@whiteslove/parsing-lexicon 0.6.4 → 0.6.6

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.6.4",
+  "version": "0.6.6",
   "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
   "repository": {
     "type": "git",

package/src/central-asia-locations.js

@@ -139,6 +139,86 @@ function semanticBounds(value, match) {
   return { start, end };
 }
 
+const NUMBERED_LOCAL_AREA_NAME_RE = /(?:^|[-–—/\s])\d{1,2}[aа]?(?=$|[^\p{L}\p{N}_])/iu;
+const NUMBERED_CHILD_SUFFIX_RE = /^\s*[-–—/]\s*\d{1,2}[aа]?(?=$|[^\p{L}\p{N}_])/iu;
+
+function isUmbrellaAreaPrefixOfNumberedBlock(value, candidate) {
+  if (candidate.type !== 'local_area') return false;
+  if (NUMBERED_LOCAL_AREA_NAME_RE.test(String(candidate.name || ''))) return false;
+  return NUMBERED_CHILD_SUFFIX_RE.test(value.slice(candidate.end));
+}
+
+const RESIDENTIAL_MARKER_TEXT = String.raw`(?:ж\.?\s*к\.?|жил(?:ой|ого)?\s+комплекс\p{L}*|residential\s+complex|turar\s+joy\s+majmuasi|tjm)`;
+const residentialMarkedAliasCache = new WeakMap();
+
+function stripResidentialMarker(value) {
+  return String(value || '')
+    .replace(new RegExp(`^\\s*${RESIDENTIAL_MARKER_TEXT}\\s*[:\\-–—]?\\s*`, 'iu'), '')
+    .replace(new RegExp(`\\s*${RESIDENTIAL_MARKER_TEXT}\\s*$`, 'iu'), '')
+    .trim();
+}
+
+function residentialMarkedAliasRegex(item) {
+  if (residentialMarkedAliasCache.has(item)) return residentialMarkedAliasCache.get(item);
+  const aliases = [...new Set([item?.name, ...(item?.aliases || [])]
+    .map(stripResidentialMarker)
+    .filter(Boolean))];
+  const re = aliases.length ? aliasesToRegex(aliases) : null;
+  residentialMarkedAliasCache.set(item, re);
+  return re;
+}
+
+function markedResidentialMatch(value, item) {
+  const aliasRe = residentialMarkedAliasRegex(item);
+  if (!aliasRe) return null;
+  const markerRe = new RegExp(`(?:^|[^\\p{L}\\p{N}_])${RESIDENTIAL_MARKER_TEXT}\\s*[:\\-–—]?\\s*`, 'giu');
+  for (const marker of value.matchAll(markerRe)) {
+    const tailStart = (marker.index ?? 0) + marker[0].length;
+    const tail = value.slice(tailStart, Math.min(value.length, tailStart + 96));
+    const aliasMatch = tail.match(aliasRe);
+    if (!aliasMatch) continue;
+    const offset = aliasMatch.index ?? 0;
+    if (offset > 0 && /\S/u.test(tail.slice(0, offset))) continue;
+    return { 0: aliasMatch[0], index: tailStart + offset };
+  }
+  return null;
+}
+
+const LOCATION_CLAUSE_BOUNDARY_RE = /[\n\r.!?;|]/u;
+const PRIMARY_LOCATION_PREFIX_RE = /(?:(?:^|[^\p{L}\p{N}_])(?:адрес|локаци(?:я|и)|расположени(?:е|я)|находится)\s*[:\-–—]?\s*|(?:^|[^\p{L}\p{N}_])(?:в|на)\s+(?:(?:ж\.?\s*к\.?|жил(?:ом|ой)\s+комплекс\p{L}*|микрорайон\p{L}*|мкр\.?|массив\p{L}*|махалл\p{L}*|mahalla|massiv|district|район\p{L}*|metro|метро)\s*)?)$/iu;
+const NEARBY_LOCATION_PREFIX_RE = /(?:^|[^\p{L}\p{N}_])(?:до|около|возле|рядом(?:\s+(?:с|со))?|недалеко\s+от|неподал[её]ку\s+от|вблизи|напротив|ориентир(?:ом)?|near|nearby|close\s+to|next\s+to|opposite|landmark|yaqin(?:ida)?|yonida|mo['’ʻ]?ljal|яқин(?:ида)?|ёнида|мўлжал|жақын|жанында|маңында)(?=$|[^\p{L}\p{N}_])[\s\S]{0,96}$/iu;
