Merge branch 'main' into clap-loca-improv-2

Author: sylvestre

.github/workflows/code-quality.yml

@@ -236,7 +236,7 @@ jobs:
         uses: Swatinem/rust-cache@v2
 
       - name: Setup Python
-        uses: actions/setup-python@v5
+        uses: actions/setup-python@v6
         with:
           python-version: '3.x'
 

.github/workflows/l10n.yml

@@ -80,7 +80,7 @@ jobs:
       with:
         persist-credentials: false
     - name: Setup Python
-      uses: actions/setup-python@v5
+      uses: actions/setup-python@v6
       with:
         python-version: '3.x'
     - name: Install Mozilla Fluent Linter

Cargo.lock

@@ -36,6 +36,8 @@ const OPT_FILE: &str = "file";
 const OPT_DEBUG: &str = "debug";
 const OPT_ISO_8601: &str = "iso-8601";
 const OPT_RFC_EMAIL: &str = "rfc-email";
+const OPT_RFC_822: &str = "rfc-822";
+const OPT_RFC_2822: &str = "rfc-2822";
 const OPT_RFC_3339: &str = "rfc-3339";
 const OPT_SET: &str = "set";
 const OPT_REFERENCE: &str = "reference";
@@ -301,6 +303,8 @@ pub fn uu_app() -> Command {
             Arg::new(OPT_RFC_EMAIL)
                 .short('R')
                 .long(OPT_RFC_EMAIL)
+                .alias(OPT_RFC_2822)
+                .alias(OPT_RFC_822)
                 .help(translate!("date-help-rfc-email"))
                 .action(ArgAction::SetTrue),
         )

src/uu/date/src/date.rs

@@ -9,9 +9,7 @@ use std::fs::File;
 use std::io::{BufRead, BufReader, Read, stdin};
 use std::path::Path;
 use uucore::error::{FromIo, UResult, USimpleError, set_exit_code};
-use uucore::translate;
-
-use uucore::{format_usage, show_error};
+use uucore::{format_usage, show_error, translate};
 
 mod helper;
 
@@ -78,7 +76,7 @@ enum NumberingStyle {
     All,
     NonEmpty,
     None,
-    Regex(Box<regex::Regex>),
+    Regex(Box<regex::bytes::Regex>),
 }
 
 impl TryFrom<&str> for NumberingStyle {
@@ -89,7 +87,7 @@ impl TryFrom<&str> for NumberingStyle {
             "a" => Ok(Self::All),
             "t" => Ok(Self::NonEmpty),
             "n" => Ok(Self::None),
-            _ if s.starts_with('p') => match regex::Regex::new(&s[1..]) {
+            _ if s.starts_with('p') => match regex::bytes::Regex::new(&s[1..]) {
                 Ok(re) => Ok(Self::Regex(Box::new(re))),
                 Err(_) => Err(translate!("nl-error-invalid-regex")),
             },
@@ -142,19 +140,30 @@ enum SectionDelimiter {
 impl SectionDelimiter {
     /// A valid section delimiter contains the pattern one to three times,
     /// and nothing else.
-    fn parse(s: &str, pattern: &str) -> Option<Self> {
-        if s.is_empty() || pattern.is_empty() {
+    fn parse(bytes: &[u8], pattern: &str) -> Option<Self> {
+        let pattern = pattern.as_bytes();
+
+        if bytes.is_empty() || pattern.is_empty() || bytes.len() % pattern.len() != 0 {
             return None;
         }
 
-        let pattern_count = s.matches(pattern).count();
-        let is_length_ok = pattern_count * pattern.len() == s.len();
+        let count = bytes.len() / pattern.len();
+        if !(1..=3).contains(&count) {
+            return None;
+        }
 
-        match (pattern_count, is_length_ok) {
-            (3, true) => Some(Self::Header),
-            (2, true) => Some(Self::Body),
-            (1, true) => Some(Self::Footer),
-            _ => None,
+        if bytes
+            .chunks_exact(pattern.len())
+            .all(|chunk| chunk == pattern)
+        {
+            match count {
+                1 => Some(Self::Footer),
+                2 => Some(Self::Body),
+                3 => Some(Self::Header),
+                _ => unreachable!(),
+            }
+        } else {
+            None
         }
     }
 }
@@ -337,9 +346,21 @@ pub fn uu_app() -> Command {
 /// `nl` implements the main functionality for an individual buffer.
 fn nl<T: Read>(reader: &mut BufReader<T>, stats: &mut Stats, settings: &Settings) -> UResult<()> {
     let mut current_numbering_style = &settings.body_numbering;
+    let mut line = Vec::new();
+
+    loop {
+        line.clear();
+        // reads up to and including b'\n'; returns 0 on EOF
+        let n = reader
+            .read_until(b'\n', &mut line)
+            .map_err_context(|| translate!("nl-error-could-not-read-line"))?;
+        if n == 0 {
+            break;
+        }
 
