@whiteslove/parsing-lexicon 0.6.2 → 0.6.4

package/src/locations.js

@@ -6,7 +6,7 @@ import {
   matchUkraineRegion,
   matchUkraineSecondaryCity,
 } from './location-data.js';
-import { LOCATION_LIST_KEYS, mergeLocationCountries } from './location-merge.js';
+import { mergeLocationCountries } from './location-merge.js';
 import { aliasesToRegex } from './normalization.js';
 import { KZ_LOCATION_EXTENSIONS } from './kz-location-extensions.js';
 import { UZ_LOCATION_EXTENSIONS } from './uz-location-extensions.js';
@@ -17,7 +17,7 @@ import { UA_METRO_LOCATION_EXTENSIONS } from './ua-location-extensions-metro.js'
 import { UA_KHARKIV_LOCATION_TRANSLATIONS } from './ua-kharkiv-location-translations.js';
 import { UA_CITY_LOCATION_EXPANSIONS } from './ua-city-location-expansions.js';
 
-export const UA_EXTRA_LOCATION_DICTIONARIES = Object.freeze({
+const UA_EXTRA_LOCATION_DICTIONARIES = Object.freeze({
   ...RAW_UA_EXTRA_LOCATION_DICTIONARIES,
   Zaporizhzhia: Object.freeze({
     ...RAW_UA_EXTRA_LOCATION_DICTIONARIES.Zaporizhzhia,
@@ -38,6 +38,46 @@ const TASHKENT_UNSUPPORTED_SEED_MICRODISTRICTS = new Set([
   'Yunusabad-22',
 ]);
 
+const TASHKENT_MAHALLA_ALIAS_ADDITIONS = Object.freeze({
+  Humoyun: Object.freeze([
+    'Khumoyun',
+    'Humoyun MFY',
+    'Ҳумоюн маҳалласи',
+    'Хумоюн',
+    'махалля Хумаюн',
+  ]),
+  "Bog'ko'cha": Object.freeze([
+    "Bog'ko'cha mahallasi",
+    "Bog'ko'cha MFY",
+    'Боғкўча маҳалласи',
+    'Богкуча махалла',
+    'махалля Богкуча',
+  ]),
+  "Bog'bon": Object.freeze([
+    "Bog'bon MFY",
+    'Богбон махалла',
+  ]),
+  Shifokorlar: Object.freeze([
+    'Shifokorlar MFY',
+    'Шифокорлар маҳалласи',
+    'Шифокорлар махалласи',
+    'махалля Шифокорлар',
+  ]),
+  "Chamanbog'": Object.freeze([
+    "Chamanbog' mahallasi",
+    "Chamanbog' MFY",
+    'Чаманбоғ маҳалласи',
+    'Чаманбог махалла',
+    'махалля Чаманбог',
+  ]),
+  Asalobod: Object.freeze([
+    'Asalobod MFY',
+    'Асалобод маҳалласи',
+    'Асалобод махалла',
+    'махалля Асалобод',
+  ]),
+});
+
 const LEGACY_SAMARKAND_SILK_ROAD_RESIDENCE = (
   BASE_LOCATION_DICTIONARIES.UZ?.Samarkand?.residentialComplexes || []
 ).find(({ name }) => name === 'Silk Road Residence');
@@ -102,10 +142,29 @@ function normalizeUzSemanticLocations(country) {
       })];
     }),
   ));
+
   const tashkent = normalized?.Tashkent;
-  const qorasuv = (tashkent?.microdistricts || []).find(({ name }) => name === 'Qorasuv');
+  if (tashkent) {
+    normalized = Object.freeze({
+      ...normalized,
+      Tashkent: Object.freeze({
+        ...tashkent,
+        mahallas: Object.freeze((tashkent.mahallas || []).map((entry) => {
+          const additions = TASHKENT_MAHALLA_ALIAS_ADDITIONS[entry.name];
+          return additions
+            ? semanticEntry(entry, entry.name, [...(entry.aliases || []), ...additions], 'mahalla')
+            : entry;
+        })),
+      }),
+    });
+  }
 
