urdu-text-utils 0.1.0

package/dist/index.cjs

@@ -0,0 +1,804 @@
+'use strict';
+
+// src/chars.ts
+var DIACRITIC_RANGES = "\\u0610-\\u061A\\u064B-\\u065F\\u0670\\u06D6-\\u06ED";
+var DIACRITICS_RE = new RegExp(`[${DIACRITIC_RANGES}]`, "gu");
+var TATWEEL_RE = /ـ/gu;
+var INVISIBLE_RE = /[​‍-‏‪-‮⁠-⁤⁦-⁩]/gu;
+var ZWNJ_RE = /‌/gu;
+var ARABIC_LETTER_RANGES = "\\u0620-\\u063F\\u0641-\\u064A\\u066E\\u066F\\u0671-\\u06D3\\u06D5\\u06EE\\u06EF\\u06FA-\\u06FF\\u0750-\\u077F\\u08A0-\\u08FF\\uFB50-\\uFDFF\\uFE70-\\uFEFC";
+var ARABIC_LETTER_RE = new RegExp(`[${ARABIC_LETTER_RANGES}]`, "u");
+var ANY_LETTER_RE_G = /\p{L}/gu;
+var URDU_DIGITS = "\u06F0\u06F1\u06F2\u06F3\u06F4\u06F5\u06F6\u06F7\u06F8\u06F9";
+var ARABIC_INDIC_DIGITS = "\u0660\u0661\u0662\u0663\u0664\u0665\u0666\u0667\u0668\u0669";
+var ASCII_DIGITS = "0123456789";
+var ANY_DIGIT_RE_G = new RegExp(`[${ASCII_DIGITS}${URDU_DIGITS}${ARABIC_INDIC_DIGITS}]`, "gu");
+var SENTENCE_SPLIT_RE = /[۔؟?!.…]+/u;
+var WORD_SPLIT_RE = /[\s،؛؟۔٫٬…!-\/:-@\[-`{-~‐-‧«»‘’“”]+/u;
+
+// src/normalize.ts
+var URDU_YEH = "\u06CC";
+var URDU_KEHEH = "\u06A9";
+var URDU_HEH_GOAL = "\u06C1";
+var ALEF = "\u0627";
+var WAW = "\u0648";
+var LETTER_MAP = {
+  // yeh family -> ی. Note ے (U+06D2 bari ye) is a distinct letter and is preserved.
+  "\u064A": URDU_YEH,
+  // ي arabic yeh
+  "\u0649": URDU_YEH,
+  // ى alef maksura
+  "\u06CD": URDU_YEH,
+  // ۍ yeh with tail
+  "\u06D0": URDU_YEH,
+  // ې pashto e
+  "\u0620": URDU_YEH,
+  // ؠ kashmiri yeh
+  // kaf family -> ک
+  "\u0643": URDU_KEHEH,
+  // ك arabic kaf
+  "\u06AA": URDU_KEHEH,
+  // ڪ swash kaf
+  // heh family -> ہ. ھ (U+06BE do-chashmi heh) is a distinct letter and is preserved.
+  "\u0647": URDU_HEH_GOAL,
+  // ه arabic heh
+  "\u06C0": URDU_HEH_GOAL,
+  // ۀ heh with yeh above
+  "\u0629": URDU_HEH_GOAL,
+  // ة teh marbuta
+  "\u06C3": URDU_HEH_GOAL,
+  // ۃ teh marbuta goal
+  "\u06D5": URDU_HEH_GOAL,
+  // ە ae
+  // alef family -> ا. آ (U+0622 alef madda) is a real Urdu letter and is preserved.
+  "\u0623": ALEF,
+  // أ alef with hamza above
+  "\u0625": ALEF,
+  // إ alef with hamza below
+  "\u0671": ALEF,
+  // ٱ alef wasla
+  "\u0672": ALEF,
+  // ٲ alef with wavy hamza above
+  "\u0673": ALEF,
+  // ٳ alef with wavy hamza below
+  // waw family -> و. ؤ (U+0624) is preserved: it carries hamza meaningfully in Urdu.
