urdu-text-utils 0.2.0 → 0.2.1

package/dist/index.cjs

@@ -15,60 +15,74 @@ 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;
+var URDU_LETTERS = [
+  { ch: "\u0627", roman: "a", vowel: true, aliases: ["\u0623", "\u0625", "\u0671", "\u0672", "\u0673"] },
+  // alef + hamza/wasla forms
+  { ch: "\u0622", roman: "aa", vowel: true },
+  // alef madda — a real Urdu letter
+  { ch: "\u0628", roman: "b" },
+  { ch: "\u067E", roman: "p" },
+  { ch: "\u062A", roman: "t" },
+  { ch: "\u0679", roman: "t" },
+  { ch: "\u062B", roman: "s" },
+  { ch: "\u062C", roman: "j" },
+  { ch: "\u0686", roman: "ch" },
+  { ch: "\u062D", roman: "h" },
+  { ch: "\u062E", roman: "kh" },
+  { ch: "\u062F", roman: "d" },
+  { ch: "\u0688", roman: "d" },
+  { ch: "\u0630", roman: "z" },
+  { ch: "\u0631", roman: "r" },
+  { ch: "\u0691", roman: "r" },
+  { ch: "\u0632", roman: "z" },
+  { ch: "\u0698", roman: "zh" },
+  { ch: "\u0633", roman: "s" },
+  { ch: "\u0634", roman: "sh" },
+  { ch: "\u0635", roman: "s" },
+  { ch: "\u0636", roman: "z" },
+  { ch: "\u0637", roman: "t" },
+  { ch: "\u0638", roman: "z" },
+  { ch: "\u0639", roman: "a", vowel: true },
+  { ch: "\u063A", roman: "gh" },
+  { ch: "\u0641", roman: "f" },
+  { ch: "\u0642", roman: "q" },
+  { ch: "\u06A9", roman: "k", aliases: ["\u0643", "\u06AA"] },
+  // arabic kaf, swash kaf
+  { ch: "\u06AF", roman: "g" },
+  { ch: "\u0644", roman: "l" },
+  { ch: "\u0645", roman: "m" },
+  { ch: "\u0646", roman: "n" },
+  { ch: "\u06BA", roman: "n" },
+  // noon ghunna
+  { ch: "\u0648", roman: "o", vowel: true, aliases: ["\u06CB", "\u06C6", "\u06C7"] },
+  // ve, oe, u
+  { ch: "\u0624", roman: "o", vowel: true, variantOf: "\u0648" },
+  // waw with hamza — kept, hamza is meaningful
+  { ch: "\u06C1", roman: "h", aliases: ["\u0647", "\u06C0", "\u0629", "\u06C3", "\u06D5"] },
+  // arabic heh, heh-yeh-above, teh marbuta, ae
+  { ch: "\u06C2", roman: "h", vowel: true, variantOf: "\u06C1" },
+  // heh goal with hamza above
+  { ch: "\u06BE", roman: "h" },
+  // do-chashmi heh — distinct letter, aspirates the consonant before it
+  { ch: "\u0621", roman: "" },
+  // hamza
+  { ch: "\u06CC", roman: "i", vowel: true, aliases: ["\u064A", "\u0649", "\u06CD", "\u06D0", "\u0620"] },
+  // arabic yeh, alef maksura, yeh-with-tail, pashto e, kashmiri yeh
+  { ch: "\u0626", roman: "i", vowel: true, variantOf: "\u06CC" },
+  // yeh with hamza above
+  { ch: "\u06D2", roman: "e", vowel: true },
+  // bari ye — distinct letter
+  { ch: "\u06D3", roman: "e", vowel: true, variantOf: "\u06D2" }
+  // bari ye with hamza
+];
 
 // 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 LETTER_MAP = {};
+for (const { ch, aliases } of URDU_LETTERS) {
+  for (const alias of aliases ?? []) {
+    LETTER_MAP[alias] = ch;
