CLDR-17535 Update likely subtags data (#3966)

unicode-org · Aug 19, 2024 · 1a914d1 · 1a914d1
1 parent b4e6abf
commit 1a914d1
Show file tree

Hide file tree

Showing 12 changed files with 8,329 additions and 9,698 deletions.
diff --git a/common/properties/scriptMetadata.txt b/common/properties/scriptMetadata.txt
@@ -180,7 +180,7 @@ Onao; 33; 1E5D0; IN; 1; EXCLUSION; NO; NO; MIN; NO; NO  # provisional data for f
 Orkh; 33; 10C00; MN; 1; EXCLUSION; YES; NO; NO; NO; NO
 Osge; 33; 104B5; US; 1; LIMITED_USE; NO; NO; NO; NO; YES
 Osma; 33; 10480; SO; 1; EXCLUSION; NO; NO; NO; NO; NO
-Ougr; 33; 10F7C; 143; 1; EXCLUSION; YES; NO; YES; NO; NO
+Ougr; 33; 10F7C; CN; 1; EXCLUSION; YES; NO; YES; NO; NO
 Palm; 33; 10873; SY; 1; EXCLUSION; YES; NO; NO; NO; NO
 Pauc; 33; 11AC0; MM; 1; EXCLUSION; NO; NO; NO; NO; NO
 Perm; 33; 1036B; RU; 1; EXCLUSION; NO; NO; NO; NO; NO

diff --git a/common/supplemental/likelySubtags.xml b/common/supplemental/likelySubtags.xml
diff --git a/common/supplemental/supplementalData.xml b/common/supplemental/supplementalData.xml
@@ -4226,7 +4226,7 @@ XXX Code for transations where no currency is involved
 			<languagePopulation type="ku" populationPercent="5.5"/>	<!--Kurdish-->
 			<languagePopulation type="apc" populationPercent="5.2" references="R1173"/>	<!--Levantine Arabic-->
 			<languagePopulation type="zza" populationPercent="1.4"/>	<!--Zaza-->
-			<languagePopulation type="kaa" populationPercent="1" references="R1199"/>	<!--Kara-Kalpak-->
+			<languagePopulation type="kaa" populationPercent="0.1" references="R1199"/>	<!--Kara-Kalpak-->
 			<languagePopulation type="kbd" populationPercent="0.77"/>	<!--Kabardian-->
 			<languagePopulation type="az" populationPercent="0.74"/>	<!--Azerbaijani-->
 			<languagePopulation type="az_Arab" populationPercent="0.65"/>	<!--Azerbaijani (Arabic)-->

diff --git a/tools/cldr-code/src/main/java/org/unicode/cldr/draft/ScriptMetadata.java b/tools/cldr-code/src/main/java/org/unicode/cldr/draft/ScriptMetadata.java
@@ -23,6 +23,9 @@
 import org.unicode.cldr.util.Containment;
 import org.unicode.cldr.util.SemiFileReader;
 import org.unicode.cldr.util.StandardCodes;
+import org.unicode.cldr.util.StandardCodes.LstrType;
+import org.unicode.cldr.util.Validity;
+import org.unicode.cldr.util.Validity.Status;
 import org.unicode.cldr.util.With;
 
 public class ScriptMetadata {
@@ -139,6 +142,12 @@ public static void addNameToCode(String type, Map<String, String> hashMap) {
 
     public static final class SkipNewUnicodeException extends ICUException {}
 
+    /**
+     * Scripts that either have no known languages as yet (Cpmn) or are used for any language
+     * (Brai).
+     */
+    public static final Set<String> SCRIPTS_WITH_NO_LANGUAGES = Set.of("Brai", "Cpmn");
+
     public static class Info implements Comparable<Info> {
         public final int rank;
         public final VersionInfo age;
@@ -173,6 +182,7 @@ private Info(String[] items) {
             ime = trinaryLookup.forString(Column.IME.getItem(items));
             hasCase = trinaryLookup.forString(Column.HAS_CASE.getItem(items));
             density = Column.DENSITY.getInt(items, -1);
+            String script = items[2];
 
             final String countryRaw = Column.ORIGIN_COUNTRY.getItem(items);
             String country = CountryCodeConverter.getCodeFromName(countryRaw, false);
@@ -191,6 +201,39 @@ private Info(String[] items) {
                 langCode = null;
             }
             likelyLanguage = langCode == null ? "und" : langCode;
+
+            // check for bad countries, bad languages
+
+            final Status scriptStatus =
+                    Validity.getInstance().getCodeToStatus(LstrType.script).get(script);
+            if (!(scriptStatus == Status.special || scriptStatus == Status.unknown)) {
+                final Status countryStatus =
+                        Validity.getInstance().getCodeToStatus(LstrType.region).get(originCountry);
+                if (countryStatus != Status.regular) {
+                    errors.add(
+                            "ScriptMetadata.java: the country ("
+                                    + originCountry
+                                    + ") for "
+                                    + script
+                                    + " is not valid: "
+                                    + countryStatus);
+                }
+                final Status languageStatus =
+                        Validity.getInstance()
+                                .getCodeToStatus(LstrType.language)
+                                .get(likelyLanguage);
+                if (languageStatus != Status.regular
+                        // make exception for scripts that has no known languages
+                        && !SCRIPTS_WITH_NO_LANGUAGES.contains(script)) {
+                    errors.add(
+                            "ScriptMetadata.java: the likely language ("
+                                    + likelyLanguage
+                                    + ") for "
+                                    + script
+                                    + " is not valid: "
+                                    + languageStatus);
+                }
+            }
         }
 
         public Info(Info other, String string, String sampleCharacter) {