Make Math_Class multivalued to honestly show its multiple values of old. (Not changing the current draft to make it multivalued, though it sounds like we should).

Author: eggrobin

unicodetools/src/main/java/org/unicode/props/PropertyParsingInfo.java

@@ -1537,31 +1537,17 @@ private static void parseFields(
                     }
                 }
                 if ((propInfo.property == UcdProperty.Math_Entity_Name
-                                || propInfo.property == UcdProperty.Math_Entity_Set)
+                                || propInfo.property == UcdProperty.Math_Entity_Set
+                                || propInfo.property == UcdProperty.Math_Class_Ex)
                         && indexUnicodeProperties.ucdVersion.compareTo(Utility.UTR25_REVISION_16)
                                 < 0) {
-                    merger = IndexUnicodeProperties.MULTIVALUED_JOINER;
+                    merger = new PropertyUtilities.RedundancyIgnoringMultivaluedJoiner();
                 }
                 if (propInfo.property == UcdProperty.Math_Descriptive_Comments
                         && indexUnicodeProperties.ucdVersion.compareTo(Utility.UTR25_REVISION_16)
                                 < 0) {
                     merger = new PropertyUtilities.NullIgnorer();
                 }
-                if (propInfo.property == UcdProperty.Math_Class_Ex
-                        && indexUnicodeProperties.ucdVersion.compareTo(Utility.UTR25_REVISION_16)
-                                < 0) {
-                    merger = new PropertyUtilities.RedundancyIgnorer();
-                }
-                if (propInfo.property == UcdProperty.Math_Class_Ex
-                        && indexUnicodeProperties.ucdVersion.compareTo(VersionInfo.UNICODE_6_0) < 0
-                        && (line.getRange().start == 0x2020 || line.getRange().start == 0x2021)
-                        && line.getRange().end == line.getRange().start
-                        && value.equals("N")) {
-                    // MathClassEx-11 had conflicting assignments for these two characters.  Instead
-                    // of making Math_Class multivalued, keep the one that stayed (R), and discard
-                    // the N.
-                    value = "R";
-                }
                 if (propInfo.property == UcdProperty.Math_Class_Ex
                         && indexUnicodeProperties.ucdVersion.compareTo(VersionInfo.UNICODE_6_1) < 0
                         && value.isEmpty()) {
@@ -1630,6 +1616,7 @@ private static void parseSimpleFieldFile(
                                 propInfo.property, defaultValue, "hardcoded", false, version);
                     }
                 }
+                Merge<String> merger = null;
                 if (line.getParts().length == 3 && propInfo.property == UcdProperty.Block) {
                     // The old Blocks files had First; Last; Block.
                     IntRange range = new IntRange();
@@ -1709,21 +1696,7 @@ private static void parseSimpleFieldFile(
                     continue;
                 } else if (propInfo.property == UcdProperty.Math_Class
                         && version.compareTo(VersionInfo.UNICODE_6_0) < 0) {
-                    // MathClass-11 had conflicting assignments for these two characters.  Instead
-                    // of making Math_Class multivalued, keep the one that stayed (R), and discard
-                    // the N.
-                    if ((line.getRange().start == 0x2020 || line.getRange().start == 0x2021)
-                            && line.getRange().start == line.getRange().end
-                            && line.getParts()[1].equals("N")) {
-                        continue;
-                    }
-                    // MathClass-9 had the same problem for U+0021 ! as well.
-                    if (version.compareTo(VersionInfo.UNICODE_5_1) < 0
-                            && line.getRange().start == 0x0021
-                            && line.getRange().start == line.getRange().end
-                            && line.getParts()[1].equals("P")) {
-                        continue;
-                    }
+                    merger = new PropertyUtilities.RedundancyIgnoringMultivaluedJoiner();
                     // MathClass-11 had a line without a value, 21EA..21F3;
                     if (line.getParts()[1].isEmpty()) {
                         line.getParts()[1] = "None";
@@ -1739,7 +1712,7 @@ private static void parseSimpleFieldFile(
                         line.getMissingSet(),
                         line.getRange(),
                         line.getParts()[1],
-                        null,
+                        merger,
                         false,
                         nextVersion);
             } else {

unicodetools/src/main/java/org/unicode/props/PropertyUtilities.java

@@ -5,6 +5,8 @@
 import com.ibm.icu.text.UnicodeSet;
 import java.util.Collection;
 import java.util.Map;
+import java.util.Set;
+
 import org.unicode.text.utility.Utility;
 
 public class PropertyUtilities {
@@ -66,6 +68,20 @@ public String merge(String first, String second) {
         }
     }
 
+    public static final class RedundancyIgnoringMultivaluedJoiner implements Merge<String> {
+        public RedundancyIgnoringMultivaluedJoiner() {}
+
+        @Override
+        public String merge(String first, String second) {
+            final Set<String> oldValues = Set.of(first.split("\\|"));
+            if (second == null || oldValues.contains(second)) {
+                return first;
+            } else {
+                return first + "|" + second;
+            }
+        }
+    }
+
     static final <K, V, M extends Map<K, V>> M putNew(M map, K key, V value) {
         final V oldValue = map.get(key);
         if (oldValue != null) {

unicodetools/src/main/java/org/unicode/props/UcdProperty.java

@@ -735,13 +735,13 @@ public enum UcdProperty {
             PropertyType.Enumerated,
             DerivedPropertyStatus.NonUCDProperty,
             Math_Class_Values.class,
-            null,
+            ValueCardinality.Ordered,
             "Math_Class"),
     Math_Class_Ex(
             PropertyType.Enumerated,
             DerivedPropertyStatus.NonUCDNonProperty,
             Math_Class_Ex_Values.class,
-            null,
+            ValueCardinality.Ordered,
             "Math_Class_Ex"),
     NFC_Quick_Check(
             PropertyType.Enumerated,

unicodetools/src/main/resources/org/unicode/props/IndexPropertyRegex.txt

@@ -251,5 +251,7 @@ Confusable_MA ;               SINGLE_VALUED ;                $codePoints
 #Emoji_Modifier ;              SINGLE_VALUED ;                <enum>
 #Emoji_Modifier_Base ;         SINGLE_VALUED ;                <enum>
 
+Math_Class       ; ORDERED ; .*
+Math_Class_Ex    ; ORDERED ; .*
 Math_Entity_Name ; MULTI_VALUED ; .*
 Math_Entity_Set  ; MULTI_VALUED ; .*

unicodetools/src/main/resources/org/unicode/props/IndexUnicodeProperties.txt

@@ -547,9 +547,9 @@ math/*/MathClassEx ; Math_Entity_Name          ; 2
 math/*/MathClassEx ; Math_Entity_Set           ; 3
 math/*/MathClassEx ; Math_Descriptive_Comments ; 4
 # Field 2 used to be the character itself.
-math/*/MathClassEx ; Math_Entity_Name          ; 3 ; v16.0
-math/*/MathClassEx ; Math_Entity_Set           ; 4 ; v16.0
-math/*/MathClassEx ; Math_Descriptive_Comments ; 5 ; v16.0
+math/*/MathClassEx ; Math_Entity_Name          ; 3 ; v17.0
+math/*/MathClassEx ; Math_Entity_Set           ; 4 ; v17.0
+math/*/MathClassEx ; Math_Descriptive_Comments ; 5 ; v17.0
 
 FileType ; Unikemet ; PropertyValue
 Unikemet ; kEH_Cat