Fix the IndexUnicodeProperties Joining_Type (#657)

unicode-org · Jan 20, 2024 · fdc9c95 · fdc9c95
1 parent bf38a00
commit fdc9c95
Show file tree

Hide file tree

Showing 7 changed files with 64 additions and 32 deletions.
diff --git a/unicodetools/src/main/java/org/unicode/props/UcdProperty.java b/unicodetools/src/main/java/org/unicode/props/UcdProperty.java
@@ -29,6 +29,7 @@
 import org.unicode.props.UcdPropertyValues.NFKC_Quick_Check_Values;
 import org.unicode.props.UcdPropertyValues.NFKD_Quick_Check_Values;
 import org.unicode.props.UcdPropertyValues.Numeric_Type_Values;
+import org.unicode.props.UcdPropertyValues.Other_Joining_Type_Values;
 import org.unicode.props.UcdPropertyValues.Script_Values;
 import org.unicode.props.UcdPropertyValues.Sentence_Break_Values;
 import org.unicode.props.UcdPropertyValues.Vertical_Orientation_Values;
@@ -241,6 +242,8 @@ public enum UcdProperty {
     NFKC_Quick_Check(PropertyType.Enumerated, NFKC_Quick_Check_Values.class, null, "NFKC_QC"),
     NFKD_Quick_Check(PropertyType.Enumerated, NFKD_Quick_Check_Values.class, null, "NFKD_QC"),
     Numeric_Type(PropertyType.Enumerated, Numeric_Type_Values.class, null, "nt"),
+    Other_Joining_Type(
+            PropertyType.Enumerated, Other_Joining_Type_Values.class, null, "Other_Joining_Type"),
     Sentence_Break(PropertyType.Enumerated, Sentence_Break_Values.class, null, "SB"),
     Vertical_Orientation(PropertyType.Enumerated, Vertical_Orientation_Values.class, null, "vo"),
     Word_Break(PropertyType.Enumerated, Word_Break_Values.class, null, "WB"),

diff --git a/unicodetools/src/main/java/org/unicode/props/UcdPropertyValues.java b/unicodetools/src/main/java/org/unicode/props/UcdPropertyValues.java
@@ -1708,6 +1708,40 @@ public static Numeric_Type_Values forName(String name) {
     }
 
     // Numeric_Value
+    public enum Other_Joining_Type_Values implements Named {
+        Join_Causing("C"),
+        Dual_Joining("D"),
+        Left_Joining("L"),
+        Right_Joining("R"),
+        Transparent("T"),
+        Non_Joining("U"),
+        Deduce_From_General_Category("Deduce_From_General_Category");
+        private final PropertyNames<Other_Joining_Type_Values> names;
+
+        private Other_Joining_Type_Values(String shortName, String... otherNames) {
+            names =
+                    new PropertyNames<Other_Joining_Type_Values>(
+                            Other_Joining_Type_Values.class, this, shortName, otherNames);
+        }
+
+        @Override
+        public PropertyNames<Other_Joining_Type_Values> getNames() {
+            return names;
+        }
+
+        @Override
+        public String getShortName() {
+            return names.getShortName();
+        }
+
+        private static final NameMatcher<Other_Joining_Type_Values> NAME_MATCHER =
+                PropertyNames.getNameToEnums(Other_Joining_Type_Values.class);
+
+        public static Other_Joining_Type_Values forName(String name) {
+            return NAME_MATCHER.get(name);
+        }
+    }
+
     public enum Script_Values implements Named {
         Adlam("Adlm"),
         Caucasian_Albanian("Aghb"),

diff --git a/unicodetools/src/main/resources/org/unicode/props/ExtraPropertyAliases.txt b/unicodetools/src/main/resources/org/unicode/props/ExtraPropertyAliases.txt
@@ -27,6 +27,9 @@ idtype ; Identifier_Type
 idns ; Idn_Status
 idn8 ; Idn_2008
 
+# Unofficial contributory property used in the derivation of Joining_Type.
+Other_Joining_Type ; Other_Joining_Type
+
 # ================================================
 # String Properties
 # ================================================

diff --git a/unicodetools/src/main/resources/org/unicode/props/ExtraPropertyValueAliases.txt b/unicodetools/src/main/resources/org/unicode/props/ExtraPropertyValueAliases.txt
@@ -96,7 +96,6 @@
 # @missing: 0000..10FFFF; kTraditionalVariant ; <none>
 
 # @missing: 0000..10FFFF; Joining_Group ; No_Joining_Group
-# @missing: 0000..10FFFF; Joining_Type ; Non_Joining
 
 # Overrides for bugs
 
@@ -124,7 +123,6 @@ idn8 ; na	 ; na
 
 # @missing: 0000..10FFFF; Idn_Mapping  ; <code point>
 
-
 # @missing: 0000..10FFFF; Identifier_Status  	; r
 
 idstatus	; r	 ; Restricted			
@@ -162,3 +160,12 @@ sc ; Zxxx                             ; Unwritten
 
