@whiteslove/parsing-lexicon 0.2.3 → 0.2.5

package/index.d.ts

@@ -372,4 +372,10 @@ export function parseExperience(value: unknown): ExperienceParseResult | null;
 export * from './src/housing-context.js';
 export * from './src/hiring-context.js';
 export * from './src/housing-intent.js';
-export * from './src/housing-structured.js';
+export * from './src/housing-structured.js';
+
+// housing semantic helper declarations added in 0.2.5
+export function resolveHousingOccupancy(value: unknown): 'wholeProperty' | 'room' | 'sharedRoom' | 'bedSpace' | null;
+export function looksHousingRoomOnly(value: unknown): boolean;
+export function resolveHousingPropertyType(value: unknown): 'flat' | 'house' | 'room' | 'studio' | 'townhouse' | 'dormitory' | null;
+

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.2.3",
+  "version": "0.2.5",
   "description": "Shared deterministic multilingual parsing lexicon for Whiteslove housing and hiring services",
   "repository": {
     "type": "git",

package/src/contact.d.ts

@@ -42,3 +42,5 @@ export function parsePhoneNumbers(value: unknown, options?: ParsePhoneOptions):
 export function normalizePhone(value: unknown, options?: ParsePhoneOptions): ParsedPhoneNumber | null;
 export function findTelegramContacts(value: unknown): readonly TelegramContact[];
 export function normalizeTelegramContact(value: unknown): TelegramContact | null;
+export declare function parsePrimaryContact(value: unknown): string | null;
+

package/src/contact.js

@@ -157,3 +157,25 @@ export function findTelegramContacts(value) {
 export function normalizeTelegramContact(value) {
   return findTelegramContacts(value)[0] || null;
 }
+
+export function parsePrimaryContact(value) {
+  const text = String(value || '');
+  if (!text) return null;
+  const intl = text.match(/\+\d[\d\s().-]{7,}\d/);
+  if (intl) {
+    const digits = intl[0].replace(/\D/g, '');
+    if (digits.length >= 10 && digits.length <= 15) return `+${digits}`;
+  }
+  const keyword = text.match(/(?:tel|тел|phone|моб|whats?app|viber|telegram|звонит|звоніть|aloqa|byla|contact)[^\d+]{0,8}(\+?\d[\d\s().-]{6,}\d)/iu);
+  if (keyword) {
+    const digits = keyword[1].replace(/\D/g, '');
+    if (digits.length >= 9 && digits.length <= 15) return keyword[1].trim();
+  }
+  const trailing = text.match(/(\+?\d[\d\s().-]{6,}\d)\s*(?:tel|тел(?:ефон)?|phone|моб|whats?app|viber|telegram|aloqa|contact)(?=$|[^\p{L}\p{N}_])/iu);
+  if (trailing) {
+    const digits = trailing[1].replace(/\D/g, '');
+    if (digits.length >= 9 && digits.length <= 15) return trailing[1].trim();
+  }
+  return findTelegramContacts(text)[0]?.handle || null;
+}
+

package/src/housing-intent.d.ts

@@ -7,3 +7,5 @@ export const HOUSING_INTENT: readonly unknown[];
 export const HOUSING_DEAL_TYPES: readonly unknown[];
 export const HOUSING_ACTION_MAP: Readonly<Record<HousingAction, Readonly<{ listingKind: HousingListingKind; dealType: Exclude<HousingDealType, 'shortRent'> }>>>;
 export function resolveHousingIntent(value: unknown): HousingIntentResult | null;
+export declare function classifyHousingDealType(value: unknown): 'sale' | 'longRent' | 'shortRent' | null;
+

package/src/housing-intent.js

@@ -130,3 +130,8 @@ export function resolveHousingIntent(value) {
     dealType: durationDeal.canonical,
   });
 }
+
+export function classifyHousingDealType(value) {
+  return resolveHousingIntent(value)?.dealType || null;
+}
+

package/src/housing-structured.js

@@ -127,8 +127,9 @@ export function parseHousingPayments(value) {
   }
 
   const noCommission = SELLER_TERMS.noCommission && findCanonical(text, [SELLER_TERMS.noCommission], { partial: true });
-  const commissionPercent = toNumber(text.match(/(?:комисси\p{L}*|commission|comision|komissiya)[^\d%]{0,16}(\d{1,3}(?:[.,]\d+)?)\s*%/iu)?.[1]);
-  const commissionMentioned = SELLER_TERMS.commission && findCanonical(text, [SELLER_TERMS.commission], { partial: true });
+  const shorthandCommission = text.match(/(?:^|[^\p{L}\p{N}_])[mм]\s*[:.\-]?\s*(\d{1,3})\s*%/iu);
+  const commissionPercent = toNumber(shorthandCommission?.[1] ?? text.match(/(?:комисси\p{L}*|commission|comision|komissiya|маклер|makler|rieltor|vositachi)[^\d%]{0,16}(\d{1,3}(?:[.,]\d+)?)\s*%/iu)?.[1]);
+  const commissionMentioned = Boolean(shorthandCommission) || (SELLER_TERMS.commission && findCanonical(text, [SELLER_TERMS.commission], { partial: true }));
 