-  if (tashkent && qorasuv) {
+  const normalizedTashkent = normalized?.Tashkent;
+  const qorasuv = (normalizedTashkent?.microdistricts || []).find(({ name }) => name === 'Qorasuv');
+
+  if (normalizedTashkent && qorasuv) {
+    // Bare Qorasuv is ambiguous with numbered Qorasuv/Karasu blocks. The
+    // umbrella area therefore requires an area/massif/daha form in free text.
     const qorasuvAliases = Object.freeze([...new Set([
       ...(qorasuv.aliases || []).filter((alias) => !/^(?:qorasuv|korasuv|корасув|карасу)$/iu.test(String(alias).trim())),
       'Qorasuv dahasi',
@@ -126,10 +185,10 @@ function normalizeUzSemanticLocations(country) {
     normalized = Object.freeze({
       ...normalized,
       Tashkent: Object.freeze({
-        ...tashkent,
-        microdistricts: Object.freeze((tashkent.microdistricts || []).filter(({ name }) => name !== 'Qorasuv')),
+        ...normalizedTashkent,
+        microdistricts: Object.freeze((normalizedTashkent.microdistricts || []).filter(({ name }) => name !== 'Qorasuv')),
         localAreas: Object.freeze([
-          ...(tashkent.localAreas || []),
+          ...(normalizedTashkent.localAreas || []),
           qorasuvArea,
         ]),
       }),
@@ -144,6 +203,10 @@ function normalizeUzSemanticLocations(country) {
     const landmarks = samarkand.landmarks || [];
     const localAreas = samarkand.localAreas || [];
     const streets = samarkand.streets || [];
+
+    // Legacy UZ seeds contain a few Samarkand listing labels as separate
+    // canonicals or under the wrong semantic collection. Normalize them to the
+    // single physical subjects already represented by geo-catalog.
     const centralPark = landmarks.find(({ name }) => name === 'Central Park');
     const alisherPark = landmarks.find(({ name }) => name === 'Alisher Navoiy Park');
     const canonicalPark = centralPark || alisherPark;
@@ -352,6 +415,8 @@ function normalizeUzSemanticLocations(country) {
     const oldCity = localAreas.find(({ name }) => name === 'Old City');
 
     if (ichanKala && oldCity) {
+      // UNESCO identifies Itchan Kala as the historic inner-city of Khiva.
+      // Keep listing-friendly Old City forms as aliases of that one place.
       const normalizedIchanKala = semanticEntry(ichanKala, 'Ichan Kala', [
         ...(ichanKala.aliases || []),
         ...(oldCity.aliases || []),
@@ -453,6 +518,57 @@ function normalizeUzSemanticLocations(country) {
     });
   }
 
+  const nukus = normalized?.Nukus;
+  if (nukus) {
+    const mahallas = nukus.mahallas || [];
+    const localAreas = nukus.localAreas || [];
+    const aliasesByCanonical = new Map([
+      ['Qizil qum', ['Qızıl qum', 'Qızılqum']],
+      ['Gúzar', ['Guzar']],
+      ['Isbilermenler aymagi', ['Isbilermenler aymaǵı']],
+      ['Darbent', ['Dárbent']],
+      ['Abat makan', ['Abat mákan']],
+      ['Jolshilar', ['Jolshılar']],
+      ['Qutli qonis', ['Qutlı qonıs']],
+      ['Amanliq guzari', ['Amanlıq gúzarı']],
+      ['Aydin jol', ['Aydın jol']],
+      ['Qum awil', ['Qum awıl']],
+      ['Qutli makan', ['Qutlı mákan']],
+      ['Shimbay shayxana', ['Shımbay shayxana']],
+      ['Shayirlar awili', ['Shayırlar awılı']],
+      ['Tungish qonis', ['Tunǵısh qonıs']],
+      ['Qurilisshi', ['Qurılısshı']],
+      ['Ornek', ['Órnek', "O'rnek"]],
+      ['Jana zaman', ['Jańa zaman']],
+      ['Baqshiliq', ['Baqshılıq']],
+      ['Tinishliq', ['Tınıshlıq']],
+      ['Xaliqlar dosligi', ['Xalıqlar doslıǵı']],
+      ['Boz awil', ['Boz awıl']],
+      ['Jas awlad', ['Jas áwlad']],
+      ['Aq jagis', ['Aq jaǵıs']],
+      ['Tele oray', ['Telecentr']],
+      ['Shadli awil', ['Shadlı']],
+      ['Qumbiz awil', ['Qumbız awıl']],
+    ]);
+
+    normalized = Object.freeze({
+      ...normalized,
+      Nukus: Object.freeze({
+        ...nukus,
+        mahallas: Object.freeze(mahallas.map((entry) => {
+          const aliases = aliasesByCanonical.get(entry.name);
+          return aliases
+            ? semanticEntry(entry, entry.name, [...(entry.aliases || []), ...aliases], 'mahalla')
+            : entry;
+        })),
+        // These names already have canonical mahalla owners. Keeping a second
+        // local-area owner makes exact bridge resolution ambiguous and causes
+        // the crawler to inspect the same physical place twice.
+        localAreas: Object.freeze(localAreas.filter(({ name }) => !['Dosliq', 'Samanbay'].includes(name))),
+      }),
+    });
+  }
+
   return normalized;
 }
 
@@ -462,6 +578,9 @@ function normalizeUaSemanticLocations(country) {
   const odesa = country?.Odesa;
   if (!odesa) return country;
 
