uutils · sylvestre · Sep 15, 2025 · Sep 24, 2025 · Sep 28, 2025
diff --git a/src/uu/sort/Cargo.toml b/src/uu/sort/Cargo.toml
@@ -34,7 +34,7 @@ self_cell = { workspace = true }
 tempfile = { workspace = true }
 thiserror = { workspace = true }
 unicode-width = { workspace = true }
-uucore = { workspace = true, features = ["fs", "parser", "version-cmp"] }
+uucore = { workspace = true, features = ["fs", "parser", "version-cmp", "i18n-collator"] }
 fluent = { workspace = true }
 
 [target.'cfg(target_os = "linux")'.dependencies]

diff --git a/src/uu/sort/locales/en-US.ftl b/src/uu/sort/locales/en-US.ftl
@@ -10,6 +10,11 @@ sort-after-help = The key format is FIELD[.CHAR][OPTIONS][,FIELD[.CHAR]][OPTIONS
 
   Valid options are: MbdfhnRrV. They override the global options for this key.
 
+  Locale-aware sorting:
+  The LC_ALL, LC_COLLATE, and LANG environment variables affect sorting order.
+  LC_ALL=C uses fast byte-wise comparison. Other locales use slower but correct Unicode collation.
+  For performance-critical scenarios with ASCII data, consider using LC_ALL=C.
+
 # Error messages
 sort-open-failed = open failed: {$path}: {$error}
 sort-parse-key-error = failed to parse key {$key}: {$msg}

diff --git a/src/uu/sort/locales/fr-FR.ftl b/src/uu/sort/locales/fr-FR.ftl
@@ -10,6 +10,11 @@ sort-after-help = Le format de clé est CHAMP[.CAR][OPTIONS][,CHAMP[.CAR]][OPTIO
 
   Les options valides sont : MbdfhnRrV. Elles remplacent les options globales pour cette clé.
 
+  Tri selon la locale :
+  Les variables d'environnement LC_ALL, LC_COLLATE et LANG affectent l'ordre de tri.
+  LC_ALL=C utilise une comparaison rapide par octets. D'autres locales utilisent une collation Unicode plus lente mais correcte.
+  Pour des scénarios critiques en performance avec des données ASCII, considérez l'utilisation de LC_ALL=C.
+
 # Messages d'erreur
 sort-open-failed = échec d'ouverture : {$path} : {$error}
 sort-parse-key-error = échec d'analyse de la clé {$key} : {$msg}

diff --git a/src/uu/sort/src/custom_str_cmp.rs b/src/uu/sort/src/custom_str_cmp.rs
@@ -8,6 +8,7 @@
 //! The goal is to compare strings without transforming them first (i.e. not allocating new strings)
 
 use std::cmp::Ordering;
+use uucore::i18n::collator::locale_cmp;
 
 fn filter_char(c: u8, ignore_non_printing: bool, ignore_non_dictionary: bool) -> bool {
     if ignore_non_dictionary && !(c.is_ascii_alphanumeric() || c.is_ascii_whitespace()) {
@@ -35,8 +36,8 @@ pub fn custom_str_cmp(
     ignore_case: bool,
 ) -> Ordering {
     if !(ignore_case || ignore_non_dictionary || ignore_non_printing) {
-        // There are no custom settings. Fall back to the default strcmp, which is faster.
-        return a.cmp(b);
+        // There are no custom settings. Fall back to locale-aware comparison.
+        return locale_cmp(a, b);
     }
     let mut a_chars = a
         .iter()

diff --git a/src/uu/sort/src/sort.rs b/src/uu/sort/src/sort.rs
@@ -46,6 +46,7 @@ use uucore::error::{FromIo, strip_errno};
 use uucore::error::{UError, UResult, USimpleError, UUsageError};
 use uucore::extendedbigdecimal::ExtendedBigDecimal;
 use uucore::format_usage;
+use uucore::i18n::collator::CollatorOptions;
 use uucore::line_ending::LineEnding;
 use uucore::parser::num_parser::{ExtendedParser, ExtendedParserError};
 use uucore::parser::parse_size::{ParseSizeError, Parser};
@@ -1037,6 +1038,28 @@ fn get_rlimit() -> UResult<usize> {
 
 const STDIN_FILE: &str = "-";
 
+/// Check if locale-aware collation will be needed based on sort settings and locale
+fn will_need_locale_collation(settings: &GlobalSettings) -> bool {
+    // First check if we're using the C locale (DEFAULT_LOCALE), which doesn't need collator
+    let (locale, _) = uucore::i18n::get_collating_locale();
+    if *locale == uucore::i18n::DEFAULT_LOCALE {
+        return false;
+    }
+
+    // Check each selector to see if any would use locale comparison
+    for selector in &settings.selectors {
+        let key_settings = &selector.settings;
+        if key_settings.mode == SortMode::Default
+            && !key_settings.ignore_case
+            && !key_settings.ignore_non_printing
+            && !key_settings.dictionary_order
+        {
+            return true;
+        }
+    }
+    false
+}
+
 #[uucore::main]
 #[allow(clippy::cognitive_complexity)]
 pub fn uumain(args: impl uucore::Args) -> UResult<()> {
@@ -1318,6 +1341,11 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
 
     settings.init_precomputed();
 
+    // Initialize locale-aware collator only if needed for string comparisons
+    if will_need_locale_collation(&settings) {
+        uucore::i18n::collator::try_init_collator(CollatorOptions::default());
+    }
+
     let result = exec(&mut files, &settings, output, &mut tmp_dir);
     // Wait here if `SIGINT` was received,
     // for signal handler to do its work and terminate the program.

diff --git a/src/uucore/src/lib/features/i18n/mod.rs b/src/uucore/src/lib/features/i18n/mod.rs
@@ -20,7 +20,7 @@ pub enum UEncoding {
     Utf8,
 }
 
-const DEFAULT_LOCALE: Locale = locale!("en-US-posix");
+pub const DEFAULT_LOCALE: Locale = locale!("en-US-posix");
 
 /// Look at 3 environment variables in the following order
 ///
@@ -64,7 +64,7 @@ fn get_locale_from_env(locale_name: &str) -> (Locale, UEncoding) {
 }
 
 /// Get the collating locale from the environment
-fn get_collating_locale() -> &'static (Locale, UEncoding) {
+pub fn get_collating_locale() -> &'static (Locale, UEncoding) {
     static COLLATING_LOCALE: OnceLock<(Locale, UEncoding)> = OnceLock::new();
 
     COLLATING_LOCALE.get_or_init(|| get_locale_from_env("LC_COLLATE"))

diff --git a/tests/by-util/test_sort.rs b/tests/by-util/test_sort.rs
@@ -1906,4 +1906,24 @@ fn test_color_environment_variables() {
     }
 }
 
+#[test]
+fn test_locale_sorting_c() {
+    // Test LC_ALL=C locale sorting (should sort by byte values)
+    new_ucmd!()
+        .env("LC_ALL", "C")
+        .pipe_in("a\no\nu\nä\nö\nü\n")
+        .succeeds()
+        .stdout_is("a\no\nu\nä\nö\nü\n");
+}
+
+#[test]
+fn test_locale_sorting_german() {
+    // Test LC_ALL=de_DE.utf-8 locale sorting (should respect German collation)
+    new_ucmd!()
+        .env("LC_ALL", "de_DE.utf-8")
+        .pipe_in("a\no\nu\nä\nö\nü\n")
+        .succeeds()
+        .stdout_is("a\nä\no\nö\nu\nü\n");
+}
+
 /* spell-checker: enable */