From 62476459df57ab91dbaee15678475617bf55c5c8 Mon Sep 17 00:00:00 2001
From: Robert Bastian <robertbastian@unicode.org>
Date: Thu, 8 Aug 2024 12:35:34 +0200
Subject: [PATCH 1/7] script fallback

---
 components/locale/src/fallback/algorithms.rs | 194 ++++++++++++++++---
 components/locale/src/fallback/mod.rs        |  11 +-
 provider/core/src/fallback.rs                |   4 +
 3 files changed, 184 insertions(+), 25 deletions(-)

diff --git a/components/locale/src/fallback/algorithms.rs b/components/locale/src/fallback/algorithms.rs
index 3b41f51a6ff..133806c378e 100644
--- a/components/locale/src/fallback/algorithms.rs
+++ b/components/locale/src/fallback/algorithms.rs
@@ -8,7 +8,7 @@ use icu_locale_core::subtags::{Language, Region, Script};
 use super::*;
 
 impl<'a> LocaleFallbackerWithConfig<'a> {
-    pub(crate) fn normalize(&self, locale: &mut DataLocale) {
+    pub(crate) fn normalize(&self, locale: &mut DataLocale, default_script: &mut Option<Script>) {
         let language = locale.language;
         // 1. Populate the region (required for region fallback only)
         if self.config.priority == LocaleFallbackPriority::Region && locale.region.is_none() {
@@ -33,8 +33,8 @@ impl<'a> LocaleFallbackerWithConfig<'a> {
             }
         }
         // 2. Remove the script if it is implied by the other subtags
-        if let Some(script) = locale.script {
-            let default_script = locale
+        if locale.script.is_some() || self.config.priority == LocaleFallbackPriority::Script {
+            *default_script = locale
                 .region
                 .and_then(|region| {
                     self.likely_subtags.language_region.get_copied(&(
@@ -48,8 +48,9 @@ impl<'a> LocaleFallbackerWithConfig<'a> {
                         .get_copied(&language.into_tinystr().to_unvalidated())
                         .map(|(s, _r)| s)
                 });
-
-            if Some(script) == default_script {
+            if locale.script == *default_script
+                && self.config.priority != LocaleFallbackPriority::Script
+            {
                 locale.script = None;
             }
         }
@@ -62,6 +63,7 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
     pub fn step(&mut self, locale: &mut DataLocale) {
         match self.config.priority {
             LocaleFallbackPriority::Language => self.step_language(locale),
+            LocaleFallbackPriority::Script => self.step_script(locale),
             LocaleFallbackPriority::Region => self.step_region(locale),
             // This case should not normally happen, but `LocaleFallbackPriority` is non_exhaustive.
             // Make it go directly to `und`.
@@ -82,8 +84,6 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
             self.backup_subdivision = Some(value);
             return;
         }
-        // 3. Assert that the locale is a language identifier
-        debug_assert!(locale.subdivision.is_none());
         // 4. Remove variants
         if let Some(single_variant) = locale.variant.take() {
             self.backup_variant = Some(single_variant);
@@ -130,8 +130,6 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
             self.backup_subdivision = Some(value);
             return;
         }
-        // 3. Assert that the locale is a language identifier
-        debug_assert!(locale.subdivision.is_none());
         // 4. Remove variants
         if let Some(variant) = locale.variant.take() {
             self.backup_variant = Some(variant);
@@ -149,6 +147,66 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
         locale.region = None;
     }
 
+    fn step_script(&mut self, locale: &mut DataLocale) {
+        // Remove the subdivision keyword
+        if let Some(value) = locale.subdivision.take() {
+            self.backup_subdivision = Some(value);
+            return;
+        }
+        // Remove variants
+        if let Some(variant) = locale.variant.take() {
+            self.backup_variant = Some(variant);
+            return;
+        }
+        // Check for parent override
+        if let Some((language, script, region)) = self.get_explicit_parent(locale) {
+            locale.language = language;
+            locale.script = script;
+            locale.region = region;
+            self.restore_subdivision_variants(locale);
+            return;
+        }
+        // Remove the region
+        if let Some(region) = locale.region {
+            self.backup_region = Some(region);
+            locale.region = None;
+            self.restore_subdivision_variants(locale);
+            return;
+        }
+
+        // Remove the script if we have a language
+        if !locale.language.is_empty() {
+            if locale.script.is_some() {
+                locale.script = None;
+                if let Some(region) = self.backup_region.take() {
+                    locale.region = Some(region);
+                    self.restore_subdivision_variants(locale);
+                }
+                // needed if more fallback is added at the end
+                #[allow(clippy::needless_return)]
+                return;
+            } else {
+                // 3. Remove the language and apply the maximized script
+                locale.language = Language::UND;
+                locale.script = self.max_script;
+                self.restore_subdivision_variants(locale);
+                // needed if more fallback is added at the end
+                #[allow(clippy::needless_return)]
+                return;
+            }
+        }
+
+        // note: UTS #35 wants us to apply "other associated scripts" now. ICU4C/J does not do this,
+        // so we don't either. They would be found here if they are ever needed:
+        // https://github.com/unicode-cldr/cldr-core/blob/master/supplemental/languageData.json
+
+        // 6. Remove script
+        if locale.script.is_some() {
+            locale.script = None;
+            self.restore_subdivision_variants(locale);
+        }
+    }
+
     fn restore_subdivision_variants(&mut self, locale: &mut DataLocale) {
         if let Some(subdivision) = self.backup_subdivision.take() {
             locale.subdivision = Some(subdivision);
@@ -178,6 +236,7 @@ mod tests {
         requires_data: bool,
         // Note: The first entry in the chain is the normalized locale
         expected_language_chain: &'static [&'static str],
+        expected_script_chain: &'static [&'static str],
         expected_region_chain: &'static [&'static str],
     }
 
