Line sorter

The snippets below sort, unique, and reverse banana, Apple, apple, banana, and Cherry — the same lines this page documents. Node calls Array.prototype.sort with no compare function — the same ECMA-262 UTF-16 code-unit order as this page. CPython sorted compares Unicode code points. Those orders match for U+0000–U+FFFF and diverge for supplementary characters. Neither snippet is localeCompare, Unix sort, or a natural/numeric sort. For code units vs locale collation, see UTF-16 vs Locale Sort: Code Units vs localeCompare.

const lines = ['banana', 'Apple', 'apple', 'banana', 'Cherry'];

lines.slice().sort().join('\n');
// "Apple\nCherry\napple\nbanana\nbanana"

[...new Set(lines)].join('\n');
// "banana\nApple\napple\nCherry"

lines.slice().reverse().join('\n');
// "Cherry\nbanana\napple\nApple\nbanana"

lines = ['banana', 'Apple', 'apple', 'banana', 'Cherry']

'\n'.join(sorted(lines))
# 'Apple\nCherry\napple\nbanana\nbanana'

'\n'.join(dict.fromkeys(lines))
# 'banana\nApple\napple\nCherry'

'\n'.join(reversed(lines))
# 'Cherry\nbanana\napple\nApple\nbanana'

Ignore-case sort on this page compares toLowerCase() keys and keeps original casing — Node lines.slice().sort((a, b) => a.toLowerCase().localeCompare(b.toLowerCase())) is locale collation, not this page. Use a UTF-16 / code-point compare on the lowercased keys instead. Python sorted(lines, key=str.lower) matches this page on the BMP sample (Apple, apple, banana, banana, Cherry). str.casefold maps ß to ss; this page leaves ß. "10" still sorts before "2". Case mapping lives on the case converter; normalize first on the Unicode normalizer.

Sort on this page is Array.prototype.sort with no compare function — UTF-16 code-unit order, not localeCompare and not Unix sort. POSIX sort reads a file or stdin and compares with the current LC_COLLATE. LC_ALL=C is byte order; on banana, Apple, apple, banana, and Cherry that matches this page (Apple, Cherry, apple, banana, banana). A UTF-8 locale uses strcoll and can regroup letters — GNU en_US.UTF-8 prints apple, Apple, banana, banana, Cherry. echo appends a newline, so use printf to feed the same five lines. sort writes a trailing newline after the last line; this page joins with LF and does not.

# Byte order (same ASCII order as this page)
printf 'banana\nApple\napple\nbanana\nCherry\n' | LC_ALL=C sort
# Apple
# Cherry
# apple
# banana
# banana

# Locale collation (GNU en_US.UTF-8)
printf 'banana\nApple\napple\nbanana\nCherry\n' | sort
# apple
# Apple
# banana
# banana
# Cherry

LC_ALL=C sort lines.txt

printf 'banana\nApple\napple\nbanana\nCherry\n' | LC_ALL=C sort

sort -u uniques after sorting (Apple, Cherry, apple, banana). Unique on this page keeps the first occurrence without sorting (banana, Apple, apple, Cherry). sort -r reverses the comparison, not the original lines (banana, banana, apple, Cherry, Apple in C). Reverse here is Array.prototype.reverse without sorting (Cherry, banana, apple, Apple, banana). GNU tac and BSD tail -r reverse lines without sorting and match that Reverse when every line ends with LF; GNU tail has no -r, and a last line with no newline can glue onto its neighbor. sort -f folds case; on this sample in C it matches Ignore case (Apple, apple, banana, banana, Cherry). GNU sort is not stable unless -s. sort -n is numeric, so 2 before 10; this page still puts "10" before "2". uniq without sort only drops adjacent duplicates, so both bananas stay on this sample. rev reverses characters inside each line, not line order. Windows has no POSIX sort unless you use WSL or Git Bash; cmd.exe sort is a different utility. Code-unit vs locale collation is in UTF-16 vs Locale Sort: Code Units vs localeCompare.