@whiteslove/parsing-lexicon 0.3.3 → 0.3.5

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.3.3",
+  "version": "0.3.5",
   "description": "Shared deterministic multilingual parsing lexicon for Whiteslove housing and hiring services",
   "repository": {
     "type": "git",

package/src/geo.js

@@ -20,7 +20,7 @@ export const TASHKENT_DISTRICTS = Object.freeze([
   entity('Yunusabad', { uzLatn: ['Yunusobod', 'Yunusobod tumani', 'Yunusabad'], uzCyrl: ['Юнусобод', 'Юнусобод тумани'], ru: ['Юнусабад', 'Юнусабадский район'], en: ['Yunusabad', 'Yunusobod'] }),
   entity('Yakkasaray', { uzLatn: ['Yakkasaroy', 'Yakkasaroy tumani', 'Yakkasaray'], uzCyrl: ['Яккасарой', 'Яккасарой тумани'], ru: ['Яккасарай', 'Яккасарайский район'], en: ['Yakkasaray', 'Yakkasaroy'] }),
   entity('Yangihayot', { uzLatn: ['Yangihayot', 'Yangihayot tumani'], uzCyrl: ['Янгиҳаёт', 'Янгиҳаёт тумани', 'Янгихаёт'], ru: ['Янгихаётский район', 'Янгихаёт'], en: ['Yangihayot'] }),
-  entity('Yashnobod', { uzLatn: ['Yashnobod', 'Yashnobod tumani'], uzCyrl: ['Яшнобод', 'Яшнобод тумани'], ru: ['Яшнабад', 'Яшнабадский район'], en: ['Yashnobod', 'Yashnabad'] }),
+  entity('Yashnobod', { uzLatn: ['Yashnobod', 'Yashnobod tumani', 'Yashnabod', 'Yashnabod tumani'], uzCyrl: ['Яшнобод', 'Яшнобод тумани'], ru: ['Яшнабад', 'Яшнабадский район'], en: ['Yashnobod', 'Yashnabad'] }),
 ]);
 
 export function canonicalTashkentDistrict(value) {

package/src/hiring-source-aliases.js

@@ -76,7 +76,6 @@ export const SOURCE_ROLE_NORMALIZATION_RULES = Object.freeze([
   sourceRoleNormalization('finance_banking_specialist', 'Finance / Banking Specialist', /^kredit\s+bo['’ʻʼ‘`]?yicha\s+mutaxa(?:s|ss)is$/iu),
   sourceRoleNormalization('finance_banking_specialist', 'Finance / Banking Specialist', /^(?:финансы?\s*[,/&+]\s*банки?|банки?\s*[,/&+]\s*финансы?|finance\s*[,/&+]\s*banking)$/iu),
   sourceRoleNormalization('welder', 'Welder', /^(?:svarchik|svarshik)$/iu),
-  sourceRoleNormalization('factory_worker', 'Factory Worker', /^(?:jizzax\s+)?kia\s+zavodidan\s+ish\s+kerak$|^zavod\s+ishlari(?:\s+.*)?$/iu),
   sourceRoleNormalization('it_specialist', 'IT Specialist', /^(?:kompyuter\s+(?:sohasida|xizmatlari\s+bo['’ʻʼ‘`]?yicha\s+ish\s+kerak)|it\s+kompyuter)$/iu),
   sourceRoleNormalization('restaurant_manager', 'Restaurant Manager', /^(?:restoran|restaurant)[^\n]*(?:boshqaruv|manager|menejer)/iu),
   sourceRoleNormalization('administrator', 'Administrator', /^virtual\s+asistent$/iu),
@@ -108,6 +107,7 @@ export const SOURCE_ROLE_NORMALIZATION_RULES = Object.freeze([
   sourceRoleNormalization('healthcare_specialist', 'Healthcare Specialist', /^mededsina$|^meditsina$|^медицина$/iu),
   sourceRoleNormalization('tourism_hospitality_specialist', 'Tourism / Hospitality Specialist', /^mehmonxona[^\n]*turfirma|^turfirma[^\n]*mehmonxona/iu),
   sourceRoleNormalization('confectioner', 'Confectioner', /qandolat|qandolatchi/iu),
+  sourceRoleNormalization('factory_worker', 'Factory Worker', /^(?:jizzax\s+)?kia\s+zavodidan\s+ish\s+kerak$|^zavod\s+ishlari(?:\s+.*)?$/iu),
 ]);
 
