@whiteslove/parsing-lexicon 0.8.0 → 0.8.1

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.8.0",
+  "version": "0.8.1",
   "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
   "repository": {
     "type": "git",

package/src/contact.js

@@ -9,6 +9,7 @@ const PHONE_LIKE_RE = /\+?\d(?:[\t \u00a0().-]*\d){9,}/g;
 const PHONE_CANDIDATE_RE = /\+?\d(?:[\t \u00a0().-]*\d){6,}(?:[\t \u00a0]*(?:ext\.?|extension|x|доб\.?|дод\.?)\s*\d{1,6})?/giu;
 const PHONE_EXTENSION_RE = /[\t \u00a0]*(?:ext\.?|extension|x|доб\.?|дод\.?)\s*(\d{1,6})$/iu;
 const DATE_LIKE_PHONE_RE = /^\d{1,2}[./-]\d{1,2}[./-](?:\d{2}|\d{4})(?:\s+\d{1,2})?$/u;
+const PRICE_LABEL_BEFORE_NUMBER_RE = /(?:цена|ціна|нарх(?:и)?|narx(?:i)?|price|стоимост[ьи]|аренд(?:а|ная\s+плата)?|rent)\s*[:=\-–—]?\s*$/iu;
 
 const TELEGRAM_USERNAME_RE = /^[A-Za-z0-9_]{5,32}$/;
 const TELEGRAM_LINK_RE = /(?:https?:\/\/)?(?:t\.me|telegram\.me|telegram\.dog)\/([A-Za-z0-9_]{5,32})(?:\/[0-9]+)?(?:[/?#][^\s]*)?/giu;
@@ -20,6 +21,18 @@ function normalizedCountryHint(value) {
   return /^[A-Z]{2}$/.test(country) ? country : undefined;
 }
 
+// PHONE_LIKE_RE deliberately accepts punctuation-separated digit sequences.
+// That also resembles a grouped monetary range such as
+// "Narxi: 950 000 - 1.000.000".  A nearby explicit price label is stronger
+// semantic evidence than the broad (unvalidated) ten-digit phone mask, so do
+// not hide that span before the money candidate parser sees it.  This does
+// not weaken validated national-phone parsing below.
+function isExplicitPriceSpan(text, start, raw) {
+  if (!/[\-–—]/u.test(raw)) return false;
+  const before = text.slice(Math.max(0, start - 48), start);
+  return PRICE_LABEL_BEFORE_NUMBER_RE.test(before);
+}
+
 function splitPhoneExtension(raw) {
   const match = String(raw || '').match(PHONE_EXTENSION_RE);
   if (!match) return { base: String(raw || '').trim(), extension: null };
@@ -38,6 +51,7 @@ export function findPhoneLikeSpans(value, options = {}) {
     const digits = raw.replace(/\D/g, '');
     if (digits.length < 10) continue;
     const start = match.index ?? 0;
+    if (isExplicitPriceSpan(text, start, raw)) continue;
     spans.push(Object.freeze({
       start,
       end: start + raw.length,