+  // “5–16 station of Velykyi Fontan” are traditional listing/locality zones
+  // anchored to the Fontanska Road transit stops, not twelve standalone city
+  // microdistricts. Keep their parsing value but expose them as local areas.
   const fontanStationAreas = Object.freeze(
     (odesa.microdistricts || [])
       .filter(({ name }) => ODESA_FONTAN_STATION_RE.test(name))
@@ -523,35 +642,4 @@ export {
   UA_SECONDARY_CITIES,
   matchUkraineRegion,
   matchUkraineSecondaryCity,
-};
-
-export function dictionaryFor(countryCode, city) {
-  return LOCATION_DICTIONARIES[countryCode]?.[city] || null;
-}
-
-export function locationCities(countryCode) {
-  return LOCATION_DICTIONARIES[countryCode] || Object.freeze({});
-}
-
-export function matchDictionaryLocation(text, countryCode, city = null) {
-  const country = locationCities(countryCode);
-  const cities = city && country[city] ? [[city, country[city]]] : Object.entries(country);
-  const value = String(text || '');
-  let best = null;
-  for (const [cityName, data] of cities) {
-    for (const type of LOCATION_LIST_KEYS) {
-      for (const entry of data[type] || []) {
-        const match = entry?.re?.exec(value);
-        if (!match) continue;
-        const start = match.index;
-        const end = start + match[0].length;
-        const containsBest = best && start <= best.start && end >= best.end && end - start > best.end - best.start;
-        if (best && !containsBest) continue;
-        best = { city: cityName, type, name: entry.name, aliases: entry.aliases, start, end };
-      }
-    }
-  }
-  if (!best) return null;
-  const { start: _start, end: _end, ...result } = best;
-  return result;
-}
+};

package/src/money-core.js

@@ -12,7 +12,7 @@ import {
 // ("4 00 000" instead of "400 000"). Accept that only when there are at least
 // two spaced groups, which keeps ordinary two-number prose out of money parsing.
 export const MONEY_NUMBER_PATTERN = '(?:\\d{1,3}(?:[ \\u00a0]\\d{2,3}){2,}(?:[.,]\\d+)?|\\d{1,3}(?:[ \\u00a0]\\d{3})+(?:[.,]\\d+)?|\\d{1,3}(?:,\\d{3})+(?:\\.\\d+)?|\\d{1,3}(?:\\.\\d{3})+(?:,\\d+)?|\\d+(?:[.,]\\d+)?)';
-export const MONEY_SCALE_PATTERN = 'k|к|тыс\\.?|тысяч(?:а|и)?|тис\\.?|thousand|ming|мың|m|м|млн\\.?|mln|million|миллион(?:ов)?|мільйон(?:ів)?|bn|млрд|mlrd|billion';
+export const MONEY_SCALE_PATTERN = 'k|к|тыс\\.?|тысяч(?:а|и)?|тис\\.?|thousand|ming|минг|мың|mii|m|м|млн\\.?|mln|million|миллион(?:ов)?|мільйон(?:ів)?|mil|milioane|bn|млрд|mlrd|billion|mld|miliard(?:e)?';
 // Each scale group needs the token-boundary guard MONEY_SINGLE_RE already has
 // below: without it, "2 до 3 месяцев" reads "м" off "месяцев" as the million
 // abbreviation and turns 3 into 3,000,000. The boundary is nested inside the

package/src/money-lexicon.js

@@ -75,5 +75,5 @@ export const SALARY_MODIFIERS = Object.freeze({
 export const NUMBER_MULTIPLIERS = Object.freeze([
   group('thousand', { ru: ['к', 'тыс', 'тыс.', 'тысяч', 'тысячи'], en: ['k', 'thousand'], uk: ['к', 'тис', 'тис.', 'тисяч'], ro: ['k', 'mii'], uzLatn: ['k', 'ming'], uzCyrl: ['минг'], kk: ['k', 'мың'] }, { multiplier: 1_000 }),
   group('million', { ru: ['м', 'млн', 'млн.', 'миллион', 'миллионов'], en: ['m', 'mln', 'million', 'millions'], uk: ['м', 'млн', 'мільйон', 'мільйонів'], ro: ['m', 'mil', 'milioane'], uzLatn: ['m', 'mln', 'million'], uzCyrl: ['млн', 'миллион'], kk: ['м', 'млн', 'миллион'] }, { multiplier: 1_000_000 }),
-  group('billion', { ru: ['млрд', 'миллиард', 'миллиардов'], en: ['bn', 'billion'], uk: ['млрд', 'мільярд'], ro: ['mld', 'miliard'], uzLatn: ['mlrd', 'billion'], uzCyrl: ['млрд', 'миллиард'], kk: ['млрд', 'миллиард'] }, { multiplier: 1_000_000_000 }),
+  group('billion', { ru: ['млрд', 'миллиард', 'миллиардов'], en: ['bn', 'billion'], uk: ['млрд', 'мільярд'], ro: ['mld', 'miliard', 'miliarde'], uzLatn: ['mlrd', 'billion'], uzCyrl: ['млрд', 'миллиард'], kk: ['млрд', 'миллиард'] }, { multiplier: 1_000_000_000 }),
 ]);