+const DISTANCE_LOCATION_PREFIX_RE = /(?:\d+(?:[.,]\d+)?\s*(?:мин(?:ут\p{L}*)?|min(?:ute)?s?|daqiqa|м|m|метр\p{L}*|км|km)\s*(?:от|до|from|to)\s+)[\s\S]{0,80}$/iu;
+const NEARBY_LOCATION_SUFFIX_RE = /^\s*(?:[-—–:]\s*)?(?:(?:\d+(?:[.,]\d+)?\s*(?:мин(?:ут\p{L}*)?|min(?:ute)?s?|daqiqa|м|m|метр\p{L}*|км|km))|(?:рядом|поблизости|nearby|yaqin(?:ida)?|яқин(?:ида)?|жақын))(?=$|[^\p{L}\p{N}_])/iu;
+
+function clauseStart(value, start) {
+  for (let index = start - 1; index >= 0; index -= 1) {
+    if (LOCATION_CLAUSE_BOUNDARY_RE.test(value[index])) return index + 1;
+  }
+  return 0;
+}
+
+function clauseEnd(value, end) {
+  for (let index = end; index < value.length; index += 1) {
+    if (LOCATION_CLAUSE_BOUNDARY_RE.test(value[index])) return index;
+  }
+  return value.length;
+}
+
+function locationMentionRole(value, candidate) {
+  const start = clauseStart(value, candidate.start);
+  const end = clauseEnd(value, candidate.end);
+  const before = value.slice(start, candidate.start);
+  const after = value.slice(candidate.end, end);
+
+  // A direct subject marker later in the same clause wins over an unrelated
+  // earlier temporal "до": "до пятницы квартира в ЖК X" is still ЖК X itself.
+  if (PRIMARY_LOCATION_PREFIX_RE.test(before)) return 'primary';
+  if (NEARBY_LOCATION_PREFIX_RE.test(before) || DISTANCE_LOCATION_PREFIX_RE.test(before) || NEARBY_LOCATION_SUFFIX_RE.test(after)) {
+    return 'nearby';
+  }
+  return 'mentioned';
+}
+
 function publicMatch(candidate) {
   const { start, end, matchedText, explicitContext, ...result } = candidate;
   return Object.freeze(result);
@@ -149,7 +229,8 @@ function findEntryMatches(text, cityName, data) {
   const raw = [];
   for (const key of LOCATION_LIST_KEYS) {
     for (const item of data?.[key] || []) {
-      const match = value.match(item?.re);
+      const match = value.match(item?.re)
+        || (key === 'residentialComplexes' ? markedResidentialMatch(value, item) : null);
       if (!match) continue;
       const { start, end } = semanticBounds(value, match);
       const candidate = {
@@ -167,8 +248,11 @@ function findEntryMatches(text, cityName, data) {
         end,
         matchedText: value.slice(start, end),
         explicitContext: false,
+        role: 'mentioned',
       };
+      if (isUmbrellaAreaPrefixOfNumberedBlock(value, candidate)) continue;
       candidate.explicitContext = hasExplicitContext(value, candidate);
+      candidate.role = locationMentionRole(value, candidate);
       raw.push(candidate);
     }
   }
@@ -274,4 +358,4 @@ export function matchCentralAsiaLocationEntities(text, countryCode, preferredCit
 export function matchCentralAsiaLocationEntity(text, countryCode, preferredCity = null, type = null) {
   const result = matchCentralAsiaLocationEntities(text, countryCode, preferredCity);
   return result.matches.find((item) => !type || item.type === type || item.key === type) || null;
-}
+}

