parsi-text

Tiny, zero-dependency toolkit for cleaning up Persian/Farsi text. It fixes the small-but-painful bugs that break search, sorting, and equality checks when text mixes Arabic and Persian characters or digit systems.

Works anywhere — Node, browser, Vue/Nuxt, React. Ships ESM + TypeScript types.

Why

ك (Arabic kaf) and ک (Persian keheh) look identical but are different code points. So do ي and ی. When users paste text from different keyboards, "کیان" === "كيان" is false, search misses results, and sort order breaks. parsi-text normalizes these away.

Install

npm install parsi-text

Usage

import {
  toPersianDigits,
  toEnglishDigits,
  normalizePersian,
  formatPersianNumber,
  normalizeZwnj,
  isRTL,
} from "parsi-text";

toPersianDigits("2026");        // "۲۰۲۶"
toEnglishDigits("۱۳۹۹");        // "1399"
toEnglishDigits("٤٥٦");         // "456"  (Arabic-Indic digits too)

normalizePersian("كيان");       // "کیان"  (Arabic kaf/yeh -> Persian)
normalizePersian("مُحَمَّد");     // "محمد"   (tashkeel stripped)

formatPersianNumber(1250000);   // "۱٬۲۵۰٬۰۰۰"  (price display)
formatPersianNumber(12.5);      // "۱۲٫۵"

normalizeZwnj("می‌‌‌شود");       // "می‌شود"  (stray half-spaces cleaned)

isRTL("سلام");                  // true
isRTL("hello");                 // false

Making text comparable

const a = normalizePersian("كيان"); // typed with an Arabic keyboard
const b = normalizePersian("کیان"); // typed with a Persian keyboard
a === b; // true

API

| Function | Description | | --- | --- | | toPersianDigits(input) | ASCII digits 0-9 → Persian ۰-۹. Accepts string or number. | | toEnglishDigits(input) | Persian ۰-۹ and Arabic-Indic ٠-٩ → ASCII 0-9. | | normalizePersian(input) | Arabic ك/ي/ى → Persian ک/ی; strips tashkeel and tatweel. Leaves digits untouched. | | formatPersianNumber(input) | Persian digits with Persian thousands (٬) and decimal (٫) separators. Accepts string or number. | | normalizeZwnj(input) | Collapses repeated ZWNJ (half-space) and drops ZWNJs next to whitespace or at string edges; keeps meaningful ones (می‌شود). | | isRTL(input) | true if the string contains any RTL (Arabic/Persian/Hebrew) character. |

All functions are pure, side-effect free, and leave non-target characters unchanged.

License

MIT © Sobhan Azimzadeh