package/src/money.js

@@ -108,6 +108,21 @@ function rangeSearchText(text) {
   return normalized;
 }
 
+function isWorkTimeRange(text, match, start, end) {
+  if (CLOCK_RANGE_RE.test(match[0])) return true;
+
+  const first = Number(String(match[1] || '').replace(',', '.'));
+  const second = Number(String(match[3] || '').replace(',', '.'));
+  if (!Number.isFinite(first) || !Number.isFinite(second) || first > 24 || second > 24) return false;
+
+  // Bare 9-19 can be a schedule too. Reject it as money only when the nearby
+  // context explicitly talks about working hours/schedule; a real salary such
+  // as "$9-19/hour" still survives because it has pay/currency context and no
+  // schedule marker.
+  const window = text.slice(Math.max(0, start - 48), Math.min(text.length, end + 48));
+  return WORK_TIME_CONTEXT_RE.test(window);
+}
+
 function bestRange(text, protectedSpans) {
   const ranges = new RegExp(MONEY_RANGE_RE.source, 'giu');
   const candidates = [];
@@ -199,4 +214,4 @@ export function salaryPeriod(value) {
 export function currencyAliases(code) {
   const entry = CURRENCY_TERMS.find((item) => item.canonical === code);
   return entry ? [entry.canonical, ...aliasesOf(entry)] : [];
-}
+}

package/src/normalization.js

@@ -48,6 +48,22 @@ const KAZAKH_SEARCH_EQUIVALENCE = Object.freeze({
   ә: 'а', ғ: 'г', қ: 'к', ң: 'н', ө: 'о', ұ: 'у', ү: 'у', і: 'и', һ: 'х',
 });
 
+// Karakalpak Latin source data commonly alternates between ASCII-friendly
+// spellings and the current Latin letters used by OSM/local sources. Treat
+// these as search equivalences only; canonical spelling remains source-owned.
+const KARAKALPAK_LATIN_SEARCH_EQUIVALENCE = Object.freeze({
+  á: 'a', ǵ: 'g', ı: 'i', ń: 'n', ó: 'o', ú: 'u',
+});
+
+const KARAKALPAK_LATIN_PATTERN_EQUIVALENCE = Object.freeze({
+  a: '[aá]', á: '[aá]',
+  g: '[gǵ]', ǵ: '[gǵ]',
+  i: '[iı]', ı: '[iı]',
+  n: '[nń]', ń: '[nń]',
+  o: '[oó]', ó: '[oó]',
+  u: '[uú]', ú: '[uú]',
+});
+
 /** Search-oriented Cyrillic folding; canonical identity still comes from explicit aliases. */
 export function foldCyrillicForSearch(value) {
   return normalizeUnicode(value)
@@ -65,9 +81,17 @@ function foldKazakhForSearch(value) {
     .join('');
 }
 
+function foldKarakalpakLatinForSearch(value) {
+  return normalizeUnicode(value)
+    .toLocaleLowerCase()
+    .split('')
+    .map((char) => KARAKALPAK_LATIN_SEARCH_EQUIVALENCE[char] ?? char)
+    .join('');
+}
+
 export function normalizedAliasKeys(value, { transliteration = true } = {}) {
   const forms = transliteration
-    ? [value, foldCyrillicForSearch(value), foldKazakhForSearch(value)]
+    ? [value, foldCyrillicForSearch(value), foldKazakhForSearch(value), foldKarakalpakLatinForSearch(value)]
     : [value];
   return [...new Set(forms.flatMap((form) => [
     normalizeForMatch(form),
@@ -103,7 +127,7 @@ function createAliasOwnersIndex(entries, { transliteration = true } = {}) {
   for (const entry of entries || []) {
     for (const sourceAlias of valuesOf(entry)) {
       const searchForms = transliteration
-        ? [sourceAlias, foldCyrillicForSearch(sourceAlias), foldKazakhForSearch(sourceAlias)]
+        ? [sourceAlias, foldCyrillicForSearch(sourceAlias), foldKazakhForSearch(sourceAlias), foldKarakalpakLatinForSearch(sourceAlias)]
         : [sourceAlias];
       for (const searchAlias of searchForms) {
         for (const key of normalizedAliasKeys(searchAlias, { transliteration: false })) {
@@ -186,8 +210,19 @@ export function escapeRegex(value) {
 }
 
 function aliasPattern(value) {
-  return escapeRegex(normalizeUnicode(value).trim())
-    .replace(/[\s\-–—'’‘`ʻʼ]+/g, "[\\s\\-–—'’‘`ʻʼ]*");
+  const source = normalizeUnicode(value).trim();
+  let pattern = '';
+  let separatorOpen = false;
+  for (const char of source) {
+    if (/[\s\-–—'’‘`ʻʼ]/u.test(char)) {
+      if (!separatorOpen) pattern += "[\\s\\-–—'’‘`ʻʼ]*";
+      separatorOpen = true;
+      continue;
+    }
+    separatorOpen = false;
+    pattern += KARAKALPAK_LATIN_PATTERN_EQUIVALENCE[char.toLocaleLowerCase()] ?? escapeRegex(char);
+  }
+  return pattern;
 }
 