+  }
+}
 var digitTargets = {
   urdu: URDU_DIGITS,
   english: ASCII_DIGITS,
@@ -626,65 +640,15 @@ function removeStopWords(text, customStopWords) {
 }
 
 // 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 PRIMARY_LETTERS = URDU_LETTERS.filter((letter) => letter.variantOf === void 0);
 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]);
+PRIMARY_LETTERS.forEach((letter, index) => WEIGHTS.set(letter.ch, [(index + 1) * 10, 0]));
+for (const letter of URDU_LETTERS) {
+  if (letter.variantOf === void 0) continue;
+  const baseWeight = WEIGHTS.get(letter.variantOf);
+  if (baseWeight) WEIGHTS.set(letter.ch, [baseWeight[0], 1]);
 }
-var NON_LETTER_BASE = (ALPHABET.length + 1) * 10;
+var NON_LETTER_BASE = (PRIMARY_LETTERS.length + 1) * 10;
 function weightsOf(text) {
   const out = [];
   for (const ch of text) {
@@ -854,7 +818,7 @@ var WORD_DICTIONARY = {
   "\u062C\u0646\u06C1\u0648\u06BA": "jinhon",
   "\u0627\u0646\u06C1\u0648\u06BA": "unhon",
   "\u0627\u067E\u0646\u0627": "apna",
-  "\u0905\u092A\u0928\u0940": "apni",
+  "\u0627\u067E\u0646\u06CC": "apni",
   "\u0627\u067E\u0646\u06D2": "apnay",
   "\u062E\u0648\u062F": "khud",
   "\u06A9\u0648\u0646": "kaun",
@@ -3759,53 +3723,12 @@ var DIGRAPHS = {
   "\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"
-};
-var VOWEL_LETTERS = /* @__PURE__ */ new Set(["\u0627", "\u0622", "\u0648", "\u06CC", "\u06D2", "\u0624", "\u0626", "\u06D3", "\u06C2", "\u0639"]);
+var LETTERS = {};
+var VOWEL_LETTERS = /* @__PURE__ */ new Set();
+for (const { ch, roman, vowel } of URDU_LETTERS) {
+  LETTERS[ch] = roman;
+  if (vowel) VOWEL_LETTERS.add(ch);
+}
 function romanizeWord(word) {
   const dictionary = WORD_DICTIONARY[word];
   if (dictionary) return dictionary;
@@ -4394,6 +4317,401 @@ function stemUrduText(text, options = {}) {
   }).join("");
 }
 
+// src/names.ts
+var URDU_FIRST_NAMES = {
+  // Male names
+  "\u0645\u062D\u0645\u062F": "Muhammad",
+  "\u0627\u062D\u0645\u062F": "Ahmed",
+  "\u0639\u0644\u06CC": "Ali",
+  "\u0639\u0645\u0631": "Umar",
+  "\u0639\u062B\u0645\u0627\u0646": "Usman",
+  "\u0627\u0628\u0648\u0628\u06A9\u0631": "Abubakar",
+  "\u06CC\u0648\u0633\u0641": "Yusuf",
+  "\u06CC\u0639\u0642\u0648\u0628": "Yaquob",
+  "\u0627\u0633\u0645\u0627\u0639\u06CC\u0644": "Ismail",
+  "\u0627\u0628\u0631\u0627\u06C1\u06CC\u0645": "Ibrahim",
+  "\u0627\u0633\u062D\u0627\u0642": "Ishaq",
+  "\u062D\u0633\u06CC\u0646": "Hussain",
+  "\u062D\u0633\u0646": "Hasan",
+  "\u0632\u06CC\u062F": "Zaid",
+  "\u062D\u0645\u0632\u06C1": "Hamza",
+  "\u0633\u0644\u0645\u0627\u0646": "Salman",
+  "\u0641\u0631\u062D\u0627\u0646": "Farhan",
+  "\u0627\u062F\u0646\u0627\u0646": "Adnan",
+  "\u0628\u0644\u0627\u0644": "Bilal",
+  "\u0637\u0627\u0631\u0642": "Tariq",
