Reverse a string (Unicode-safe) snippet

-- PostgreSQL: split per character, remember positions, re-aggregate DESC.
WITH graphemes AS (
  SELECT c.ch, c.ord
  FROM regexp_split_to_table(E'cafe\u0301 Foo', '') WITH ORDINALITY AS c(ch, ord)
)
SELECT string_agg(ch, '' ORDER BY ord DESC) AS reversed;
-- ooF ´efac — chars here are code points, not clusters: see notes

// Intl.Segmenter (Node 16+, all modern browsers) segments into grapheme
// clusters; [...s] iterates code points and strands combining marks.
const reverse = (s) =>
  [...new Intl.Segmenter('en', { granularity: 'grapheme' }).segment(s)]
    .map(({ segment }) => segment)
    .reverse()
    .join('');

console.log(reverse('cafe\u0301 Foo')); // ooF éfac — the é stayed glued

function reverseGraphemes(s: string): string {
  const segments = new Intl.Segmenter('en', { granularity: 'grapheme' }).segment(s);
  return [...segments].map(({ segment }) => segment).reverse().join('');
}

// the decomposed é (e + U+0301) must survive as ONE cluster
console.log(reverseGraphemes('cafe\u0301 Foo') === 'ooF e\u0301fac'); // true

package main

import (
	"fmt"
	"strings"

	"github.com/rivo/uniseg"
)

func reverseGraphemes(s string) string {
	var clusters []string
	g := uniseg.NewGraphemes(s)
	for g.Next() {
		clusters = append(clusters, g.Str())
	}
	for i, j := 0, len(clusters)-1; i < j; i, j = i+1, j-1 {
		clusters[i], clusters[j] = clusters[j], clusters[i]
	}
	return strings.Join(clusters, "")
}

func main() {
	fmt.Println(reverseGraphemes("cafe\u0301 Foo")) // ooF éfac
}

use unicode_segmentation::UnicodeSegmentation;

fn main() {
    let s = "cafe\u{0301} Foo";

    // .chars().rev().collect() is code-point-wise: valid UTF-8, but
    // U+0301 detaches from its base. graphemes(true) = extended clusters.
    let reversed: String = s.graphemes(true).rev().collect();
    assert_eq!(reversed, "ooF e\u{0301}fac");
    println!("{reversed}");
}

<?php
function reverseGraphemes(string $s): string {
    $clusters = [];
    $offset = 0; // byte position, advanced by reference each call
    while ($offset < strlen($s)) {
        $clusters[] = grapheme_extract($s, 1, GRAPHEME_EXTR_COUNT, $offset, $offset);
    }
    return implode('', array_reverse($clusters));
}

echo reverseGraphemes("cafe\u{0301} Foo"), PHP_EOL; // ooF éfac

import grapheme

def reverse_graphemes(s: str) -> str:
    # graphemes() yields clusters; reversed() needs a sequence,
    # so materialize the generator before reversing it.
    return "".join(reversed(list(grapheme.graphemes(s))))

assert reverse_graphemes("cafe\u0301 Foo") == "ooF e\u0301fac"
print(reverse_graphemes("cafe\u0301 Foo")) # ooF éfac

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

/* Return a malloc'd copy of s with its CODE POINTS (not bytes)
   reversed — caller frees. A UTF-8 code point begins at any byte
   that is not a 0b10xxxxxx continuation byte; walk backwards. */
char *reverse_utf8(const char *s) {
    size_t len = strlen(s);
    char *out = malloc(len + 1);
    if (out == NULL) return NULL;
    size_t o = 0;
    size_t i = len;
    while (i > 0) {
        size_t start = i - 1;
        while (start > 0 && (((unsigned char)s[start] & 0xC0) == 0x80))
            start--;                    /* skip continuation bytes */
        memcpy(out + o, s + start, i - start);
        o += i - start;
        i = start;
    }
    out[o] = '\0';
    return out;
}

int main(void) {
    char *r = reverse_utf8("cafe\u0301 Foo");
    if (r != NULL) {
        printf("%s\n", r); /* ooF ´efac — the accent strands: see notes */
        free(r);
    }
    return 0;
}