 function matcherFor(entries, { transliteration = true } = {}) {

package/src/tashkent-colloquial.js

@@ -14,6 +14,10 @@ export const TASHKENT_AREA_ADDITIONS = Object.freeze({
   Almazar: Object.freeze([
     area('Shifokorlar-1', ['шифокорлар 1', 'shifokorlar 1']),
     area('Shifokorlar-2', ['шифокорлар 2', 'shifokorlar 2']),
+    area('Shifokorlar-3', ['шифокорлар 3', 'shifokorlar 3']),
+    area('Shifokorlar-4', ['шифокорлар 4', 'shifokorlar 4', 'toshgu toshpi', 'тошгу тошпи']),
+    area('Shifokorlar-5', ['шифокорлар 5', 'shifokorlar 5']),
+    area('Shifokorlar-6', ['шифокорлар 6', 'shifokorlar 6']),
     area('Beruni-3', ['беруни 3', 'beruniy 3', 'beruni 3']),
     area('Hislat', ['хислат', 'hislat', 'xislat']),
     ...[1, 2, 3, 4].map((n) => area(`Beshkurgan-${n}`, [`бешкурган ${n}`, `бешқўрғон ${n}`, `beshqo rg on ${n}`, `beshkurgan ${n}`])),
@@ -21,6 +25,9 @@ export const TASHKENT_AREA_ADDITIONS = Object.freeze({
   Bektemir: Object.freeze([
     area('Suvsoz-1', ['сувсоз 1', 'водник 1', 'suvsoz 1']),
     area('Suvsoz-2', ['сувсоз 2', 'водник 2', 'suvsoz 2']),
+    area('Suvsoz-3', ['сувсоз 3', 'suvsoz 3']),
+    area('Suvsoz-4', ['сувсоз 4', 'suvsoz 4']),
+    area('Suvsoz-5', ['сувсоз 5', 'suvsoz 5']),
     area('Binokor', ['бинокор', 'binokor']),
     area('Binokor-2', ['бинокор 2', 'binokor 2']),
     area('Majnuntol', ['мажнунтол', 'majnuntol']),
@@ -79,4 +86,4 @@ export const FULL_TASHKENT_AREAS = Object.freeze(Object.fromEntries(
       ...(TASHKENT_AREAS[district] || []),
       ...(TASHKENT_AREA_ADDITIONS[district] || []),
     ])]),
-));
+));

package/src/tashkent-gastronomic-streets.js

@@ -0,0 +1,108 @@
+import { aliasesToRegex } from './normalization.js';
+
+function gastronomicStreet(canonical, aliases = [], district = null) {
+  const all = Object.freeze([...new Set([canonical, ...aliases].filter(Boolean))]);
+  return Object.freeze({
+    canonical,
+    name: canonical,
+    category: 'gastronomic_street',
+    entityType: 'street',
+    country: 'UZ',
+    city: 'Tashkent',
+    district,
+    aliases: all,
+    re: aliasesToRegex(all),
+  });
+}
+
+export const TASHKENT_GASTRONOMIC_STREETS = Object.freeze([
+  gastronomicStreet('Taras Shevchenko Street', [
+    'Taras Shevchenko Gastronomic Street',
+    'Taras Shevchenko nomidagi gastronomik ko‘cha',
+    "Taras Shevchenko nomidagi gastronomik ko'cha",
+    'Гастрономическая улица имени Тараса Шевченко',
+    'улица Тараса Шевченко',
+    'Taras Shevchenko ko‘chasi',
+  ], 'Mirobod'),
+  gastronomicStreet('Mehrgiyo Street', [
+    '"Mehrgiyo" Gastronomic Street',
+    'Mehrgiyo Gastronomic Street',
+    '"Mehrgiyo" gastronomik ko‘chasi',