+  "\u06CB": WAW,
+  // ۋ ve
+  "\u06C6": WAW,
+  // ۆ oe
+  "\u06C7": WAW
+  // ۇ u
+};
+var digitTargets = {
+  urdu: URDU_DIGITS,
+  english: ASCII_DIGITS,
+  arabic: ARABIC_INDIC_DIGITS
+};
+var PUNCTUATION_MAP = {
+  ",": "\u060C",
+  // ،
+  ";": "\u061B",
+  // ؛
+  "?": "\u061F"
+  // ؟
+};
+var DIGIT_LOOKUP = /* @__PURE__ */ new Map();
+for (const set of [ASCII_DIGITS, URDU_DIGITS, ARABIC_INDIC_DIGITS]) {
+  for (let i = 0; i < 10; i++) DIGIT_LOOKUP.set(set[i], i);
+}
+function normalizeUrdu(input, options = {}) {
+  if (!input) return "";
+  const {
+    compatibility = true,
+    stripDiacritics = false,
+    stripTatweel = true,
+    stripZwnj = false,
+    collapseWhitespace = true,
+    digits = "preserve",
+    urduPunctuation = false
+  } = options;
+  let text = compatibility ? input.normalize("NFKC") : input.normalize("NFC");
+  text = text.replace(INVISIBLE_RE, "");
+  if (stripZwnj) text = text.replace(ZWNJ_RE, "");
+  if (stripTatweel) text = text.replace(TATWEEL_RE, "");
+  if (stripDiacritics) text = text.replace(DIACRITICS_RE, "");
+  const digitTarget = digits === "preserve" ? null : digitTargets[digits];
+  let out = "";
+  for (const ch of text) {
+    const mapped = LETTER_MAP[ch];
+    if (mapped !== void 0) {
+      out += mapped;
+      continue;
+    }
+    if (digitTarget) {
+      const value = DIGIT_LOOKUP.get(ch);
+      if (value !== void 0) {
+        out += digitTarget[value];
+        continue;
+      }
+    }
+    if (urduPunctuation) {
+      const punct = PUNCTUATION_MAP[ch];
+      if (punct !== void 0) {
+        out += punct;
+        continue;
+      }
+    }
+    out += ch;
+  }
+  if (collapseWhitespace) out = out.replace(/\s+/gu, " ").trim();
+  return out;
+}
+function removeDiacritics(input) {
+  if (!input) return "";
+  return input.normalize("NFC").replace(DIACRITICS_RE, "");
+}
+function foldUrdu(input) {
+  return normalizeUrdu(input, {
+    stripDiacritics: true,
+    stripZwnj: true,
+    collapseWhitespace: true
+  }).toLowerCase();
+}
+
+// src/numbers.ts
+var SETS = {
+  urdu: URDU_DIGITS,
+  english: ASCII_DIGITS,
+  arabic: ARABIC_INDIC_DIGITS
+};
+var VALUE_OF = /* @__PURE__ */ new Map();
+for (const set of Object.values(SETS)) {
+  for (let i = 0; i < 10; i++) VALUE_OF.set(set[i], i);
+}
+function convert(input, target) {
+  if (!input) return "";
+  const digits = SETS[target];
+  let out = "";
+  for (const ch of input) {
+    const value = VALUE_OF.get(ch);
+    out += value === void 0 ? ch : digits[value];
+  }
+  return out;
+}
+function toUrduDigits(input) {
+  return convert(input, "urdu");
+}
+function toEnglishDigits(input) {
+  return convert(input, "english");
+}
+function toArabicIndicDigits(input) {
+  return convert(input, "arabic");
+}
+function convertNumbers(input, to = "urdu") {
+  return convert(input, to);
+}
+function parseUrduNumber(input) {
+  if (!input) return NaN;
+  const ascii = toEnglishDigits(input).replace(/[٬,\s]/gu, "").replace(/٫/gu, ".");
+  if (!/^[+-]?(\d+(\.\d*)?|\.\d+)$/u.test(ascii)) return NaN;