package/src/housing-card-fields.js

@@ -39,6 +39,7 @@ const PLACE_KINDS = Object.freeze([
 const NEARBY_BLOCK_RE = /(?:^|\n)[^\n]{0,30}?(?:что\s+(?:есть|находится)\s+р[яa]дом|р[яa]дом(?:\s+(?:есть|находятся|расположены|с\s+домом))?|непода[лл][её]ку|поблизости|в\s+шаговой\s+доступности|окружени[ие]|ориентир(?:ы|ами)?|yaqinida|atrofida|yonida|nearby|landmarks?)\s*[:—-]?\s*([^\n]{4,500})/iu;
 const NEARBY_NOISE_RE = /^(?:и|а|в|на|у|до|от|все|вс[её]|есть|рядом|близко|недалеко|минут\p{L}*|пешком|транспорт|остановк\p{L}*|магазин|магазины|аптека|аптеки|садик|садики|всё\s+необходимое|развит\p{L}*)$/iu;
 const NEARBY_FILLER_RE = /(?:инфраструктур\p{L}*|все\s+необходимо|вс[её]\s+рядом|шаговой\s+доступност|развитый\s+район)/iu;
+const QUARTER_NEARBY_PREFIX_RE = /(?:^|[\s,;:—–-])(?:до|рядом\s+(?:с|со)|недалеко\s+от|возле|около|ориентир\s*[:—–-]?|near(?:by)?|close\s+to|lângă|aproape\s+de)\s*$/iu;
 
 function enumeratedNearby(text) {
   const blocks = [...text.matchAll(new RegExp(NEARBY_BLOCK_RE.source, `${NEARBY_BLOCK_RE.flags}g`))]
@@ -61,18 +62,33 @@ function enumeratedNearby(text) {
   return items;
 }
 
+function isNearbyQuarterReference(text, match) {
+  const start = match?.index ?? -1;
+  if (start < 0) return false;
+  const before = text.slice(Math.max(0, start - 64), start);
+  return QUARTER_NEARBY_PREFIX_RE.test(before);
+}
+
+function firstPrimaryQuarterMatch(text, pattern) {
+  const flags = pattern.flags.includes('g') ? pattern.flags : `${pattern.flags}g`;
+  for (const match of text.matchAll(new RegExp(pattern.source, flags))) {
+    if (!isNearbyQuarterReference(text, match)) return match;
+  }
+  return null;
+}
+
 export function parseHousingQuarterLabel(value) {
   const text = String(value || '');
   if (!text) return null;
   if (/(?:^|[^\p{L}\p{N}_])(?:глинк[аи]?|glinka)(?:$|[^\p{L}\p{N}_])/iu.test(text)) return 'Glinka';
-  const match = text.match(/(\d{1,3})\s*(?:-?\s*(?:chi|чи))?\s*[-\s]?\s*(?:квартал|кв-?л(?=$|[^\p{L}\p{N}_])|kvartal(?:i)?|квартил|kvartil|мкр(?=$|[^\p{L}\p{N}_])|микрорайон|массив|massiv|daha(?:si|dan)?|hudud|худуд)/iu)
-    || text.match(/(?:квартал|kvartal(?:i)?|квартил|kvartil|мкр|микрорайон|массив|massiv|daha(?:si|dan)?|hudud|худуд)\s*[-№#]?\s*(\d{1,3})/iu)
-    || text.match(/(?:чиланзар|chilonzor|chilanzar)\s*[-№#]?\s*(\d{1,2})(?!\d)/iu);
+  const match = firstPrimaryQuarterMatch(text, /(\d{1,3})\s*(?:-?\s*(?:chi|чи))?\s*[-\s]?\s*(?:квартал|кв-?л(?=$|[^\p{L}\p{N}_])|kvartal(?:i)?|квартил|kvartil|мкр(?=$|[^\p{L}\p{N}_])|микрорайон|массив|massiv|daha(?:si|dan)?|hudud|худуд)/iu)
+    || firstPrimaryQuarterMatch(text, /(?:квартал|kvartal(?:i)?|квартил|kvartil|мкр|микрорайон|массив|massiv|daha(?:si|dan)?|hudud|худуд)\s*[-№#]?\s*(\d{1,3})/iu)
+    || firstPrimaryQuarterMatch(text, /(?:чиланзар|chilonzor|chilanzar)\s*[-№#]?\s*(\d{1,2})(?!\d)/iu);
   if (match) {
     const label = `${Number(match[1])} kvartal`;
     return STREET_LIKE_PLACE_RE.test(label) ? null : label;
   }
-  const centralBlock = text.match(/(?:^|[^\p{L}\p{N}_])(?:ц|c)\s*[-–]?\s*(\d{1,2})(?:$|[^\p{L}\p{N}_])/iu);
+  const centralBlock = firstPrimaryQuarterMatch(text, /(?:^|[^\p{L}\p{N}_])(?:ц|c)\s*[-–]?\s*(\d{1,2})(?:$|[^\p{L}\p{N}_])/iu);
   return centralBlock ? `C-${Number(centralBlock[1])}` : null;
 }
 