+    "Mehrgiyo gastronomik ko'chasi",
+    'Гастрономическая улица «Мехргиё»',
+    'Улица Мехригийо',
+    'Mehrgiyo ko‘chasi',
+  ], 'Yangihayot'),
+  gastronomicStreet('Al-Khwarizmi Street', [
+    'Al-Khwarizmi Gastronomic Street',
+    'Al-Xorazmiy gastronomik ko‘chasi',
+    "Al-Xorazmiy gastronomik ko'chasi",
+    'Гастрономическая улица Аль-Хоразмий',
+    'Гастрономическая улица Аль-Хорезми',
+    'улица Ал-Хоразмий',
+    'Al-Xorazmiy ko‘chasi',
+  ], 'Chilanzar'),
+  gastronomicStreet('Dilsaroy Street', [
+    '“Dilsaroy” Gastronomic Street',
+    'Dilsaroy Gastronomic Street',
+    '“Dilsaroy” gastronomik ko‘chasi',
+    "Dilsaroy gastronomik ko'chasi",
+    'Гастрономическая улица «Дильсарой»',
+    'улица Дильсарой',
+    'Dilsaroy ko‘chasi',
+  ], 'Almazar'),
+  gastronomicStreet('Rihsili Street', [
+    'Rihsili Gastronomic Street',
+    'Rihsiliy gastronomik ko‘chasi',
+    'Rixsiliy gastronomik ko‘chasi',
+    'Гастрономическая улица Рихсили',
+    'Улица Рихсили',
+    'Rihsili Street',
+    'Rixsiliy ko‘chasi',
+  ], 'Yunusabad'),
+  gastronomicStreet('Shota Rustaveli Street', [
+    'Shota Rustaveli Gastronomic Street',
+    'Shota Rustaveli gastronomik ko‘chasi',
+    'Гастрономическая улица Шота Руставели',
+    'улица Шота Руставели',
+    'Shota Rustaveli ko‘chasi',
+  ], 'Yakkasaray'),
+  gastronomicStreet('Chigatoy-Darvoza Street', [
+    'Chigatoy–Darvoza gastronomic street',
+    'Chig‘atoy–Darvoza gastronomik ko‘chasi',
+    'Chigatay–Darvoza Street',
+    'Chigatoy–Darvoza Street',
+    'Гастрономическая улица Чигатой–Дарвоза',
+    'Улица Чигатой-Дарвоза',
+  ], 'Almazar'),
+  gastronomicStreet('Gulkhaniy Street', [
+    'Gulkhani Gastronomic Street',
+    'Gulxaniy gastronomik ko‘chasi',
+    'Гастрономическая улица Гулханий',
+    'Улица Гулханий',
+    'Gulkhaniy Street',
+    'Gulxaniy ko‘chasi',
+  ], 'Shaykhantahur'),
+  gastronomicStreet('Farhod Street', [
+    'Farhod Gastronomic Street',
+    'Farhod gastronomik ko‘chasi',
+    "Farhod gastronomik ko'chasi",
+    'Фархадская гастрономическая улица',
+    'Улица Фарход',
+    'Farhod ko‘chasi',
+  ], 'Uchtepa'),
+  gastronomicStreet('Sogdiyona Street', [
+    'Sogdiyona Gastronomic Street',
+    'So‘g‘diyona gastronomik ko‘chasi',
+    "So'g'diyona gastronomik ko'chasi",
+    'Гастрономическая улица Согдиёна',
+    'Улица Согдиёна',
+    'So‘g‘diyona ko‘chasi',
+  ], 'Sergeli'),
+]);
+
+export function resolveTashkentGastronomicStreet(value) {
+  const text = String(value || '').trim();
+  if (!text) return null;
+  return TASHKENT_GASTRONOMIC_STREETS.find((entry) => entry.re.test(text)) || null;
+}

