CLDR-10024 Add tests for RBNF by grhoten · Pull Request #5458 · unicode-org/cldr

grhoten · 2026-03-11T23:26:44Z

We're adding 2 types of new tests. This is a lot easier to do with the flattened RBNF rules.

One test involves conformance tests. These were generated from the existing rules. I didn't check them for accuracy.
One test involves roundtripping every public ruleset in every language with a set of values.

The current set of tests depend on several other open pull requests being merged before merging these changes. That includes important fixes for Serbian (#5456), and relevant new rules.

This PR completes the ticket.

ALLOW_MANY_COMMITS=true

grhoten · 2026-03-11T23:27:06Z

common/rbnf/ar.xml

 %spellout-cardinal-masculine:
 -x: ناقص >>;
-x.x: <%%spellout-numbering-m< فاصل >> ;
+x.x: <%%spellout-numbering-m< فاصل >>;


Whitespace typo getting fixed.

grhoten · 2026-03-11T23:27:43Z

common/rbnf/el.xml

 1000000000000000000: =#,##0=;
 %spellout-ordinal-neuter:
 -x: μείον >>;
+x.x: =0.#=;


The exact rule doesn't matter. It just needs to sort of roundtrip.

grhoten · 2026-03-11T23:28:01Z

common/rbnf/he.xml

 0: =%spellout-numbering=;
 %spellout-construct-masculine:
 -x: מינוס >>;
+x.x: << נקודה >>;


Oops! The rule was missing.

grhoten · 2026-03-11T23:29:12Z

common/rbnf/hr.xml

+1000000: <%spellout-cardinal-masculine< $(cardinal,one{milijun}few{milijuna}other{milijuna})$[ >>];
+1000000000: <%spellout-cardinal-feminine< $(cardinal,one{milijarda}few{milijarde}other{milijardi})$[ >>];
+1000000000000: <%spellout-cardinal-masculine< $(cardinal,one{bilijun}few{bilijuna}other{bilijuna})$[ >>];
+1000000000000000: <%spellout-cardinal-feminine< $(cardinal,one{bilijarda}few{bilijarde}other{bilijardi})$[ >>];


Known roundtrip issue that is fixed by switching to cardinal plural rules.

grhoten · 2026-03-11T23:30:05Z

common/rbnf/ja.xml

        <rulesetGrouping type="SpelloutRules">
            <rbnfRules><![CDATA[
 %spellout-numbering-year-latn:
+-x: マイナス>>;


Not an important rule, but it's needed for roundtripping.

grhoten · 2026-03-11T23:30:20Z

common/rbnf/vi.xml

 1000000000: << tỷ[ >%%after-hundred>];
 1000000000000000000: =#,##0=;
 %spellout-ordinal:
+-x: âm >>;


Not an important rule, but it's needed for roundtripping.

grhoten · 2026-03-11T23:40:04Z

tools/cldr-code/src/test/java/org/unicode/cldr/unittest/TestRBNF.java

+        ROUNDTRIP_VALUES.put(
+                "SpelloutRules",
+                new Number[] {
+                        -1L, 0L, 0.2, 1L, 1.1, 2L, 3L, 10L, 99L, 100L, 101L, 999L, 1000L, 1001L,
+                        1999L, 2000L, 2001L, 2100L, 2200L, 10000L, 20000L, 100000L, 200000L, 1000000L,
+                        2000000L
+                });


These are values that we're roundtripping for the spellout rules.

grhoten · 2026-03-11T23:51:05Z

tools/cldr-code/src/test/java/org/unicode/cldr/unittest/TestRBNF.java

+    private static final Set<String> ALIASES =
+            new TreeSet<>(Arrays.asList("nb", "en_001"));
+
+    private static final Number[] ROUNDTRIP_LONG_VALUES = {


These are values that we're roundtripping for the non-spellout rules. Those are normally positive integers.

grhoten · 2026-03-12T01:19:00Z

common/rbnf/ak.xml

 30: aduasa[->%%spellout-cardinal-tens>];
 40: adu<<[->%%spellout-cardinal-tens>];
-100: ɔha[-na->>];
+100: ɔha[-na->>];


Unnecessary soft hyphens are here. I'm not sure why the tests sometimes failed on this locale, but also they shouldn't be present in this scenario. I was using a slightly different version of ICU with these tests.

grhoten · 2026-03-12T04:46:50Z

tools/cldr-code/src/test/java/org/unicode/cldr/unittest/TestRBNF.java

+        return type + ";" + ruleSetName + ";" + numStr + ";" + formatted;
+    }
+
+    public static void main(String[] args) {


If you need to seed the conformance tests, just run this class, and it will automatically generate some sample conformance tests.

