Fix the handling of multivalued Unihan properties (#652)

unicode-org · Jan 18, 2024 · 77e3ce6 · 77e3ce6
1 parent b07cf2e
commit 77e3ce6
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 15 deletions.
diff --git a/unicodetools/src/main/java/org/unicode/props/UnicodeProperty.java b/unicodetools/src/main/java/org/unicode/props/UnicodeProperty.java
@@ -286,14 +286,21 @@ public List<String> getValueAliases(String valueAlias, List<String> result) {
         if (result == null) result = new ArrayList<>(1);
         result = _getValueAliases(valueAlias, result);
         if (!result.contains(valueAlias)) { // FIX && type < NUMERIC
-            result = _getValueAliases(valueAlias, result); // for debugging
-            throw new IllegalArgumentException(
-                    "Internal error: "
-                            + getName()
-                            + " doesn't contain "
-                            + valueAlias
-                            + ": "
-                            + new BagFormatter().join(result));
+            if (type == MISC) {
+                // Unihan has multivalued properties but does not use aliases.
+                result.add(valueAlias);
+            } else {
+                result = _getValueAliases(valueAlias, result); // for debugging
+                throw new IllegalArgumentException(
+                        "Internal error: "
+                                + getName()
+                                + " ("
+                                + getTypeName()
+                                + ") doesn't contain "
+                                + valueAlias
+                                + ": "
+                                + new BagFormatter().join(result));
+            }
         }
         return result;
     }

diff --git a/unicodetools/src/main/resources/org/unicode/text/UCD/UnicodeInvariantTest.txt b/unicodetools/src/main/resources/org/unicode/text/UCD/UnicodeInvariantTest.txt
@@ -892,14 +892,13 @@ Let $ideohack = [〆 〇 〡-〩]
 # 10.0 added Nushu
 # 13.0 added Khitan_Small_Script
 
-# Following don't work yet. Tested independently in TestInvariants
-# \p{Unified_Ideograph} ⊇ \P{kRSUnicode=∅}
+\p{Unified_Ideograph} ⊂ \p{kRSUnicode=/./}
+\p{kRSUnicode=/./} = [\p{Block=/^CJK.(Unified|Compatibility).Ideographs/} - \p{gc=Cn}]
+\p{kRSUnicode=/./} = \p{kTotalStrokes=/./}
 
-# \p{Unified_Ideograph} = \P{kRSUnicode=∅}
-
-# \P{kRSUnicode=∅} = \P{kTotalStrokes=∅}
-
-# \P{kHanyuPinyin=∅} ⊇ \P{kMandarin=∅}
+# TODO(eggrobin): Should those two have a kMandarin, or this not actually an invariant?
+# TODO(macchiati): The kHanyuPinyin UnicodeSet is excruciatingly slow.
+# \p{kHanyuPinyin=/./} - \p{kMandarin=/./} = [\{228F5}\x{2574C}]
 
 # InPC-InSC-gc invariants
 # See https://www.unicode.org/L2/L2023/23200-category-invariants.pdf.