+  return Number(ascii);
+}
+var ONES = ["", "\u0627\u06CC\u06A9", "\u062F\u0648", "\u062A\u06CC\u0646", "\u0686\u0627\u0631", "\u067E\u0627\u0646\u0686", "\u0686\u06BE", "\u0633\u0627\u062A", "\u0622\u0679\u06BE", "\u0646\u0648"];
+var TEENS_AND_TENS = {
+  10: "\u062F\u0633",
+  11: "\u06AF\u06CC\u0627\u0631\u06C1",
+  12: "\u0628\u0627\u0631\u06C1",
+  13: "\u062A\u06CC\u0631\u06C1",
+  14: "\u0686\u0648\u062F\u06C1",
+  15: "\u067E\u0646\u062F\u0631\u06C1",
+  16: "\u0633\u0648\u0644\u06C1",
+  17: "\u0633\u062A\u0631\u06C1",
+  18: "\u0627\u0679\u06BE\u0627\u0631\u06C1",
+  19: "\u0627\u0646\u06CC\u0633",
+  20: "\u0628\u06CC\u0633",
+  30: "\u062A\u06CC\u0633",
+  40: "\u0686\u0627\u0644\u06CC\u0633",
+  50: "\u067E\u0686\u0627\u0633",
+  60: "\u0633\u0627\u0679\u06BE",
+  70: "\u0633\u062A\u0631",
+  80: "\u0627\u0633\u06CC",
+  90: "\u0646\u0648\u06D2",
+  100: "\u0633\u0648"
+};
+function numberToUrduWords(value) {
+  if (!Number.isFinite(value)) return "";
+  if (!Number.isInteger(value)) throw new TypeError("numberToUrduWords expects an integer");
+  if (value < 0) return `\u0645\u0646\u0641\u06CC ${numberToUrduWords(-value)}`;
+  if (value === 0) return "\u0635\u0641\u0631";
+  const scales = [
+    [1e7, "\u06A9\u0631\u0648\u0691"],
+    [1e5, "\u0644\u0627\u06A9\u06BE"],
+    [1e3, "\u06C1\u0632\u0627\u0631"],
+    [100, "\u0633\u0648"]
+  ];
+  const parts = [];
+  let rest = value;
+  for (const [size, name] of scales) {
+    if (rest >= size) {
+      const count = Math.floor(rest / size);
+      rest %= size;
+      parts.push(`${numberToUrduWords(count)} ${name}`);
+    }
+  }
+  if (rest > 0) {
+    if (TEENS_AND_TENS[rest]) {
+      parts.push(TEENS_AND_TENS[rest]);
+    } else if (rest < 10) {
+      parts.push(ONES[rest]);
+    } else {
+      const tens = Math.floor(rest / 10) * 10;
+      parts.push(`${TEENS_AND_TENS[tens] ?? ""} ${ONES[rest % 10] ?? ""}`.trim());
+    }
+  }
+  return parts.join(" ").replace(/\s+/gu, " ").trim();
+}
+
+// src/detect.ts
+function urduRatio(input) {
+  if (!input) return 0;
+  let letters = 0;
+  let urdu = 0;
+  for (const ch of input.match(ANY_LETTER_RE_G) ?? []) {
+    letters++;
+    if (ARABIC_LETTER_RE.test(ch)) urdu++;
+  }
+  return letters === 0 ? 0 : urdu / letters;
+}
+function isUrdu(input, options = {}) {
+  const { threshold = 0.5, minLetters = 1 } = options;
+  if (!input) return false;
+  let letters = 0;
+  let urdu = 0;
+  for (const ch of input.match(ANY_LETTER_RE_G) ?? []) {
+    letters++;
+    if (ARABIC_LETTER_RE.test(ch)) urdu++;
+  }
+  if (urdu < minLetters) return false;
+  return letters > 0 && urdu / letters >= threshold;
+}
+var URDU_ONLY = /* @__PURE__ */ new Set([
+  "\u0679",
+  // ٹ tteh
+  "\u0688",
+  // ڈ ddal
+  "\u0691",
+  // ڑ rreh
+  "\u06BA",
+  // ں noon ghunna
+  "\u06D2",
+  // ے bari ye
+  "\u06D3",
+  // ۓ bari ye with hamza
+  "\u06C1",
+  // ہ heh goal
+  "\u06BE",
+  // ھ do-chashmi heh
+  "\u06A9",
+  // ک keheh
+  "\u06AF",
+  // گ gaf
+  "\u0686",
+  // چ tcheh
+  "\u067E",
+  // پ peh
+  "\u0698",