 # TODO: there is no Unicode 13.1, see https://github.com/unicode-org/unicodetools/issues/100
 age; 13.1                             ; V13_1
+
+# @missing: 0000..10FFFF; Other_Joining_Type ; Deduce_From_General_Category
+Other_Joining_Type ; C                                ; Join_Causing
+Other_Joining_Type ; D                                ; Dual_Joining
+Other_Joining_Type ; L                                ; Left_Joining
+Other_Joining_Type ; R                                ; Right_Joining
+Other_Joining_Type ; T                                ; Transparent
+Other_Joining_Type ; U                                ; Non_Joining
+Other_Joining_Type ; Deduce_From_General_Category     ; Deduce_From_General_Category
diff --git a/unicodetools/src/main/resources/org/unicode/props/IndexUnicodeProperties.txt b/unicodetools/src/main/resources/org/unicode/props/IndexUnicodeProperties.txt
@@ -92,7 +92,10 @@ UnicodeData; Simple_Lowercase_Mapping ; 13
 UnicodeData; Simple_Titlecase_Mapping ; 14
 UnicodeData; Unicode_1_Name ; 10
 UnicodeData; ISO_Comment ; 11
-ArabicShaping; Joining_Type; 2
+# Handle the complex default of ArabicShaping.txt by introducing an unofficial
+# contributory property, to be used when deriving Joining_Type.
+ArabicShaping; Other_Joining_Type; 2
+DerivedJoiningType; Joining_Type; 1
 ArabicShaping; Joining_Group; 3
 BidiMirroring; Bidi_Mirroring_Glyph;
 Blocks ; Block

diff --git a/unicodetools/src/main/resources/org/unicode/text/UCD/UnicodeInvariantTest.txt b/unicodetools/src/main/resources/org/unicode/text/UCD/UnicodeInvariantTest.txt
@@ -546,6 +546,17 @@ Let $nonAlphabeticDependentVowels = [\N{ORIYA SIGN OVERLINE}\N{THAI CHARACTER MA
 Let $nonDiacriticNuktas = [\u1BE6\U00010A38\U00010A39\U00010A3A\U0001133B]
 [\p{InSc=Nukta} - \p{Diacritic}] = $nonDiacriticNuktas
 
+## Joining_Type and Joining_Group
+# Where defined, the Joining_Group refines the Joining_Type.
+EquivalencesOf \P{Joining_Group=No_Joining_Group} Joining_Group ⇒ Joining_Type
+\p{gc=Mn} ⊆ \p{Joining_Type=Transparent}
+\p{gc=Me} ⊆ \p{Joining_Type=Transparent}
+
+# Derivation of Joining_Type from the second column of ArabicShaping.txt (unofficially Other_Joining_Type).
+In \P{Other_Joining_Type=Deduce_From_General_Category} Joining_Type = Other_Joining_Type
+[ \p{Other_Joining_Type=Deduce_From_General_Category} & [\p{gc=Mn}\p{gc=Me}\p{gc=Cf}] ] ⊆ \p{Joining_Type=Transparent}
+[ \p{Other_Joining_Type=Deduce_From_General_Category} - [\p{gc=Mn}\p{gc=Me}\p{gc=Cf}] ] ⊆ \p{Joining_Type=Non_Joining}
+
 ##########################
 # LineBreak property
 ##########################

diff --git a/unicodetools/src/test/java/org/unicode/propstest/TestProperties.java b/unicodetools/src/test/java/org/unicode/propstest/TestProperties.java
@@ -214,35 +214,6 @@ public void TestAAScripts() {
         }
     }
 
-    @Test
-    public void TestJoiningGroupConsistency() {
-        // TODO(egg): I would like to be able to put that in the invariants tests as « the partition
-        // defined by Joining_Group is finer than that defined by Joining_Type ».
-        UnicodeMap<String> joiningGroup = iup.load(UcdProperty.Joining_Group);
-        UnicodeMap<String> joiningType = iup.load(UcdProperty.Joining_Type);
-        var charactersByJoiningGroup = new HashMap<String, UnicodeSet>();
-        joiningGroup.addInverseTo(charactersByJoiningGroup).remove("No_Joining_Group");
-        charactersByJoiningGroup.forEach(
-                (group, set) -> {
-                    final int first = set.getRangeStart(0);
-                    final String firstType = joiningType.get(first);
-                    set.forEach(
-                            (c) -> {
-                                assertEquals(
-                                        "U+"
-                                                + getCodeAndName(Character.toString(first))
-                                                + "\nand\nU+"
-                                                + getCodeAndName(c)
-                                                + "\nhave different joining types but are in the"
-                                                + " same joining group ("
-                                                + group
-                                                + ")\n",
-                                        firstType,
-                                        joiningType.get(c));
-                            });
-                });
-    }
-
     @Test
     public void TestScripts() {