@whiteslove/parsing-lexicon 0.4.1 → 0.4.4

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.4.1",
-  "description": "Shared deterministic multilingual parsing lexicon for Whiteslove housing and hiring services",
+  "version": "0.4.4",
+  "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
   "repository": {
     "type": "git",
     "url": "git+https://github.com/AmoneMisa/parsing-lexicon.git"

package/src/geography-display.js

@@ -67,6 +67,12 @@ export const GEOGRAPHY_DISPLAY_NAMES = Object.freeze({
     }),
     microdistrict: Object.freeze({
       Tsukrovyi: 'Сахарный',
+      Karakamysh: 'Каракамыш', Sebzar: 'Себзар', Tashselmash: 'Ташсельмаш', Aviasozlar: 'Авиасозлар',
+      Kuylyuk: 'Куйлюк', Sergeli: 'Сергели массив', Sputnik: 'Спутник', 'Yangi Choshtepa': 'Янги Чоштепа',
+      // Same places as Choshtepa/Takhtapul above, but under the spelling the
+      // mahalla registry (uz-location-extensions.js) actually uses as its
+      // canonical name -- exact-key lookup means both spellings must be here.
+      Chashtepa: 'Чаштепа', Taxtapul: 'Тахтапул',
     }),
     metro: Object.freeze({
       'Buyuk Ipak Yoli': 'Буюк Ипак Йули', Pushkin: 'Пушкин', 'Hamid Olimjon': 'Хамид Алимджан',

package/src/housing-address.js

@@ -33,7 +33,11 @@ function escapeRegExp(value) {
 
 function compactStreet(value) {
   return clean(value)
-    .replace(new RegExp(`^${PREFIX_STREET_MARKER}\\s*`, 'iu'), '')
+    // The bare abbreviations in PREFIX_STREET_MARKER (e.g. "пр" for "пр-т")
+    // must not be followed by a letter, or this would strip a false-positive
+    // prefix off an unrelated word that merely starts the same way (e.g.
+    // "проживания" -> "оживания").
+    .replace(new RegExp(`^${PREFIX_STREET_MARKER}(?!\\p{L})\\s*`, 'iu'), '')
     .replace(new RegExp(`\\s+${POSTFIX_STREET_MARKER}$`, 'iu'), '')
     .replace(/[\s,;:.\-–—]+$/gu, '')
     .trim() || null;
@@ -167,7 +171,7 @@ function explicitStreetAddress(text) {
     if (prefixTyped) return prefixTyped;
 
     const boundedPrefix = line.match(new RegExp(
-      `(?:^|[\\s,;])${PREFIX_STREET_MARKER}\\s*((?:${STREET_WORD}\\s+){0,3}${STREET_WORD})(?=$|[,;])`,
+      `(?:^|[\\s,;])${PREFIX_STREET_MARKER}(?!\\p{L})\\s*((?:${STREET_WORD}\\s+){0,3}${STREET_WORD})(?=$|[,;])`,
       'iu',
     ));
     if (boundedPrefix) return result(boundedPrefix[0], boundedPrefix[1], null, null, 0.9);

package/src/housing-display.js

@@ -12,6 +12,7 @@ const DISPLAY = Object.freeze({
     Market: 'Market', Cafe: 'Cafe', Restaurant: 'Restaurant', Playground: 'Playground', Pharmacy: 'Pharmacy',
     Mosque: 'Mosque', Church: 'Church', 'Railway station': 'Railway station', Airport: 'Airport',
     'Bobur Park': 'Bobur Park', 'Independence Square': 'Independence Square', 'Mega Planet': 'Mega Planet',
+    'Bek Baraka': 'Bek Baraka', Khadra: 'Khadra',
   }),
   ru: Object.freeze({
     Park: 'Парк', Metro: 'Метро', 'Bus stop': 'Автобусная остановка', 'Public transport': 'Общественный транспорт',
@@ -21,6 +22,7 @@ const DISPLAY = Object.freeze({
     Market: 'Рынок', Cafe: 'Кафе', Restaurant: 'Ресторан', Playground: 'Детская площадка', Pharmacy: 'Аптека',
     Mosque: 'Мечеть', Church: 'Церковь', 'Railway station': 'Железнодорожный вокзал', Airport: 'Аэропорт',
     'Bobur Park': 'Парк Бобура', 'Independence Square': 'Площадь Независимости', 'Mega Planet': 'Mega Planet',
+    'Bek Baraka': 'Бек-Барака', Khadra: 'Хадра',
   }),
 });
 
@@ -51,15 +53,28 @@ export function housingSemanticCanonical(value) {
   return exactPoi(text)?.name || text;
 }
 
+function cyrillicPoiAlias(poiEntry) {
+  return (poiEntry?.aliases || []).find((alias) => /[а-яёіїґ]/iu.test(alias)) || null;
+}
+
 export function housingSemanticDisplayName(value, locale = 'en') {
   const text = String(value || '').trim();
   if (!text) return '';
   const language = languageKey(locale);
   const extension = extensionEntity(text);
   if (extension?.display?.[language]) return extension.display[language];
+  const poiEntry = exactPoi(text);
   const canonical = extension?.canonical
     || findCanonical(text, GENERIC_LANDMARK_TERMS)?.canonical
-    || exactPoi(text)?.name
+    || poiEntry?.name
     || text;
+  // Named landmarks (markets, squares, parks) are added one at a time in
+  // tashkent-pois.js and often only carry a Russian alias, no dedicated
+  // DISPLAY entry. Fall back to that alias for ru rather than leaking the
+  // Latin canonical name, instead of requiring every POI to be re-listed here.
+  if (language === 'ru' && !DISPLAY.ru[canonical]) {
+    const fallback = cyrillicPoiAlias(poiEntry);
+    if (fallback) return fallback;
+  }
   return DISPLAY[language]?.[canonical] || canonical;
 }

package/src/locations.js

@@ -172,11 +172,21 @@ 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) {
-      const match = (data[type] || []).find((entry) => entry?.re?.test(value));
-      if (match) return { city: cityName, type, name: match.name, aliases: match.aliases };
+      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 };
+      }
     }
   }
-  return null;
+  if (!best) return null;
+  const { start: _start, end: _end, ...result } = best;
+  return result;
 }

package/src/uz-location-extensions.js

@@ -210,6 +210,8 @@ export const UZ_LOCATION_EXTENSIONS = Object.freeze({
       ['Tashkent City', 'Ташкент Сити', 'Toshkent City', 'Tashkent City IBC'],
     ],
     streets: [
+      { name: 'Gulobod Street', aliases: ['Gulobod ko‘chasi', "Gulobod ko'chasi", 'улица Гулобод', 'ул. Гулобод'], parent: 'Shaykhantahur' },
+      { name: 'Sebzor Street', aliases: ['Sebzor ko‘chasi', "Sebzor ko'chasi", 'улица Себзор', 'ул. Себзор'], parent: 'Almazar' },
       { name: 'Lolazor Street', aliases: ['Lolazor ko‘chasi', "Lolazor ko'chasi", 'улица Лолазор', 'ул. Лолазор'], parent: 'Uchtepa' },
       { name: 'Shohimardon Street', aliases: ['Shohimardon ko‘chasi', "Shohimardon ko'chasi", 'улица Шохимардон', 'ул. Шохимардон'], parent: 'Yashnobod' },
       { name: 'Shohimardon Passage 1', aliases: ['Shohimardon 1-tor ko‘chasi', "Shohimardon 1-tor ko'chasi", '1-й проезд Шохимардон', '1 проезд Шохимардон'], parent: 'Yashnobod' },