urdu-text-utils 0.3.1 → 0.4.0

package/dist/index.js

@@ -191,8 +191,23 @@ function parseUrduNumber(input) {
   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 = {
+var SEPARATORS = {
+  urdu: { group: "\u066C", decimal: "\u066B" },
+  english: { group: ",", decimal: "." }
+};
+var URDU_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 URDU_TEENS_AND_TENS = {
   10: "\u062F\u0633",
   11: "\u06AF\u06CC\u0627\u0631\u06C1",
   12: "\u0628\u0627\u0631\u06C1",
@@ -213,20 +228,104 @@ var TEENS_AND_TENS = {
   90: "\u0646\u0648\u06D2",
   100: "\u0633\u0648"
 };
+var URDU_TENS_UNITS = {
+  21: "\u0627\u06A9\u06CC\u0633",
+  22: "\u0628\u0627\u0626\u06CC\u0633",
+  23: "\u062A\u0626\u06CC\u0633",
+  24: "\u0686\u0648\u0628\u06CC\u0633",
+  25: "\u067E\u0686\u06CC\u0633",
+  26: "\u0686\u06BE\u0628\u06CC\u0633",
+  27: "\u0633\u062A\u0627\u0626\u06CC\u0633",
+  28: "\u0627\u0679\u06BE\u0627\u0626\u06CC\u0633",
+  29: "\u0627\u0646\u062A\u06CC\u0633",
+  31: "\u0627\u06A9\u062A\u06CC\u0633",
+  32: "\u0628\u062A\u06CC\u0633",
+  33: "\u062A\u06CC\u0646\u062A\u06CC\u0633",
+  34: "\u0686\u0648\u0646\u062A\u06CC\u0633",
+  35: "\u067E\u06CC\u0646\u062A\u06CC\u0633",
+  36: "\u0686\u06BE\u062A\u06CC\u0633",
+  37: "\u0633\u06CC\u0646\u062A\u06CC\u0633",
+  38: "\u0627\u0691\u062A\u06CC\u0633",
+  39: "\u0627\u0646\u062A\u0627\u0644\u06CC\u0633",
+  41: "\u0627\u06A9\u062A\u0627\u0644\u06CC\u0633",
+  42: "\u0628\u06CC\u0627\u0644\u06CC\u0633",
+  43: "\u062A\u06CC\u0646\u062A\u0627\u0644\u06CC\u0633",
+  44: "\u0686\u0648\u0627\u0644\u06CC\u0633",
+  45: "\u067E\u06CC\u0646\u062A\u0627\u0644\u06CC\u0633",
+  46: "\u0686\u06BE\u06CC\u0627\u0644\u06CC\u0633",
+  47: "\u0633\u06CC\u0646\u062A\u0627\u0644\u06CC\u0633",
+  48: "\u0627\u0691\u062A\u0627\u0644\u06CC\u0633",
+  49: "\u0627\u0646\u0686\u0627\u0633",
+  51: "\u0627\u06A9\u06CC\u0627\u0648\u0646",
+  52: "\u0628\u0627\u0648\u0646",
+  53: "\u062A\u0631\u067E\u0646",
+  54: "\u0686\u0648\u0646",
+  55: "\u067E\u0686\u067E\u0646",
+  56: "\u0686\u06BE\u067E\u0646",
+  57: "\u0633\u062A\u0627\u0648\u0646",
+  58: "\u0627\u0679\u06BE\u0627\u0648\u0646",
+  59: "\u0627\u0646\u0633\u0679\u06BE",
+  61: "\u0627\u06A9\u0633\u0679\u06BE",
+  62: "\u0628\u0627\u0633\u0679\u06BE",
+  63: "\u062A\u0631\u0633\u0679\u06BE",
+  64: "\u0686\u0648\u0646\u0633\u0679\u06BE",
+  65: "\u067E\u06CC\u0646\u0633\u0679\u06BE",
+  66: "\u0686\u06BE\u06CC\u0627\u0633\u0679\u06BE",
+  67: "\u0633\u0691\u0633\u0679\u06BE",
+  68: "\u0627\u0691\u0633\u0679\u06BE",
+  69: "\u0627\u0646\u06C1\u062A\u0631",
+  71: "\u0627\u06A9\u06C1\u062A\u0631",
+  72: "\u0628\u06C1\u062A\u0631",
+  73: "\u062A\u06C1\u062A\u0631",
+  74: "\u0686\u0648\u06C1\u062A\u0631",
+  75: "\u067E\u0686\u06C1\u062A\u0631",
+  76: "\u0686\u06BE\u06C1\u062A\u0631",