@@ -187,12 +246,20 @@ mod tests {
             input: "en-u-sd-usca",
             requires_data: false,
             expected_language_chain: &["en-u-sd-usca", "en"],
+            expected_script_chain: &["en-u-sd-usca", "en", "und-u-sd-usca"],
             expected_region_chain: &["en-u-sd-usca", "en", "und-u-sd-usca"],
         },
         TestCase {
             input: "en-US-u-sd-usca",
             requires_data: false,
             expected_language_chain: &["en-US-u-sd-usca", "en-US", "en-u-sd-usca", "en"],
+            expected_script_chain: &[
+                "en-US-u-sd-usca",
+                "en-US",
+                "en-u-sd-usca",
+                "en",
+                "und-u-sd-usca",
+            ],
             expected_region_chain: &["en-US-u-sd-usca", "en-US", "und-US-u-sd-usca", "und-US"],
         },
         TestCase {
@@ -206,6 +273,16 @@ mod tests {
                 "en-fonipa",
                 "en",
             ],
+            expected_script_chain: &[
+                "en-US-fonipa-u-sd-usca",
+                "en-US-fonipa",
+                "en-US",
+                "en-fonipa-u-sd-usca",
+                "en-fonipa",
+                "en",
+                "und-fonipa-u-sd-usca",
+                "und-fonipa",
+            ],
             expected_region_chain: &[
                 "en-US-fonipa-u-sd-usca",
                 "en-US-fonipa",
@@ -219,30 +296,60 @@ mod tests {
             input: "en-u-sd-usca",
             requires_data: true,
             expected_language_chain: &["en-u-sd-usca", "en"],
+            expected_script_chain: &[
+                "en-u-sd-usca",
+                "en",
+                "und-Latn-u-sd-usca",
+                "und-Latn",
+                "und-u-sd-usca",
+            ],
             expected_region_chain: &["en-US-u-sd-usca", "en-US", "und-US-u-sd-usca", "und-US"],
         },
         TestCase {
             input: "en-Latn-u-sd-usca",
             requires_data: true,
             expected_language_chain: &["en-u-sd-usca", "en"],
+            expected_script_chain: &[
+                "en-Latn-u-sd-usca",
+                "en-Latn",
+                "en",
+                "und-Latn-u-sd-usca",
+                "und-Latn",
+                "und-u-sd-usca",
+            ],
             expected_region_chain: &["en-US-u-sd-usca", "en-US", "und-US-u-sd-usca", "und-US"],
         },
         TestCase {
             input: "en-Latn-US-u-sd-usca",
             requires_data: true,
             expected_language_chain: &["en-US-u-sd-usca", "en-US", "en-u-sd-usca", "en"],
+            expected_script_chain: &[
+                "en-Latn-US-u-sd-usca",
+                "en-Latn-US",
+                "en-Latn-u-sd-usca",
+                "en-Latn",
+                "en-US-u-sd-usca",
+                "en-US",
+                "en-u-sd-usca",
+                "en",
+                "und-Latn-u-sd-usca",
+                "und-Latn",
+                "und-u-sd-usca",
+            ],
             expected_region_chain: &["en-US-u-sd-usca", "en-US", "und-US-u-sd-usca", "und-US"],
         },
         TestCase {
             input: "sr-ME",
             requires_data: true,
             expected_language_chain: &["sr-ME", "sr-Latn-ME", "sr-Latn"],
+            expected_script_chain: &["sr-ME", "sr", "und-Latn"],
             expected_region_chain: &["sr-ME", "und-ME"],
         },
         TestCase {
             input: "sr-Latn-ME",
             requires_data: true,
             expected_language_chain: &["sr-ME", "sr-Latn-ME", "sr-Latn"],
+            expected_script_chain: &["sr-Latn-ME", "sr-Latn", "sr-ME", "sr", "und-Latn"],
             expected_region_chain: &["sr-ME", "und-ME"],
         },
         TestCase {
@@ -256,54 +363,79 @@ mod tests {
                 "sr-Latn-fonipa",
                 "sr-Latn",
             ],
+            expected_script_chain: &[
+                "sr-ME-fonipa",
+                "sr-ME",
+                "sr-fonipa",
+                "sr",
+                "und-Latn-fonipa",
+                "und-Latn",
+                "und-fonipa",
+            ],
             expected_region_chain: &["sr-ME-fonipa", "sr-ME", "und-ME-fonipa", "und-ME"],
         },
         TestCase {
             input: "sr-RS",
             requires_data: true,
             expected_language_chain: &["sr-RS", "sr"],
+            expected_script_chain: &["sr-RS", "sr", "und-Cyrl"],
             expected_region_chain: &["sr-RS", "und-RS"],
         },
         TestCase {
             input: "sr-Cyrl-RS",
             requires_data: true,
             expected_language_chain: &["sr-RS", "sr"],
+            expected_script_chain: &["sr-Cyrl-RS", "sr-Cyrl", "sr-RS", "sr", "und-Cyrl"],
             expected_region_chain: &["sr-RS", "und-RS"],
         },
         TestCase {
             input: "sr-Latn-RS",
             requires_data: true,
             expected_language_chain: &["sr-Latn-RS", "sr-Latn"],
+            expected_script_chain: &["sr-Latn-RS", "sr-Latn", "sr-RS", "sr", "und-Latn"],
             expected_region_chain: &["sr-Latn-RS", "und-RS"],
         },
         TestCase {
             input: "de-Latn-LI",
             requires_data: true,
             expected_language_chain: &["de-LI", "de"],
+            expected_script_chain: &["de-Latn-LI", "de-Latn", "de-LI", "de", "und-Latn"],
             expected_region_chain: &["de-LI", "und-LI"],
         },
         TestCase {
             input: "ca-ES-valencia",
             requires_data: true,
             expected_language_chain: &["ca-ES-valencia", "ca-ES", "ca-valencia", "ca"],
+            expected_script_chain: &[
+                "ca-ES-valencia",
+                "ca-ES",
+                "ca-valencia",
+                "ca",
+                "und-Latn-valencia",
+                "und-Latn",
+                "und-valencia",
+            ],
             expected_region_chain: &["ca-ES-valencia", "ca-ES", "und-ES-valencia", "und-ES"],
         },
         TestCase {
             input: "es-AR",
             requires_data: true,
             expected_language_chain: &["es-AR", "es-419", "es"],
+            expected_script_chain: &["es-AR", "es-419", "es", "und-Latn"],
             expected_region_chain: &["es-AR", "und-AR"],
         },
         TestCase {
             input: "hi-IN",
             requires_data: true,
             expected_language_chain: &["hi-IN", "hi"],
+            expected_script_chain: &["hi-IN", "hi", "und-Deva"],
             expected_region_chain: &["hi-IN", "und-IN"],
         },
         TestCase {
             input: "hi-Latn-IN",
             requires_data: true,
             expected_language_chain: &["hi-Latn-IN", "hi-Latn", "en-IN", "en-001", "en"],
+            expected_script_chain: &["hi-Latn-IN", "hi-Latn", "en-IN", "en-001", "en", "und-Latn"],
             expected_region_chain: &["hi-Latn-IN", "und-IN"],
         },
         TestCase {
@@ -313,26 +445,50 @@ mod tests {
             // The fallback algorithm does not visit the language-script bundle when the
             // script is the default for the language
             expected_language_chain: &["zh-CN", "zh"],
+            expected_script_chain: &["zh-CN", "zh", "und-Hans"],
             expected_region_chain: &["zh-CN", "und-CN"],
         },
         TestCase {
             input: "zh-TW",
             requires_data: true,
             expected_language_chain: &["zh-TW", "zh-Hant-TW", "zh-Hant"],
+            expected_script_chain: &["zh-TW", "zh", "und-Hant"],
             expected_region_chain: &["zh-TW", "und-TW"],
         },
         TestCase {
             input: "yue-HK",
             requires_data: true,
             expected_language_chain: &["yue-HK", "yue"],
+            expected_script_chain: &["yue-HK", "yue", "und-Hant"],
             expected_region_chain: &["yue-HK", "und-HK"],
         },
         TestCase {
             input: "yue-HK",
             requires_data: true,
             expected_language_chain: &["yue-HK", "yue"],
+            expected_script_chain: &["yue-HK", "yue", "und-Hant"],
             expected_region_chain: &["yue-HK", "und-HK"],
-            // TODO(#3867): script fallback should do zh-Hant or und-Hant as well
+        },
+        TestCase {
+            input: "az-Arab-IR",
+            requires_data: true,
+            expected_language_chain: &["az-IR", "az-Arab-IR", "az-Arab"],
+            expected_script_chain: &["az-Arab-IR", "az-Arab", "az-IR", "az", "und-Arab"],
+            expected_region_chain: &["az-IR", "und-IR"],
+        },
+        TestCase {
+            input: "az-IR",
+            requires_data: true,
+            expected_language_chain: &["az-IR", "az-Arab-IR", "az-Arab"],
+            expected_script_chain: &["az-IR", "az", "und-Arab"],
+            expected_region_chain: &["az-IR", "und-IR"],
+        },
+        TestCase {
+            input: "az-Arab",
+            requires_data: true,
+            expected_language_chain: &["az-Arab"],
+            expected_script_chain: &["az-Arab", "az", "und-Arab"],
+            expected_region_chain: &["az-IR", "und-IR"],
         },
     ];
 