package/src/housing-listing-enrichment.js

@@ -3,6 +3,7 @@ import { findAllCanonical, normalizeUnicode } from './normalization.js';
 import { GENERIC_LANDMARK_TERMS } from './landmarks.js';
 import { matchTashkentHousingDistrict, matchTashkentHousingMetro, matchTashkentHousingQuarter } from './tashkent-housing-geography.js';
 import { TASHKENT_LANDMARKS } from './tashkent-pois.js';
+import { TASHKENT_RESIDENTIAL_COMPLEXES, matchTashkentResidentialComplex } from './tashkent-residential-complexes.js';
 import { parseHousingRoomCount, parseHousingFloor, parseHousingAreas } from './housing-structured.js';
 import { parseHousingListingFields } from './housing-listing-fields.js';
 import { parseHousingResidentialComplex } from './housing-text.js';
@@ -38,6 +39,9 @@ const AIR_CONDITIONER_RE = /(?:кондицион|air\s*con|konditsioner|kandit(
 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;
 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|ориентир\s*[:—–-]?|near(?:by)?|close\s+to|lângă|aproape\s+de)(?!\p{L})[^.!?\r\n;]*/giu;
+const NEARBY_TRAVEL_TAIL_RE = /(?<!\p{L})(?:до|până\s+la)(?!\p{L})[^.!?\r\n;]{0,96}(?<!\p{L})\d{1,3}\s*(?:мин(?:ут(?:ы|а|ах)?|\.?)?|min(?:ute)?s?|дақиқ\p{L}*|daqiqa|км|km|метр(?:а|ов)?|m)(?!\p{L})[^.!?\r\n;]*/giu;
+const RESIDENTIAL_CONTEXT_RE = /(?:ж\.?\s*к\.?|жил(?:ой|ого)\s+комплекс|новострой(?:ка|ки)?|residential\s+complex|residence|turar\s+joy|uy[-\s]?joy|majmua|массив)/iu;
 
 function unique(values) {
   return [...new Set(values.filter(Boolean))];
@@ -47,6 +51,42 @@ function categoryOf(entry) {
   return entry?.category || GENERIC_CATEGORY[entry?.canonical] || 'landmark';
 }
 
+function nearbyReferenceRanges(text) {
+  const ranges = [];
+  for (const pattern of [NEARBY_RELATION_TAIL_RE, NEARBY_TRAVEL_TAIL_RE]) {
+    const regex = new RegExp(pattern.source, pattern.flags);
+    for (const match of text.matchAll(regex)) {
+      const start = match.index ?? 0;
+      ranges.push({ start, end: start + match[0].length });
+    }
+  }
+  return ranges;
+}
+
+function insideNearbyReference(match, ranges) {
+  return ranges.some((range) => match.start >= range.start && match.end <= range.end);
+}
+
+function withoutNearbyLocationReferences(text) {
+  return String(text || '')
+    .replace(NEARBY_RELATION_TAIL_RE, ' ')
+    .replace(NEARBY_TRAVEL_TAIL_RE, ' ');
+}
+
+function hasResidentialContext(text, match) {
+  if (!match.entry?.ambiguous) return true;
+  const context = text.slice(Math.max(0, match.start - 36), Math.min(text.length, match.end + 36));
+  return RESIDENTIAL_CONTEXT_RE.test(context);
+}
+
+function residentialComplexNearbyMatches(text) {
+  const ranges = nearbyReferenceRanges(text);
+  if (!ranges.length) return [];
+  return findAllCanonical(text, TASHKENT_RESIDENTIAL_COMPLEXES)
+    .filter((match) => insideNearbyReference(match, ranges) && hasResidentialContext(text, match))
+    .map((match) => ({ canonical: match.canonical, category: 'residential_complex', start: match.start }));
+}
+
 function extensionMatches(text) {
   return findAllCanonical(text, HOUSING_POI_EXTENSIONS).map((match) => ({
     canonical: match.canonical,
@@ -83,7 +123,7 @@ function genericMatches(text) {
 export function parseHousingNearby(value) {
   const text = normalizeUnicode(value ?? '');
   if (!text) return deepFreeze([]);
-  const specific = [...extensionMatches(text), ...catalogPoiMatches(text)].sort((a, b) => a.start - b.start);
+  const specific = [...extensionMatches(text), ...catalogPoiMatches(text), ...residentialComplexNearbyMatches(text)].sort((a, b) => a.start - b.start);
   const specificCategories = new Set(specific.map((item) => item.category));
   const generic = genericMatches(text)
     .filter((item) => !specificCategories.has(item.category))
@@ -201,7 +241,10 @@ export function parseHousingListingEnrichment(value, { country = '' } = {}) {
   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 primaryResidentialText = withoutNearbyLocationReferences(text);
+  const parsedRc = specificResidentialComplex(primaryResidentialText)
+    || matchTashkentResidentialComplex(primaryResidentialText)?.name
+    || parseHousingResidentialComplex(primaryResidentialText);
   const commission = parseHousingCommission(text);
   const commissionAmount = parseHousingCommissionAmount(text);
   const address = parseHousingAddress(text);

