@whiteslove/parsing-lexicon 0.10.0 → 0.10.2

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.10.0",
+  "version": "0.10.2",
   "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
   "repository": {
     "type": "git",

package/src/geo.js

@@ -19,7 +19,7 @@ export const TASHKENT_DISTRICTS = Object.freeze([
   entity('Shaykhantahur', { uzLatn: ['Shayxontohur', 'Shayxontohur tumani', 'Shaykhantahur'], uzCyrl: ['Шайхонтоҳур', 'Шайхонтоҳур тумани', 'Шайхонтохур'], ru: ['Шайхантахур', 'Шайхантахурский район'], en: ['Shaykhantahur', 'Shayxontohur'] }),
   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('Yangihayot', { uzLatn: ['Yangihayot', 'Yangihayot tumani'], uzCyrl: ['Янгиҳаёт', 'Янгиҳаёт тумани', 'Янгихаёт'], ru: ['Янгихаётский район', 'Янгихаёт', 'Янгихаетский район', 'Янгихаетский', 'Янгихает'], en: ['Yangihayot'] }),
   entity('Yashnobod', { uzLatn: ['Yashnobod', 'Yashnobod tumani', 'Yashnabod', 'Yashnabod tumani', 'Yashnobot', 'Yashnobot tumani'], uzCyrl: ['Яшнобод', 'Яшнобод тумани'], ru: ['Яшнабад', 'Яшнабадский район'], en: ['Yashnobod', 'Yashnabad'] }),
 ]);
 
@@ -28,17 +28,21 @@ export function canonicalTashkentDistrict(value) {
 }
 
 function station(name, ru, en, line, aliases = []) {
-  const allAliases = [...new Set([name, ru, en, ...aliases])];
-  return deepFreeze({
+  const allAliases = Object.freeze([...new Set([name, ru, en, ...aliases])]);
+  let re = null;
+  return Object.freeze({
     canonical: name,
     name,
     type: 'metro',
     country: 'UZ',
     city: 'Tashkent',
     line,
-    labels: { ru, en },
+    labels: Object.freeze({ ru, en }),
     aliases: allAliases,
-    re: aliasesToRegex(allAliases),
+    get re() {
+      re ||= aliasesToRegex(allAliases);
+      return re;
+    },
   });
 }
 

package/src/geography-detection.js

@@ -14,10 +14,18 @@ function cityInflectionRegex(aliases) {
 
 const CITY_MATCHERS = CITIES.map((item) => {
   const aliases = [item.canonical, ...aliasesOf(item)];
+  let exact = null;
+  let inflected;
   return Object.freeze({
     item,
-    exact: aliasesToRegex(aliases),
-    inflected: cityInflectionRegex(aliases),
+    get exact() {
+      exact ||= aliasesToRegex(aliases);
+      return exact;
+    },
+    get inflected() {
+      if (inflected === undefined) inflected = cityInflectionRegex(aliases);
+      return inflected;
+    },
   });
 });
 
@@ -27,7 +35,14 @@ const COUNTRY_MATCHERS = COUNTRIES.map((item) => {
   // are still handled by canonicalCountryCode().
   const aliases = [item.canonical, ...aliasesOf(item)]
     .filter((alias) => normalizeForMatch(alias).replace(/\s/g, '').length > 2);
-  return Object.freeze({ item, re: aliasesToRegex(aliases) });
+  let re = null;
+  return Object.freeze({
+    item,
+    get re() {
+      re ||= aliasesToRegex(aliases);
+      return re;
+    },
+  });
 });
 
 const US_STATE_RE = /\b[A-Z][A-Za-z.'-]+(?:\s+[A-Z][A-Za-z.'-]+){0,3}\s*,?\s+(?:AL|AK|AZ|AR|CA|CO|CT|DE|FL|GA|HI|ID|IL|IN|IA|KS|KY|LA|ME|MD|MA|MI|MN|MS|MO|MT|NE|NV|NH|NJ|NM|NY|NC|ND|OH|OK|OR|PA|RI|SC|SD|TN|TX|UT|VT|VA|WA|WV|WI|WY|DC)\b/;