package/src/housing-listing-enrichment.js

@@ -32,11 +32,15 @@ const APPLIANCE_PATTERNS = Object.freeze([
 const FIRST_RENT_UZ_RE = /(?:hali\s+hech\s+kim\s+(?:yashamagan|turmagan)|ҳали\s+ҳеч\s+ким\s+(?:яшамаган|турмаган))/iu;
 const LANDLORD_PRESENT_RE = /(?:xozaykali|hojaykali|xo['’]?jaykali|с\s+хозяйк(?:ой|ой\s+в\s+квартире)|хозяйк\p{L}*\s+(?:жив[её]т|прожива\p{L}*)|with\s+(?:the\s+)?(?:landlord|owner)\s+(?:present|living\s+in)|cu\s+proprietar(?:ul)?\s+în\s+cas(?:ă|a)|үй\s*иесі\s+(?:тұрады|бірге\s+тұрады))/iu;
 const STUDENT_RE = /(?:studentlar\s+uchun|talabalar\s+uchun|студент(?:ам|ы|ок|ов)?\s+(?:можно|для)|для\s+студент|students?\s+(?:only|welcome)|for\s+students|pentru\s+studen[țt]i|studen[țt]i(?:lor)?|студенттерге|студенттер\s+үшін|(?:oila|oyla)(?:ga|lar|li)?\s+yoki\s+\d{1,2}\s+ta\s+bola(?:lar)?(?:ga)?\s+(?:ijara(?:ga)?\s+)?(?:beril|topshiril))/iu;
-const NO_BROKER_RE = /(?:bez\s+makler|maklersiz|vositachisiz|без\s+(?:маклер|посредник|риелтор|риэлтор|комисси)|no\s+(?:broker|agent|commission|agency\s+fee)|f[ăa]r[ăa]\s+(?:comision|agen[țt]ie|intermediari)|делдалсыз|комиссиясыз)/iu;
+const NO_BROKER_RE = /(?:bez\s*makler|maklersiz|vositachisiz|без\s+(?:маклер|посредник|риелтор|риэлтор|комисси)|no\s+(?:broker|agent|commission|agency\s+fee)|f[ăa]r[ăa]\s+(?:comision|agen[țt]ie|intermediari)|делдалсыз|комиссиясыз)/iu;
 const BROKER_RE = /(?:makler|vositachi|макл(?:ер[а-яё]*)?|ри[еэ]лтор[а-яё]*|агентств[а-яё]*|комисси[а-яё]*|broker|realtor|commission|comision(?:ul)?|agen[țt]ie|delda[lл]\p{L}*|делдал\p{L}*)/iu;
 const MEN_RE = /(?:o['’ʻʼ‘`]?g['’ʻʼ‘`]?il\s+bola(?:lar)?(?:ga)?|ogil\s+bola(?:lar)?(?:ga)?|sherik\s+bola|эркак(?:лар)?|erkak(?:lar)?(?:ga)?|только\s+(?:мужчин|парн)|\bmen\s+only\b|b[ăa]rba[țt]i(?:lor)?|b[ăa]ie[țt]i(?:lor)?|жігіттерге|жігіттер(?:ге)?|хлопц(?:ям|і|ів)?|чоловік(?:ам|и)?)/iu;
 const WOMEN_RE = /(?:qiz(?:lar)?(?:ga)?|ayol(?:lar)?(?:ga|ni)?|киз(?:лар)?(?:га)?|аёл(?:лар)?(?:га|ни)?|девушк\p{L}*|женщин\p{L}*|girls?\s+only|women\s+only|fete(?:lor)?|femei(?:lor)?|қыздарға|қыздар(?:ға)?|дівчат(?:ам|а|ок)?|жінк(?:ам|и)?)/iu;
 const FAMILY_RE = /(?:семь\p{L}*|family|oila(?:ga|lar|li)?|oyla(?:ga|lar|li)?|oila\s+uchun|оилага|оелага|оилавий|oelaga|famil(?:ie|ia)|cuplu(?:ri)?|отбасына|отбасылы|жанұяға|сім['’ʼ]?[яїі](?:ям|ям[иі])?|сімейн\p{L}*)/iu;
+// A category word in an exclusion clause is negative evidence, not a tenant
+// preference.  Keep this narrow and local to avoid rejecting ordinary
+// "oila uchun" / "для семьи" invitations elsewhere in the listing.
+const FAMILY_EXCLUSION_RE = /(?:oila|oyla|семь\p{L}*|family|famil(?:ie|ia)|cuplu(?:ri)?|отбас\p{L}*|жанұя\p{L}*)[^\r\n.!?]{0,48}(?:bezovta\s+qilmasin|murojaat\s+qilmasin|kerak\s+emas|qabul\s+qilinmaydi|не\s+(?:беспокоить|принимаем)|not\s+(?:allowed|welcome))/iu;
 const ROOM_SHARE_RE = /(?:sherik(?:ka|lik|likga)?|шерик(?:ка|лик)?|roommate|flatmate|подселени|койко[-\s]?мест|место\s+в\s+(?:комнат|квартир)|birga\s+yashash(?:ga)?|kvartira(?:ga|da)?[^\r\n.!?]{0,36}(?:\d+|bitta|1)\s*(?:ta\s*)?(?:qiz|ayol)[^\r\n.!?]{0,20}(?:ijarachi\s*)?(?:kerak|kere)|coleg\s+de\s+(?:apartament|camer[ăa])|bed\s+space|бөлмелес(?:\s+керек)?|көрші\s+керек)/iu;
 const AIR_CONDITIONER_RE = /(?:кондицион|air\s*con|konditsioner|kandit(?:s|c)?aner|kanditsaner|кандитсанер)/iu;
 const PER_PERSON_PRICE_RE = /(?:kishi\s+boshiga|киши\s+бошига)\s*(\d{1,3}(?:[\s.,]\d{3})*|\d+(?:[.,]\d+)?)\s*(ming|минг|million|mln|млн)?(?:dan|дан)?/iu;
@@ -176,7 +180,7 @@ export function parseHousingNearby(value) {
 
 export function parseHousingAudience(value) {
   const text = normalizeUnicode(value ?? '');
-  const family = FAMILY_RE.test(text);
+  const family = FAMILY_RE.test(text) && !FAMILY_EXCLUSION_RE.test(text);
   const women = WOMEN_RE.test(text);
   const men = MEN_RE.test(text);
   const students = STUDENT_RE.test(text);