package/src/tashkent-pois.js

@@ -11,10 +11,12 @@ function poi(name, category, aliases = [], options = {}) {
     re: aliasesToRegex(all),
     contextRequired: Boolean(options.contextRequired),
     contextRe: options.context ? new RegExp(options.context, 'iu') : null,
+    ...(options.parent ? { parent: options.parent } : {}),
   });
 }
 
 const MALL_CONTEXT = '(?:mall|молл|трц|тц|торгов(?:ый|ого)\\s+центр|savdo\\s+markazi|shopping\\s+(?:center|centre))';
+const PARK_CONTEXT = '(?:парк|bog|бог|боғ|park|national|milliy|navoi|наво)';
 
 export const TASHKENT_PARKS = Object.freeze([
   poi('Tashkent City Park', 'park', ['Ташкент Сити парк', 'Парк Tashkent City']),
@@ -22,21 +24,28 @@ export const TASHKENT_PARKS = Object.freeze([
   poi('Ecopark', 'park', ['Eco Park', 'Экопарк', 'Центральный экопарк']),
   poi('Central Park Mirzo Ulugbek', 'park', ['Центральный парк имени Мирзо Улугбека', 'Центральный парк', 'Central Park', 'Парк Мирзо Улугбека']),
   poi('Japanese Garden', 'park', ['Японский сад', "Yapon bog'i", 'Yapon bog‘i', 'Yapon bogʻI']),
-  poi('Milliy Bog Park', 'park', ['Миллий бог', "Milliy Bog'", 'Milliy Bog‘', 'Milliy Bogʻ', 'Национальный парк'], { contextRequired: true, context: '(?:парк|bog|боғ|park|national)' }),
-  poi('Alisher Navoi National Park', 'park', ['Парк имени Алишера Навои']),
-  poi('Dream Park', 'park', ['Dream Park имени Гафура Гуляма', 'Парк имени Гафура Гуляма', 'Парк Гафура Гуляма']),
-  poi('Anhor Park', 'park', ['Anhor', 'Анхор']),
-  poi('Lokomotiv Park', 'park', ['Anhor Lokomotiv', 'Анхор Локомотив', 'Парк Локомотив']),
+  poi('Alisher Navoi National Park', 'park', [
+    'Парк имени Алишера Навои',
+    'Национальный парк Узбекистана имени Алишера Навои',
+    'Milliy Bog Park',
+    "Milliy Bog'",
+    'Milliy Bog‘',
+    'Milliy Bogʻ',
+    'Миллий бог',
+    'Национальный парк',
+  ], { contextRequired: true, context: PARK_CONTEXT }),
+  poi('Dream Park', 'park', ['Dream Park имени Гафура Гуляма', 'Парк имени Гафура Гуляма', 'Парк Гафура Гуляма', 'Gafur Gulyam Park', 'Gafur Gulam Dream Park', "G'afur G'ulom bog'i", 'Парк культуры и отдыха имени Гафура Гуляма']),
+  poi('Anhor Lokomotiv Park', 'park', ['Anhor Lokomotiv', 'Anhor-Lokomotiv', 'Анхор Локомотив', 'Парк Анхор Локомотив', 'Парк Локомотив']),
   poi('Tashkentland', 'park', ['Ташкентленд']),
   poi('Victory Park', 'park', ['Парк Победы', "G'alaba Bog'i", 'G‘alaba Bog‘i', 'Gʻalaba Bogʻi']),
-  poi('Ashgabat Park', 'park', ['Ашхабад парк', 'Ashxobod Park']),
-  poi('Dostlik Park', 'park', ['Парк Дустлик', "Do'stlik Park", 'Do‘stlik Park', 'Doʻstlik Park']),
-  poi('Navruz Park', 'park', ['Парк Навруз']),
-  poi('Bobur Park', 'park', ['Парк Бабура', 'Парк Бобура', "Bobur bog'i", 'Bobur bog‘i']),
-  poi('Furqat Park', 'park', ['Парк Фурката']),
+  poi('Ashgabat Park', 'park', ['Ашхабад парк', 'Ashxobod Park', "Ashxobod bog'i", 'Ashxobod bog‘i']),
+  poi('Dostlik Park', 'park', ['Парк Дустлик', "Do'stlik Park", 'Do‘stlik Park', 'Doʻstlik Park', 'Bobur Park', 'Парк Бабура', 'Парк Бобура', "Bobur bog'i", 'Bobur bog‘i']),
+  poi('Navruz Park', 'park', ['Парк Навруз', 'Navroʻz bogʻi', "Navro'z bog'i"]),
+  poi('Furqat Park', 'park', ['Парк Фурката', "Furqat bog'i", 'Furqat bog‘i']),
   poi('Yakub Kolas Park', 'park', ['Парк Якуба Коласа', 'Парк имени Якуба Коласа']),
-  poi('Friendship of Peoples Park', 'park', ['Парк Дружбы народов']),
-  poi('Yangi Ozbekiston Park', 'park', ["Yangi O'zbekiston Bog'i", 'Yangi O‘zbekiston Bog‘i', 'Парк Новый Узбекистан', 'Новый Узбекистан', "Yangi O'zbekiston Park"]),
+  poi('Friendship of Peoples Park', 'park', ['Парк Дружбы народов', "Xalqlar Do'stligi Park", 'Xalqlar Do‘stligi Park']),
+  poi('Yangi Ozbekiston Park', 'park', ["Yangi O'zbekiston Bog'i", 'Yangi O‘zbekiston Bog‘i', 'Yangi Oʻzbekiston bogʻi', 'Парк Новый Узбекистан', 'Новый Узбекистан', "Yangi O'zbekiston Park"]),
+  poi('Tashkent Botanical Garden', 'park', ['Botanical Garden named after Fyodor Rusanov', 'Botanical Garden named after Fyodor Rusananov', 'Ботанический сад имени Фёдора Русанова', 'Ташкентский ботанический сад', 'Fyodor Rusanov nomidagi Botanika bog‘i', 'Toshkent botanika bog‘i']),
 ]);
 
 export const TASHKENT_SQUARES = Object.freeze([
@@ -65,15 +74,17 @@ export const TASHKENT_MARKETS = Object.freeze([
   poi('Ippodrom Bazaar', 'market', ['Ипподром']),
   poi('Abu Sahiy', 'market', ['Абу Сахий']),
   poi('Orikzor Bazaar', 'market', ['Урикзор', "O'rikzor", 'O‘rikzor', 'Oʻrikzor']),
+  poi('Stroygorod', 'market', ['Стройгород', 'рынок Стройгород', 'Стройгород рынок', 'Stroygorod market', 'Stroygorod bazaar'], { parent: 'Uchtepa' }),
   poi('Kadysheva Bazaar', 'market', ['Кадышева рынок', 'Кадышева базар']),
   poi('Yangiobod Bazaar', 'market', ['Янгиабадский рынок', 'Янгиабад', 'Yangiobod bozori']),
 ]);
 