 export function normalizeSourceRole(value) {

package/src/housing-listing-enrichment.d.ts

@@ -44,6 +44,7 @@ export interface HousingListingEnrichment {
   nearby?: readonly string[];
   amenities?: readonly string[];
   district?: string | null;
+  quarter?: { number: number; suffix: string } | null;
   metro?: string | null;
   residenceComplex?: string | null;
   address?: string | null;

package/src/housing-listing-enrichment.js

@@ -1,7 +1,7 @@
 import { deepFreeze } from './lexicon-core.js';
 import { findAllCanonical, normalizeUnicode } from './normalization.js';
 import { GENERIC_LANDMARK_TERMS } from './landmarks.js';
-import { matchTashkentHousingDistrict, matchTashkentHousingMetro } from './tashkent-housing-geography.js';
+import { matchTashkentHousingDistrict, matchTashkentHousingMetro, matchTashkentHousingQuarter } from './tashkent-housing-geography.js';
 import { TASHKENT_LANDMARKS } from './tashkent-pois.js';
 import { parseHousingRoomCount, parseHousingFloor, parseHousingAreas } from './housing-structured.js';
 import { parseHousingListingFields } from './housing-listing-fields.js';
@@ -18,10 +18,10 @@ const GENERIC_CATEGORY = Object.freeze({
 });
 
 const APPLIANCE_PATTERNS = Object.freeze([
-  ['Washing machine', /(?:кир\s*машин|стиральн\p{L}*\s+машин|washing\s+machine|kir\s*moshina|kirmoshina|kir\s*yuvish\s+mashin)/iu],
-  ['Microwave', /(?:микроволнов|microwave|mikrovolnov|mikravolnof)/iu],
-  ['Vacuum cleaner', /(?:пылесос|vacuum\s+cleaner|pilesos|pedesos)/iu],
-  ['Oven', /(?:духовк|oven|duxovka|duhofka|dxofka)/iu],
+  ['Washing machine', /(?:кир\s*машин|кирмошин\p{L}*|стиральн\p{L}*\s+машин|washing\s+machine|kir\s*moshina|kirmoshina|kir\s*yuvish\s+mashin)/iu],
+  ['Microwave', /(?:микроволнов|микравалноф\p{L}*|microwave|mikrovolnov|mikravolnof)/iu],
+  ['Vacuum cleaner', /(?:пылесос|педесос|vacuum\s+cleaner|pilesos|pedesos)/iu],
+  ['Oven', /(?:духовк|дхофк\p{L}*|oven|duxovka|duhofka|dxofka)/iu],
   ['Kitchen', /(?:кухн|kitchen|oshxona|kuxn\p{L}*)/iu],
 ]);
 
@@ -141,6 +141,11 @@ const COMMISSION_AMOUNT_LEADING_RE = /(\d{1,3}(?:[\s.,]\d{3})*|\d+(?:[.,]\d+)?)\
 