grhoten · 2026-03-12T17:10:25Z

common/rbnf/ky.xml

 1001: << миң[>%%and>];
 2000: << миң[>%%and>];
-100000: << миң[>%%commas>];
+100000/1000: << миң[>%%commas>];


Oops! The original syntax probably wasn't intended, since the word is the same as the thousand. Now 1000 and 100000 don't conflict on the value.

grhoten · 2026-03-12T17:11:12Z

common/rbnf/ky.xml

-1000000000000000000: =#,##0='inci;
+1000000000000001: <%spellout-numbering< квадриллион[ >>|унчу];
+2000000000000000: <%spellout-numbering< квадриллион[ >>|унчу];
+1000000000000000000: =#,##0='инчи;


Oops! Wrong script. It now matches the digits-ordinal ruleset. Also modernizing the syntax magically fixed the roundtripping issue.

grhoten · 2026-03-12T17:13:42Z

common/rbnf/su.xml

-100: <%%spellout-cardinal-large<ratus[ >>];
-1000: <%%spellout-cardinal-large<rebu[ >>];
-1000000: <%%spellout-cardinal-large<juta[ >>];
-1000000000: <%%spellout-cardinal-large<miliar[ >>];
+100: saratus[ >>];
+200: << ratus[ >>];
+1000: sarebu[ >>];
+2000: << rebu[ >>];
+1000000: sajuta[ >>];
+2000000: << juta[ >>];
+1000000000: samiliar[ >>];
+2000000000: << miliar[ >>];
 1000000000000: =#,##0=;
-%%spellout-cardinal-large:
-1: sa;
-2: =%spellout-cardinal= ;


The extra space at the end was confusing the ICU parsing for some reason. Making it structured more like the other languages fixed the issue.

grhoten · 2026-03-12T17:15:03Z

tools/cldr-code/src/test/java/org/unicode/cldr/unittest/TestRBNF.java

+    private static final Set<String> KNOWN_BROKEN_LOCALES =
+            new TreeSet<>(Arrays.asList("ga", "lt"));
+    private static final Set<String> ALIASES = new TreeSet<>(Arrays.asList("nb", "en_001"));


The known broken locales have roundtrip issues. We will just warn about them for now instead of failing on them.

We skip over the known aliases where we are acknowledging their existence, but we don't really care that they're empty.

grhoten · 2026-03-12T19:46:24Z

common/rbnf/hr.xml

-100: st[ >>];
-200: dvest[ >>];
-300: trist[ >>];
-400: četrist[ >>];
-500: petst[ >>];
-600: šest[ >>];
-700: sedamst[ >>];
-800: osamst[ >>];
-900: devetst[ >>];
+100: stot[ >>];
+200: dvestot[ >>];
+300: tristot[ >>];
+400: četristot[ >>];
+500: petstot[ >>];
+600: šestot[ >>];
+700: sedamstot[ >>];
+800: osamstot[ >>];
+900: devetstot[ >>];


This is a big ol' truncation typo. Number Format Tester confirms that these were wrong.

This source reference confirms the correct way to spell 3 examples.
https://www.lets-learn.eu/croatian/guide/ordinal-numbers-in-croatian

CLDR-10024 Add tests for RBNF

6a9cc49

grhoten requested a review from macchiati March 11, 2026 23:26

github-actions bot assigned grhoten Mar 11, 2026

grhoten commented Mar 11, 2026

View reviewed changes

AEApple requested review from echeran and sffc March 11, 2026 23:57

CLDR-10024 Add tests for RBNF

6375c59

grhoten commented Mar 12, 2026

View reviewed changes

CLDR-10024 Add tests for RBNF

748b432

grhoten commented Mar 12, 2026

View reviewed changes

grhoten added 3 commits March 12, 2026 06:23

CLDR-10024 Add tests for RBNF

51816d8

CLDR-10024 Add tests for RBNF

bb951e2

CLDR-10024 Add tests for RBNF

8ccde97

grhoten commented Mar 12, 2026

View reviewed changes

CLDR-10024 Add tests for RBNF

11361b9

grhoten commented Mar 12, 2026

View reviewed changes

CLDR-10024 Add tests for RBNF

44cb9b2

Conversation

grhoten commented Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

grhoten Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

grhoten Mar 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

grhoten Mar 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

grhoten Mar 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

grhoten Mar 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

grhoten commented Mar 11, 2026 •

edited

Loading

grhoten Mar 11, 2026 •

edited

Loading

grhoten Mar 12, 2026 •

edited

Loading

grhoten Mar 12, 2026 •

edited

Loading

grhoten Mar 12, 2026 •

edited

Loading

grhoten Mar 12, 2026 •

edited

Loading