@@ -347,6 +503,7 @@ mod tests {
                     LocaleFallbackPriority::Language,
                     cas.expected_language_chain,
                 ),
+                (LocaleFallbackPriority::Script, cas.expected_script_chain),
                 (LocaleFallbackPriority::Region, cas.expected_region_chain),
             ] {
                 let mut config = LocaleFallbackConfig::default();
@@ -359,22 +516,15 @@ mod tests {
                 let mut it = fallbacker
                     .for_config(config)
                     .fallback_for(cas.input.parse().unwrap());
-                for &expected in expected_chain {
-                    assert_eq!(
-                        expected,
-                        &*it.get().write_to_string(),
-                        "{:?} ({:?})",
-                        cas.input,
-                        priority
-                    );
+                let mut actual_chain = Vec::new();
+                while !it.get().is_und() {
+                    actual_chain.push(it.get().write_to_string().into_owned());
                     it.step();
                 }
                 assert_eq!(
-                    "und",
-                    &*it.get().write_to_string(),
+                    expected_chain, &actual_chain,
                     "{:?} ({:?})",
-                    cas.input,
-                    priority
+                    cas.input, priority
                 );
             }
         }
diff --git a/components/locale/src/fallback/mod.rs b/components/locale/src/fallback/mod.rs
index cb7c9dca77a..bcf6995fb0a 100644
--- a/components/locale/src/fallback/mod.rs
+++ b/components/locale/src/fallback/mod.rs
@@ -6,8 +6,7 @@
 //! locale with data.
 
 use crate::provider::*;
-use icu_locale_core::subtags::Subtag;
-use icu_locale_core::subtags::Variant;
+use icu_locale_core::subtags::*;
 use icu_provider::prelude::*;
 
 #[doc(inline)]
@@ -85,6 +84,8 @@ struct LocaleFallbackIteratorInner<'a> {
     config: LocaleFallbackConfig,
     backup_subdivision: Option<Subtag>,
     backup_variant: Option<Variant>,
+    backup_region: Option<Region>,
+    max_script: Option<Script>,
 }
 
 /// Iteration type for locale fallback operations.
@@ -209,7 +210,9 @@ impl<'a> LocaleFallbackerWithConfig<'a> {
     ///
     /// When first initialized, the locale is normalized according to the fallback algorithm.
     pub fn fallback_for(&self, mut locale: DataLocale) -> LocaleFallbackIterator<'a, 'static> {
-        self.normalize(&mut locale);
+        let mut default_script = None;
+        self.normalize(&mut locale, &mut default_script);
+        let max_script = locale.script.or(default_script);
         LocaleFallbackIterator {
             current: locale,
             inner: LocaleFallbackIteratorInner {
@@ -218,6 +221,8 @@ impl<'a> LocaleFallbackerWithConfig<'a> {
                 config: self.config,
                 backup_subdivision: None,
                 backup_variant: None,
+                backup_region: None,
+                max_script,
             },
             phantom: core::marker::PhantomData,
         }
diff --git a/provider/core/src/fallback.rs b/provider/core/src/fallback.rs
index a52319a8735..1133a0c74e7 100644
--- a/provider/core/src/fallback.rs
+++ b/provider/core/src/fallback.rs
@@ -18,6 +18,10 @@ pub enum LocaleFallbackPriority {
     ///
     /// For example, `"en-US"` should go to `"en"` and then `"und"`.
     Language,
+    /// Prioritize the script.
+    ///
+    /// For example, `"en-US"` should go to `"en"` and then `"und-Latn"` and then `"und"`.
+    Script,
     /// Prioritize the region.
     ///
     /// For example, `"en-US"` should go to `"und-US"` and then `"und"`.

From c55d72dfaea63458f3cba57bf2fc5945da41d885 Mon Sep 17 00:00:00 2001
From: Robert Bastian <robertbastian@unicode.org>
Date: Fri, 9 Aug 2024 10:46:15 +0200
Subject: [PATCH 2/7] normalize script

---
 components/locale/src/fallback/algorithms.rs | 19 ++++++-------------
 1 file changed, 6 insertions(+), 13 deletions(-)