   return deepFreeze({
     deposit: {
@@ -150,7 +151,7 @@ export function parseHousingSeller(value) {
   const text = normalizeUnicode(value ?? '');
   if (!text) return deepFreeze({ type: null, confidence: 0 });
   const owner = SELLER_TERMS.owner && findCanonical(text, [SELLER_TERMS.owner], { partial: true });
-  const agency = SELLER_TERMS.agency && findCanonical(text, [SELLER_TERMS.agency], { partial: true });
+  const agency = (SELLER_TERMS.agency && findCanonical(text, [SELLER_TERMS.agency], { partial: true })) || /(?:^|[^\p{L}\p{N}_])[mм]\s*\d{1,3}\s*%/iu.test(text);
   if (owner && !agency) return deepFreeze({ type: 'owner', confidence: 1 });
   if (agency && !owner) return deepFreeze({ type: 'agency', confidence: 1 });
   if (owner && agency) return deepFreeze({ type: null, confidence: 0.45 });

package/src/housing-text.d.ts

@@ -1,2 +1,7 @@
 export declare function parseHousingRoomsFromText(value: unknown): number | null;
 export declare function parseHousingResidentialComplex(value: unknown): string | null;
+export declare function parseHousingAreaFromText(value: unknown): number | null;
+export declare function parseHousingFloorFromText(value: unknown): { floor: number | null; totalFloors: number | null };
+export declare function parseHousingAudience(value: unknown): 'family' | 'women' | 'men' | null;
+export declare function parseHousingAmenities(value: unknown): readonly ('dishwasher' | 'separateRooms' | 'washingMachine' | 'television' | 'bedLinen' | 'towels')[];
+