+  77: "\u0633\u062A\u062A\u0631",
+  78: "\u0627\u0679\u06BE\u06C1\u062A\u0631",
+  79: "\u0627\u0646\u0627\u0633\u06CC",
+  81: "\u0627\u06A9\u06CC\u0627\u0633\u06CC",
+  82: "\u0628\u06CC\u0627\u0633\u06CC",
+  83: "\u062A\u0631\u0627\u0633\u06CC",
+  84: "\u0686\u0648\u0631\u0627\u0633\u06CC",
+  85: "\u067E\u0686\u0627\u0633\u06CC",
+  86: "\u0686\u06BE\u06CC\u0627\u0633\u06CC",
+  87: "\u0633\u062A\u0627\u0633\u06CC",
+  88: "\u0627\u0679\u06BE\u0627\u0633\u06CC",
+  89: "\u0646\u0648\u0627\u0633\u06CC",
+  91: "\u0627\u06A9\u0627\u0646\u0648\u06D2",
+  92: "\u0628\u0627\u0646\u0648\u06D2",
+  93: "\u062A\u0631\u0627\u0646\u0648\u06D2",
+  94: "\u0686\u0648\u0631\u0627\u0646\u0648\u06D2",
+  95: "\u067E\u0686\u0627\u0646\u0648\u06D2",
+  96: "\u0686\u06BE\u06CC\u0627\u0646\u0648\u06D2",
+  97: "\u0633\u062A\u0627\u0646\u0648\u06D2",
+  98: "\u0627\u0679\u06BE\u0627\u0646\u0648\u06D2",
+  99: "\u0646\u0627\u0646\u0648\u06D2"
+};
+var URDU_SCALES = [
+  [1e11, "\u06A9\u06BE\u0631\u0628"],
+  [1e9, "\u0627\u0631\u0628"],
+  [1e7, "\u06A9\u0631\u0648\u0691"],
+  [1e5, "\u0644\u0627\u06A9\u06BE"],
+  [1e3, "\u06C1\u0632\u0627\u0631"],
+  [100, "\u0633\u0648"]
+];
+var URDU_ORDINAL_IRREGULARS = {
+  1: { masculine: "\u067E\u06C1\u0644\u0627", feminine: "\u067E\u06C1\u0644\u06CC" },
+  2: { masculine: "\u062F\u0648\u0633\u0631\u0627", feminine: "\u062F\u0648\u0633\u0631\u06CC" },
+  3: { masculine: "\u062A\u06CC\u0633\u0631\u0627", feminine: "\u062A\u06CC\u0633\u0631\u06CC" },
+  4: { masculine: "\u0686\u0648\u062A\u06BE\u0627", feminine: "\u0686\u0648\u062A\u06BE\u06CC" },
+  5: { masculine: "\u067E\u0627\u0646\u0686\u0648\u0627\u06BA", feminine: "\u067E\u0627\u0646\u0686\u0648\u06CC\u06BA" },
+  6: { masculine: "\u0686\u06BE\u0679\u0627", feminine: "\u0686\u06BE\u0679\u06CC" }
+};
 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 ${numberToUrduWords(-value)}`;
   if (value === 0) return "\u0635\u0641\u0631";
-  const scales = [
-    [1e7, "\u06A9\u0631\u0648\u0691"],
-    [1e5, "\u0644\u0627\u06A9\u06BE"],
-    [1e3, "\u06C1\u0632\u0627\u0631"],
-    [100, "\u0633\u0648"]
-  ];
   const parts = [];
   let rest = value;
-  for (const [size, name] of scales) {
+  for (const [size, name] of URDU_SCALES) {
     if (rest >= size) {
       const count = Math.floor(rest / size);
       rest %= size;
@@ -234,17 +333,83 @@ function numberToUrduWords(value) {
     }
   }
   if (rest > 0) {
-    if (TEENS_AND_TENS[rest]) {
-      parts.push(TEENS_AND_TENS[rest]);
+    const irregular = URDU_TENS_UNITS[rest];
+    if (irregular !== void 0) {
+      parts.push(irregular);
+    } else if (URDU_TEENS_AND_TENS[rest]) {
+      parts.push(URDU_TEENS_AND_TENS[rest]);
     } else if (rest < 10) {
-      parts.push(ONES[rest]);
+      parts.push(URDU_ONES[rest]);