 export function parseHousingCommissionAmount(value) {
   const text = normalizeUnicode(value ?? '');
+  // A listing that explicitly says "no broker/commission" cannot also have a
+  // commission amount — without this guard, an unrelated number near the
+  // negated broker word (e.g. a room count in "2 xonali ... maklersiz")
+  // could be misread as the commission amount.
+  if (NO_BROKER_RE.test(text)) return null;
   const match = text.match(COMMISSION_AMOUNT_RE) || text.match(COMMISSION_AMOUNT_LEADING_RE);
   if (!match) return null;
   const amount = Number(String(match[1]).replace(/\s+/g, '').replace(',', '.'));
@@ -193,7 +198,8 @@ export function parseHousingListingEnrichment(value, { country = '' } = {}) {
   const audience = parseHousingAudience(text);
   const perPersonPrice = parseHousingPerPersonPrice(text, { country });
   const observedAmenities = parseHousingObservedAmenities(text);
-  const district = matchTashkentHousingDistrict(text)?.name || null;
+  const quarter = matchTashkentHousingQuarter(text);
+  const district = matchTashkentHousingDistrict(text)?.name || quarter?.district || null;
   const metro = matchTashkentHousingMetro(text)?.name || null;
   const parsedRc = specificResidentialComplex(text) || parseHousingResidentialComplex(text);
   const commission = parseHousingCommission(text);
@@ -243,6 +249,7 @@ export function parseHousingListingEnrichment(value, { country = '' } = {}) {
     nearby: parseHousingNearby(text),
     amenities: observedAmenities,
     district: district || null,
+    quarter: quarter ? { number: quarter.number, suffix: quarter.suffix } : null,
     metro: metro || null,
     residenceComplex: parsedRc || null,
     address: address.address,

package/src/housing-structured.js

@@ -49,6 +49,18 @@ function currencyNear(text) {
 export function parseHousingRoomCount(value) {
   const text = normalizeUnicode(value ?? '');
   if (!text) return null;
+
+  // Uzbek market shorthand: a unit formally registered as N rooms that was
+  // physically converted into a different count ("1 xonali ... 2 xona
+  // qilingan" — nominally 1-room, made into 2). The converted count is the
+  // one that matters and must win over the nominal count mentioned earlier
+  // in the same text.
+  const converted = text.match(/(\d{1,2})\s*(?:ta\s*)?xona(?:ga)?\s+(?:qilingan|aylantirilgan|bo['’ʻʼ‘`]?lingan)(?=$|[^\p{L}\p{N}_])/iu);
+  if (converted) {
+    const rooms = toNumber(converted[1]);
+    if (rooms != null && rooms >= 1 && rooms <= 20) return rooms;
+  }
+
   const describedRooms = [...text.matchAll(/(?:^|[^\p{L}\p{N}_])(\d{1,2})\s*ta\s+(?:katta|kichkina)\s+xona(?=$|[^\p{L}\p{N}_])/giu)];
   if (describedRooms.length > 1) {
     const total = describedRooms.reduce((sum, match) => sum + Number(match[1]), 0);

package/src/tashkent-housing-geography.d.ts

@@ -44,5 +44,6 @@ export function matchTashkentNumberedArea(value: unknown, canonical: string): Ta
 export function hasTashkentAreaAlias(value: unknown, canonical: string): boolean;
 export function hasExplicitTashkentDistrict(value: unknown, canonical: string): boolean;
 export function matchTashkentHousingDistrict(value: unknown): TashkentDistrictEntry | null;
+export function matchTashkentHousingQuarter(value: unknown): Readonly<{ district: string; number: number; suffix: string }> | null;
 export function matchTashkentHousingMetro(value: unknown): TashkentMetroEntry | null;
 export function matchTashkentHousingTransit(value: unknown): TashkentHousingLocationEntry | null;

package/src/tashkent-housing-geography.js

@@ -252,6 +252,23 @@ export function matchTashkentHousingDistrict(value) {
   return null;
 }
 
