Sorting

import { sortUrdu, compareUrdu } from "urdu-text-utils";

sortUrdu(["گل", "آم", "بادام"]);
// ["آم", "بادام", "گل"]

Why not Intl

["ٹماٹر", "تربوز", "پپیتا"].sort();
// wrong — ٹ (U+0679) sorts nowhere near ت (U+062A)

["ٹماٹر", "تربوز", "پپیتا"].sort((a, b) => a.localeCompare(b, "ur"));
// still wrong in most runtimes

The ur locale is not implemented as its own collation in most JavaScript engines; it falls back to Arabic root collation. Arabic has no ٹ, ڈ, ڑ, ں, ے, ک or گ in its alphabet, so their relative order is undefined rather than correct.

This library ships the alphabet explicitly:

ا آ ب پ ت ٹ ث ج چ ح خ د ڈ ذ ر ڑ ز ژ س ش ص ض ط ظ
ع غ ف ق ک گ ل م ن ں و ہ ھ ء ی ے

Behaviour

sortUrdu(["گھر", "کتاب"]); // ["کتاب", "گھر"]  — ک before گ
sortUrdu(["ماں", "مان"]);   // ["مان", "ماں"]   — ن before ں
sortUrdu(["کتابیں", "کتاب"]); // ["کتاب", "کتابیں"] — prefix first

Diacritics and Unicode variants are folded before comparison, so مُحَمَّد sorts exactly where محمد would, and Arabic-spelled input sorts with its Urdu equivalent:

compareUrdu("كتاب", "کتاب"); // 0

Variant letters sort next to their base rather than at the end of the alphabet: ؤ with و, ئ with ی, ۂ with ہ, ۓ with ے. A secondary weight keeps them stable and distinct.

Spaces sort before letters, so ابو بکر comes before ابوبکر. Characters outside the alphabet — Latin, digits, punctuation — sort after every Urdu letter, ordered by codepoint.

Options

sortUrdu(rows, {
  getText: (row) => row.name,
  descending: true,
});

sortUrdu never mutates its input; it returns a new array. Use compareUrdu directly when you need the comparator itself:

rows.sort((a, b) => compareUrdu(a.name, b.name));

Try it

normalizeUrdu(text) stripDiacritics

Input

كيا حال ہے

Output

کیا حال ہے