diff --git a/components/locale/src/fallback/algorithms.rs b/components/locale/src/fallback/algorithms.rs
index 133806c378e..e8227de4731 100644
--- a/components/locale/src/fallback/algorithms.rs
+++ b/components/locale/src/fallback/algorithms.rs
@@ -48,9 +48,7 @@ impl<'a> LocaleFallbackerWithConfig<'a> {
                         .get_copied(&language.into_tinystr().to_unvalidated())
                         .map(|(s, _r)| s)
                 });
-            if locale.script == *default_script
-                && self.config.priority != LocaleFallbackPriority::Script
-            {
+            if locale.script == *default_script {
                 locale.script = None;
             }
         }
@@ -310,8 +308,7 @@ mod tests {
             requires_data: true,
             expected_language_chain: &["en-u-sd-usca", "en"],
             expected_script_chain: &[
-                "en-Latn-u-sd-usca",
-                "en-Latn",
+                "en-u-sd-usca",
                 "en",
                 "und-Latn-u-sd-usca",
                 "und-Latn",
@@ -324,10 +321,6 @@ mod tests {
             requires_data: true,
             expected_language_chain: &["en-US-u-sd-usca", "en-US", "en-u-sd-usca", "en"],
             expected_script_chain: &[
-                "en-Latn-US-u-sd-usca",
-                "en-Latn-US",
-                "en-Latn-u-sd-usca",
-                "en-Latn",
                 "en-US-u-sd-usca",
                 "en-US",
                 "en-u-sd-usca",
@@ -349,7 +342,7 @@ mod tests {
             input: "sr-Latn-ME",
             requires_data: true,
             expected_language_chain: &["sr-ME", "sr-Latn-ME", "sr-Latn"],
-            expected_script_chain: &["sr-Latn-ME", "sr-Latn", "sr-ME", "sr", "und-Latn"],
+            expected_script_chain: &["sr-ME", "sr", "und-Latn"],
             expected_region_chain: &["sr-ME", "und-ME"],
         },
         TestCase {
@@ -385,7 +378,7 @@ mod tests {
             input: "sr-Cyrl-RS",
             requires_data: true,
             expected_language_chain: &["sr-RS", "sr"],
-            expected_script_chain: &["sr-Cyrl-RS", "sr-Cyrl", "sr-RS", "sr", "und-Cyrl"],
+            expected_script_chain: &["sr-RS", "sr", "und-Cyrl"],
             expected_region_chain: &["sr-RS", "und-RS"],
         },
         TestCase {
@@ -399,7 +392,7 @@ mod tests {
             input: "de-Latn-LI",
             requires_data: true,
             expected_language_chain: &["de-LI", "de"],
-            expected_script_chain: &["de-Latn-LI", "de-Latn", "de-LI", "de", "und-Latn"],
+            expected_script_chain: &["de-LI", "de", "und-Latn"],
             expected_region_chain: &["de-LI", "und-LI"],
         },
         TestCase {
@@ -473,7 +466,7 @@ mod tests {
             input: "az-Arab-IR",
             requires_data: true,
             expected_language_chain: &["az-IR", "az-Arab-IR", "az-Arab"],
-            expected_script_chain: &["az-Arab-IR", "az-Arab", "az-IR", "az", "und-Arab"],
+            expected_script_chain: &["az-IR", "az", "und-Arab"],
             expected_region_chain: &["az-IR", "und-IR"],
         },
         TestCase {

From 1896f0b691a7ede408f69135e6c62b6d10ce47fd Mon Sep 17 00:00:00 2001
From: Robert Bastian <robertbastian@unicode.org>
Date: Tue, 13 Aug 2024 15:36:14 +0200
Subject: [PATCH 3/7] don't cross script boundaries when removing region

---
 components/locale/src/fallback/algorithms.rs | 141 +++++++++++--------
 1 file changed, 80 insertions(+), 61 deletions(-)

diff --git a/components/locale/src/fallback/algorithms.rs b/components/locale/src/fallback/algorithms.rs
index 65d40b53ad4..87e6ef613b8 100644
--- a/components/locale/src/fallback/algorithms.rs
+++ b/components/locale/src/fallback/algorithms.rs
@@ -167,6 +167,14 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
         // Remove the region
         if let Some(region) = locale.region {
             self.backup_region = Some(region);
+            let language_implied_script = self
+                .likely_subtags
+                .language
+                .get_copied(&locale.language.into_tinystr().to_unvalidated())
+                .map(|(s, _r)| s);
+            if language_implied_script != self.max_script {
+                locale.script = self.max_script;
+            }
             locale.region = None;
             self.restore_subdivision_variants(locale);
             return;
@@ -174,7 +182,12 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
 
         // Remove the script if we have a language
         if !locale.language.is_default() {
-            if locale.script.is_some() {
+            let language_implied_script = self
+                .likely_subtags
+                .language
+                .get_copied(&locale.language.into_tinystr().to_unvalidated())
+                .map(|(s, _r)| s);
+            if locale.script.is_some() && language_implied_script == locale.script {
                 locale.script = None;
                 if let Some(region) = self.backup_region.take() {
                     locale.region = Some(region);
@@ -240,56 +253,56 @@ mod tests {
 