package/src/hiring-professions.js

@@ -331,6 +331,8 @@ function profession(canonical, group, strongAliases, weakAliases = []) {
   const meta = GROUP_META.get(group) || { family: 'other', aliases: [] };
   const strong = [...new Set([...strongAliases, ...(ROMANIAN_PROFESSION_ALIASES[canonical] || [])].filter(Boolean))];
   const weak = [...new Set(weakAliases.filter(Boolean))];
+  let strongRe = null;
+  let weakRe;
   return Object.freeze({
     id: canonical,
     taxonomyId: `profession.${group}.${canonical}`,
@@ -340,8 +342,14 @@ function profession(canonical, group, strongAliases, weakAliases = []) {
     aliases: Object.freeze(strong),
     strongAliases: Object.freeze(strong),
     weakAliases: Object.freeze(weak),
-    strongRe: aliasesToRegex(strong),
-    weakRe: weak.length ? aliasesToRegex(weak) : null,
+    get strongRe() {
+      strongRe ||= aliasesToRegex(strong);
+      return strongRe;
+    },
+    get weakRe() {
+      if (weakRe === undefined) weakRe = weak.length ? aliasesToRegex(weak) : null;
+      return weakRe;
+    },
   });
 }
 
@@ -369,7 +377,15 @@ export const SENIORITY_LEVELS = Object.freeze([
   { canonical: 'chief', aliases: ['chief officer','chief executive','chief technology','chief product','chief financial','chief marketing','chief operating'], score: 1 },
 ].map((entry) => {
   const aliases = [...new Set([...entry.aliases, ...(ROMANIAN_SENIORITY_ALIASES[entry.canonical] || [])])];
-  return Object.freeze({ ...entry, aliases: Object.freeze(aliases), re: aliasesToRegex(aliases) });
+  let re = null;
+  return Object.freeze({
+    ...entry,
+    aliases: Object.freeze(aliases),
+    get re() {
+      re ||= aliasesToRegex(aliases);
+      return re;
+    },
+  });
 }));
 
 export function matchProfessions(value, { limit = 5, allowWeak = true } = {}) {

package/src/hiring-source-aliases.js

@@ -1,13 +1,20 @@
 import { aliasesToRegex } from './normalization.js';
 import { matchProfessions } from './hiring-professions.js';
 
-const sourceRole = (canonical, label, aliases, group = 'other') => Object.freeze({
-  canonical,
-  label,
-  group,
-  aliases: Object.freeze([...new Set(aliases)]),
-  re: aliasesToRegex([...new Set(aliases)]),
-});
+function sourceRole(canonical, label, aliases, group = 'other') {
+  const all = Object.freeze([...new Set(aliases)]);
+  let re = null;
+  return Object.freeze({
+    canonical,
+    label,
+    group,
+    aliases: all,
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
+  });
+}
 