-    for line in reader.lines() {
-        let line = line.map_err_context(|| translate!("nl-error-could-not-read-line"))?;
+        if line.last().copied() == Some(b'\n') {
+            line.pop();
+        }
 
         if line.is_empty() {
             stats.consecutive_empty_lines += 1;
@@ -386,11 +407,12 @@ fn nl<T: Read>(reader: &mut BufReader<T>, stats: &mut Stats, settings: &Settings
                     ));
                 };
                 println!(
-                    "{}{}{line}",
+                    "{}{}{}",
                     settings
                         .number_format
                         .format(line_number, settings.number_width),
                     settings.number_separator.to_string_lossy(),
+                    String::from_utf8_lossy(&line),
                 );
                 // update line number for the potential next line
                 match line_number.checked_add(settings.line_increment) {
@@ -399,7 +421,7 @@ fn nl<T: Read>(reader: &mut BufReader<T>, stats: &mut Stats, settings: &Settings
                 }
             } else {
                 let spaces = " ".repeat(settings.number_width + 1);
-                println!("{spaces}{line}");
+                println!("{spaces}{}", String::from_utf8_lossy(&line));
             }
         }
     }

src/uu/nl/src/nl.rs

@@ -140,26 +140,20 @@ fn create_bundle(
     // Disable Unicode directional isolate characters
     bundle.set_use_isolating(false);
 
-    // Load common strings from uucore locales directory
-    if let Some(common_dir) = find_uucore_locales_dir(locales_dir) {
-        let common_locale_path = common_dir.join(format!("{locale}.ftl"));
-        if let Ok(common_ftl) = fs::read_to_string(&common_locale_path) {
-            if let Ok(common_resource) = FluentResource::try_new(common_ftl) {
-                bundle.add_resource_overriding(common_resource);
-            }
+    let mut try_add_resource_from = |dir_opt: Option<std::path::PathBuf>| {
+        if let Some(resource) = dir_opt
+            .map(|dir| dir.join(format!("{locale}.ftl")))
+            .and_then(|locale_path| fs::read_to_string(locale_path).ok())
+            .and_then(|ftl| fluent_bundle::FluentResource::try_new(ftl).ok())
+        {
+            bundle.add_resource_overriding(resource);
         }
-    }
+    };
 
+    // Load common strings from uucore locales directory
+    try_add_resource_from(find_uucore_locales_dir(locales_dir));
     // Then, try to load utility-specific strings from the utility's locale directory
-    let util_locales_dir = get_locales_dir(util_name).ok();
-    if let Some(util_dir) = util_locales_dir {
-        let util_locale_path = util_dir.join(format!("{locale}.ftl"));
-        if let Ok(util_ftl) = fs::read_to_string(&util_locale_path) {
-            if let Ok(util_resource) = FluentResource::try_new(util_ftl) {
-                bundle.add_resource_overriding(util_resource);
-            }
-        }
-    }
+    try_add_resource_from(get_locales_dir(util_name).ok());
 
     // If we have at least one resource, return the bundle
     if bundle.has_message("common-error") || bundle.has_message(&format!("{util_name}-about")) {

src/uucore/src/lib/mods/locale.rs

@@ -374,6 +374,20 @@ fn test_output_delimiter_with_adjacent_ranges() {
         .stdout_only("ab:cd\n");
 }
 
+#[test]
+fn test_emoji_delim() {
+    new_ucmd!()
+        .args(&["-d🗿", "-f1"])
+        .pipe_in("💐🗿🌹\n")
+        .succeeds()
+        .stdout_only("💐\n");
+    new_ucmd!()
+        .args(&["-d🗿", "-f2"])
+        .pipe_in("💐🗿🌹\n")
+        .succeeds()
+        .stdout_only("🌹\n");
+}
+
 #[cfg(target_os = "linux")]
 #[test]
 fn test_failed_write_is_reported() {

tests/by-util/test_cut.rs

@@ -19,7 +19,7 @@ fn test_invalid_arg() {
 
 #[test]
 fn test_date_email() {
-    for param in ["--rfc-email", "--rfc-e", "-R"] {
+    for param in ["--rfc-email", "--rfc-e", "-R", "--rfc-2822", "--rfc-822"] {
         new_ucmd!().arg(param).succeeds();
     }
 }

tests/by-util/test_date.rs

@@ -566,3 +566,17 @@ fn test_join_non_utf8_paths() {
             .stdout_only("a 1 2\n");
     }
 }
+
+#[test]
+fn join_emoji_delim_inner_key() {
+    let ts = TestScenario::new(util_name!());
+    let at = &ts.fixtures;
+
+    at.write("file1", "a🗿b\n");
+    at.write("file2", "u🗿b\n");
+
+    ts.ucmd()
+        .args(&["-t🗿", "-1", "2", "-2", "2", "file1", "file2"])
+        .succeeds()
+        .stdout_only("b🗿a🗿u\n");
+}

tests/by-util/test_join.rs

@@ -6147,3 +6147,19 @@ fn test_acl_display_symlink() {
         .succeeds()
         .stdout_matches(&re_with_acl);
 }
+
+#[test]
+fn ls_emoji_alignment() {
+    let scene = TestScenario::new(util_name!());
+    let at = &scene.fixtures;
+
+    at.touch("a");
+    at.touch("💐");
+    at.touch("漢");
+    scene
+        .ucmd()
+        .succeeds()
+        .stdout_contains("a")
+        .stdout_contains("💐")
+        .stdout_contains("漢");
+}