     // TODO: Consider loading these from a JSON file
     const TEST_CASES: &[TestCase] = &[
-        TestCase {
-            input: "en-u-sd-usca",
-            requires_data: false,
-            expected_language_chain: &["en-u-sd-usca", "en"],
-            expected_script_chain: &["en-u-sd-usca", "en", "und-u-sd-usca"],
-            expected_region_chain: &["en-u-sd-usca", "en", "und-u-sd-usca"],
-        },
-        TestCase {
-            input: "en-US-u-sd-usca",
-            requires_data: false,
-            expected_language_chain: &["en-US-u-sd-usca", "en-US", "en-u-sd-usca", "en"],
-            expected_script_chain: &[
-                "en-US-u-sd-usca",
-                "en-US",
-                "en-u-sd-usca",
-                "en",
-                "und-u-sd-usca",
-            ],
-            expected_region_chain: &["en-US-u-sd-usca", "en-US", "und-US-u-sd-usca", "und-US"],
-        },
-        TestCase {
-            input: "en-US-fonipa-u-sd-usca",
-            requires_data: false,
-            expected_language_chain: &[
-                "en-US-fonipa-u-sd-usca",
-                "en-US-fonipa",
-                "en-US",
-                "en-fonipa-u-sd-usca",
-                "en-fonipa",
-                "en",
-            ],
-            expected_script_chain: &[
-                "en-US-fonipa-u-sd-usca",
-                "en-US-fonipa",
-                "en-US",
-                "en-fonipa-u-sd-usca",
-                "en-fonipa",
-                "en",
-                "und-fonipa-u-sd-usca",
-                "und-fonipa",
-            ],
-            expected_region_chain: &[
-                "en-US-fonipa-u-sd-usca",
-                "en-US-fonipa",
-                "en-US",
-                "und-US-fonipa-u-sd-usca",
-                "und-US-fonipa",
-                "und-US",
-            ],
-        },
+        // TestCase {
+        //     input: "en-u-sd-usca",
+        //     requires_data: false,
+        //     expected_language_chain: &["en-u-sd-usca", "en"],
+        //     expected_script_chain: &["en-u-sd-usca", "en", "und-u-sd-usca"],
+        //     expected_region_chain: &["en-u-sd-usca", "en", "und-u-sd-usca"],
+        // },
+        // TestCase {
+        //     input: "en-US-u-sd-usca",
+        //     requires_data: false,
+        //     expected_language_chain: &["en-US-u-sd-usca", "en-US", "en-u-sd-usca", "en"],
+        //     expected_script_chain: &[
+        //         "en-US-u-sd-usca",
+        //         "en-US",
+        //         "en-u-sd-usca",
+        //         "en",
+        //         "und-u-sd-usca",
+        //     ],
+        //     expected_region_chain: &["en-US-u-sd-usca", "en-US", "und-US-u-sd-usca", "und-US"],
+        // },
+        // TestCase {
+        //     input: "en-US-fonipa-u-sd-usca",
+        //     requires_data: false,
+        //     expected_language_chain: &[
+        //         "en-US-fonipa-u-sd-usca",
+        //         "en-US-fonipa",
+        //         "en-US",
+        //         "en-fonipa-u-sd-usca",
+        //         "en-fonipa",
+        //         "en",
+        //     ],
+        //     expected_script_chain: &[
+        //         "en-US-fonipa-u-sd-usca",
+        //         "en-US-fonipa",
+        //         "en-US",
+        //         "en-fonipa-u-sd-usca",
+        //         "en-fonipa",
+        //         "en",
+        //         "und-fonipa-u-sd-usca",
+        //         "und-fonipa",
+        //     ],
+        //     expected_region_chain: &[
+        //         "en-US-fonipa-u-sd-usca",
+        //         "en-US-fonipa",
+        //         "en-US",
+        //         "und-US-fonipa-u-sd-usca",
+        //         "und-US-fonipa",
+        //         "und-US",
+        //     ],
+        // },
         TestCase {
             input: "en-u-sd-usca",
             requires_data: true,
@@ -335,14 +348,14 @@ mod tests {
             input: "sr-ME",
             requires_data: true,
             expected_language_chain: &["sr-ME", "sr-Latn-ME", "sr-Latn"],
-            expected_script_chain: &["sr-ME", "sr", "und-Latn"],
+            expected_script_chain: &["sr-ME", "sr-Latn", "und-Latn"],
             expected_region_chain: &["sr-ME", "und-ME"],
         },
         TestCase {
             input: "sr-Latn-ME",
             requires_data: true,
             expected_language_chain: &["sr-ME", "sr-Latn-ME", "sr-Latn"],
-            expected_script_chain: &["sr-ME", "sr", "und-Latn"],
+            expected_script_chain: &["sr-ME", "sr-Latn", "und-Latn"],
             expected_region_chain: &["sr-ME", "und-ME"],
         },
         TestCase {
@@ -359,8 +372,8 @@ mod tests {
             expected_script_chain: &[
                 "sr-ME-fonipa",
                 "sr-ME",
-                "sr-fonipa",
-                "sr",
+                "sr-Latn-fonipa",
+                "sr-Latn",
                 "und-Latn-fonipa",
                 "und-Latn",
                 "und-fonipa",
@@ -385,7 +398,7 @@ mod tests {
             input: "sr-Latn-RS",
             requires_data: true,
             expected_language_chain: &["sr-Latn-RS", "sr-Latn"],
-            expected_script_chain: &["sr-Latn-RS", "sr-Latn", "sr-RS", "sr", "und-Latn"],
+            expected_script_chain: &["sr-Latn-RS", "sr-Latn", "und-Latn"],
             expected_region_chain: &["sr-Latn-RS", "und-RS"],
         },
         TestCase {
@@ -445,7 +458,7 @@ mod tests {
             input: "zh-TW",
             requires_data: true,
             expected_language_chain: &["zh-TW", "zh-Hant-TW", "zh-Hant"],
-            expected_script_chain: &["zh-TW", "zh", "und-Hant"],
+            expected_script_chain: &["zh-TW", "zh-Hant", "und-Hant"],
             expected_region_chain: &["zh-TW", "und-TW"],
         },
         TestCase {
@@ -466,21 +479,21 @@ mod tests {
             input: "az-Arab-IR",
             requires_data: true,
             expected_language_chain: &["az-IR", "az-Arab-IR", "az-Arab"],
-            expected_script_chain: &["az-IR", "az", "und-Arab"],
+            expected_script_chain: &["az-IR", "az-Arab", "und-Arab"],
             expected_region_chain: &["az-IR", "und-IR"],
         },
         TestCase {
             input: "az-IR",
             requires_data: true,
             expected_language_chain: &["az-IR", "az-Arab-IR", "az-Arab"],
-            expected_script_chain: &["az-IR", "az", "und-Arab"],
+            expected_script_chain: &["az-IR", "az-Arab", "und-Arab"],
             expected_region_chain: &["az-IR", "und-IR"],
         },
         TestCase {
             input: "az-Arab",
             requires_data: true,
             expected_language_chain: &["az-Arab"],
-            expected_script_chain: &["az-Arab", "az", "und-Arab"],
+            expected_script_chain: &["az-Arab", "und-Arab"],
             expected_region_chain: &["az-IR", "und-IR"],
         },
     ];
@@ -510,7 +523,13 @@ mod tests {
                     .for_config(config)
                     .fallback_for(cas.input.parse().unwrap());
                 let mut actual_chain = Vec::new();
-                while !it.get().is_default() {
+                for i in 0..20 {
+                    if i == 19 {
+                        eprintln!("20 iterations reached!");
+                    }
+                    if it.get().is_default() {
+                        break;
+                    }
                     actual_chain.push(it.get().write_to_string().into_owned());
                     it.step();
                 }

From 805b15dcc4eade0b9b328720cc16f21148eb41b4 Mon Sep 17 00:00:00 2001
From: Robert Bastian <robertbastian@unicode.org>
Date: Tue, 13 Aug 2024 15:38:16 +0200
Subject: [PATCH 4/7] no-data tests

---
 components/locale/src/fallback/algorithms.rs | 100 +++++++++----------
 1 file changed, 50 insertions(+), 50 deletions(-)

diff --git a/components/locale/src/fallback/algorithms.rs b/components/locale/src/fallback/algorithms.rs
index 87e6ef613b8..083bd8494b3 100644
--- a/components/locale/src/fallback/algorithms.rs
+++ b/components/locale/src/fallback/algorithms.rs
@@ -253,56 +253,56 @@ mod tests {
 
