sort: deduplicate SI unit constants to shared uucore module

Author: sylvestre

src/uu/sort/src/buffer_hint.rs

@@ -9,6 +9,8 @@ use std::ffi::OsString;
 use crate::{
     FALLBACK_AUTOMATIC_BUF_SIZE, MAX_AUTOMATIC_BUF_SIZE, MIN_AUTOMATIC_BUF_SIZE, STDIN_FILE,
 };
+#[cfg(test)]
+use uucore::parser::parse_size::MEGA;
 
 // Heuristics to size the external sort buffer without overcommit memory.
 pub(crate) fn automatic_buffer_size(files: &[OsString]) -> usize {
@@ -135,15 +137,15 @@ mod tests {
 
     #[test]
     fn desired_buffer_matches_total_when_small() {
-        let six_mebibytes = 6 * 1024 * 1024;
+        let six_mebibytes = 6 * MEGA;
         let expected = ((six_mebibytes as u128) * 12)
             .clamp(six_mebibytes as u128, crate::MAX_AUTOMATIC_BUF_SIZE as u128);
         assert_eq!(desired_file_buffer_bytes(six_mebibytes as u128), expected);
     }
 
     #[test]
     fn desired_buffer_caps_at_max_for_large_inputs() {
-        let large = 256 * 1024 * 1024; // 256 MiB
+        let large = 256 * MEGA; // 256 MiB
         assert_eq!(
             desired_file_buffer_bytes(large as u128),
             crate::MAX_AUTOMATIC_BUF_SIZE as u128

src/uu/sort/src/chunks.rs

@@ -22,8 +22,9 @@ use uucore::error::{UResult, USimpleError};
 use crate::{
     GeneralBigDecimalParseResult, GlobalSettings, Line, SortMode, numeric_str_cmp::NumInfo,
 };
+use uucore::parser::parse_size::MEGA;
 
-const MAX_TOKEN_BUFFER_BYTES: usize = 4 * 1024 * 1024;
+const MAX_TOKEN_BUFFER_BYTES: usize = 4 * MEGA;
 const MAX_TOKEN_BUFFER_ELEMS: usize = MAX_TOKEN_BUFFER_BYTES / std::mem::size_of::<Range<usize>>();
 
 self_cell!(
@@ -374,7 +375,7 @@ fn read_to_buffer<T: Read>(
 
                     // We need to read more lines
                     let len = buffer.len();
-                    let grow_by = (len / 2).max(1024 * 1024);
+                    let grow_by = (len / 2).max(MEGA);
                     buffer.resize(len + grow_by, 0);
                     read_target = &mut buffer[len..];
                 } else {

src/uu/sort/src/ext_sort.rs

@@ -35,6 +35,7 @@ use crate::{
     compare_by, merge, sort_by,
 };
 use crate::{Line, print_sorted};
+use uucore::parser::parse_size::MEGA;
 
 // Note: update `test_sort::test_start_buffer` if this size is changed
 const START_BUFFER_SIZE: usize = 8_000;
@@ -116,11 +117,11 @@ fn reader_writer<
     // Cap oversized buffer requests to avoid unnecessary allocations and give the automatic
     // heuristic room to grow when the user does not provide an explicit value.
     let mut buffer_size = match settings.buffer_size {
-        size if size <= 512 * 1024 * 1024 => size,
+        size if size <= 512 * MEGA => size,
         size => size / 2,
     };
     if !settings.buffer_size_is_explicit {
-        buffer_size = buffer_size.max(8 * 1024 * 1024);
+        buffer_size = buffer_size.max(8 * MEGA);
     }
     let read_result: ReadResult<Tmp> = read_write_loop(
         files,

src/uu/sort/src/sort.rs

@@ -52,6 +52,9 @@ use uucore::i18n::collator::locale_cmp;
 use uucore::i18n::decimal::locale_decimal_separator;
 use uucore::line_ending::LineEnding;
 use uucore::parser::num_parser::{ExtendedParser, ExtendedParserError};
+#[cfg(test)]
+use uucore::parser::parse_size::{EXA, TERA};
+use uucore::parser::parse_size::{GIGA, KILO, MEGA};
 use uucore::parser::parse_size::{ParseSizeError, Parser};
 use uucore::parser::shortcut_value_parser::ShortcutValueParser;
 use uucore::posix::{MODERN, TRADITIONAL};
@@ -122,18 +125,6 @@ fn locale_decimal_pt() -> u8 {
 const NEGATIVE: &u8 = &b'-';
 const POSITIVE: &u8 = &b'+';
 
-// Non-breaking space constants
-const UTF8_NBSP: &[u8] = &[0xc2, 0xa0]; // UTF-8 encoding of non-breaking space (U+00A0)
-const ISO_NBSP: u8 = 0xa0; // ISO 8859-1 non-breaking space
-
-// SI unit constants for byte parsing
-const KILO: usize = 1024;
-const MEGA: usize = 1024 * 1024;
-const GIGA: usize = 1024 * 1024 * 1024;
-const TERA: usize = 1024 * 1024 * 1024 * 1024;
-const PETA: usize = 1024 * 1024 * 1024 * 1024 * 1024;
-const EXA: usize = 1024 * 1024 * 1024 * 1024 * 1024 * 1024;
-
 // The automatic buffer heuristics clamp to this range to avoid
 // over-committing memory on constrained systems while still keeping
 // reasonably large chunks for typical workloads.
@@ -3207,7 +3198,7 @@ mod tests {
             ("1b", 1),
             ("1024b", KILO),
             ("1024Mb", KILO * MEGA), // NOTE: This might not be how GNU `sort` behaves for 'Mb'
-            ("1", KILO),                    // K is default
+            ("1", KILO),             // K is default
             ("50", 50 * KILO),
             ("K", KILO),
             ("k", KILO),

src/uucore/src/lib/features/parser/parse_size.rs

@@ -6,6 +6,14 @@
 
 //! Parser for sizes in SI or IEC units (multiples of 1000 or 1024 bytes).
 
+// SI unit constants for byte parsing (powers of 1024)
+pub const KILO: usize = 1024;
+pub const MEGA: usize = 1024 * 1024;
+pub const GIGA: usize = 1024 * 1024 * 1024;
+pub const TERA: usize = 1024 * 1024 * 1024 * 1024;
+pub const PETA: usize = 1024 * 1024 * 1024 * 1024 * 1024;
+pub const EXA: usize = 1024 * 1024 * 1024 * 1024 * 1024 * 1024;
+
 use std::error::Error;
 use std::fmt;
 use std::num::{IntErrorKind, ParseIntError};