package/src/housing-text.js

@@ -73,3 +73,89 @@ export function parseHousingResidentialComplex(value) {
   const name = words.join(' ').trim();
   return /[a-zA-Zа-яёіїґ]{2,}/i.test(name) ? name : null;
 }
+
+export function parseHousingAreaFromText(value) {
+  const text = String(value || '');
+  if (!text) return null;
+  const match =
+    text.match(/(\d{2,4})\s*(?:m2|m²|мкв|м2|м²|sq ?m|кв\.?\s*м|квадрат[а-яё]*)/i) ||
+    text.match(/(?:^|\n)[^\d\r\n]{0,8}[1-9]\s*[¹²³⁴⁵⁶⁷⁸⁹]?\s*\/\s*[0-9]{1,2}\s*\/\s*[0-9]{1,2}\s+(\d{2,4})\s*кв(?=\s|$)/im);
+  return match ? Number(match[1]) : null;
+}
+
+export function parseHousingFloorFromText(value) {
+  const text = String(value || '');
+  if (!text) return { floor: null, totalFloors: null };
+  const t = text.toLowerCase();
+  const floorWord = '(?:этаж(?:да)?|поверх|qavat|қабат|қабатт|etaj|floor|эт\\.)';
+  const valid = (floor, total) => floor >= 0 && floor <= 200 && (total == null || (total >= floor && total <= 200));
+
+  if (/(?:^|\n)[^\d\r\n]{0,8}[1-9]\s*\/\s*0\s*\/\s*-1\s*(?:этаж|эт\.?)?[^\r\n]*(?:подвал|цоколь)/im.test(t)) return { floor: -1, totalFloors: null };
+
+  const compact = t.match(/(?:^|\n)[^\d\r\n]{0,8}[1-9]\s*[¹²³⁴⁵⁶⁷⁸⁹]?\s*\/\s*([0-9]{1,2})\s*\/\s*([0-9]{1,2})[.,;:]?(?=\s|$)/m);
+  if (compact) {
+    const floor = Number(compact[1]);
+    const total = Number(compact[2]);
+    if (floor >= 1 && floor <= 40 && total >= 2 && total <= 40 && floor <= total) return { floor, totalFloors: total };
+  }
+
+  const labelledPair = t.match(/([1-9]\d?)\s*-?\s*(?:qavat|этаж|поверх|қабат)\s*(?:\/|из|iz|of)\s*([1-9]\d?)\s*-?\s*(?:qavatli|qavat|этаж(?:ей|ный)?|поверх(?:ів|овий)?|қабатты?)/i);
+  if (labelledPair) {
+    const floor = Number(labelledPair[1]);
+    const total = Number(labelledPair[2]);
+    if (floor <= 40 && total <= 40 && floor <= total) return { floor, totalFloors: total };
+  }
+
+  const sep = '(?:\\/|из|iz|of)';
+  const pair = t.match(new RegExp(`${floorWord}\\D{0,4}(\\d{1,2})\\s*${sep}\\s*(\\d{1,2})`)) || t.match(new RegExp(`(\\d{1,2})\\s*${sep}\\s*(\\d{1,2})\\s*${floorWord}`));
+  if (pair) {
+    const floor = Number(pair[1]);
+    const total = Number(pair[2]);
+    if (valid(floor, total)) return { floor, totalFloors: total };
+  }
+
+  const notLetter = '(?!н|ей|ів|ност|ка|ки|s)';
+  const single = t.match(new RegExp(`(\\d{1,2})[^\\S\\r\\n]*-?[^\\S\\r\\n]*(?:го|ом|ым|ой|ий|nd|rd|th|st|й|м|е)?[^\\S\\r\\n]*${floorWord}${notLetter}`)) || t.match(new RegExp(`${floorWord}\\s*[:№#]?\\s*(\\d{1,2})\\b`));
+  if (single) {
+    const floor = Number(single[1]);
+    if (valid(floor, null)) {
+      const explicitTotal = t.match(/(?:этажность|этажей|поверхови|поверховість|qavatlar(?:\s*soni)?|qavatli|қабатты?)\D{0,6}(\d{1,2})/);
+      const leadingTotal = t.match(/([1-9]\d?)\s*-?\s*(?:этажн[а-яё]*|поверхов[а-яіїґ]*|qavatli|қабатты?)\s*(?:дом|здани|будин|uy|bino)?/i);
+      const total = explicitTotal ? Number(explicitTotal[1]) : leadingTotal ? Number(leadingTotal[1]) : null;
+      return { floor, totalFloors: total && total >= floor && total <= 200 ? total : null };
+    }
+  }
+
+  const bare = t.match(/(?<![\d/])([1-9]\d?)\s*\/\s*([1-9]\d?)(?![\d/])/);
+  if (bare) {
+    const floor = Number(bare[1]);
+    const total = Number(bare[2]);
+    if (floor >= 1 && floor <= 40 && total >= 2 && total <= 40 && floor <= total) return { floor, totalFloors: total };
+  }
+  return { floor: null, totalFloors: null };
+}
+
+export function parseHousingAudience(value) {
+  const text = String(value || '');
+  if (!text) return null;
+  const t = text.toLowerCase();
+  if (/(?:семейн|сімейн|для семь|для сім)[^.\n]{0,80}(?:одиноч|мужчин|женщин|чоловік|жінок)|(?:одиноч|мужчин|женщин|чоловік|жінок)[^.\n]{0,80}(?:семейн|сімейн|для семь|для сім)/u.test(t)) return null;
+  if (/(для семь|семейн|сімейн|для сім|для родин|for famil|families?|pentru famil|oila(?:ga| uchun|\s+qo['’`]?yiladi|\s+quyiladi)|оила|отбасы)/u.test(t)) return 'family';
+  if (/(девуш|девоч|для дівч|дівчат|for girls|for women|only girls|doar fete|\bfete\b|qiz(?:lar|la)?(?:ga| uchun)?|(?:қ|к)из(?:лар|ла)?|қыздар)/u.test(t)) return 'women';
+  if (/(парн(ей|ям)|для мужчин|мужчинам|для хлопц|for men\b|for boys|doar b[aă]ie[țt]i|yigit(lar)?(ga| uchun)?|(?:ў|у)гил\s*бол|жігіт|ер адам)/u.test(t)) return 'men';
+  return null;
+}
+
+export function parseHousingAmenities(value) {
+  const text = String(value || '');
+  if (!text) return Object.freeze([]);
+  const amenities = [];
+  if (/(?:посудомо|посудомийн|dishwasher|idish\s*yuvish|idishyuvg|ma[șs]ina de sp[ăa]lat vase)/iu.test(text)) amenities.push('dishwasher');
+  if (/(?:комнат\p{L}*\s+раздельн|изолированн\p{L}*\s+комнат|separate\s+rooms?)/iu.test(text)) amenities.push('separateRooms');
+  if (/(?:стиральн\p{L}*\s+машин|washing\s+machine|kir\s*yuvish\s*mashin|kirmoshina)/iu.test(text)) amenities.push('washingMachine');
+  if (/(?:телевизор|телевизион|televizor|television|\btv\b)/iu.test(text)) amenities.push('television');
+  if (/(?:постельн\p{L}*\s+бель|bed\s*linen|toza\s+choyshab|yostiq\s+jild)/iu.test(text)) amenities.push('bedLinen');
+  if (/(?:полотенц|towels?|sochiq)/iu.test(text)) amenities.push('towels');
+  return Object.freeze(amenities);
+}
+