#include <iostream>
#include <string>
#include <string_view>
#include <vector>

// Reverse by code point: a code point begins at any byte that is
// not a 0b10xxxxxx continuation byte. std::string iterators are
// BYTE-wise — std::reverse would emit invalid UTF-8.
std::string reverse_utf8(std::string_view s) {
    std::vector<std::string_view> cps;
    for (std::size_t i = 0; i < s.size();) {
        unsigned char c = static_cast<unsigned char>(s[i]);
        std::size_t n = c >= 0xF0 ? 4 : c >= 0xE0 ? 3 : c >= 0xC0 ? 2 : 1;
        cps.push_back(s.substr(i, n));
        i += n;
    }
    std::string out;
    for (auto it = cps.rbegin(); it != cps.rend(); ++it) out += *it;
    return out;
}

int main() {
    std::cout << reverse_utf8("cafe\u0301 Foo") << '\n'; // ooF ´efac
}

using System.Globalization;

static string Reverse(string s)
{
    var clusters = new List<string>();
    var it = StringInfo.GetTextElementEnumerator(s);
    while (it.MoveNext())
    {
        clusters.Add((string)it.Current!); // text element = grapheme cluster
    }
    clusters.Reverse();
    return string.Concat(clusters);
}

Console.WriteLine(Reverse("cafe\u0301 Foo")); // ooF éfac

import java.text.BreakIterator;
import java.util.ArrayList;
import java.util.Collections;
import java.util.List;

public class ReverseUnicode {
    // StringBuilder.reverse() keeps surrogate PAIRS intact but still
    // detaches combining marks — the character BreakIterator segments
    // user-perceived characters (grapheme clusters).
    static String reverse(String s) {
        BreakIterator it = BreakIterator.getCharacterInstance();
        it.setText(s);
        List<String> clusters = new ArrayList<>();
        int start = it.first();
        for (int end = it.next(); end != BreakIterator.DONE; start = end, end = it.next()) {
            clusters.add(s.substring(start, end));
        }
        Collections.reverse(clusters);
        return String.join("", clusters);
    }

    public static void main(String[] args) {
        System.out.println(reverse("cafe\u0301 Foo")); // ooF éfac
    }
}

let s = "cafe\u{301} Foo"

// Swift's Character IS an extended grapheme cluster, so the obvious
// one-liner is already Unicode-correct — the one language where
// naive wins.
print(String(s.reversed())) // ooF éfac

import java.text.BreakIterator

fun reverseGraphemes(s: String): String {
    val it = BreakIterator.getCharacterInstance().apply { setText(s) }
    val clusters = buildList {
        var start = it.first()
        var end = it.next()
        while (end != BreakIterator.DONE) {
            add(s.substring(start, end))
            start = end
            end = it.next()
        }
    }
    return clusters.asReversed().joinToString("")
}

fun main() {
    println(reverseGraphemes("cafe\u0301 Foo")) // ooF éfac
}

s = "cafe\u0301 Foo"

# .reverse is code-point-wise: U+0301 lands BEFORE its 'e'.
# grapheme_clusters (Ruby 2.5+) splits on extended clusters.
puts s.grapheme_clusters.reverse.join # ooF éfac

const std = @import("std");

pub fn main() void {
    const s = "cafe\u{301} Foo";

    // Walk the UTF-8 bytes backwards: a code point begins at any
    // byte that is NOT a 0b10xxxxxx continuation byte.
    var out: [128]u8 = undefined;
    var len: usize = 0;
    var i: usize = s.len;
    while (i > 0) {
        var start = i - 1;
        while (start > 0 and (s[start] & 0b1100_0000) == 0b1000_0000) start -= 1;
        const cp = s[start..i];
        @memcpy(out[len .. len + cp.len], cp);
        len += cp.len;
        i = start;
    }
    std.debug.print("{s}\n", .{out[0..len]}); // ooF ´efac
}