     } else {
       const tens = Math.floor(rest / 10) * 10;
-      parts.push(`${TEENS_AND_TENS[tens] ?? ""} ${ONES[rest % 10] ?? ""}`.trim());
+      parts.push(`${URDU_TEENS_AND_TENS[tens] ?? ""} ${URDU_ONES[rest % 10] ?? ""}`.trim());
     }
   }
   return parts.join(" ").replace(/\s+/gu, " ").trim();
 }
+function toDigitString(n) {
+  const plain = String(n);
+  const exponential = /^(\d)(?:\.(\d+))?e\+(\d+)$/u.exec(plain);
+  if (!exponential) return plain;
+  const lead = exponential[1];
+  const fraction = exponential[2] ?? "";
+  const zeros = Number(exponential[3]) - fraction.length;
+  return `${lead}${fraction}${"0".repeat(zeros)}`;
+}
+function groupDigits(intStr, grouping, separator) {
+  if (grouping === "none" || separator === "") return intStr;
+  if (grouping === "western") {
+    return intStr.replace(/\B(?=(\d{3})+(?!\d))/gu, separator);
+  }
+  const lastThree = intStr.slice(-3);
+  const rest = intStr.slice(0, -3);
+  return rest === "" ? lastThree : `${rest.replace(/\B(?=(\d{2})+(?!\d))/gu, separator)}${separator}${lastThree}`;
+}
+function formatUrduNumber(value, options = {}) {
+  if (!Number.isFinite(value)) return "";
+  const digits = options.digits ?? "urdu";
+  const grouping = options.grouping ?? "southAsian";
+  const decimals = Math.min(15, Math.max(0, Math.floor(options.decimals ?? 2)));
+  const separators = SEPARATORS[digits];
+  const negative = value < 0;
+  const abs = Math.abs(value);
+  const scale = 10 ** decimals;
+  let intPart = Math.floor(abs);
+  let scaledFraction = Math.floor((abs - intPart) * scale + 0.5);
+  if (scaledFraction >= scale) {
+    intPart += 1;
+    scaledFraction -= scale;
+  }
+  const intStr = groupDigits(toDigitString(intPart), grouping, separators.group);
+  const fraction = String(scaledFraction).padStart(decimals, "0").replace(/0+$/u, "");
+  const body = fraction === "" ? intStr : `${intStr}${separators.decimal}${fraction}`;
+  const sign = negative ? "-" : "";
+  return digits === "urdu" ? `${sign}${toUrduDigits(body)}` : `${sign}${body}`;
+}
+function formatUrduCurrency(value, options = {}) {
+  if (!Number.isFinite(value)) return "";
+  const style = options.style ?? "digits";
+  const unit = options.unit ?? "\u0631\u0648\u067E\u06D2";
+  const singularUnit = options.singularUnit ?? "\u0631\u0648\u067E\u06CC\u06C1";
+  const amount = style === "words" ? numberToUrduWords(value) : formatUrduNumber(value, options);
+  if (amount === "") return "";
+  return `${amount} ${Math.abs(value) === 1 ? singularUnit : unit}`;
+}
+function toUrduOrdinal(value, options = {}) {
+  if (!Number.isFinite(value)) return "";
+  if (!Number.isInteger(value)) throw new TypeError("toUrduOrdinal expects an integer");
+  const gender = options.gender ?? "masculine";