 /**
  * Source spellings seen in real CV feeds which are too source-specific or

package/src/housing-listing-enrichment.js

@@ -32,7 +32,7 @@ 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 STUDENT_RE = /(?:studentlar\s+uchun|talabalar\s+uchun|(?<![\p{L}\p{N}_])(?:student(?:ka)?|talaba)(?:lar)?(?:ga)?\s+(?:qiz|yigit|o['’ʻʼ‘`]?g['’ʻʼ‘`]?il\s+bola|bola)(?:lar)?(?:ga)?(?![\p{L}\p{N}_])|(?<![\p{L}\p{N}_])(?:studentlar|talabalar)ga(?![\p{L}\p{N}_])|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 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;
@@ -45,6 +45,10 @@ const FAMILY_EXCLUSION_RE = /(?:oila|oyla|семь\p{L}*|family|famil(?:ie|ia)|c
 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;
+// Shared-flat posts price each tenant with the ablative "-dan" ("800 000
+// so'mdan" = "800 000 so'm from each") once they ask for several people.
+const PER_PERSON_SUFFIX_PRICE_RE = /(\d{1,3}(?:[  ]\d{3})+|\d+)\s*(ming|минг|million|mln|млн)?\s*(?:so['’ʻʼ‘`]?m|сўм|сум)(?:dan|дан)(?![\p{L}\p{N}_])/iu;
+const SEVERAL_OCCUPANTS_RE = /(?<![\p{L}\p{N}_])([2-9])\s*ta\s+(?:qiz|yigit|bola|kishi|odam|talaba|student)\p{L}*\s+(?:kerak|kere|olinadi)/iu;
 const WALK_MINUTES_RE = /(?:yayov|piyoda|пешком)\s*(\d{1,2})\s*(?:daqiqa|min(?:ute)?s?|минут)/iu;
 const TRANSIT_ROUTES_RE = /(?:aftobuslar|avtobuslar|автобуслар|автобусы)[^\r\n\d]{0,24}((?:\d{1,4}[\s,;/]*){1,10})/iu;
 const NEARBY_RELATION_TAIL_RE = /(?<!\p{L})(?:рядом\s+(?:с|со)|недалеко\s+от|возле|около|напротив|навпроти|ориентир\s*[:—–-]?|ор[-–—]?р\.?\s*[:—–-]?|near(?:by)?|close\s+to|next\s+to|opposite|behind|in\s+front\s+of|yaqin(?:ida)?|lângă|aproape\s+de|în\s+apropiere\s+de|vizavi\s+de|în\s+spatele|în\s+fața)(?!\p{L})[^.!?\r\n;]*/giu;
@@ -194,7 +198,9 @@ export function parseHousingAudience(value) {
   const students = STUDENT_RE.test(text);
   if (family && students) return deepFreeze({ primary: 'family', alternatives: ['family', 'students'] });
   if (family && women) return deepFreeze({ primary: 'family', alternatives: ['family', 'women'] });
-  if (students) return deepFreeze({ primary: 'students', alternatives: ['students'] });
+  // "Student qizlarga" narrows students to girls: the gender is the binding
+  // restriction, and studentTarget still reports the student part.
+  if (students && !women && !men) return deepFreeze({ primary: 'students', alternatives: ['students'] });
   if (family) return deepFreeze({ primary: 'family', alternatives: ['family'] });
   if (women) return deepFreeze({ primary: 'women', alternatives: ['women'] });
   if (men) return deepFreeze({ primary: 'men', alternatives: ['men'] });
@@ -222,14 +228,15 @@ export function parseHousingCommission(value) {
 
 export function parseHousingPerPersonPrice(value, { country = '' } = {}) {
   const text = normalizeUnicode(value ?? '');
-  const match = text.match(PER_PERSON_PRICE_RE);
+  const match = text.match(PER_PERSON_PRICE_RE)
+    || (SEVERAL_OCCUPANTS_RE.test(text) ? text.match(PER_PERSON_SUFFIX_PRICE_RE) : null);
   if (!match) return null;
   let amount = Number(String(match[1]).replace(/\s+/g, '').replace(',', '.'));
   if (!Number.isFinite(amount) || amount <= 0) return null;
   const scale = String(match[2] || '').toLowerCase();
   if (scale === 'ming' || scale === 'минг') amount *= 1000;
   else if (scale === 'million' || scale === 'mln' || scale === 'млн') amount *= 1_000_000;
-  const currency = String(country || '').toUpperCase() === 'UZ' || /(?:ming|минг)/iu.test(match[0]) ? 'UZS' : null;
+  const currency = String(country || '').toUpperCase() === 'UZ' || /(?:ming|минг|so['’ʻʼ‘`]?m|сўм|сум)/iu.test(match[0]) ? 'UZS' : null;
   const approximate = /около|примерно|~|≈/iu.test(text);
   return deepFreeze({ amount, currency, approximate, scope: 'person' });
 }

package/src/housing-money.js

@@ -602,7 +602,10 @@ export function parseHousingPrice(value, context = '') {
   const hasExplicitUzbekThousandScale = country === 'UZ' && /\b(?:ming|минг)\b/iu.test(priceText);
   if (!explicit && !hasExplicitUzbekThousandScale && fallbackCurrency === 'UZS' && price != null) {
     const dailyUzbek = /(?:kunlik|sutkaga|kecha[- ]?kunduz|посуточн|суточн)/i.test(priceText);
-    currency = price >= 1_000_000 || (dailyUzbek && price >= 10_000) ? 'UZS' : 'USD';
+    // A monthly rent of 100,000+ is only plausible in so'm; the USD reading
+    // turned "800 000" per person into an 800,000 USD rent.
+    const monthlyRent = dealType === 'longRent' && price >= 100_000;
+    currency = price >= 1_000_000 || monthlyRent || (dailyUzbek && price >= 10_000) ? 'UZS' : 'USD';
   }
 