     // TODO: Consider loading these from a JSON file
     const TEST_CASES: &[TestCase] = &[
-        // TestCase {
-        //     input: "en-u-sd-usca",
-        //     requires_data: false,
-        //     expected_language_chain: &["en-u-sd-usca", "en"],
-        //     expected_script_chain: &["en-u-sd-usca", "en", "und-u-sd-usca"],
-        //     expected_region_chain: &["en-u-sd-usca", "en", "und-u-sd-usca"],
-        // },
-        // TestCase {
-        //     input: "en-US-u-sd-usca",
-        //     requires_data: false,
-        //     expected_language_chain: &["en-US-u-sd-usca", "en-US", "en-u-sd-usca", "en"],
-        //     expected_script_chain: &[
-        //         "en-US-u-sd-usca",
-        //         "en-US",
-        //         "en-u-sd-usca",
-        //         "en",
-        //         "und-u-sd-usca",
-        //     ],
-        //     expected_region_chain: &["en-US-u-sd-usca", "en-US", "und-US-u-sd-usca", "und-US"],
-        // },
-        // TestCase {
-        //     input: "en-US-fonipa-u-sd-usca",
-        //     requires_data: false,
-        //     expected_language_chain: &[
-        //         "en-US-fonipa-u-sd-usca",
-        //         "en-US-fonipa",
-        //         "en-US",
-        //         "en-fonipa-u-sd-usca",
-        //         "en-fonipa",
-        //         "en",
-        //     ],
-        //     expected_script_chain: &[
-        //         "en-US-fonipa-u-sd-usca",
-        //         "en-US-fonipa",
-        //         "en-US",
-        //         "en-fonipa-u-sd-usca",
-        //         "en-fonipa",
-        //         "en",
-        //         "und-fonipa-u-sd-usca",
-        //         "und-fonipa",
-        //     ],
-        //     expected_region_chain: &[
-        //         "en-US-fonipa-u-sd-usca",
-        //         "en-US-fonipa",
-        //         "en-US",
-        //         "und-US-fonipa-u-sd-usca",
-        //         "und-US-fonipa",
-        //         "und-US",
-        //     ],
-        // },
+        TestCase {
+            input: "en-u-sd-usca",
+            requires_data: false,
+            expected_language_chain: &["en-u-sd-usca", "en"],
+            expected_script_chain: &["en-u-sd-usca", "en", "und-u-sd-usca"],
+            expected_region_chain: &["en-u-sd-usca", "en", "und-u-sd-usca"],
+        },
+        TestCase {
+            input: "en-US-u-sd-usca",
+            requires_data: false,
+            expected_language_chain: &["en-US-u-sd-usca", "en-US", "en-u-sd-usca", "en"],
+            expected_script_chain: &[
+                "en-US-u-sd-usca",
+                "en-US",
+                "en-u-sd-usca",
+                "en",
+                "und-u-sd-usca",
+            ],
+            expected_region_chain: &["en-US-u-sd-usca", "en-US", "und-US-u-sd-usca", "und-US"],
+        },
+        TestCase {
+            input: "en-US-fonipa-u-sd-usca",
+            requires_data: false,
+            expected_language_chain: &[
+                "en-US-fonipa-u-sd-usca",
+                "en-US-fonipa",
+                "en-US",
+                "en-fonipa-u-sd-usca",
+                "en-fonipa",
+                "en",
+            ],
+            expected_script_chain: &[
+                "en-US-fonipa-u-sd-usca",
+                "en-US-fonipa",
+                "en-US",
+                "en-fonipa-u-sd-usca",
+                "en-fonipa",
+                "en",
+                "und-fonipa-u-sd-usca",
+                "und-fonipa",
+            ],
+            expected_region_chain: &[
+                "en-US-fonipa-u-sd-usca",
+                "en-US-fonipa",
+                "en-US",
+                "und-US-fonipa-u-sd-usca",
+                "und-US-fonipa",
+                "und-US",
+            ],
+        },
         TestCase {
             input: "en-u-sd-usca",
             requires_data: true,

From 442418a36a89a332daf01f9dea49a4c587254727 Mon Sep 17 00:00:00 2001
From: Robert Bastian <robertbastian@unicode.org>
Date: Tue, 13 Aug 2024 16:50:47 +0200
Subject: [PATCH 5/7] correctly handle -u-sd

---
 components/locale/src/fallback/algorithms.rs | 99 +++++++++-----------
 1 file changed, 42 insertions(+), 57 deletions(-)

diff --git a/components/locale/src/fallback/algorithms.rs b/components/locale/src/fallback/algorithms.rs
index 083bd8494b3..a4fe6457696 100644
--- a/components/locale/src/fallback/algorithms.rs
+++ b/components/locale/src/fallback/algorithms.rs
@@ -9,6 +9,17 @@ use super::*;
 
 impl<'a> LocaleFallbackerWithConfig<'a> {
     pub(crate) fn normalize(&self, locale: &mut DataLocale, default_script: &mut Option<Script>) {
+        // 0. If there is an invalid "sd" subtag, drop it
+        if let Some(subdivision) = locale.subdivision.take() {
+            if let Some(region) = locale.region {
+                if subdivision
+                    .as_str()
+                    .starts_with(region.into_tinystr().to_ascii_lowercase().as_str())
+                {
+                    locale.subdivision = Some(subdivision);
+                }
+            }
+        }
         let language = locale.language;
         // 1. Populate the region (required for region fallback only)
         if self.config.priority == LocaleFallbackPriority::Region && locale.region.is_none() {
@@ -52,8 +63,6 @@ impl<'a> LocaleFallbackerWithConfig<'a> {
                 locale.script = None;
             }
         }
-        // 4. If there is an invalid "sd" subtag, drop it
-        // For now, ignore it, and let fallback do it for us
     }
 }
 
@@ -92,7 +101,7 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
             locale.language = language;
             locale.script = script;
             locale.region = region;
-            self.restore_subdivision_variants(locale);
+            locale.variant = self.backup_variant.take();
             return;
         }
         // 6. Add the script subtag if necessary
@@ -104,7 +113,7 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
                     region.into_tinystr().to_unvalidated(),
                 )) {
                     locale.script = Some(script);
-                    self.restore_subdivision_variants(locale);
+                    locale.variant = self.backup_variant.take();
                     return;
                 }
             }
@@ -112,7 +121,7 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
         // 7. Remove region
         if locale.region.is_some() {
             locale.region = None;
-            self.restore_subdivision_variants(locale);
+            locale.variant = self.backup_variant.take();
             return;
         }
         // 8. Remove language+script
@@ -137,7 +146,8 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
         if !locale.language.is_default() || locale.script.is_some() {
             locale.script = None;
             locale.language = Language::UND;
-            self.restore_subdivision_variants(locale);
+            locale.variant = self.backup_variant.take();
+            locale.subdivision = self.backup_subdivision.take();
             return;
         }
         // 6. Remove region
@@ -161,7 +171,7 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
             locale.language = language;
             locale.script = script;
             locale.region = region;
-            self.restore_subdivision_variants(locale);
+            locale.variant = self.backup_variant.take();
             return;
         }
         // Remove the region
@@ -176,7 +186,7 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
                 locale.script = self.max_script;
             }
             locale.region = None;
-            self.restore_subdivision_variants(locale);
+            locale.variant = self.backup_variant.take();
             return;
         }
 
