UnicodeLabel {
 
     /*
      * Name: Unicode_1_Name Name: ISO_Comment Name: Name Name: Unicode_1_Name
-     * 
+     *
      */
 
     public static final int UNKNOWN = 0, BINARY = 2, EXTENDED_BINARY = 3,
@@ -269,6 +269,14 @@ public int getMaxWidth(boolean getShortest) {
         return maxValueWidth;
     }
 
+    public final UnicodeSet getTrueSet() {
+	if (!isType(BINARY)) {
+	    throw new IllegalArgumentException("Only applicable to binary (boolean) properties, not " + getName() +
+            " which is of type " + getTypeName());
+	}
+        return getSet("Yes", null);
+    }
+
     public final UnicodeSet getSet(String propertyValue) {
         return getSet(propertyValue, null);
     }
@@ -324,7 +332,7 @@ public UnicodeSet getSet(PatternMatcher matcher, UnicodeSet result) {
     /*
      * public UnicodeSet getMatchSet(UnicodeSet result) { if (result == null)
      * result = new UnicodeSet(); addAll(matchIterator, result); return result; }
-     * 
+     *
      * public void setMatchSet(UnicodeSet set) { matchIterator = new
      * UnicodeSetIterator(set); }
      */
@@ -453,7 +461,7 @@ protected UnicodeMap _getUnicodeMap() {
     }
 
     UnicodeSet sameValueRanges = null;
-    
+
     private UnicodeSet getSameValueRanges(boolean uniformUnassigned) {
         if (sameValueRanges == null) {
             sameValueRanges = new UnicodeSet();
@@ -508,7 +516,7 @@ public int compare(Object o1, Object o2) {
 
     /**
      * Utility for managing property & non-string value aliases
-     * 
+     *
      */
     // TODO optimize
     public static boolean equalNames(String a, String b) {
@@ -642,7 +650,7 @@ public static String regularize(String source, boolean titlecaseStart) {
     /**
      * Utility function for comparing codepoint to string without generating new
      * string.
-     * 
+     *
      * @param codepoint
      * @param other
      * @return true if the codepoint equals the string
@@ -670,7 +678,7 @@ public static final <T extends Object> boolean equals(T a, T b) {
 
     /**
      * Utility that should be on UnicodeSet
-     * 
+     *
      * @param source
      * @param result
      */
@@ -1140,7 +1148,7 @@ public static abstract class BaseProperty extends UnicodeProperty {
         private static final String[] YES_VALUES = {"Yes", "Y", "T", "True"};
 
         /**
-         * 
+         *
          */
         private static final String[][] YES_NO_ALIASES = new String[][] {YES_VALUES, NO_VALUES};
 
@@ -1344,7 +1352,7 @@ public UnicodeMapProperty set(UnicodeMap map) {
             unicodeMap = map.freeze();
             return this;
         }
-        
+
         @Override
         protected UnicodeMap _getUnicodeMap() {
             return unicodeMap;
@@ -1400,7 +1408,7 @@ public static UnicodeSet addUntested(UnicodeSet result, boolean uniformUnassigne
         if (uniformUnassigned && result.contains(UnicodeProperty.SAMPLE_UNASSIGNED)) {
             result.addAll(UnicodeProperty.UNASSIGNED);
         }
-        
+
         if (result.contains(UnicodeProperty.SAMPLE_HIGH_SURROGATE)) {
             result.addAll(UnicodeProperty.HIGH_SURROGATES);
         }
@@ -1410,7 +1418,7 @@ public static UnicodeSet addUntested(UnicodeSet result, boolean uniformUnassigne
         if (result.contains(UnicodeProperty.SAMPLE_LOW_SURROGATE)) {
             result.addAll(UnicodeProperty.LOW_SURROGATES);
         }
-        
+
         if (result.contains(UnicodeProperty.SAMPLE_PRIVATE_USE_AREA)) {
             result.addAll(UnicodeProperty.PRIVATE_USE_AREA);
         }
@@ -1458,7 +1466,7 @@ public boolean isDefault(int cp) {
             return equals(cp, value);
         }
         String defaultValue = getValue(SAMPLE_UNASSIGNED);
-        return defaultValue == null ? value == null : defaultValue.equals(value);   
+        return defaultValue == null ? value == null : defaultValue.equals(value);
     }
 
     public boolean hasUniformUnassigned() {

diff --git a/UnicodeJsps/src/main/java/org/unicode/jsp/UnicodeUtilities.java b/UnicodeJsps/src/main/java/org/unicode/jsp/UnicodeUtilities.java
@@ -350,13 +350,7 @@ public static void showSet(UnicodeSet inputSetRaw, CodePointShower codePointShow
                 out.append("</td></tr>");
             }
         } else if (codePointShower.abbreviate) {
-            if (codePointShower.doTable) {
-                out.append("<tr><td colSpan='4'>");
-            }
             codePointShower.showAbbreviated(inputSetRaw, out);
-            if (codePointShower.doTable) {
-                out.append("</td></tr>");
-            }
         } else {
             LinkedHashMap<String,UnicodeSet> items = new LinkedHashMap<String, UnicodeSet>();
             String specials = "Unassigned, Private use, or Surrogates";
@@ -392,7 +386,7 @@ public static void showSet(UnicodeSet inputSetRaw, CodePointShower codePointShow
             for (String newBlock : items.keySet()) {
                 UnicodeSet set = items.get(newBlock);
                 if (codePointShower.doTable) {
-                    out.append("<tr><td colSpan='4'>");
+                    out.append("<tr><td colSpan='3'>");
                 }
                 out.append("<h3>" + newBlock + "</b> <div class='ri'>items: " + numberFormat.format(set.size()) + "</div></h3>\n");
                 if (codePointShower.doTable) {
@@ -627,10 +621,15 @@ List<UnicodeProperty> getProps(String input) {
         private void showAbbreviated(UnicodeSet a, Appendable out) throws IOException {
             UnicodeUtilities.CodePointShower codePointShower = this;
 