   return Object.freeze({ amount: price, currency, approximate: price != null && APPROXIMATE_RE.test(priceText) });

package/src/housing-structured.js

@@ -87,7 +87,9 @@ 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*(?:ta\s*)?(?:(?:-\s*)?комнат\p{L}*|(?:-\s*)?к(?:\.|\b)|(?:-\s*)?xona(?:li|si|lari)?|(?:-\s*)?хона(?:лик|ли|си|лари)?|бөлмелі|rooms?)(?=$|[^\p{L}\p{N}])/iu);
+  // "5-х комн.кв" / "3-ех комнатная": a Russian case ending may sit between
+  // the number and the room word, which may itself be abbreviated.
+  const numeric = text.match(/(?:^|[^\p{L}\p{N}])(\d{1,2})\s*(?:-\s*(?:х|ти|и|ух|[её]х)\s*)?(?:ta\s*)?(?:(?:-\s*)?комн(?:ат)?\p{L}*|(?:-\s*)?к(?:\.|\b)|(?:-\s*)?xona(?:li|si|lari)?|(?:-\s*)?хона(?:лик|ли|си|лари)?|бөлмелі|rooms?)(?=$|[^\p{L}\p{N}])/iu);
   if (numeric) {
     const rooms = toNumber(numeric[1]);
     if (rooms != null && rooms >= 1 && rooms <= 20) return rooms;
@@ -127,7 +129,8 @@ export function parseHousingFloor(value) {
     }
   }
 
-  const beforeMarker = text.match(/(?:^|[^\d])(\d{1,3})\s*-?\s*(?:(?:chi|чи)\s*)?(?:этаж(?:да)?|поверх|floor|etaj|qavat(?:i(?:da(?:gi)?)?|da)?|қабат(?:ы(?:нда(?:ғы)?)?|та)?|кават(?:и(?:да(?:ги)?)?|да)?|қават(?:и(?:да(?:ги)?)?|да)?)(?=$|[^\p{L}\p{N}_])/iu);
+  // "3-й этаж" / "5-м этаже" carry a Russian ordinal ending like Uzbek "-chi".
+  const beforeMarker = text.match(/(?:^|[^\d])(\d{1,3})\s*-?\s*(?:(?:chi|чи|й|ий|ый|ой|м|ом|ем)\s*)?(?:этаж(?:да)?|поверх|floor|etaj|qavat(?:i(?:da(?:gi)?)?|da)?|қабат(?:ы(?:нда(?:ғы)?)?|та)?|кават(?:и(?:да(?:ги)?)?|да)?|қават(?:и(?:да(?:ги)?)?|да)?)(?=$|[^\p{L}\p{N}_])/iu);
   let floor = toNumber(beforeMarker?.[1]);
   let totalFloors = null;
 
@@ -140,7 +143,7 @@ export function parseHousingFloor(value) {
   if (floor == null && FIRST_FLOOR_WORD_RE.test(text)) floor = 1;
 
   if (totalFloors == null) {
-    const total = text.match(/(?:дом\s*)?(\d{1,3})\s*(?:[- ]?этаж(?:н\p{L}*|лик)|поверхов\p{L}*|storey|story|floors?\s+total|qavatli|қабатты|каватли|қаватли)/iu);
+    const total = text.match(/(?:дом\s*)?(\d{1,3})\s*(?:-\s*(?:ти|х|и|ми)\s*)?(?:[- ]?этаж(?:н\p{L}*|лик)|поверхов\p{L}*|storey|story|floors?\s+total|qavatli|қабатты|каватли|қаватли)/iu);
     totalFloors = toNumber(total?.[1]);
   }
 