+  if (value < 0) return `\u0645\u0646\u0641\u06CC ${toUrduOrdinal(-value, options)}`;
+  if (value === 0) return "\u0635\u0641\u0631";
+  const irregular = URDU_ORDINAL_IRREGULARS[value];
+  if (irregular) return irregular[gender];
+  const words = numberToUrduWords(value).split(" ");
+  const last = words.pop();
+  const stem = last.endsWith("\u0648") ? last.slice(0, -1) : last;
+  const ordinated = `${stem}${gender === "feminine" ? "\u0648\u06CC\u06BA" : "\u0648\u0627\u06BA"}`;
+  const prefix = words.length === 1 && words[0] === "\u0627\u06CC\u06A9" ? [] : words;
+  return [...prefix, ordinated].join(" ");
+}
 
 // src/detect.ts
 function urduRatio(input) {
@@ -4076,6 +4241,186 @@ function timeAgoUrdu(date, relativeTo = /* @__PURE__ */ new Date(), options = {}
   if (!addSuffix) return `${count} \u0633\u0627\u0644`;
   return isPast ? `${count} \u0633\u0627\u0644 \u067E\u06C1\u0644\u06D2` : `${count} \u0633\u0627\u0644 \u0628\u0639\u062F`;
 }
+var ENGLISH_MONTH_NAMES = [
+  "january",
+  "february",
+  "march",
+  "april",
+  "may",
+  "june",
+  "july",
+  "august",
+  "september",
+  "october",
+  "november",
+  "december"
+];
+var ENGLISH_WEEKDAY_NAMES = [
+  "sunday",
+  "monday",
+  "tuesday",
+  "wednesday",
+  "thursday",
+  "friday",
+  "saturday"
+];
+var MONTH_LOOKUP = (() => {
+  const lookup = /* @__PURE__ */ Object.create(null);
+  URDU_MONTHS_GREGORIAN.forEach((name, index) => {
+    lookup[name] = index;
+  });
+  ENGLISH_MONTH_NAMES.forEach((name, index) => {
+    lookup[name] = index;
+    lookup[name.slice(0, 3)] = index;
+  });
+  return lookup;
+})();
+var WEEKDAY_TOKENS = /* @__PURE__ */ new Set();
+for (const name of URDU_WEEKDAYS) WEEKDAY_TOKENS.add(name);
+for (const name of ENGLISH_WEEKDAY_NAMES) {
+  WEEKDAY_TOKENS.add(name);
+  WEEKDAY_TOKENS.add(name.slice(0, 3));
+}
+var URDU_PERIODS = Object.assign(
+  /* @__PURE__ */ Object.create(null),
+  {
+    \u0635\u0628\u062D: "morning",
+    \u062F\u0648\u067E\u06C1\u0631: "afternoon",
+    \u0634\u0627\u0645: "evening",
+    \u0631\u0627\u062A: "night"
+  }
+);
+function resolveHour(hour, period) {
+  switch (period) {
+    case "am":
+    case "morning":
+      return hour === 12 ? 0 : hour;
+    case "pm":
+    case "afternoon":
+    case "evening":
+      return hour < 12 ? hour + 12 : hour;
+    case "night":
+      if (hour === 12) return 0;
+      return hour >= 8 && hour <= 11 ? hour + 12 : hour;
+    default:
+      return hour;
+  }
+}
+function buildLocalDate(year, month, day, hour, minute, second) {
+  if (!Number.isInteger(year) || year < 100) return null;
+  if (!Number.isInteger(month) || month < 0 || month > 11) return null;
+  if (!Number.isInteger(day) || day < 1 || day > 31) return null;
+  if (!Number.isInteger(hour) || hour < 0 || hour > 23) return null;