+  // ژ jeh
+  "\u06CC"
+  // ی farsi yeh
+]);
+function hasUrduSpecificLetters(input) {
+  for (const ch of input) if (URDU_ONLY.has(ch)) return true;
+  return false;
+}
+
+// src/stats.ts
+function countWords(input) {
+  return splitWords(input).length;
+}
+function splitWords(input) {
+  if (!input) return [];
+  return input.split(WORD_SPLIT_RE).map((w) => w.trim()).filter((w) => w.length > 0);
+}
+function countSentences(input) {
+  return splitSentences(input).length;
+}
+function splitSentences(input) {
+  if (!input) return [];
+  return input.split(SENTENCE_SPLIT_RE).map((s) => s.trim()).filter((s) => s.length > 0);
+}
+function analyzeUrdu(input) {
+  const text = input ?? "";
+  const characters = [...text].length;
+  const charactersNoSpaces = [...text.replace(/\s/gu, "")].length;
+  const words = countWords(text);
+  const sentences = countSentences(text);
+  const paragraphs = text.split(/\n\s*\n/u).map((p) => p.trim()).filter((p) => p.length > 0).length;
+  return {
+    characters,
+    charactersNoSpaces,
+    words,
+    sentences,
+    paragraphs,
+    urduPercentage: Math.round(urduRatio(text) * 100),
+    diacritics: (text.match(DIACRITICS_RE) ?? []).length,
+    digits: (text.match(ANY_DIGIT_RE_G) ?? []).length,
+    averageWordsPerSentence: sentences === 0 ? 0 : Math.round(words / sentences * 10) / 10,
+    readingTimeMinutes: Math.round(words / 180 * 10) / 10
+  };
+}
+
+// src/collate.ts
+var ALPHABET = [
+  "\u0627",
+  "\u0622",
+  "\u0628",
+  "\u067E",
+  "\u062A",
+  "\u0679",
+  "\u062B",
+  "\u062C",
+  "\u0686",
+  "\u062D",
+  "\u062E",
+  "\u062F",
+  "\u0688",
+  "\u0630",
+  "\u0631",
+  "\u0691",
+  "\u0632",
+  "\u0698",
+  "\u0633",
+  "\u0634",
+  "\u0635",
+  "\u0636",
+  "\u0637",
+  "\u0638",
+  "\u0639",
+  "\u063A",
+  "\u0641",
+  "\u0642",
+  "\u06A9",
+  "\u06AF",
+  "\u0644",
+  "\u0645",
+  "\u0646",
+  "\u06BA",
+  "\u0648",
+  "\u06C1",
+  "\u06BE",
+  "\u0621",
+  "\u06CC",
+  "\u06D2"
+];
+var VARIANTS = {
+  "\u0624": ["\u0648", 1],
+  // ؤ waw with hamza
+  "\u0626": ["\u06CC", 1],
+  // ئ yeh with hamza
+  "\u06C2": ["\u06C1", 1],
+  // ۂ heh goal with hamza
+  "\u06D3": ["\u06D2", 1]
+  // ۓ bari ye with hamza
+};
+var WEIGHTS = /* @__PURE__ */ new Map();
+ALPHABET.forEach((ch, index) => WEIGHTS.set(ch, [(index + 1) * 10, 0]));
+for (const [ch, [base, secondary]] of Object.entries(VARIANTS)) {
+  const baseWeight = WEIGHTS.get(base);
+  if (baseWeight) WEIGHTS.set(ch, [baseWeight[0], secondary]);
+}
+var NON_LETTER_BASE = (ALPHABET.length + 1) * 10;
+function weightsOf(text) {
+  const out = [];
+  for (const ch of text) {
+    const weight = WEIGHTS.get(ch);
+    if (weight) {
+      out.push(weight[0], weight[1]);
+    } else if (ch === " ") {
+      out.push(1, 0);
+    } else {
+      out.push(NON_LETTER_BASE + (ch.codePointAt(0) ?? 0), 0);
+    }
+  }
+  return out;
+}
+function compareUrdu(a, b) {
+  const left = weightsOf(foldUrdu(a));
+  const right = weightsOf(foldUrdu(b));
+  const length = Math.min(left.length, right.length);