+  "\u062C\u0627\u0648\u06CC\u062F": "Javed",
+  "\u0639\u0631\u0641\u0627\u0646": "Irfan",
+  "\u06A9\u0627\u0634\u0641": "Kashif",
+  "\u0646\u062F\u06CC\u0645": "Nadeem",
+  "\u06CC\u0627\u0633\u0631": "Yasir",
+  "\u0631\u0636\u0648\u0627\u0646": "Rizwan",
+  "\u0641\u06CC\u0635\u0644": "Faisal",
+  "\u062E\u0627\u0644\u062F": "Khalid",
+  "\u0631\u0627\u0634\u062F": "Rashid",
+  "\u0634\u0627\u06C1\u062F": "Shahid",
+  "\u0633\u0639\u06CC\u062F": "Saeed",
+  "\u0646\u0639\u06CC\u0645": "Naeem",
+  "\u0648\u0633\u06CC\u0645": "Waseem",
+  "\u0631\u0641\u06CC\u0642": "Rafiq",
+  "\u062C\u0645\u06CC\u0644": "Jamil",
+  "\u0632\u0628\u06CC\u0631": "Zubair",
+  "\u0637\u0627\u06C1\u0631": "Tahir",
+  "\u0639\u0627\u0631\u0641": "Arif",
+  "\u0646\u0639\u0645\u0627\u0646": "Noman",
+  "\u062B\u0627\u0642\u0628": "Saqib",
+  "\u0639\u062F\u06CC\u0644": "Adeel",
+  "\u0639\u0645\u0631\u0627\u0646": "Imran",
+  "\u0641\u0648\u0627\u062F": "Fawad",
+  "\u0622\u0635\u0641": "Asif",
+  "\u0634\u0639\u06CC\u0628": "Shoaib",
+  "\u0632\u0627\u06C1\u062F": "Zahid",
+  "\u0639\u0627\u0645\u0631": "Amer",
+  "\u0645\u062C\u06CC\u062F": "Majid",
+  "\u0633\u0631\u0641\u0631\u0627\u0632": "Sarfraz",
+  "\u062D\u0645\u06CC\u062F": "Hamid",
+  "\u0646\u0638\u06CC\u0631": "Nazir",
+  "\u0642\u06CC\u0635\u0631": "Qaiser",
+  "\u0644\u0637\u06CC\u0641": "Latif",
+  "\u062A\u0646\u0648\u06CC\u0631": "Tanveer",
+  "\u0633\u0644\u06CC\u0645\u0627\u0646": "Suleman",
+  "\u0627\u0641\u062A\u062E\u0627\u0631": "Iftikhar",
+  "\u0634\u0628\u06CC\u0631": "Shabbir",
+  "\u0633\u0627\u062C\u062F": "Sajid",
+  "\u0645\u0646\u0635\u0648\u0631": "Mansoor",
+  "\u0633\u0644\u0637\u0627\u0646": "Sultan",
+  "\u0646\u0635\u06CC\u0631": "Nasir",
+  "\u062D\u0628\u06CC\u0628": "Habib",
+  "\u0645\u0642\u0635\u0648\u062F": "Maqsood",
+  "\u0645\u0633\u0639\u0648\u062F": "Masood",
+  "\u0633\u06C1\u06CC\u0644": "Sohail",
+  "\u0627\u0634\u0631\u0641": "Ashraf",
+  "\u0645\u0628\u0627\u0631\u06A9": "Mubarak",
+  "\u0641\u06C1\u062F": "Fahad",
+  "\u0627\u0638\u06C1\u0631": "Azhar",
+  "\u0648\u06A9\u06CC\u0644": "Wakeel",
+  "\u0631\u0624\u0641": "Rauf",
+  "\u062F\u0627\u0646\u06CC\u0627\u0644": "Daniyal",
+  "\u0630\u06CC\u0634\u0627\u0646": "Zeeshan",
+  "\u0645\u0635\u0637\u0641\u06CC": "Mustafa",
+  // Female names
+  "\u0641\u0627\u0637\u0645\u06C1": "Fatima",
+  "\u0639\u0627\u0626\u0634\u06C1": "Ayesha",
+  "\u062E\u062F\u06CC\u062C\u06C1": "Khadija",
+  "\u062D\u0633\u0646\u0627": "Husna",
+  "\u0632\u06CC\u0646\u0628": "Zainab",
+  "\u0622\u0645\u0646\u06C1": "Amina",
+  "\u0633\u0627\u0631\u0627": "Sara",
+  "\u0627\u062F\u06CC\u06C1": "Adia",
+  "\u0646\u0648\u0631": "Noor",
+  "\u062D\u0631\u0627": "Hira",
+  "\u0627\u0642\u0635\u06CC": "Aqsa",