package/src/housing-poi-extensions.js

@@ -108,6 +108,12 @@ export const HOUSING_POI_EXTENSIONS = Object.freeze([
     uzLatn: ['Assalom Jomiy', 'Assalom Jomiy turar joy majmuasi'],
     uzCyrl: ['Ассалом Жомий'],
   }, { category: 'residential_complex', display: { ru: 'ЖК «Ассалом Жомий»', en: 'Assalom Jomiy' } }),
+  group('Assalom Sohil', {
+    ru: ['Ассалом Сохил', 'ЖК Ассалом Сохил', 'жилой комплекс Ассалом Сохил'],
+    en: ['Assalom Sohil', 'Assalom Sohil residential complex'],
+    uzLatn: ['Assalom Sohil', 'Assalom Sohil turar joy majmuasi'],
+    uzCyrl: ['Ассалом Сохил'],
+  }, { category: 'residential_complex', display: { ru: 'ЖК «Ассалом Сохил»', en: 'Assalom Sohil' } }),
   group('Jomiy Roundabout', {
     ru: ['круг Жомий', 'круг жомий', 'кольцо Жомий'],
     en: ['Jomiy Roundabout', 'Jomiy circle'],

package/src/housing-structured.js

@@ -84,6 +84,23 @@ export function parseHousingFloor(value) {
   const text = normalizeUnicode(value ?? '');
   if (!text) return deepFreeze({ floor: null, totalFloors: null });
 
+  // Common classifieds shorthand is rooms/floor/total floors, e.g. 2/10/16.
+  // It must be checked before the generic floor/total fraction parser or the
+  // first two numbers would incorrectly become floor=2, totalFloors=10.
+  // Requiring the triple to occupy its own line avoids treating dates as this
+  // housing-specific shorthand inside ordinary prose.
+  const compactTriple = text.match(/(?:^|[\r\n])\s*(\d{1,2})\s*[\/\\]\s*(\d{1,3})\s*[\/\\]\s*(\d{1,3})\s*(?=$|[\r\n])/u);
+  if (compactTriple) {
+    const rooms = toNumber(compactTriple[1]);
+    const floor = toNumber(compactTriple[2]);
+    const totalFloors = toNumber(compactTriple[3]);
+    if (rooms != null && rooms >= 1 && rooms <= 20
+      && floor != null && totalFloors != null
+      && floor >= 1 && floor <= totalFloors && totalFloors <= 200) {
+      return deepFreeze({ floor, totalFloors });
+    }
+  }
+
   const fraction = text.match(/(?:^|[^\d])(\d{1,3})\s*[\/\\]\s*(\d{1,3})(?=$|[^\d])/u);
   if (fraction) {
     const floor = toNumber(fraction[1]);
@@ -142,7 +159,7 @@ const AREA_LABELS = Object.freeze([
   ['total', /(?:общая\s+площадь|площадь|total\s+area|surface\s+area|suprafa(?:ță|ta)(?:\s+total(?:ă|a))?|umumiy\s+maydon|жалпы\s+аудан)/iu],
 ]);
 