+  for (let i = 0; i < length; i++) {
+    const diff = left[i] - right[i];
+    if (diff !== 0) return diff;
+  }
+  return left.length - right.length;
+}
+function sortUrdu(items, options = {}) {
+  const { descending = false, getText } = options;
+  const key = getText ?? ((item) => String(item));
+  const sorted = [...items].sort((a, b) => compareUrdu(key(a), key(b)));
+  return descending ? sorted.reverse() : sorted;
+}
+
+// src/search.ts
+function editDistance(a, b, limit = Infinity) {
+  if (a === b) return 0;
+  if (Math.abs(a.length - b.length) > limit) return limit + 1;
+  let previous = Array.from({ length: b.length + 1 }, (_, i) => i);
+  let current = new Array(b.length + 1);
+  for (let i = 1; i <= a.length; i++) {
+    current[0] = i;
+    let rowMin = current[0];
+    for (let j = 1; j <= b.length; j++) {
+      const cost = a[i - 1] === b[j - 1] ? 0 : 1;
+      current[j] = Math.min(current[j - 1] + 1, previous[j] + 1, previous[j - 1] + cost);
+      rowMin = Math.min(rowMin, current[j]);
+    }
+    if (rowMin > limit) return limit + 1;
+    [previous, current] = [current, previous];
+  }
+  return previous[b.length];
+}
+function scoreOf(query, text, options) {
+  if (!query) return 0;
+  if (text === query) return 1;
+  if (text.startsWith(query)) return 0.9;
+  if (text.includes(query)) return 0.8;
+  if (!options.fuzzy) return 0;
+  const queryWords = splitWords(query);
+  const textWords = splitWords(text);
+  if (queryWords.length === 0 || textWords.length === 0) return 0;
+  let matched = 0;
+  for (const qw of queryWords) {
+    const limit = Math.min(options.maxDistance, Math.max(1, Math.floor(qw.length / 3)));
+    const hit = textWords.some(
+      (tw) => tw.includes(qw) || editDistance(qw, tw, limit) <= limit
+    );
+    if (hit) matched++;
+  }
+  if (matched === 0) return 0;
+  return matched / queryWords.length * 0.7;
+}
+function searchUrdu(query, items, options = {}) {
+  return searchUrduRanked(query, items, options).map((r) => r.item);
+}
+function searchUrduRanked(query, items, options = {}) {
+  const { getText, fuzzy = false, maxDistance = 1, limit, sortByScore = true } = options;
+  const read = getText ?? ((item) => String(item));
+  const folded = foldUrdu(query);
+  if (!folded) return [];
+  const results = [];
+  for (const item of items) {
+    const score = scoreOf(folded, foldUrdu(read(item)), { fuzzy, maxDistance });
+    if (score > 0) results.push({ item, score });
+  }
+  if (sortByScore) results.sort((a, b) => b.score - a.score);
+  return limit === void 0 ? results : results.slice(0, limit);
+}
+function highlightUrdu(text, query, wrap = (m) => `<mark>${m}</mark>`) {
+  const foldedQuery = foldUrdu(query);
+  if (!foldedQuery || !text) return text;
+  const sourceChars = [...text];
+  const foldedChars = [];
+  const sourceIndexOf = [];
+  sourceChars.forEach((ch, i) => {
+    const folded = normalizeUrdu(ch, {
+      stripDiacritics: true,
+      stripZwnj: true,
+      collapseWhitespace: false
+    }).toLowerCase();
+    for (const f of folded) {