+  "\u0645\u0627\u06C1\u0645": "Maham",
+  "\u0627\u06CC\u0645\u0627\u0646": "Iman",
+  "\u0644\u0627\u0626\u0628\u06C1": "Laiba",
+  "\u06C1\u0627\u0646\u06CC\u06C1": "Hania",
+  "\u0631\u0645\u0634\u06C1": "Rimsha",
+  "\u0645\u06CC\u0645\u0648\u0646\u06C1": "Memoona",
+  "\u062B\u0646\u0627": "Sana",
+  "\u062D\u0646\u0627": "Hina",
+  "\u0627\u0642\u0631\u0627": "Iqra",
+  "\u0641\u06CC\u0632\u0627": "Fiza",
+  "\u0641\u0636\u06C1": "Fizza",
+  "\u0633\u062F\u0631\u06C1": "Sidra",
+  "\u0631\u062F\u0627": "Rida",
+  "\u0646\u062F\u0627": "Nida",
+  "\u0645\u0648\u0645\u06CC\u0646\u06C1": "Momina",
+  "\u062B\u0648\u0628\u06CC\u06C1": "Sobia",
+  "\u0646\u0627\u0626\u0644\u06C1": "Naila",
+  "\u0628\u0634\u0631\u06CC": "Bushra",
+  "\u0631\u062E\u0633\u0627\u0646\u06C1": "Rukhsana",
+  "\u0634\u0627\u0632\u06CC\u06C1": "Shazia",
+  "\u0646\u0627\u06C1\u06CC\u062F": "Naheed",
+  "\u067E\u0631\u0648\u06CC\u0646": "Parveen",
+  "\u0641\u0631\u0632\u0627\u0646\u06C1": "Farzana",
+  "\u0646\u0627\u0632\u06CC\u06C1": "Nazia",
+  "\u0635\u0628\u06CC\u062D\u06C1": "Sabiha",
+  "\u0631\u0628\u0627\u0628": "Rubab",
+  "\u0635\u0627\u0626\u0645\u06C1": "Saima",
+  "\u0645\u06C1\u0631\u06CC\u0646": "Mehreen",
+  "\u0639\u0646\u0628\u0631\u06CC\u0646": "Ambreen",
+  "\u06A9\u0627\u0626\u0646\u0627\u062A": "Kainat",
+  "\u0637\u0627\u06C1\u0631\u06C1": "Tahira",
+  "\u0633\u0645\u06CC\u0639\u06C1": "Samia",
+  "\u0633\u0627\u0645\u06CC\u06C1": "Samiya",
+  "\u0633\u0644\u0645\u06CC": "Salma",
+  "\u0631\u0648\u0632\u06CC\u0646\u06C1": "Rozina",
+  "\u0645\u0631\u06CC\u0645": "Maryam",
+  "\u0631\u0627\u0628\u0639\u06C1": "Rabia",
+  "\u062A\u0633\u0644\u06CC\u0645": "Tasleem",
+  "\u0627\u0633\u0645\u0627\u0621": "Asma",
+  "\u0627\u0631\u0645": "Iram",
+  "\u0646\u0627\u0632": "Naaz",
+  "\u0641\u0631\u062D": "Farah",
+  "\u0646\u0627\u062F\u06CC\u06C1": "Nadia",
+  "\u0633\u06CC\u0645\u0627": "Seema",
+  "\u06C1\u0645\u0627": "Huma",
+  "\u0646\u0632\u06C1\u062A": "Nuzhat",
+  "\u0639\u0638\u0645\u06CC": "Uzma",
+  "\u06A9\u0631\u0646": "Kiran",
+  "\u0633\u0645\u06CC\u0631\u0627": "Sumaira",
+  "\u0641\u0631\u062E\u0646\u062F\u06C1": "Farkhanda",
+  "\u062A\u0633\u0646\u06CC\u0645": "Tasneem",
+  "\u0632\u0628\u06CC\u062F\u06C1": "Zubaida",
+  "\u062C\u0648\u06CC\u0631\u06CC\u06C1": "Javeria",
+  "\u0645\u06C1\u0648\u0634": "Mehwish",
+  "\u0639\u0627\u0644\u06CC\u06C1": "Alia"
+};
+var URDU_FAMILY_NAMES = {
+  "\u062E\u0627\u0646": "Khan",
+  "\u0634\u0627\u06C1": "Shah",
+  "\u0686\u0648\u06C1\u062F\u0631\u06CC": "Chaudhry",
+  "\u0645\u0644\u06A9": "Malik",
+  "\u0634\u06CC\u062E": "Sheikh",
+  "\u0642\u0631\u06CC\u0634\u06CC": "Qureshi",
+  "\u0633\u06CC\u062F": "Syed",
+  "\u0646\u06CC\u0627\u0632\u06CC": "Niazi",