+            boolean haveStrings = false;
             for (UnicodeSetIterator it = new UnicodeSetIterator(a); it.nextRange();) {
                 int s = it.codepoint;
                 if (s == UnicodeSetIterator.IS_STRING) {
-                    out.append(UnicodeUtilities.showCodePoint(it.string)).append("<br>\n");
+                    if (!haveStrings && codePointShower.doTable) {
+                	out.append("<tr><td colSpan='3'>");
+                	haveStrings = true;
+                    }
+                    out.append(UnicodeUtilities.showCodePoint(it.string)).append("\n");
                 } else {        
                     int end = it.codepointEnd;
                     if (end == s) {
@@ -655,6 +654,9 @@ private void showAbbreviated(UnicodeSet a, Appendable out) throws IOException {
                     }
                 }
             }
+            if (haveStrings && codePointShower.doTable) {
+        	out.append("</td></tr>");
+            }
         }
 
         String getPropString(List<UnicodeProperty> props, String codePoints, boolean shortName) {

diff --git a/UnicodeJsps/src/main/java/org/unicode/jsp/XPropertyFactory.java b/UnicodeJsps/src/main/java/org/unicode/jsp/XPropertyFactory.java
@@ -39,7 +39,7 @@ public static synchronized XPropertyFactory make() {
         singleton = new XPropertyFactory();
         return singleton;
     }
-    
+
     public final Factory add2(UnicodeProperty sp) {
         UnicodeProperty already = getProperty(sp.getName());
         if (already== null) {
@@ -49,8 +49,8 @@ public final Factory add2(UnicodeProperty sp) {
             return this;
         }
     }
-    
-    {   
+
+    {
         ICUPropertyFactory base = ICUPropertyFactory.make();
         for (String propertyAlias : (List<String>)base.getInternalAvailablePropertyAliases(new ArrayList())) {
             add(base.getProperty(propertyAlias));
@@ -150,23 +150,26 @@ public String transform(Integer source) {
         .setMain("Script_Extensions", "scx", UnicodeProperty.ENUMERATED, "1.1")
         .addValueAliases(ScriptTester.getScriptSpecialsAlternates(), false)
                 );
-        
+
         CachedProps cp = CachedProps.CACHED_PROPS;
         for (String prop : cp.getAvailable()) {
             add2(cp.getProperty(prop));
         }
-        UnicodeSet Basic_Emoji = cp.getProperty("Basic_Emoji").getSet("Yes");
-        UnicodeSet RGI_Emoji_Modifier_Sequence = cp.getProperty("RGI_Emoji_Modifier_Sequence").getSet("Yes");
-        UnicodeSet RGI_Emoji_Tag_Sequence = cp.getProperty("RGI_Emoji_Tag_Sequence").getSet("Yes");
-        UnicodeSet RGI_Emoji_Zwj_Sequence = cp.getProperty("RGI_Emoji_Zwj_Sequence").getSet("Yes");
+        UnicodeSet Basic_Emoji = cp.getProperty("Basic_Emoji").getSet("Yes", null); // TODO: was .getTrueSet();
+        UnicodeSet Emoji_Keycap_Sequence = cp.getProperty("RGI_Emoji_Keycap_Sequence").getSet("Yes", null); // TODO: was .getTrueSet();
+        UnicodeSet RGI_Emoji_Modifier_Sequence = cp.getProperty("RGI_Emoji_Modifier_Sequence").getSet("Yes", null); // TODO: was .getTrueSet();
+        UnicodeSet RGI_Emoji_Tag_Sequence = cp.getProperty("RGI_Emoji_Tag_Sequence").getSet("Yes", null); // TODO: was .getTrueSet();
+        UnicodeSet RGI_Emoji_Flag_Sequence = cp.getProperty("RGI_Emoji_Flag_Sequence").getSet("Yes", null); // TODO: was .getTrueSet();
+        UnicodeSet RGI_Emoji_Zwj_Sequence = cp.getProperty("RGI_Emoji_Zwj_Sequence").getSet("Yes", null); // TODO: was .getTrueSet();
         UnicodeSet RGI_Emoji = new UnicodeSet()
         	.add(Basic_Emoji)
+        	.add(Emoji_Keycap_Sequence)
         	.add(RGI_Emoji_Modifier_Sequence)
+        	.add(RGI_Emoji_Flag_Sequence)
         	.add(RGI_Emoji_Tag_Sequence)
         	.add(RGI_Emoji_Zwj_Sequence)
         	.freeze();
         add(new UnicodeSetProperty().set(RGI_Emoji).setMain("RGI_Emoji", "RGI_Emoji", UnicodeProperty.BINARY, "13.0"));
-
     }
 
     private void addCollationProperty() {
@@ -261,7 +264,7 @@ private void addBytes(StringBuilder builder, int bytes) {
                 builder.append(' ');
             }
             first = false;
-            builder.append(hex);    
+            builder.append(hex);
         }
     }
 

diff --git a/UnicodeJsps/src/main/resources/org/unicode/jsp/props/RGI_Emoji_Modifier_Sequence.bin b/UnicodeJsps/src/main/resources/org/unicode/jsp/props/RGI_Emoji_Modifier_Sequence.bin
diff --git a/UnicodeJsps/src/main/resources/org/unicode/jsp/props/RGI_Emoji_Zwj_Sequence.bin b/UnicodeJsps/src/main/resources/org/unicode/jsp/props/RGI_Emoji_Zwj_Sequence.bin