@@ -191,7 +201,8 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
                 locale.script = None;
                 if let Some(region) = self.backup_region.take() {
                     locale.region = Some(region);
-                    self.restore_subdivision_variants(locale);
+                    locale.subdivision = self.backup_subdivision.take();
+                    locale.variant = self.backup_variant.take();
                 }
                 // needed if more fallback is added at the end
                 #[allow(clippy::needless_return)]
@@ -200,7 +211,7 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
                 // 3. Remove the language and apply the maximized script
                 locale.language = Language::UND;
                 locale.script = self.max_script;
-                self.restore_subdivision_variants(locale);
+                locale.variant = self.backup_variant.take();
                 // needed if more fallback is added at the end
                 #[allow(clippy::needless_return)]
                 return;
@@ -214,16 +225,7 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
         // 6. Remove script
         if locale.script.is_some() {
             locale.script = None;
-            self.restore_subdivision_variants(locale);
-        }
-    }
-
-    fn restore_subdivision_variants(&mut self, locale: &mut DataLocale) {
-        if let Some(subdivision) = self.backup_subdivision.take() {
-            locale.subdivision = Some(subdivision);
-        }
-        if let Some(variant) = self.backup_variant.take() {
-            locale.variant = Some(variant);
+            locale.variant = self.backup_variant.take();
         }
     }
 
@@ -254,23 +256,17 @@ mod tests {
     // TODO: Consider loading these from a JSON file
     const TEST_CASES: &[TestCase] = &[
         TestCase {
-            input: "en-u-sd-usca",
+            input: "en-fonipa",
             requires_data: false,
-            expected_language_chain: &["en-u-sd-usca", "en"],
-            expected_script_chain: &["en-u-sd-usca", "en", "und-u-sd-usca"],
-            expected_region_chain: &["en-u-sd-usca", "en", "und-u-sd-usca"],
+            expected_language_chain: &["en-fonipa", "en"],
+            expected_script_chain: &["en-fonipa", "en", "und-fonipa"],
+            expected_region_chain: &["en-fonipa", "en", "und-fonipa"],
         },
         TestCase {
             input: "en-US-u-sd-usca",
             requires_data: false,
-            expected_language_chain: &["en-US-u-sd-usca", "en-US", "en-u-sd-usca", "en"],
-            expected_script_chain: &[
-                "en-US-u-sd-usca",
-                "en-US",
-                "en-u-sd-usca",
-                "en",
-                "und-u-sd-usca",
-            ],
+            expected_language_chain: &["en-US-u-sd-usca", "en-US", "en"],
+            expected_script_chain: &["en-US-u-sd-usca", "en-US", "en"],
             expected_region_chain: &["en-US-u-sd-usca", "en-US", "und-US-u-sd-usca", "und-US"],
         },
         TestCase {
@@ -280,7 +276,6 @@ mod tests {
                 "en-US-fonipa-u-sd-usca",
                 "en-US-fonipa",
                 "en-US",
-                "en-fonipa-u-sd-usca",
                 "en-fonipa",
                 "en",
             ],
@@ -288,10 +283,8 @@ mod tests {
                 "en-US-fonipa-u-sd-usca",
                 "en-US-fonipa",
                 "en-US",
-                "en-fonipa-u-sd-usca",
                 "en-fonipa",
                 "en",
-                "und-fonipa-u-sd-usca",
                 "und-fonipa",
             ],
             expected_region_chain: &[
@@ -304,44 +297,36 @@ mod tests {
             ],
         },
         TestCase {
-            input: "en-u-sd-usca",
+            input: "en-fonipa",
             requires_data: true,
-            expected_language_chain: &["en-u-sd-usca", "en"],
+            expected_language_chain: &["en-fonipa", "en"],
             expected_script_chain: &[
-                "en-u-sd-usca",
+                "en-fonipa",
                 "en",
-                "und-Latn-u-sd-usca",
+                "und-Latn-fonipa",
                 "und-Latn",
-                "und-u-sd-usca",
+                "und-fonipa",
             ],
-            expected_region_chain: &["en-US-u-sd-usca", "en-US", "und-US-u-sd-usca", "und-US"],
+            expected_region_chain: &["en-US-fonipa", "en-US", "und-US-fonipa", "und-US"],
         },
         TestCase {
-            input: "en-Latn-u-sd-usca",
+            input: "en-Latn-fonipa",
             requires_data: true,
-            expected_language_chain: &["en-u-sd-usca", "en"],
+            expected_language_chain: &["en-fonipa", "en"],
             expected_script_chain: &[
-                "en-u-sd-usca",
+                "en-fonipa",
                 "en",
-                "und-Latn-u-sd-usca",
+                "und-Latn-fonipa",
                 "und-Latn",
-                "und-u-sd-usca",
+                "und-fonipa",
             ],
-            expected_region_chain: &["en-US-u-sd-usca", "en-US", "und-US-u-sd-usca", "und-US"],
+            expected_region_chain: &["en-US-fonipa", "en-US", "und-US-fonipa", "und-US"],
         },
         TestCase {
             input: "en-Latn-US-u-sd-usca",
             requires_data: true,
-            expected_language_chain: &["en-US-u-sd-usca", "en-US", "en-u-sd-usca", "en"],
-            expected_script_chain: &[
-                "en-US-u-sd-usca",
-                "en-US",
-                "en-u-sd-usca",
-                "en",
-                "und-Latn-u-sd-usca",
-                "und-Latn",
-                "und-u-sd-usca",
-            ],
+            expected_language_chain: &["en-US-u-sd-usca", "en-US", "en"],
+            expected_script_chain: &["en-US-u-sd-usca", "en-US", "en", "und-Latn"],
             expected_region_chain: &["en-US-u-sd-usca", "en-US", "und-US-u-sd-usca", "und-US"],
         },
         TestCase {

From c67d60a28506d9909ed13d60591c485413f54c3d Mon Sep 17 00:00:00 2001
From: Robert Bastian <robertbastian@unicode.org>
Date: Tue, 13 Aug 2024 18:12:03 +0200
Subject: [PATCH 6/7] as discussed

---
 components/locale/src/fallback/algorithms.rs | 36 +++++++-------------
 1 file changed, 13 insertions(+), 23 deletions(-)

diff --git a/components/locale/src/fallback/algorithms.rs b/components/locale/src/fallback/algorithms.rs
index a4fe6457696..144a933ad59 100644
--- a/components/locale/src/fallback/algorithms.rs
+++ b/components/locale/src/fallback/algorithms.rs
@@ -146,8 +146,11 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
         if !locale.language.is_default() || locale.script.is_some() {
             locale.script = None;
             locale.language = Language::UND;
-            locale.variant = self.backup_variant.take();
-            locale.subdivision = self.backup_subdivision.take();
+            // Don't produce und-variant
+            if locale.region.is_some() {
+                locale.variant = self.backup_variant.take();
+                locale.subdivision = self.backup_subdivision.take();
+            }
             return;
         }
         // 6. Remove region
@@ -211,7 +214,10 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
                 // 3. Remove the language and apply the maximized script
                 locale.language = Language::UND;
                 locale.script = self.max_script;
-                locale.variant = self.backup_variant.take();
+                // Don't produce und-variant
+                if locale.script.is_some() {
+                    locale.variant = self.backup_variant.take();
+                }
                 // needed if more fallback is added at the end
                 #[allow(clippy::needless_return)]
                 return;
@@ -225,7 +231,6 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
         // 6. Remove script
         if locale.script.is_some() {
             locale.script = None;
-            locale.variant = self.backup_variant.take();
         }
     }
 
