@whiteslove/parsing-lexicon 0.4.1 → 0.4.3

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.4.1",
+  "version": "0.4.3",
   "description": "Shared deterministic multilingual parsing lexicon for Whiteslove housing and hiring services",
   "repository": {
     "type": "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;
 }