 export const TASHKENT_MALLS = Object.freeze([
   poi('Tashkent City Mall', 'mall', ['Ташкент Сити Молл']),
-  poi('Samarqand Darvoza', 'mall', ['Самарканд Дарвоза']),
+  poi('Samarqand Darvoza', 'mall', ['Самарканд Дарвоза', 'Samarkand Darvoza']),
   poi('Compass Mall', 'mall', ['Compass', 'ТРЦ Compass', 'ТЦ Compass', 'Compass savdo markazi'], { contextRequired: true, context: MALL_CONTEXT }),
   poi('Mega Planet', 'mall', ['MegaPlanet', 'Мега Планет', 'ТРЦ Mega Planet', 'Mega Planet savdo markazi']),
+  poi('Anhor Park Mall', 'mall', ['Anhor Park', 'Anhor Mall', 'Anhor Park Savdo Markazi', 'Анхор Парк', 'Анхор Молл', 'ТРЦ Анхор', 'ТЦ Анхор']),
   poi('Riviera Mall', 'mall', ['Riviera', 'Ривьера Молл', 'ТРЦ Riviera', 'Riviera savdo markazi'], { contextRequired: true, context: MALL_CONTEXT }),
   poi('NEXT Mall', 'mall', ['Next', 'NEXT', 'Некст Молл', 'ТРЦ NEXT', 'NEXT savdo markazi'], { contextRequired: true, context: MALL_CONTEXT }),
   poi('Atlas Mall', 'mall', ['Atlas', 'Атлас Молл', 'ТРЦ Atlas', 'Atlas savdo markazi'], { contextRequired: true, context: MALL_CONTEXT }),
@@ -81,13 +92,15 @@ export const TASHKENT_MALLS = Object.freeze([
   poi('Depo Mall', 'mall', ['Depo', 'Депо Молл', 'ТРЦ Depo', 'Depo savdo markazi'], { contextRequired: true, context: MALL_CONTEXT }),
   poi('Vega Centre', 'mall', ['Vega Center', 'Вега Центр', 'Vega savdo markazi']),
   poi('Yunusabad Gallery', 'mall', ['Юнусабад Галерея', 'Yunusobod Galereya', 'Yunusobod Gallery']),
-  poi('Poytaxt Mall', 'mall', ['Poytaxt', 'Пойтахт Молл', 'Пойтахт', 'Poytaxt savdo markazi'], { contextRequired: true, context: MALL_CONTEXT }),
+  poi('Poytaxt Mall', 'mall', ['Poytaxt', 'Пойтахт Молл', 'Пойтахт', 'Poytaxt Shopping Center', 'Poytaxt savdo markazi'], { contextRequired: true, context: MALL_CONTEXT }),
   poi('Alfraganus Mall', 'mall', ['Alfraganus', 'Альфраганус Молл', 'Alfraganus savdo markazi'], { contextRequired: true, context: MALL_CONTEXT }),
-  poi('HT Mall', 'mall', ['HT Молл', 'HT savdo markazi']),
+  poi('High Town Mall', 'mall', ['HT Mall', 'HT Молл', 'HT savdo markazi']),
   poi('Seoul Mun Mall', 'mall', ['Seoul Mun', 'Сеул Мун Молл', 'Seoul Mun savdo markazi'], { contextRequired: true, context: MALL_CONTEXT }),
   poi('Ecobozor', 'mall', ['Eco Bozor', 'Ekobozor', 'Эко Бозор', 'Экобозор']),
-  poi('Magnum Samarkand Darvoza', 'mall', ['Magnum super', 'Magnum Supermarket', 'Магнум Самарканд Дарвоза']),
-  poi('Baraka Market', 'mall', ['Baraka market', 'Baraka Supermarket', 'Baraka Savdo', 'Барака маркет', 'Барака Савдо']),
+  poi('Chimgan Shopping Center', 'mall', ['Chimgan', 'Chimgan Mall', 'ТРЦ Чимган', 'ТЦ Чимган', 'Chimgan savdo markazi'], { contextRequired: true, context: MALL_CONTEXT }),
+  poi('Golden Life', 'mall', ['Golden Life Mall', 'Голден Лайф', 'ТРЦ Golden Life', 'ТЦ Golden Life']),
+  poi("Chig'atoy Mall", 'mall', ["Chig'atoy", 'Chigatoy', "ТРЦ Chig'atoy", 'ТРЦ Чигатай', 'ТЦ Чигатай']),
+  poi('Scopus Mall', 'mall', ['Scopus', 'Скопус Молл', 'ТРЦ Scopus']),
 ]);
 
 export const TASHKENT_ATTRACTIONS = Object.freeze([
@@ -180,6 +193,13 @@ export const TASHKENT_MEDICAL_POIS = Object.freeze([
   poi('Republic Specialized Nephrology and Transplantation Centre', 'medical', ['Republican Nephrology Center', 'Республиканский нефрологический центр', 'Центр нефрологии и трансплантации']),
   poi('Republican Perinatal Center', 'medical', ['Республиканский перинатальный центр', 'Перинатальный центр']),
   poi('Institute of Obstetrics and Gynecology', 'medical', ['Институт акушерства и гинекологии']),
+  poi('AKFA Medline', 'medical', ['AKFA MEDLINE', 'Akfa Medline', 'Акфа Медлайн']),
+  poi('Shifo Nur', 'medical', ['Шифо Нур']),
+  poi('Nano Medical Clinic', 'medical', ['Клиника Нано', 'Nano Clinic', 'Nano Klinikasi', 'Nano tibbiy klinikasi']),
+  poi('Medas Medical Center', 'medical', ['Medas Group', 'MEDAS', 'Медицинский центр MEDAS']),
+  poi('Prof Med Clinic', 'medical', ['ProfMedService', 'Prof Med Klinikasi', 'Клиника Профмедсервис']),
+  poi('Dr. Akshay Kumar Eye Clinic', 'medical', ['Глазная клиника доктора Акшая Кумара', 'Vedanta Medical']),
+  poi('Shox International Hospital', 'medical', ['Shox Hospital', 'Международная больница Шоx']),
 ]);
 
 export const TASHKENT_LEGACY_LANDMARKS = Object.freeze([
@@ -231,4 +251,4 @@ export function matchTashkentPoi(value, category = null) {
   }
   matches.sort((a, b) => b.length - a.length || a.start - b.start);
   return matches[0]?.entry ?? null;
-}
+}