@@ -259,8 +264,8 @@ mod tests {
             input: "en-fonipa",
             requires_data: false,
             expected_language_chain: &["en-fonipa", "en"],
-            expected_script_chain: &["en-fonipa", "en", "und-fonipa"],
-            expected_region_chain: &["en-fonipa", "en", "und-fonipa"],
+            expected_script_chain: &["en-fonipa", "en"],
+            expected_region_chain: &["en-fonipa", "en"],
         },
         TestCase {
             input: "en-US-u-sd-usca",
@@ -285,7 +290,6 @@ mod tests {
                 "en-US",
                 "en-fonipa",
                 "en",
-                "und-fonipa",
             ],
             expected_region_chain: &[
                 "en-US-fonipa-u-sd-usca",
@@ -300,26 +304,14 @@ mod tests {
             input: "en-fonipa",
             requires_data: true,
             expected_language_chain: &["en-fonipa", "en"],
-            expected_script_chain: &[
-                "en-fonipa",
-                "en",
-                "und-Latn-fonipa",
-                "und-Latn",
-                "und-fonipa",
-            ],
+            expected_script_chain: &["en-fonipa", "en", "und-Latn-fonipa", "und-Latn"],
             expected_region_chain: &["en-US-fonipa", "en-US", "und-US-fonipa", "und-US"],
         },
         TestCase {
             input: "en-Latn-fonipa",
             requires_data: true,
             expected_language_chain: &["en-fonipa", "en"],
-            expected_script_chain: &[
-                "en-fonipa",
-                "en",
-                "und-Latn-fonipa",
-                "und-Latn",
-                "und-fonipa",
-            ],
+            expected_script_chain: &["en-fonipa", "en", "und-Latn-fonipa", "und-Latn"],
             expected_region_chain: &["en-US-fonipa", "en-US", "und-US-fonipa", "und-US"],
         },
         TestCase {
@@ -361,7 +353,6 @@ mod tests {
                 "sr-Latn",
                 "und-Latn-fonipa",
                 "und-Latn",
-                "und-fonipa",
             ],
             expected_region_chain: &["sr-ME-fonipa", "sr-ME", "und-ME-fonipa", "und-ME"],
         },
@@ -404,7 +395,6 @@ mod tests {
                 "ca",
                 "und-Latn-valencia",
                 "und-Latn",
-                "und-valencia",
             ],
             expected_region_chain: &["ca-ES-valencia", "ca-ES", "und-ES-valencia", "und-ES"],
         },

From 51779560dbb6a1b1e527c627f7399178872c0c15 Mon Sep 17 00:00:00 2001
From: Robert Bastian <robertbastian@unicode.org>
Date: Tue, 13 Aug 2024 18:23:57 +0200
Subject: [PATCH 7/7] skip fully expanded locales

---
 components/locale/src/fallback/algorithms.rs | 31 +++++++-------------
 1 file changed, 10 insertions(+), 21 deletions(-)

diff --git a/components/locale/src/fallback/algorithms.rs b/components/locale/src/fallback/algorithms.rs
index 144a933ad59..702215c526e 100644
--- a/components/locale/src/fallback/algorithms.rs
+++ b/components/locale/src/fallback/algorithms.rs
@@ -104,22 +104,18 @@ impl<'a> LocaleFallbackIteratorInner<'a> {
             locale.variant = self.backup_variant.take();
             return;
         }
-        // 6. Add the script subtag if necessary
-        if locale.script.is_none() {
-            if let Some(region) = locale.region {
+        // 7. Remove region
+        if let Some(region) = locale.region {
+            // 6. Add the script subtag if necessary
+            if locale.script.is_none() {
                 let language = locale.language;
                 if let Some(script) = self.likely_subtags.language_region.get_copied(&(
                     language.into_tinystr().to_unvalidated(),
                     region.into_tinystr().to_unvalidated(),
                 )) {
                     locale.script = Some(script);
-                    locale.variant = self.backup_variant.take();
-                    return;
                 }
             }
-        }
-        // 7. Remove region
-        if locale.region.is_some() {
             locale.region = None;
             locale.variant = self.backup_variant.take();
             return;
@@ -324,28 +320,21 @@ mod tests {
         TestCase {
             input: "sr-ME",
             requires_data: true,
-            expected_language_chain: &["sr-ME", "sr-Latn-ME", "sr-Latn"],
+            expected_language_chain: &["sr-ME", "sr-Latn"],
             expected_script_chain: &["sr-ME", "sr-Latn", "und-Latn"],
             expected_region_chain: &["sr-ME", "und-ME"],
         },
         TestCase {
             input: "sr-Latn-ME",
             requires_data: true,
-            expected_language_chain: &["sr-ME", "sr-Latn-ME", "sr-Latn"],
+            expected_language_chain: &["sr-ME", "sr-Latn"],
             expected_script_chain: &["sr-ME", "sr-Latn", "und-Latn"],
             expected_region_chain: &["sr-ME", "und-ME"],
         },
         TestCase {
             input: "sr-ME-fonipa",
             requires_data: true,
-            expected_language_chain: &[
-                "sr-ME-fonipa",
-                "sr-ME",
-                "sr-Latn-ME-fonipa",
-                "sr-Latn-ME",
-                "sr-Latn-fonipa",
-                "sr-Latn",
-            ],
+            expected_language_chain: &["sr-ME-fonipa", "sr-ME", "sr-Latn-fonipa", "sr-Latn"],
             expected_script_chain: &[
                 "sr-ME-fonipa",
                 "sr-ME",
@@ -432,7 +421,7 @@ mod tests {
         TestCase {
             input: "zh-TW",
             requires_data: true,
-            expected_language_chain: &["zh-TW", "zh-Hant-TW", "zh-Hant"],
+            expected_language_chain: &["zh-TW", "zh-Hant"],
             expected_script_chain: &["zh-TW", "zh-Hant", "und-Hant"],
             expected_region_chain: &["zh-TW", "und-TW"],
         },
@@ -453,14 +442,14 @@ mod tests {
         TestCase {
             input: "az-Arab-IR",
             requires_data: true,
-            expected_language_chain: &["az-IR", "az-Arab-IR", "az-Arab"],
+            expected_language_chain: &["az-IR", "az-Arab"],
             expected_script_chain: &["az-IR", "az-Arab", "und-Arab"],
             expected_region_chain: &["az-IR", "und-IR"],
         },
         TestCase {
             input: "az-IR",
             requires_data: true,
-            expected_language_chain: &["az-IR", "az-Arab-IR", "az-Arab"],
+            expected_language_chain: &["az-IR", "az-Arab"],
             expected_script_chain: &["az-IR", "az-Arab", "und-Arab"],
             expected_region_chain: &["az-IR", "und-IR"],
         },