+      foldedChars.push(f);
+      sourceIndexOf.push(i);
+    }
+  });
+  const haystack = foldedChars.join("");
+  let out = "";
+  let cursor = 0;
+  let from = 0;
+  for (; ; ) {
+    const hit = haystack.indexOf(foldedQuery, from);
+    if (hit === -1) break;
+    const startSource = sourceIndexOf[hit];
+    const endSource = (sourceIndexOf[hit + foldedQuery.length - 1] ?? startSource) + 1;
+    out += sourceChars.slice(cursor, startSource).join("");
+    out += wrap(sourceChars.slice(startSource, endSource).join(""));
+    cursor = endSource;
+    from = hit + foldedQuery.length;
+  }
+  out += sourceChars.slice(cursor).join("");
+  return out;
+}
+
+// src/transliterate.ts
+var WORD_DICTIONARY = {
+  "\u0622\u067E": "aap",
+  "\u0622\u067E\u06A9\u0627": "aapka",
+  "\u0627\u0648\u0631": "aur",
+  "\u0627\u06CC\u06A9": "aik",
+  "\u0627\u062D\u0645\u062F": "ahmed",
+  "\u0627\u0686\u06BE\u0627": "acha",
+  "\u0628\u06C1\u062A": "bohat",
+  "\u067E\u0627\u06A9\u0633\u062A\u0627\u0646": "pakistan",
+  "\u067E\u0627\u0646\u06CC": "paani",
+  "\u067E\u0631": "par",
+  "\u067E\u06C1\u0644\u0627": "pehla",
+  "\u062A\u0645": "tum",
+  "\u062A\u06BE\u0627": "tha",
+  "\u062A\u06BE\u06CC": "thi",
+  "\u062D\u0627\u0644": "haal",
+  "\u062E\u0627\u0646": "khan",
+  "\u062E\u0648\u0628\u0635\u0648\u0631\u062A": "khoobsurat",
+  "\u062F\u0646": "din",
+  "\u0631\u0627\u062A": "raat",
+  "\u0632\u06CC\u062F": "zaid",
+  "\u0633\u06D2": "se",
+  "\u0634\u06C1\u0631": "shehar",
+  "\u0639\u0644\u06CC": "ali",
+  "\u06A9\u0627": "ka",
+  "\u06A9\u062A\u0627\u0628": "kitaab",
+  "\u06A9\u0631": "kar",
+  "\u06A9\u0631\u0646\u0627": "karna",
+  "\u06A9\u06D2": "ke",
+  "\u06A9\u06CC": "ki",
+  "\u06A9\u06CC\u0627": "kya",
+  "\u06A9\u06CC\u0633\u06D2": "kaisay",
+  "\u0644\u0691\u06A9\u0627": "larka",
+  "\u0644\u0691\u06A9\u06CC": "larki",
+  "\u0644\u06CC\u06D2": "liye",
+  "\u0645\u062D\u0645\u062F": "muhammad",
+  "\u0645\u0636\u0645\u0648\u0646": "mazmoon",
+  "\u0645\u0644\u06A9": "mulk",
+  "\u0645\u06CC\u06BA": "mein",
+  "\u0645\u06CC\u0631\u0627": "mera",
+  "\u0645\u06CC\u0631\u06CC": "meri",
+  "\u0646\u0627\u0645": "naam",
+  "\u0646\u06C1\u06CC\u06BA": "nahi",
+  "\u0648\u06C1": "woh",
+  "\u06C1\u06D2": "hai",
+  "\u06C1\u06CC\u06BA": "hain",
+  "\u06C1\u0648": "ho",
+  "\u06C1\u0648\u06BA": "hoon",
+  "\u06CC\u06C1": "yeh"
+};
+var DIGRAPHS = {
+  "\u0628\u06BE": "bh",
+  "\u067E\u06BE": "ph",
+  "\u062A\u06BE": "th",
+  "\u0679\u06BE": "th",
+  "\u062C\u06BE": "jh",
+  "\u0686\u06BE": "chh",
+  "\u062F\u06BE": "dh",
+  "\u0688\u06BE": "dh",
+  "\u0691\u06BE": "rh",
+  "\u06A9\u06BE": "kh",
+  "\u06AF\u06BE": "gh",
+  "\u0644\u06BE": "lh",
+  "\u0645\u06BE": "mh",
+  "\u0646\u06BE": "nh",
+  "\u0631\u06BE": "rh"
+};
+var LETTERS = {
+  "\u0627": "a",
+  "\u0622": "aa",
+  "\u0628": "b",
+  "\u067E": "p",
+  "\u062A": "t",
+  "\u0679": "t",
+  "\u062B": "s",
+  "\u062C": "j",
+  "\u0686": "ch",
+  "\u062D": "h",
+  "\u062E": "kh",
+  "\u062F": "d",
+  "\u0688": "d",
+  "\u0630": "z",