+/**
+ * Resolve a numbered massif/quarter mention (e.g. "Юнусабад-7 кв",
+ * "Yunusabad-7") that matchTashkentHousingDistrict() deliberately defers,
+ * returning the district it belongs to alongside the quarter number.
+ */
+export function matchTashkentHousingQuarter(value) {
+  const text = String(value ?? '');
+  if (!text) return null;
+  for (const canonical of Object.keys(TASHKENT_NUMBERED_AREA_ALIASES)) {
+    const numbered = matchTashkentNumberedArea(text, canonical);
+    if (!numbered) continue;
+    const district = TASHKENT_HOUSING_DISTRICTS.find((entry) => entry.name === canonical) || null;
+    return Object.freeze({ district: district?.name || canonical, number: numbered.number, suffix: numbered.suffix });
+  }
+  return null;
+}
+
 const EXTRA_METRO_ALIASES = Object.freeze({
   Sergeli: Object.freeze(['Sergile', 'Sergele']),
 });
@@ -272,6 +289,11 @@ export function matchTashkentHousingMetro(value) {
     const match = text.match(station.re);
     if (!match) continue;
     if (hasExplicitMetroContext(text, match)) return station;
+    // "Toshkent" is both a metro station and the city's own name, so a bare
+    // mention ("Toshkent shahri") is not evidence of the station the way a
+    // bare mention of any other station name would be. Require an explicit
+    // metro context for this one station specifically.
+    if (station.name === 'Toshkent') continue;
     if (station.name === 'Qoyliq' && QOYLIQ_MASSIF_RE.test(text)) continue;
     const areaCanonical = METRO_NUMBERED_AREA[station.name];
     if (areaCanonical && matchTashkentNumberedArea(text, areaCanonical)) continue;

package/src/ua-location-extensions-regional.js

@@ -177,10 +177,10 @@ export const UA_REGIONAL_LOCATION_EXTENSIONS = Object.freeze({
 
   Chernivtsi: city({
     microdistricts: [
-      ['Tsentr','Центр'],['Prospekt','Проспект'],['Komarova','Комарова'],['Pivdenno-Kiltseva','Південно-Кільцева','Южно-Кольцевая'],['Hraviton','Гравітон','Гравитон'],['Kalichanka','Калічанка','Каличанка'],['Sadgora','Садгора','Садгірський','Садгірський район'],['Roscha','Роша'],['Klokuchka','Клокучка'],['Roscha-Stynka','Роша-Стинка'],['Tsetsyno','Цецино'],['Lenkivtsi','Ленківці','Ленковцы'],['Ruska','Руська'],['Fastivska','Фастівська'],
+      ['Tsentr','Центр'],['Prospekt','Проспект'],['Komarova','Комарова'],['Pivdenno-Kiltseva','Південно-Кільцева','Южно-Кольцевая'],['Hraviton','Гравітон','Гравитон'],['Kalichanka','Калічанка','Каличанка'],['Sadgora','Садгора','Садгірський','Садгірський район'],['Roscha','Роша'],['Klokuchka','Клокучка'],['Roscha-Stynka','Роша-Стинка'],['Tsetsyno','Цецино'],['Lenkivtsi','Ленківці','Ленковцы'],['Fastivska','Фастівська'],
     ],
     residentialComplexes: [
-      ['Vodohrai','ЖК Водограй'],['Compass','ЖК Compass','ЖК Компас'],['Grand Royal','ЖК Grand Royal'],['Family House','ЖК Family House'],['Comfort Hall','ЖК Comfort Hall'],['Kyivskyi','ЖК Київський','ЖК Киевский'],['Rodynnyi','ЖК Родинний','ЖК Родинный'],['Panorama','ЖК Panorama'],['Park Avenue','ЖК Park Avenue'],['Central','ЖК Central'],
+      ['Vodohrai','ЖК Водограй'],['Compass','ЖК Compass','ЖК Компас'],['Comfort Hall','ЖК Comfort Hall'],['Kyivskyi','ЖК Київський','ЖК Киевский'],['Rodynnyi','ЖК Родинний','ЖК Родинный'],['Panorama','ЖК Panorama'],['Park Avenue','ЖК Park Avenue'],['Central','ЖК Central'],
     ],
     landmarks: [
       ['Metropolitan Residence','Резиденція митрополитів','Резиденция митрополитов'],['Chernivtsi University','ЧНУ','університет Федьковича'],['Teatralna Square','Театральна площа'],['Central Square','Центральна площа'],['Turkish Square','Турецька площа','Турецька криниця'],['Kobylianska Street','Ольги Кобилянської','вулиця Кобилянської'],['Shevchenko Park','парк Шевченка'],['Botanical Garden','Ботанічний сад'],['City Hall','Ратуша'],