+  if (!Number.isInteger(minute) || minute < 0 || minute > 59) return null;
+  if (!Number.isInteger(second) || second < 0 || second > 59) return null;
+  const date = new Date(year, month, day, hour, minute, second);
+  if (date.getFullYear() !== year || date.getMonth() !== month || date.getDate() !== day || date.getHours() !== hour) {
+    return null;
+  }
+  return date;
+}
+function expandYear(year) {
+  if (year >= 100) return year;
+  return year < 70 ? 2e3 + year : 1900 + year;
+}
+var INVISIBLE_CHARS_RE = /[\u200B-\u200F\u202A-\u202E\u2066-\u2069\uFEFF]/gu;
+function parseNumericDate(input) {
+  const iso = /^(\d{4})-(\d{1,2})-(\d{1,2})(?:[T ](\d{1,2}):(\d{2})(?::(\d{2}))?)?$/u.exec(input);
+  if (iso) {
+    return buildLocalDate(
+      Number(iso[1]),
+      Number(iso[2]) - 1,
+      Number(iso[3]),
+      iso[4] === void 0 ? 0 : Number(iso[4]),
+      iso[5] === void 0 ? 0 : Number(iso[5]),
+      iso[6] === void 0 ? 0 : Number(iso[6])
+    );
+  }
+  const dayFirst = /^(\d{1,2})[/\-.](\d{1,2})[/\-.](\d{2,4})(?:[ ,T]+(\d{1,2}):(\d{2})(?::(\d{2}))?)?$/u.exec(input);
+  if (dayFirst) {
+    return buildLocalDate(
+      expandYear(Number(dayFirst[3])),
+      Number(dayFirst[2]) - 1,
+      Number(dayFirst[1]),
+      dayFirst[4] === void 0 ? 0 : Number(dayFirst[4]),
+      dayFirst[5] === void 0 ? 0 : Number(dayFirst[5]),
+      dayFirst[6] === void 0 ? 0 : Number(dayFirst[6])
+    );
+  }
+  return null;
+}
+function parseUrduDate(input, options = {}) {
+  if (!input) return null;
+  const normalized = toEnglishDigits(input).replace(INVISIBLE_CHARS_RE, "").replace(/[،۔]/gu, " ").replace(/\s+/gu, " ").trim();
+  if (normalized === "") return null;
+  const numeric = parseNumericDate(normalized);
+  if (numeric) return numeric;
+  let month = null;
+  let hour = null;
+  let minute = 0;
+  let second = 0;
+  let period;
+  const numbers = [];
+  for (const rawToken of normalized.split(" ")) {
+    const token = rawToken.replace(/^[^\p{L}\p{N}]+/u, "").replace(/[^\p{L}\p{N}]+$/u, "");
+    if (token === "") continue;
+    const urduPeriod = URDU_PERIODS[token];
+    if (urduPeriod) {
+      period = urduPeriod;
+      continue;
+    }
+    const lower = token.toLowerCase();
+    if (lower === "am" || lower === "pm") {
+      period = lower;
+      continue;
+    }
+    if (WEEKDAY_TOKENS.has(token) || WEEKDAY_TOKENS.has(lower)) continue;
+    const time = /^(\d{1,2}):(\d{2})(?::(\d{2}))?$/u.exec(token);
+    if (time) {
+      hour = Number(time[1]);
+      minute = Number(time[2]);
+      second = time[3] === void 0 ? 0 : Number(time[3]);
+      continue;
+    }
+    const monthIndex = MONTH_LOOKUP[token] ?? MONTH_LOOKUP[lower];
+    if (monthIndex !== void 0) {
+      month = monthIndex;
+      continue;
+    }
+    if (/^\d+$/u.test(token)) numbers.push(Number(token));