+  "\u0631": "r",
+  "\u0691": "r",
+  "\u0632": "z",
+  "\u0698": "zh",
+  "\u0633": "s",
+  "\u0634": "sh",
+  "\u0635": "s",
+  "\u0636": "z",
+  "\u0637": "t",
+  "\u0638": "z",
+  "\u0639": "a",
+  "\u063A": "gh",
+  "\u0641": "f",
+  "\u0642": "q",
+  "\u06A9": "k",
+  "\u06AF": "g",
+  "\u0644": "l",
+  "\u0645": "m",
+  "\u0646": "n",
+  "\u06BA": "n",
+  "\u0648": "o",
+  "\u06C1": "h",
+  "\u06BE": "h",
+  "\u0621": "",
+  "\u06CC": "i",
+  "\u0626": "i",
+  "\u06D2": "e",
+  "\u06D3": "e",
+  "\u0624": "o",
+  "\u06C2": "h"
+};
+function romanizeWord(word) {
+  const dictionary = WORD_DICTIONARY[word];
+  if (dictionary) return dictionary;
+  const chars = [...word];
+  let out = "";
+  for (let i = 0; i < chars.length; i++) {
+    const pair = chars[i] + (chars[i + 1] ?? "");
+    const digraph = DIGRAPHS[pair];
+    if (digraph) {
+      out += digraph;
+      i++;
+      continue;
+    }
+    const ch = chars[i];
+    if (i === 0 && ch === "\u0627") {
+      const next = chars[1];
+      if (next === "\u0648") {
+        out += "o";
+        i++;
+        continue;
+      }
+      if (next === "\u06CC") {
+        out += "i";
+        i++;
+        continue;
+      }
+      out += "a";
+      continue;
+    }
+    out += LETTERS[ch] ?? ch;
+  }
+  return out;
+}
+function romanize(input, options = {}) {
+  if (!input) return "";
+  const clean = removeDiacritics(normalizeUrdu(input));
+  const roman = clean.split(/(\s+)/u).map((chunk) => /^\s+$/u.test(chunk) ? chunk : romanizeWord(chunk)).join("");
+  if (!options.capitalize) return roman;
+  return roman.charAt(0).toUpperCase() + roman.slice(1);
+}
+var ROMAN_DICTIONARY = {};
+for (const [urduWord, roman] of Object.entries(WORD_DICTIONARY)) {
+  ROMAN_DICTIONARY[roman] = urduWord;
+}
+Object.assign(ROMAN_DICTIONARY, {
+  hai: "\u06C1\u06D2",
+  hay: "\u06C1\u06D2",
+  he: "\u06C1\u06D2",
+  hein: "\u06C1\u06CC\u06BA",
+  hoon: "\u06C1\u0648\u06BA",
+  hun: "\u06C1\u0648\u06BA",
+  ap: "\u0622\u067E",
+  kaise: "\u06A9\u06CC\u0633\u06D2",
+  kaisa: "\u06A9\u06CC\u0633\u0627",
+  nam: "\u0646\u0627\u0645",
+  naam: "\u0646\u0627\u0645",
+  mai: "\u0645\u06CC\u06BA",
+  main: "\u0645\u06CC\u06BA",
+  nahin: "\u0646\u06C1\u06CC\u06BA",
+  kia: "\u06A9\u06CC\u0627",
+  bahut: "\u0628\u06C1\u062A",
+  bhot: "\u0628\u06C1\u062A",
+  acha: "\u0627\u0686\u06BE\u0627",
+  achha: "\u0627\u0686\u06BE\u0627",
+  shukriya: "\u0634\u06A9\u0631\u06CC\u06C1",
+  salam: "\u0633\u0644\u0627\u0645"
+});
+var ROMAN_RULES = [
+  [/^kh/u, "\u06A9\u06BE"],
+  [/^gh/u, "\u06AF\u06BE"],
+  [/^chh/u, "\u0686\u06BE"],
+  [/^ch/u, "\u0686"],
+  [/^sh/u, "\u0634"],
+  [/^th/u, "\u062A\u06BE"],
+  [/^ph/u, "\u067E\u06BE"],
+  [/^bh/u, "\u0628\u06BE"],
+  [/^aa/u, "\u0622"],
+  [/^oo/u, "\u0648"],
+  [/^ee/u, "\u06CC"],
+  [/^ai/u, "\u06CC"],
+  [/^b/u, "\u0628"],
+  [/^p/u, "\u067E"],
+  [/^t/u, "\u062A"],
+  [/^j/u, "\u062C"],
+  [/^d/u, "\u062F"],
+  [/^r/u, "\u0631"],
+  [/^z/u, "\u0632"],
+  [/^s/u, "\u0633"],
+  [/^f/u, "\u0641"],
+  [/^q/u, "\u0642"],
+  [/^k/u, "\u06A9"],