package/src/housing-money.d.ts

@@ -10,7 +10,7 @@ export type HousingMoneyParseContext = Readonly<{
   fallbackCurrency?: string;
   dealType?: 'sale' | 'longRent' | 'shortRent' | string | null;
 }>;
-export type HousingMoneyCandidate = Readonly<{ amount: number; currency: string; start: number; end: number; explicitCurrency: boolean; scale: string | null; priceKeyword: boolean; paymentRole: string; approximate: boolean; confidenceBoost: number; confidence: number }>;
+export type HousingMoneyCandidate = Readonly<{ amount: number; currency: string; start: number; end: number; explicitCurrency: boolean; scale: string | null; priceKeyword: boolean; range: Readonly<{ minimum: number; maximum: number }> | null; paymentRole: string; approximate: boolean; confidenceBoost: number; confidence: number }>;
 export function extractHousingMoneyCandidates(value: unknown, context?: string | HousingMoneyParseContext): readonly HousingMoneyCandidate[];
 export function rankHousingPriceCandidates(candidates: readonly HousingMoneyCandidate[]): readonly HousingMoneyCandidate[];
 

package/src/housing-money.js

@@ -199,7 +199,7 @@ export function extractHousingMoneyCandidates(value, context = '') {
   const text = maskPhoneLikeSpans(String(value || ''), ' ', { country });
   const candidates = [];
   const seen = new Set();
-  const addCandidate = ({ amount, currency, start, end, explicitCurrency, scale = null, priceKeyword = false, confidenceBoost = 0 }) => {
+  const addCandidate = ({ amount, currency, start, end, explicitCurrency, scale = null, priceKeyword = false, range = null, confidenceBoost = 0 }) => {
     if (amount == null || amount < 1 || amount > 5_000_000_000 || seen.has(`${start}:${end}`)) return;
     seen.add(`${start}:${end}`);
     candidates.push(Object.freeze({
@@ -210,6 +210,7 @@ export function extractHousingMoneyCandidates(value, context = '') {
       explicitCurrency,
       scale,
       priceKeyword,
+      range,
       paymentRole: candidatePaymentRole(text, start, end),
       approximate: APPROXIMATE_RE.test(text.slice(Math.max(0, start - 12), end)),
       confidenceBoost,
@@ -239,6 +240,33 @@ export function extractHousingMoneyCandidates(value, context = '') {
     });
   }
 
+  // A price-labelled range is common in Telegram rentals even when the author
+  // omits "sum".  In Uzbek country context, grouped endpoints are an explicit
+  // UZS signal; preserve both endpoints on the candidate while exposing the
+  // lower bound through the legacy single-price result.  It must be collected
+  // before generic amounts so the second endpoint cannot be selected merely
+  // because it is larger.
+  const labelledPriceRangeRe = new RegExp(
+    `${PRICE_KEYWORD}\\s*[:=\\-–—]?\\s*(${MONEY_NUMBER_PATTERN})\\s*(?:-{1,3}|–|—|to|до|dan\\s+gacha)\\s*(${MONEY_NUMBER_PATTERN})(?=$|[^\\p{L}\\p{N}_])`,
+    'igu',
+  );
+  for (const match of text.matchAll(labelledPriceRangeRe)) {
+    const minimum = parseNumericAmount(match[1]);
+    const maximum = parseNumericAmount(match[2]);
+    if (minimum == null || maximum == null || minimum > maximum) continue;
+    const start = match.index ?? 0;
+    addCandidate({
+      amount: Math.round(minimum),
+      currency: country === 'UZ' ? 'UZS' : fallbackCurrency || '',
+      start,
+      end: start + match[0].length,
+      explicitCurrency: false,
+      priceKeyword: true,
+      range: Object.freeze({ minimum: Math.round(minimum), maximum: Math.round(maximum) }),
+      confidenceBoost: 0.2,
+    });
+  }
+
   const splitMillionRe = /(?:^|[^\p{L}\p{N}_])(\d{1,3})\s*(?:млн\.?|mln\.?|миллион(?:а|ов)?|million(?:s)?)\s+(\d{1,3})(?=$|[^\p{L}\p{N}_])/giu;
   for (const match of text.matchAll(splitMillionRe)) {
     const start = match.index ?? 0;

package/src/housing-structured.js

@@ -87,7 +87,7 @@ export function parseHousingRoomCount(value) {
     if (total >= 1 && total <= 20) return total;
   }
   for (const [re, rooms] of NUMBER_WORDS) if (re.test(text)) return rooms;
-  const numeric = text.match(/(?:^|[^\p{L}\p{N}])(\d{1,2})\s*(?:(?:-\s*)?комнат\p{L}*|(?:-\s*)?к(?:\.|\b)|(?:-\s*)?xona(?:li)?|(?:-\s*)?хона(?:лик|ли)?|бөлмелі|rooms?)(?=$|[^\p{L}\p{N}])/iu);
+  const numeric = text.match(/(?:^|[^\p{L}\p{N}])(\d{1,2})\s*(?:ta\s*)?(?:(?:-\s*)?комнат\p{L}*|(?:-\s*)?к(?:\.|\b)|(?:-\s*)?xona(?:li|si|lari)?|(?:-\s*)?хона(?:лик|ли|си|лари)?|бөлмелі|rooms?)(?=$|[^\p{L}\p{N}])/iu);
   if (numeric) {
     const rooms = toNumber(numeric[1]);
     if (rooms != null && rooms >= 1 && rooms <= 20) return rooms;