@@ -167,7 +170,7 @@ const AREA_UNIT_RE = String.raw`(?:м²|м2|m²|m2|sqm|sq\.?\s*m|mp|кв\.?\s*м
 const AREA_UNIT_AFTER_LABEL_RE = String.raw`(?:${AREA_UNIT_RE}|[mм])`;
 
 function areaAfterLabel(text, labelRe) {
-  const re = new RegExp(`${labelRe.source}\\s*[:=-]?\\s*(\\d{1,4}(?:[.,]\\d{1,2})?)\\s*${AREA_UNIT_AFTER_LABEL_RE}(?=$|[^\\p{L}\\p{N}_])`, 'iu');
+  const re = new RegExp(`${labelRe.source}\\s*[:=-]?\\s*(\\d{1,4}(?:[.,]\\d{1,2})?)\\s*-?\\s*${AREA_UNIT_AFTER_LABEL_RE}(?=$|[^\\p{L}\\p{N}_])`, 'iu');
   return toNumber(text.match(re)?.[1]);
 }
 

package/src/kg-location-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function entry(name, aliases = [], meta = {}) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     ...meta,
     canonical: name,
@@ -10,7 +11,10 @@ function entry(name, aliases = [], meta = {}) {
     entityType: meta.entityType || meta.type,
     country: 'KG',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/kz-address-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function entry(name, aliases, entityType) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
@@ -9,7 +10,10 @@ function entry(name, aliases, entityType) {
     entityType,
     country: 'KZ',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/kz-almaty-cleaned-residential-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function residential(name, aliases = []) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
@@ -9,7 +10,10 @@ function residential(name, aliases = []) {
     entityType: 'residential_complex',
     country: 'KZ',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/kz-almaty-street-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function street(name, aliases = []) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
@@ -9,7 +10,10 @@ function street(name, aliases = []) {
     entityType: 'street',
     country: 'KZ',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/kz-astana-cleaned-residential-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function residential(name, aliases = []) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
@@ -9,7 +10,10 @@ function residential(name, aliases = []) {
     entityType: 'residential_complex',
     country: 'KZ',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/kz-location-extensions.js

@@ -2,7 +2,18 @@ import { aliasesToRegex } from './normalization.js';
 
 function entry(name, aliases = [], meta = {}) {
   const all = [...new Set([name, ...aliases].filter(Boolean))];
-  return Object.freeze({ ...meta, canonical: name, name, type: meta.type || meta.entityType, aliases: Object.freeze(all), re: aliasesToRegex(all) });
+  let re = null;
+  return Object.freeze({
+    ...meta,
+    canonical: name,
+    name,
+    type: meta.type || meta.entityType,
+    aliases: Object.freeze(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
+  });
 }
 
 function entries(rows = [], defaults = {}) {

package/src/kz-primary-address-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function entry(name, aliases, entityType) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
@@ -9,7 +10,10 @@ function entry(name, aliases, entityType) {
     entityType,
     country: 'KZ',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/kz-regional-address-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function street(name, aliases = []) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
@@ -9,7 +10,10 @@ function street(name, aliases = []) {
     entityType: 'street',
     country: 'KZ',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/kz-regional-residential-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function residential(name, aliases = []) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
@@ -9,7 +10,10 @@ function residential(name, aliases = []) {
     entityType: 'residential_complex',
     country: 'KZ',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/kz-residential-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function residential(name, aliases = []) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
@@ -9,7 +10,10 @@ function residential(name, aliases = []) {
     entityType: 'residential_complex',
     country: 'KZ',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/kz-shymkent-cleaned-address-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function entry(name, aliases = [], entityType) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
@@ -9,7 +10,10 @@ function entry(name, aliases = [], entityType) {
     entityType,
     country: 'KZ',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/location-data.js

@@ -5,14 +5,35 @@ import { UA_REGIONS as UA_REGION_LEXICON } from './geography.js';
 function entries(rows) {
   return Object.freeze(rows.map(([name, ...aliases]) => {
     const all = [...new Set([name, ...aliases].filter(Boolean))];
-    return Object.freeze({ canonical: name, name, aliases: Object.freeze(all), re: aliasesToRegex(all) });
+    let re = null;
+    return Object.freeze({
+      canonical: name,
+      name,
+      aliases: Object.freeze(all),
+      get re() {
+        re ||= aliasesToRegex(all);
+        return re;
+      },
+    });
   }));
 }
 
 function lexiconEntries(items) {
   return Object.freeze((items || []).map((item) => {
     const all = [...new Set([item.canonical, ...aliasesOf(item)].filter(Boolean))];
-    return Object.freeze({ canonical: item.canonical, name: item.canonical, type: item.type, country: item.country, city: item.city, aliases: Object.freeze(all), re: aliasesToRegex(all) });
+    let re = null;
+    return Object.freeze({
+      canonical: item.canonical,
+      name: item.canonical,
+      type: item.type,
+      country: item.country,
+      city: item.city,
+      aliases: Object.freeze(all),
+      get re() {
+        re ||= aliasesToRegex(all);
+        return re;
+      },
+    });
   }));
 }
 