+  [/^g/u, "\u06AF"],
+  [/^l/u, "\u0644"],
+  [/^m/u, "\u0645"],
+  [/^n/u, "\u0646"],
+  [/^v|^w/u, "\u0648"],
+  [/^h/u, "\u06C1"],
+  [/^y/u, "\u06CC"],
+  [/^a/u, "\u0627"],
+  [/^e/u, "\u06CC"],
+  [/^i/u, "\u06CC"],
+  [/^o/u, "\u0648"],
+  [/^u/u, "\u0648"]
+];
+function romanWordToUrdu(word) {
+  const lower = word.toLowerCase();
+  const dictionary = ROMAN_DICTIONARY[lower];
+  if (dictionary) return dictionary;
+  let rest = lower;
+  let out = "";
+  let position = 0;
+  while (rest.length > 0) {
+    let matched = false;
+    for (const [pattern, replacement] of ROMAN_RULES) {
+      const hit = pattern.exec(rest);
+      if (!hit) continue;
+      const isShortVowel = /^[aeiou]$/u.test(hit[0]);
+      const skip = isShortVowel && position > 0 && rest.length > 1;
+      if (!skip) out += replacement;
+      rest = rest.slice(hit[0].length);
+      position++;
+      matched = true;
+      break;
+    }
+    if (!matched) {
+      out += rest[0];
+      rest = rest.slice(1);
+      position++;
+    }
+  }
+  return out;
+}
+function romanToUrdu(input) {
+  if (!input) return "";
+  return input.split(/(\s+)/u).map((chunk) => /^\s+$/u.test(chunk) ? chunk : romanWordToUrdu(chunk)).join("");
+}
+function urduSlug(input, options = {}) {
+  const { separator = "-", maxLength, preserveUrdu = false } = options;
+  if (!input) return "";
+  const words = splitWords(removeDiacritics(normalizeUrdu(input)));
+  const parts = preserveUrdu ? words : words.map(romanizeWord);
+  const allowed = preserveUrdu ? /[^\p{L}\p{N}]+/gu : /[^a-z0-9]+/gu;
+  let slug = parts.map((part) => part.toLowerCase().replace(allowed, " ").trim()).flatMap((part) => part.split(/\s+/u)).filter((part) => part.length > 0).join(separator);
+  if (maxLength && slug.length > maxLength) {
+    slug = slug.slice(0, maxLength);
+    const lastSeparator = slug.lastIndexOf(separator);
+    if (lastSeparator > 0) slug = slug.slice(0, lastSeparator);
+  }
+  return slug;
+}
+
+exports.analyzeUrdu = analyzeUrdu;
+exports.compareUrdu = compareUrdu;
+exports.convertNumbers = convertNumbers;
+exports.countSentences = countSentences;
+exports.countWords = countWords;
+exports.editDistance = editDistance;
+exports.foldUrdu = foldUrdu;
+exports.hasUrduSpecificLetters = hasUrduSpecificLetters;
+exports.highlightUrdu = highlightUrdu;
+exports.isUrdu = isUrdu;
+exports.normalizeUrdu = normalizeUrdu;
+exports.numberToUrduWords = numberToUrduWords;
+exports.parseUrduNumber = parseUrduNumber;
+exports.removeDiacritics = removeDiacritics;
+exports.romanToUrdu = romanToUrdu;
+exports.romanize = romanize;
+exports.searchUrdu = searchUrdu;
+exports.searchUrduRanked = searchUrduRanked;
+exports.sortUrdu = sortUrdu;
+exports.splitSentences = splitSentences;
+exports.splitWords = splitWords;
+exports.toArabicIndicDigits = toArabicIndicDigits;
+exports.toEnglishDigits = toEnglishDigits;
+exports.toUrduDigits = toUrduDigits;
+exports.urduRatio = urduRatio;
+exports.urduSlug = urduSlug;
+//# sourceMappingURL=index.cjs.map
+//# sourceMappingURL=index.cjs.map