+  }
+  if (month === null || numbers.length === 0) return null;
+  let day;
+  let year;
+  if (numbers.length === 1) {
+    const only = numbers[0];
+    if (only > 31) {
+      day = 1;
+      year = only;
+    } else {
+      day = only;
+      year = options.defaultYear ?? (/* @__PURE__ */ new Date()).getFullYear();
+    }
+  } else {
+    const first = numbers[0];
+    const last = numbers[numbers.length - 1];
+    if (first > 31 && last <= 31) {
+      day = last;
+      year = first;
+    } else {
+      day = first;
+      year = last;
+    }
+  }
+  return buildLocalDate(
+    expandYear(year),
+    month,
+    day,
+    hour === null ? 0 : resolveHour(hour, period),
+    minute,
+    second
+  );
+}
 
 // src/stemmer.ts
 var URDU_PREFIXES = [
@@ -4710,6 +5055,6 @@ function extractNameParts(fullName) {
   return { honorific, firstName, familyName, suffix };
 }
 
-export { ENGLISH_NAME_ALIASES, HONORIFICS, NAME_PREFIXES, PROTECTED_WORDS, SENTENCE_ABBREVIATIONS, URDU_FAMILY_NAMES, URDU_FIRST_NAMES, URDU_MONTHS_GREGORIAN, URDU_MONTHS_HIJRI, URDU_PREFIXES, URDU_STOP_WORDS, URDU_SUFFIXES, URDU_WEEKDAYS, analyzeUrdu, compareUrdu, convertNumbers, countSentences, countWords, editDistance, extractNameParts, filterStopWords, foldUrdu, formatUrduDate, getAffixes, getUrduMonthName, getUrduWeekdayName, hasUrduSpecificLetters, highlightUrdu, isStopWord, isUrdu, normalizeUrdu, numberToUrduWords, parseUrduNumber, removeDiacritics, removeStopWords, romanToUrdu, romanize, searchUrdu, searchUrduRanked, sortUrdu, splitSentences, splitWords, stemUrdu, stemUrduText, timeAgoUrdu, toArabicIndicDigits, toEnglishDigits, toUrduDigits, transliterateNameToEnglish, transliterateNameToUrdu, urduRatio, urduSlug };
+export { ENGLISH_NAME_ALIASES, HONORIFICS, NAME_PREFIXES, PROTECTED_WORDS, SENTENCE_ABBREVIATIONS, URDU_FAMILY_NAMES, URDU_FIRST_NAMES, URDU_MONTHS_GREGORIAN, URDU_MONTHS_HIJRI, URDU_ONES, URDU_ORDINAL_IRREGULARS, URDU_PREFIXES, URDU_SCALES, URDU_STOP_WORDS, URDU_SUFFIXES, URDU_TEENS_AND_TENS, URDU_TENS_UNITS, URDU_WEEKDAYS, analyzeUrdu, compareUrdu, convertNumbers, countSentences, countWords, editDistance, extractNameParts, filterStopWords, foldUrdu, formatUrduCurrency, formatUrduDate, formatUrduNumber, getAffixes, getUrduMonthName, getUrduWeekdayName, hasUrduSpecificLetters, highlightUrdu, isStopWord, isUrdu, normalizeUrdu, numberToUrduWords, parseUrduDate, parseUrduNumber, removeDiacritics, removeStopWords, romanToUrdu, romanize, searchUrdu, searchUrduRanked, sortUrdu, splitSentences, splitWords, stemUrdu, stemUrduText, timeAgoUrdu, toArabicIndicDigits, toEnglishDigits, toUrduDigits, toUrduOrdinal, transliterateNameToEnglish, transliterateNameToUrdu, urduRatio, urduSlug };
 //# sourceMappingURL=index.js.map
 //# sourceMappingURL=index.js.map