-const AREA_UNIT_RE = String.raw`(?:м²|м2|m²|m2|sqm|sq\.?\s*m|mp|кв\.?\s*м|квадрат\p{L}*)`;
+const AREA_UNIT_RE = String.raw`(?:м²|м2|m²|m2|sqm|sq\.?\s*m|mp|кв\.?\s*м|кв2|квадрат\p{L}*)`;
 
 function areaAfterLabel(text, labelRe) {
   const re = new RegExp(`${labelRe.source}\\s*[:=-]?\\s*(\\d{1,4}(?:[.,]\\d{1,2})?)\\s*${AREA_UNIT_RE}`, 'iu');

package/src/locations-runtime.js

@@ -8,11 +8,16 @@ import {
 import { LOCATION_LIST_KEYS, mergeLocationCountries } from './location-merge.js';
 import { KG_LOCATION_EXTENSIONS } from './kg-location-extensions.js';
 import { KZ_DISTRICT_EXTENSIONS } from './kz-district-extensions.js';
+import { UZ_TASHKENT_CONTEXT_EXTENSIONS } from './uz-tashkent-context-extensions.js';
 
 export { UA_REGION_ENTRIES, UA_SECONDARY_CITIES, matchUkraineRegion, matchUkraineSecondaryCity };
 
 export const LOCATION_DICTIONARIES = Object.freeze({
   ...BASE_LOCATION_DICTIONARIES,
+  UZ: mergeLocationCountries(
+    BASE_LOCATION_DICTIONARIES.UZ || {},
+    UZ_TASHKENT_CONTEXT_EXTENSIONS,
+  ),
   KZ: mergeLocationCountries(
     BASE_LOCATION_DICTIONARIES.KZ || {},
     KZ_DISTRICT_EXTENSIONS,

package/src/ua-city-location-expansions.js

@@ -152,6 +152,14 @@ export const UA_CITY_LOCATION_EXPANSIONS = Object.freeze({
     ['Nizhyn Railway Station', 'Залізничний вокзал Ніжина', 'станція Ніжин'],
     ['Ніжинський краєзнавчий музей імені Івана Спаського', 'Ніжинський краєзнавчий музей ім. І. Спаського', 'Ніжинський краєзнавчий музей ім. Спаського І.Г.'],
     ['Ніжинська поштова станція', 'Музей «Ніжинська поштова станція»', 'Ніжинський музей "Поштова станція"'],
+    ['Ніжинський державний університет імені Миколи Гоголя', 'НДУ ім. М. Гоголя', 'Nizhyn Mykola Gogol State University'],
+    ['Свято-Миколаївський кафедральний собор', 'Миколаївський кафедральний собор (Ніжин)', 'Миколаївський собор', 'Собор Святого Миколая Чудотворця'],
+  ]),
+
+  Pryluky: landmarks([
+    ['Pryluky Railway Station'],
+    ['Прилуцький краєзнавчий музей імені В. І. Маслова', 'Прилуцький краєзнавчий музей ім. В. І. Маслова', 'Прилуцький краєзнавчий музей ім. В.І. Маслова'],
+    ['Собор Різдва Пресвятої Богородиці', 'Собор Різдва Богородиці', 'Собор Різдва Богородиці (Прилуки)'],
   ]),
 
   Volodymyr: landmarks([

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

@@ -0,0 +1,37 @@
+import { TASHKENT_AREAS } from './geo.js';
+import { aliasesToRegex } from './normalization.js';
+
+function legacyAreaEntry(district, canonical) {
+  const source = (TASHKENT_AREAS[district] || []).find((item) => item.name === canonical);
+  if (!source) throw new Error(`Missing canonical Tashkent area: ${district}/${canonical}`);
+
+  const aliases = Object.freeze([...new Set([
+    source.name,
+    ...(source.aliases || []),
+  ].filter(Boolean))]);
+
+  return Object.freeze({
+    canonical: source.name,
+    name: source.name,
+    type: source.type || 'local_area',
+    entityType: source.type || 'local_area',
+    country: 'UZ',
+    city: 'Tashkent',
+    district,
+    parent: district,
+    aliases,
+    re: aliasesToRegex(aliases),
+  });
+}
+
+// C-1 / Ц-1 is a very common Tashkent housing reference. The historical
+// TASHKENT_AREAS catalog already owns its canonical identity (`Buyuk Ipak Yuli`),
+// so expose that same entity to the runtime location matcher instead of creating
+// a second C-1 canonical or consumer-local regex.
+const C1 = legacyAreaEntry('Mirzo Ulugbek', 'Buyuk Ipak Yuli');
+
+export const UZ_TASHKENT_CONTEXT_EXTENSIONS = Object.freeze({
+  Tashkent: Object.freeze({
+    localAreas: Object.freeze([C1]),
+  }),
+});