+  "\u062C\u062F\u0648\u0646": "Jadoon",
+  "\u0633\u0648\u0627\u062A\u06CC": "Swati",
+  "\u06A9\u06BE\u0679\u06A9": "Khattak",
+  "\u06CC\u0648\u0633\u0641\u0632\u0626\u06CC": "Yusufzai",
+  "\u06A9\u06BE\u0648\u0633\u06C1": "Khosa",
+  "\u062A\u0627\u0644\u067E\u0648\u0631": "Talpur",
+  "\u062C\u0648\u0646\u06CC\u062C\u0648": "Junejo",
+  "\u0628\u06BE\u0679\u06CC": "Bhatti",
+  "\u0631\u0627\u062C\u067E\u0648\u062A": "Rajput",
+  "\u0633\u0648\u0645\u0631\u0648": "Soomro",
+  "\u062F\u06C1\u0627\u0631": "Dahar",
+  "\u06AF\u06CC\u0644\u0627\u0646\u06CC": "Gillani",
+  "\u0679\u0648\u0627\u0646\u0627\u0633": "Tiwanas",
+  "\u0679\u0648\u0627\u0646\u06C1": "Tiwana",
+  "\u062F\u0648\u0644\u062A\u0627\u0646\u06CC": "Doltani",
+  "\u062F\u0627\u0644\u062A\u0627\u0646\u06CC": "Daltani",
+  "\u06AF\u0648\u0631\u0686\u0627\u0646\u06CC": "Gurchani",
+  "\u0631\u0646\u062F": "Rind",
+  "\u0628\u06AF\u0679\u06CC": "Bugti",
+  "\u0645\u0631\u06CC": "Marri",
+  "\u0628\u0632\u06CC\u0646\u062C\u0648": "Bizenjo",
+  "\u0645\u06CC\u0646\u06AF\u0644": "Mengal",
+  "\u0632\u0631\u062F\u0627\u0631\u06CC": "Zardari",
+  "\u0628\u06BE\u0679\u0648": "Bhutto",
+  "\u0634\u0631\u06CC\u0641": "Sharif"
+};
+var HONORIFICS = {
+  "\u062C\u0646\u0627\u0628": "Janab",
+  "\u0635\u0627\u062D\u0628": "Sahib",
+  "\u0627\u0633\u062A\u0627\u062F": "Ustad",
+  "\u062D\u06A9\u06CC\u0645": "Hakim",
+  "\u0688\u0627\u06A9\u0679\u0631": "Dr.",
+  "\u067E\u0631\u0648\u0641\u06CC\u0633\u0631": "Prof.",
+  "\u0627\u0646\u062C\u06CC\u0646\u0626\u0631": "Eng.",
+  "\u0627\u06CC\u0688\u0648\u0648\u06A9\u06CC\u0679": "Adv.",
+  "\u0631\u06CC\u0679\u0627\u0626\u0631\u0688": "Retd.",
+  "\u0634\u06C1\u06CC\u062F": "Shaheed"
+};
+var NAME_PREFIXES = {
+  "\u0633\u06CC\u062F": "Syed",
+  "\u0634\u06CC\u062E": "Sheikh",
+  "\u0686\u0648\u06C1\u062F\u0631\u06CC": "Chowdhury",
+  "\u0645\u0644\u06A9": "Malik",
+  "\u0645\u06CC\u0627\u06BA": "Mian",
+  "\u062E\u0627\u0646": "Khan",
+  "\u0628\u06CC\u06AF\u0645": "Begum",
+  "\u0635\u0627\u062D\u0628": "Sahib"
+};
+var ENGLISH_TO_URDU = /* @__PURE__ */ new Map();
+for (const table of [HONORIFICS, URDU_FIRST_NAMES, URDU_FAMILY_NAMES, NAME_PREFIXES]) {
+  for (const [urduWord, englishWord] of Object.entries(table)) {
+    ENGLISH_TO_URDU.set(englishWord.toLowerCase(), urduWord);
+  }
+}
+var ENGLISH_NAME_ALIASES = {
+  hassan: "\u062D\u0633\u0646",
+  // vs. canonical Hasan
+  omar: "\u0639\u0645\u0631",
+  // vs. canonical Umar
+  omer: "\u0639\u0645\u0631",
+  // vs. canonical Umar
+  yaqoob: "\u06CC\u0639\u0642\u0648\u0628",
+  // vs. canonical Yaquob
+  jameel: "\u062C\u0645\u06CC\u0644",
+  // vs. canonical Jamil
+  majeed: "\u0645\u062C\u06CC\u062F",
+  // vs. canonical Majid
+  amna: "\u0622\u0645\u0646\u06C1",
+  // vs. canonical Amina