package/src/locations.js

@@ -115,6 +115,7 @@ const UZ_BASE_LOCATION_DICTIONARIES = Object.freeze({
 
 function semanticEntry(entry, name, aliases, entityType) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     ...entry,
     canonical: name,
@@ -122,7 +123,10 @@ function semanticEntry(entry, name, aliases, entityType) {
     type: entityType,
     entityType,
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 
@@ -179,13 +183,17 @@ function normalizeUzSemanticLocations(country) {
       'Корасув дахаси',
       'Карасу даха',
     ])]);
+    let qorasuvRe = null;
     const qorasuvArea = Object.freeze({
       ...qorasuv,
       type: 'local_area',
       entityType: 'local_area',
       parent: 'Mirzo Ulugbek',
       aliases: qorasuvAliases,
-      re: aliasesToRegex(qorasuvAliases),
+      get re() {
+        qorasuvRe ||= aliasesToRegex(qorasuvAliases);
+        return qorasuvRe;
+      },
     });
 
     normalized = Object.freeze({

package/src/money-lexicon.js

@@ -5,7 +5,7 @@ const group = (canonical, aliases, extra = {}) => lexiconEntity(canonical, alias
 export const CURRENCY_TERMS = Object.freeze([
   group('USD', { ru: ['$', 'usd', 'доллар', 'доллара', 'долларов', 'бакс', 'у.е.', 'у е'], en: ['$', 'usd', 'dollar', 'dollars', 'us dollar'], uk: ['$', 'usd', 'долар', 'долари', 'доларів'], ro: ['$', 'usd', 'dolar', 'dolari'], uzLatn: ['$', 'usd', 'dollar'], uzCyrl: ['$', 'доллар'], kk: ['$', 'usd', 'доллар'] }),
   group('EUR', { ru: ['€', 'eur', 'евро'], en: ['€', 'eur', 'euro', 'euros'], uk: ['€', 'eur', 'євро'], ro: ['€', 'eur', 'euro'], uzLatn: ['€', 'eur', 'yevro'], uzCyrl: ['€', 'евро'], kk: ['€', 'eur', 'еуро'] }),
-  group('UZS', { ru: ['сум', 'сумов', 'узс'], en: ['uzs', 'uzbek som'], uk: ['uzs', 'сум'], ro: ['uzs'], uzLatn: ["so'm", 'so‘m', 'soʻm', 'sum', 'uzs'], uzCyrl: ['сўм', 'сум'], kk: ['uzs'] }),
+  group('UZS', { ru: ['сум', 'сумов', 'узс'], en: ['uzs', 'uzbek som'], uk: ['uzs', 'сум'], ro: ['uzs'], uzLatn: ["so'm", 'so‘m', 'soʻm', 'sum', 'uzs', "so'mdan", 'so‘mdan', 'soʻmdan', 'somdan', 'sumdan', "so'mgacha", 'so‘mgacha', 'soʻmgacha'], uzCyrl: ['сўм', 'сум', 'сўмдан', 'сумдан', 'сўмгача'], kk: ['uzs'] }),
   group('KZT', { ru: ['₸', 'тенге', 'тг', 'kzt'], en: ['₸', 'kzt', 'tenge'], uk: ['₸', 'тенге', 'kzt'], ro: ['₸', 'kzt', 'tenge'], uzLatn: ['₸', 'kzt', 'tenge'], uzCyrl: ['₸', 'kzt', 'тенге'], kk: ['₸', 'теңге', 'тенге', 'тг', 'kzt'] }),
   group('UAH', { ru: ['грн', 'гривна', 'гривны', 'гривен', '₴', 'uah'], en: ['uah', 'hryvnia', 'hryvnias', '₴'], uk: ['грн', 'гривня', 'гривні', 'гривень', '₴', 'uah'], ro: ['uah', 'hrivne', 'grivne'], uzLatn: ['uah'], uzCyrl: ['uah'], kk: ['uah'] }),
   group('RUB', { ru: ['₽', 'руб', 'руб.', 'рубль', 'рубля', 'рублей', 'rub', 'rur'], en: ['₽', 'rub', 'rur', 'ruble', 'rubles'], uk: ['₽', 'руб', 'рублів', 'rub'], ro: ['₽', 'rub', 'ruble'], uzLatn: ['₽', 'rub', 'rubl'], uzCyrl: ['₽', 'рубль'], kk: ['₽', 'rub', 'рубль'] }),