package/src/housing.js

@@ -1,4 +1,5 @@
 import { lexiconEntity } from './lexicon-core.js';
+import { findCanonical } from './normalization.js';
 import { HOUSING_DEAL_TYPES } from './housing-intent.js';
 const group = (canonical, aliases, extra = {}) => lexiconEntity(canonical, aliases, extra);
 
@@ -76,7 +77,7 @@ export const PROPERTY_TYPES = Object.freeze([
   }),
   group('house', {
     ru: ['дом', 'частный дом', 'коттедж'], en: ['house', 'home', 'cottage'], uk: ['будинок', 'приватний будинок', 'котедж'], ro: ['casă', 'casa', 'vilă', 'vila'],
-    uzLatn: ['uy', 'hovli', 'xovli'], uzCyrl: ['уй', 'ҳовли', 'ховли'], kk: ['үй', 'жеке үй', 'коттедж'],
+    uzLatn: ['hovli', 'xovli'], uzCyrl: ['ҳовли', 'ховли'], kk: ['үй', 'жеке үй', 'коттедж'],
   }),
   group('room', {
     ru: ['комната', 'комнату'], en: ['room'], uk: ['кімната'], ro: ['cameră', 'camera'], uzLatn: ['xona', 'hona'], uzCyrl: ['хона'], kk: ['бөлме'],
@@ -198,3 +199,28 @@ export function flattenAliases(item) {
   if (!item?.aliases) return [];
   return Object.values(item.aliases).flat();
 }
+
+export function resolveHousingOccupancy(value) {
+  const match = findCanonical(value, HOUSING_OCCUPANCY_TYPES, { partial: true });
+  return match?.canonical || null;
+}
+
+export function looksHousingRoomOnly(value) {
+  const text = String(value || '');
+  if (!text) return false;
+  const occupancy = resolveHousingOccupancy(text);
+  if (occupancy === 'room' || occupancy === 'sharedRoom' || occupancy === 'bedSpace') return true;
+  return /подселени|підселен|комнату\s+в|кімнату\s+в|сда[её]тся\s+комната|сдается\s+комната|сдам\s+комнату|здам\s+кімнат|room\s+in\s+a\s+(?:shared\s+)?flat|room\s+for\s+rent|shared\s+(?:flat|apartment|room)|roommate|flatmate|xona\s+ijaraga|xona\s+beriladi|sherik(?:ka|lik)|шерик(?:ка|лик)|(?:1|бир)\s*та\s*(?:бола|киши|қиз|киз)\s*керак|1\s*хонага[^\r\n]{0,40}(?:киши|одам)\s*турилади|бөлме\s+жалға|închiriez\s+camer[ăa]|ищу[^\r\n]{0,60}сосед|ищем[^\r\n]{0,60}сосед|нужен[^\r\n]{0,60}сосед|нужна[^\r\n]{0,60}сосед|шукаю[^\r\n]{0,60}сусід|шукаємо[^\r\n]{0,60}сусід|потрібен[^\r\n]{0,60}сусід|потрібна[^\r\n]{0,60}сусід|співмешкан|співжител|соседк|сусідк/iu.test(text);
+}
+
+export function resolveHousingPropertyType(value) {
+  const text = String(value || '');
+  if (!text) return null;
+  const flat = PROPERTY_TYPES.find((entry) => entry.canonical === 'flat');
+  if (flat && findCanonical(text, [flat], { partial: true })) return 'flat';
+  const genericUzbekHome = /(?:^|[^\p{L}\p{N}_])(?:uy|уй)(?=$|[^\p{L}\p{N}_])/iu.test(text);
+  const explicitHouse = /(?:hovli|xovli|ҳовли|ховли|house|casa|dom|villa|будин|коттедж|вілл|вилл|(?:^|[^\p{L}\p{N}_])(?:дом|үй)(?=$|[^\p{L}\p{N}_]))/iu.test(text);
+  if (genericUzbekHome && !explicitHouse) return null;
+  return findCanonical(text, PROPERTY_TYPES, { partial: true })?.canonical || null;
+}
+