CLDR-17367 v45 JSON: improve localeRules processing (#3611)

(cherry picked from commit fc51e78)
unicode-org · Apr 22, 2024 · bfb4982 · bfb4982
1 parent d6fdabd
commit bfb4982
Show file tree

Hide file tree

Showing 2 changed files with 31 additions and 2 deletions.
diff --git a/tools/cldr-code/src/main/java/org/unicode/cldr/json/Ldml2JsonConverter.java b/tools/cldr-code/src/main/java/org/unicode/cldr/json/Ldml2JsonConverter.java
@@ -1855,7 +1855,30 @@ private JsonElement startNonleafNode(JsonElement out, final CldrNode node) throw
                     o.getAsJsonObject().addProperty(attrAsKey, v);
                 } // else, omit
             } else {
-                o.getAsJsonObject().addProperty(attrAsKey, value);
+                // hack for localeRules
+                if (attrAsKey.equals("_localeRules")) {
+                    // find the _localeRules object, add if it didn't exist
+                    JsonElement localeRules = out.getAsJsonObject().get(attrAsKey);
+                    if (localeRules == null) {
+                        localeRules = new JsonObject();
+                        out.getAsJsonObject().add(attrAsKey, localeRules);
+                    }
+                    // find the sibling object, add if it did't exist ( this will be parentLocale or
+                    // collations etc.)
+                    JsonElement sibling = localeRules.getAsJsonObject().get(name);
+                    if (sibling == null) {
+                        sibling = new JsonObject();
+                        localeRules.getAsJsonObject().add(name, sibling);
+                    }
+                    // get the 'parent' attribute, which wil be the value
+                    final String parent =
+                            XPathParts.getFrozenInstance(node.getUntransformedPath())
+                                    .getAttributeValue(-1, "parent");
+                    // finally, we add something like "nonLikelyScript: und"
+                    sibling.getAsJsonObject().addProperty(value, parent);
+                } else {
+                    o.getAsJsonObject().addProperty(attrAsKey, value);
+                }
             }
         }
         return o;
@@ -2312,7 +2335,9 @@ private void writeLeafNode(
             String attrValue = escapeValue(rawAttrValue);
             // attribute is prefixed with "_" when being used as key.
             String attrAsKey = "_" + key;
-            logger.finest(() -> "Leaf Node: " + node.getUntransformedPath() + " ." + key);
+            if (node != null) {
+                logger.finest(() -> "Leaf Node: " + node.getUntransformedPath() + " ." + key);
+            }
             if (LdmlConvertRules.ATTRVALUE_AS_ARRAY_SET.contains(key)) {
                 String[] strings = attrValue.trim().split("\\s+");
                 JsonArray a = new JsonArray();

diff --git a/tools/cldr-code/src/main/java/org/unicode/cldr/json/LdmlConvertRules.java b/tools/cldr-code/src/main/java/org/unicode/cldr/json/LdmlConvertRules.java
@@ -289,6 +289,8 @@ public static class SplittableAttributeSpec {
         new SplittableAttributeSpec("parentLocale", "locales", "parent"),
         new SplittableAttributeSpec(
                 "collations", "locales", "parent"), // parentLocale component=collations
+        new SplittableAttributeSpec(
+                "plurals", "locales", "parent"), // parentLocale component=plurals
         new SplittableAttributeSpec(
                 "segmentations", "locales", "parent"), // parentLocale component=segmentations
         new SplittableAttributeSpec("hours", "regions", null),
@@ -298,6 +300,8 @@ public static class SplittableAttributeSpec {
         new SplittableAttributeSpec("unitPreference", "regions", null),
         new SplittableAttributeSpec("grammaticalFeatures", "locales", null),
         new SplittableAttributeSpec("grammaticalDerivations", "locales", null),
+        // this will cause EMPTY parentLocales elements to work properly
+        new SplittableAttributeSpec("parentLocales", "component", "" /* Not null */),
     };
 
     /** The set that contains all timezone type of elements. */