sort: add locale-aware month parsing using ICU

Author: ChrisDryden

Cargo.lock

@@ -326,6 +326,7 @@ glob = "0.3.1"
 half = "2.4.1"
 hostname = "0.4"
 icu_collator = "2.0.0"
+icu_datetime = "2.0.1"
 icu_decimal = "2.0.0"
 icu_locale = "2.0.0"
 icu_provider = "2.0.0"

Cargo.toml

@@ -34,7 +34,12 @@ self_cell = { workspace = true }
 tempfile = { workspace = true }
 thiserror = { workspace = true }
 unicode-width = { workspace = true }
-uucore = { workspace = true, features = ["fs", "parser-size", "version-cmp"] }
+uucore = { workspace = true, features = [
+  "fs",
+  "parser-size",
+  "version-cmp",
+  "i18n-month",
+] }
 fluent = { workspace = true }
 
 [target.'cfg(unix)'.dependencies]
@@ -49,6 +54,7 @@ uucore = { workspace = true, features = [
   "parser-size",
   "version-cmp",
   "i18n-collator",
+  "i18n-month",
 ] }
 
 [[bin]]

src/uu/sort/Cargo.toml

@@ -45,6 +45,7 @@ use uucore::error::{FromIo, strip_errno};
 use uucore::error::{UError, UResult, USimpleError, UUsageError};
 use uucore::extendedbigdecimal::ExtendedBigDecimal;
 use uucore::format_usage;
+use uucore::i18n::month::month_parse as locale_month_parse;
 use uucore::line_ending::LineEnding;
 use uucore::parser::num_parser::{ExtendedParser, ExtendedParserError};
 use uucore::parser::parse_size::{ParseSizeError, Parser};
@@ -647,7 +648,7 @@ impl<'a> Line<'a> {
                         .enumerate()
                         .skip_while(|(_, c)| c.is_ascii_whitespace());
 
-                    let month = if month_parse(initial_selection) == Month::Unknown {
+                    let month = if locale_month_parse(initial_selection) == 0 {
                         // We failed to parse a month, which is equivalent to matching nothing.
                         // Add the "no match for key" marker to the first non-whitespace character.
                         let first_non_whitespace = month_chars.next();
@@ -2116,49 +2117,8 @@ fn random_shuffle(a: &[u8], b: &[u8], salt: &[u8]) -> Ordering {
     da.cmp(&db)
 }
 
-#[derive(Eq, Ord, PartialEq, PartialOrd, Clone, Copy)]
-enum Month {
-    Unknown,
-    January,
-    February,
-    March,
-    April,
-    May,
-    June,
-    July,
-    August,
-    September,
-    October,
-    November,
-    December,
-}
-
-/// Parse the beginning string into a Month, returning [`Month::Unknown`] on errors.
-fn month_parse(line: &[u8]) -> Month {
-    let line = line.trim_ascii_start();
-
-    match line.get(..3).map(|x| x.to_ascii_uppercase()).as_deref() {
-        Some(b"JAN") => Month::January,
-        Some(b"FEB") => Month::February,
-        Some(b"MAR") => Month::March,
-        Some(b"APR") => Month::April,
-        Some(b"MAY") => Month::May,
-        Some(b"JUN") => Month::June,
-        Some(b"JUL") => Month::July,
-        Some(b"AUG") => Month::August,
-        Some(b"SEP") => Month::September,
-        Some(b"OCT") => Month::October,
-        Some(b"NOV") => Month::November,
-        Some(b"DEC") => Month::December,
-        _ => Month::Unknown,
-    }
-}
-
 fn month_compare(a: &[u8], b: &[u8]) -> Ordering {
-    let ma = month_parse(a);
-    let mb = month_parse(b);
-
-    ma.cmp(&mb)
+    locale_month_parse(a).cmp(&locale_month_parse(b))
 }
 
 fn print_sorted<'a, T: Iterator<Item = &'a Line<'a>>>(

src/uu/sort/src/sort.rs

@@ -73,6 +73,9 @@ selinux = { workspace = true, optional = true }
 icu_collator = { workspace = true, optional = true, features = [
   "compiled_data",
 ] }
+icu_datetime = { workspace = true, optional = true, features = [
+  "compiled_data",
+] }
 icu_decimal = { workspace = true, optional = true, features = [
   "compiled_data",
 ] }
@@ -143,10 +146,11 @@ format = [
   "quoting-style",
   "unit-prefix",
 ]
-i18n-all = ["i18n-collator", "i18n-decimal"]
+i18n-all = ["i18n-collator", "i18n-decimal", "i18n-month"]
 i18n-common = ["icu_locale"]
 i18n-collator = ["i18n-common", "icu_collator"]
 i18n-decimal = ["i18n-common", "icu_decimal", "icu_provider"]
+i18n-month = ["i18n-common", "icu_datetime", "icu_provider", "libc"]
 mode = ["libc"]
 perms = ["entries", "libc", "walkdir"]
 buf-copy = []

src/uucore/Cargo.toml

@@ -11,6 +11,8 @@ use icu_locale::{Locale, locale};
 pub mod collator;
 #[cfg(feature = "i18n-decimal")]
 pub mod decimal;
+#[cfg(feature = "i18n-month")]
+pub mod month;
 
 /// The encoding specified by the locale, if specified
 /// Currently only supports ASCII and UTF-8 for the sake of simplicity.
@@ -77,6 +79,13 @@ pub fn get_numeric_locale() -> &'static (Locale, UEncoding) {
     NUMERIC_LOCALE.get_or_init(|| get_locale_from_env("LC_NUMERIC"))
 }
 
+/// Get the time locale from the environment (used for month names, etc.)
+pub fn get_time_locale() -> &'static (Locale, UEncoding) {
+    static TIME_LOCALE: OnceLock<(Locale, UEncoding)> = OnceLock::new();
+
+    TIME_LOCALE.get_or_init(|| get_locale_from_env("LC_TIME"))
+}
+
 /// Return the encoding deduced from the locale environment variable.
 pub fn get_locale_encoding() -> UEncoding {
     get_collating_locale().1