+  gilani: "\u06AF\u06CC\u0644\u0627\u0646\u06CC",
+  // vs. canonical Gillani
+  sahab: "\u0635\u0627\u062D\u0628",
+  // vs. canonical Sahib
+  doctor: "\u0688\u0627\u06A9\u0679\u0631",
+  // vs. canonical Dr.
+  professor: "\u067E\u0631\u0648\u0641\u06CC\u0633\u0631",
+  // vs. canonical Prof.
+  engineer: "\u0627\u0646\u062C\u06CC\u0646\u0626\u0631",
+  // vs. canonical Eng.
+  advocate: "\u0627\u06CC\u0688\u0648\u0648\u06A9\u06CC\u0679",
+  // vs. canonical Adv.
+  retd: "\u0631\u06CC\u0679\u0627\u0626\u0631\u0688"
+  // vs. canonical Retd.
+};
+function transliterateSingleName(urduName) {
+  const normalized = removeDiacritics(normalizeUrdu(urduName));
+  const words = normalized.split(/\s+/).filter((w) => w.length > 0);
+  const result = [];
+  for (const word of words) {
+    const honorific = HONORIFICS[word];
+    if (honorific) {
+      result.push(honorific);
+      continue;
+    }
+    const firstName = URDU_FIRST_NAMES[word];
+    if (firstName) {
+      result.push(firstName);
+      continue;
+    }
+    const familyName = URDU_FAMILY_NAMES[word];
+    if (familyName) {
+      result.push(familyName);
+      continue;
+    }
+    const prefix = NAME_PREFIXES[word];
+    if (prefix) {
+      result.push(prefix);
+      continue;
+    }
+    result.push(transliterateNameWord(word));
+  }
+  return result.join(" ");
+}
+function transliterateNameWord(word) {
+  if (!word) return "";
+  const consonants = {
+    "\u0628": "b",
+    "\u067E": "p",
+    "\u062A": "t",
+    "\u0679": "t",
+    "\u062C": "j",
+    "\u0686": "ch",
+    "\u062D": "h",
+    "\u062E": "kh",
+    "\u062F": "d",
+    "\u0688": "d",
+    "\u0631": "r",
+    "\u0691": "r",
+    "\u0632": "z",
+    "\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",
+    "\u0648": "w",
+    "\u06C1": "h",
+    "\u06BE": "h",
+    "\u06CC": "y"
+  };
+  let result = "";
+  const chars = [...word];
+  for (let i = 0; i < chars.length; i++) {
+    const ch = chars[i];
+    const next = chars[i + 1];
+    if (next === "\u06BE" && consonants[ch]) {
+      const base = consonants[ch];
+      result += base + "h";
+      i++;
+      continue;
+    }
+    if (ch === "\u0622") {
+      result += "Aa";
+      continue;
+    }
+    if (ch === "\u0627" && i === 0) {
+      result += "A";
+      continue;
+    }
+    if (ch === "\u0627") {
+      result += "a";
+      continue;
+    }
+    if (ch === "\u06CC" && i === chars.length - 1) {
+      result += "i";
+      continue;
+    }
+    if (ch === "\u06CC" && next === "\u06BA") {
+      result += "ain";
+      i++;
+      continue;
+    }
+    if (ch === "\u0648" && i === 0) {
+      result += "W";
+      continue;
+    }
+    if (ch === "\u0648") {
+      result += "o";
+      continue;
+    }