package/src/odesa-metropolitan.js

@@ -2,12 +2,16 @@ import { aliasesToRegex } from './normalization.js';
 
 function entity(name, type, aliases = [], options = {}) {
   const all = [...new Set([name, ...aliases].filter(Boolean))];
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
     type,
     aliases: Object.freeze(all),
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
     parent: options.parent || null,
     cluster: options.cluster || null,
     contextRequired: Boolean(options.contextRequired),

package/src/tashkent-gastronomic-streets.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function gastronomicStreet(canonical, aliases = [], district = null) {
   const all = Object.freeze([...new Set([canonical, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical,
     name: canonical,
@@ -11,7 +12,10 @@ function gastronomicStreet(canonical, aliases = [], district = null) {
     city: 'Tashkent',
     district,
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/tashkent-housing-geography.js

@@ -3,6 +3,7 @@ import { TASHKENT_DISTRICTS, TASHKENT_METRO } from './geo.js';
 
 function locationEntry(name, category, aliases = [], options = {}) {
   const all = [...new Set([name, ...aliases].filter(Boolean))];
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
@@ -10,7 +11,10 @@ function locationEntry(name, category, aliases = [], options = {}) {
     country: 'UZ',
     city: 'Tashkent',
     aliases: Object.freeze(all),
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
     contextRequired: Boolean(options.contextRequired),
     contextRe: options.context ? new RegExp(options.context, 'iu') : null,
   });
@@ -18,6 +22,7 @@ function locationEntry(name, category, aliases = [], options = {}) {
 
 const TASHKENT_HOUSING_DISTRICTS = Object.freeze(TASHKENT_DISTRICTS.map((district) => {
   const aliases = [...new Set([district.canonical, ...aliasesOf(district)].filter(Boolean))];
+  let re = null;
   return Object.freeze({
     canonical: district.canonical,
     name: district.canonical,
@@ -25,7 +30,10 @@ const TASHKENT_HOUSING_DISTRICTS = Object.freeze(TASHKENT_DISTRICTS.map((distric
     country: 'UZ',
     city: 'Tashkent',
     aliases: Object.freeze(aliases),
-    re: aliasesToRegex(aliases),
+    get re() {
+      re ||= aliasesToRegex(aliases);
+      return re;
+    },
   });
 }));
 
@@ -167,7 +175,7 @@ export const TASHKENT_NUMBERED_AREA_ALIASES = Object.freeze({
   Kuylyuk: Object.freeze(['куйлюк', 'куйлик', 'kuylyuk', 'kuyliq', 'qoyliq', 'qo yliq']),
   Sergeli: Object.freeze(['сергели', 'sergeli', 'sergile', 'sergele']),
   Yunusabad: Object.freeze(['юнусабад', 'yunusabad', 'yunusobod']),
-  Yangihayot: Object.freeze(['янгихаёт', 'янгихаят', 'yangihayot']),
+  Yangihayot: Object.freeze(['янгихаёт', 'янгихает', 'янгихаят', 'yangihayot']),
 });
 
 const REVERSE_NUMBERED_ALIASES = Object.freeze({
@@ -241,15 +249,24 @@ export function hasExplicitTashkentDistrict(value, canonical) {
 export function matchTashkentHousingDistrict(value) {
   const text = String(value ?? '');
   if (!text) return null;
+  let explicit = null;
+  let implicit = null;
   for (const district of TASHKENT_HOUSING_DISTRICTS) {
     const match = text.match(district.re);
     if (!match) continue;
-    if (hasExplicitDistrictContext(text, match)) return district;
+    // The earliest explicitly marked district wins. In "Янгихаетский район
+    // Сергели-1" the "район" also precedes Сергели, but it belongs to the
+    // adjective before it; Sergeli-1 is only the quarter name.
+    if (hasExplicitDistrictContext(text, match)) {
+      if (!explicit || match.index < explicit.index) explicit = { district, index: match.index };
+      continue;
+    }
+    if (implicit) continue;
     if (hasExplicitMetroContext(text, match) || hasExplicitAreaContext(text, match) || hasExplicitMahallaContext(text, match)) continue;
     if (matchTashkentNumberedArea(text, district.name)) continue;
-    return district;
+    implicit = district;
   }
-  return null;
+  return explicit?.district || implicit;
 }
 
 /**

package/src/tashkent-pois.js

@@ -2,13 +2,17 @@ import { aliasesToRegex } from './normalization.js';
 
 function poi(name, category, aliases = [], options = {}) {
   const all = [...new Set([name, ...aliases].filter(Boolean))];
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
     category,
     categories: Object.freeze([...new Set([category, ...(options.categories || [])])]),
     aliases: Object.freeze(all),
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
     contextRequired: Boolean(options.contextRequired),
     contextRe: options.context ? new RegExp(options.context, 'iu') : null,
     ...(options.parent ? { parent: options.parent } : {}),

package/src/tashkent-residential-complexes.js

@@ -12,12 +12,16 @@ const AMBIGUOUS_TASHKENT_COMPLEXES = new Set([
 
 function complex(name, aliases = []) {
   const all = [...new Set([name, ...aliases].filter(Boolean))];
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
     aliases: Object.freeze(all),
     ambiguous: AMBIGUOUS_TASHKENT_COMPLEXES.has(name),
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/ua-chernihiv-residential-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function residential(name, aliases = []) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
@@ -9,7 +10,10 @@ function residential(name, aliases = []) {
     entityType: 'residential_complex',
     country: 'UA',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/ua-residential-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function residential(name, aliases = []) {
   const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical: name,
     name,
@@ -9,7 +10,10 @@ function residential(name, aliases = []) {
     entityType: 'residential_complex',
     country: 'UA',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/uz-location-extensions.js

@@ -2,7 +2,18 @@ import { aliasesToRegex } from './normalization.js';
 
 function entry(name, aliases = [], meta = {}) {
   const all = [...new Set([name, ...aliases].filter(Boolean))];
-  return Object.freeze({ ...meta, canonical: name, name, type: meta.type || meta.entityType, aliases: Object.freeze(all), re: aliasesToRegex(all) });
+  let re = null;
+  return Object.freeze({
+    ...meta,
+    canonical: name,
+    name,
+    type: meta.type || meta.entityType,
+    aliases: Object.freeze(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
+  });
 }
 
 function entries(rows = [], defaults = {}) {

package/src/uz-samarkand-context-extensions.js

@@ -2,6 +2,7 @@ import { aliasesToRegex } from './normalization.js';
 
 function entry(canonical, type, aliases = [], labels = null) {
   const all = Object.freeze([...new Set([canonical, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical,
     name: canonical,
@@ -11,7 +12,10 @@ function entry(canonical, type, aliases = [], labels = null) {
     city: 'Samarkand',
     ...(labels ? { labels: Object.freeze(labels) } : {}),
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }
 

package/src/uz-tashkent-context-extensions.js

@@ -11,6 +11,7 @@ function legacyAreaEntry(district, canonical) {
     ...(source.aliases || []),
   ].filter(Boolean))]);
 
+  let re = null;
   return Object.freeze({
     canonical: source.name,
     name: source.name,
@@ -21,12 +22,16 @@ function legacyAreaEntry(district, canonical) {
     district,
     parent: district,
     aliases,
-    re: aliasesToRegex(aliases),
+    get re() {
+      re ||= aliasesToRegex(aliases);
+      return re;
+    },
   });
 }
 
 function residentialEntry(canonical, aliases = []) {
   const all = Object.freeze([...new Set([canonical, ...aliases].filter(Boolean))]);
+  let re = null;
   return Object.freeze({
     canonical,
     name: canonical,
@@ -35,7 +40,10 @@ function residentialEntry(canonical, aliases = []) {
     country: 'UZ',
     city: 'Tashkent',
     aliases: all,
-    re: aliasesToRegex(all),
+    get re() {
+      re ||= aliasesToRegex(all);
+      return re;
+    },
   });
 }