+    if (consonants[ch]) {
+      result += consonants[ch];
+      continue;
+    }
+    result += ch;
+  }
+  return result.charAt(0).toUpperCase() + result.slice(1);
+}
+function transliterateNameToEnglish(urduName, options = {}) {
+  if (!urduName) return "";
+  const { preserveCase = true, includeHonorifics = true } = options;
+  const result = transliterateSingleName(urduName);
+  if (!includeHonorifics) {
+    return result.split(/\s+/).filter((word) => !Object.values(HONORIFICS).includes(word)).join(" ");
+  }
+  return preserveCase ? result : result.toLowerCase();
+}
+function transliterateNameToUrdu(englishName) {
+  if (!englishName) return "";
+  const words = englishName.split(/\s+/).filter((w) => w.length > 0);
+  return words.map((word) => reverseTransliterateName(word)).join(" ");
+}
+function reverseTransliterateName(englishName) {
+  const lower = englishName.toLowerCase();
+  return ENGLISH_TO_URDU.get(lower) ?? ENGLISH_NAME_ALIASES[lower] ?? englishName;
+}
+function extractNameParts(fullName) {
+  const normalized = removeDiacritics(normalizeUrdu(fullName));
+  const words = normalized.split(/\s+/).filter((w) => w.length > 0);
+  const honorifics = Object.keys(HONORIFICS);
+  const suffixes = ["sahib", "sahab", "\u0628\u06CC\u06AF\u0645", "begum"];
+  let honorific;
+  let suffix;
+  const nameWords = [];
+  for (const word of words) {
+    if (honorifics.includes(word)) {
+      if (!honorific) {
+        honorific = word;
+      } else {
+        suffix = word;
+      }
+    } else if (suffixes.includes(word)) {
+      suffix = word;
+    } else {
+      nameWords.push(word);
+    }
+  }
+  const familyName = nameWords.length > 1 ? nameWords[nameWords.length - 1] : void 0;
+  const firstName = nameWords.length > 1 ? nameWords.slice(0, -1).join(" ") : nameWords[0] || "";
+  return { honorific, firstName, familyName, suffix };
+}
+
 exports.URDU_MONTHS_GREGORIAN = URDU_MONTHS_GREGORIAN;
 exports.URDU_MONTHS_HIJRI = URDU_MONTHS_HIJRI;
 exports.URDU_PREFIXES = URDU_PREFIXES;
@@ -4406,6 +4724,7 @@ exports.convertNumbers = convertNumbers;
 exports.countSentences = countSentences;
 exports.countWords = countWords;
 exports.editDistance = editDistance;
+exports.extractNameParts = extractNameParts;
 exports.filterStopWords = filterStopWords;
 exports.foldUrdu = foldUrdu;
 exports.formatUrduDate = formatUrduDate;
@@ -4434,6 +4753,8 @@ exports.timeAgoUrdu = timeAgoUrdu;
 exports.toArabicIndicDigits = toArabicIndicDigits;
 exports.toEnglishDigits = toEnglishDigits;
 exports.toUrduDigits = toUrduDigits;
+exports.transliterateNameToEnglish = transliterateNameToEnglish;
+exports.transliterateNameToUrdu = transliterateNameToUrdu;
 exports.urduRatio = urduRatio;
 exports.urduSlug = urduSlug;
 //# sourceMappingURL=index.cjs.map