@whiteslove/parsing-lexicon 0.5.8 → 0.5.9

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.5.8",
+  "version": "0.5.9",
   "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
   "repository": {
     "type": "git",
@@ -38,7 +38,8 @@
     "./central-asia": "./src/central-asia.js",
     "./ukraine": "./src/ukraine.js",
     "./location-merge": "./src/location-merge.js",
-    "./locations": "./src/locations.js",
+    "./locations": "./src/locations-runtime.js",
+    "./kg-location-extensions": "./src/kg-location-extensions.js",
     "./kz-location-extensions": "./src/kz-location-extensions.js",
     "./uz-location-extensions": "./src/uz-location-extensions.js",
     "./central-asia-locations": "./src/central-asia-locations.js",

package/src/hiring-source-semantics.js

@@ -22,11 +22,14 @@ const FIELD_EXTRA_ALIASES = Object.freeze({
 });
 
 const APPLICATION_MODAL_RE = /(?:отправить\s+резюме\s+)?пол\s+мужской\s+женский\s+образование[\s\S]{0,500}?выберите\s+вакансию[\s\S]*/iu;
-const LEAKED_CSS_RULE_RE = /\s*[.#][\w\\-]+(?:\s+[.#][\w\\-]+)?(?::[\w-]+)?\s*\{\s*(?:[\w-]+\s*:\s*[^;{}]+;?\s*)+\}\s*/gu;
+const LEAKED_STYLESHEET_START_RE = /:root\s*\{\s*--[\w-]+\s*:/iu;
+const LEAKED_CSS_RULE_RE = /\s*(?:[.#][\w\\-]+|:root)(?:[^{}\n]{0,300})\{\s*(?:(?:--)?[\w-]+\s*:\s*[^;{}]+;?\s*)+\}\s*/giu;
 
-/** Remove source-page controls and stylesheet fragments copied into vacancy text. */
+/** Remove source-page controls and stylesheet fragments copied into hiring text. */
 export function cleanHiringSourceText(value) {
-  return String(value || '')
+  const text = String(value || '');
+  const stylesheetStart = text.search(LEAKED_STYLESHEET_START_RE);
+  return (stylesheetStart >= 0 ? text.slice(0, stylesheetStart) : text)
     .replace(APPLICATION_MODAL_RE, '')
     .replace(LEAKED_CSS_RULE_RE, ' ')
     .replace(/[ \t]+(?=\n)/g, '')

package/src/index.js

@@ -18,7 +18,8 @@ export * from './geography-detection.js';
 export * from './central-asia.js';
 export * from './ukraine.js';
 export * from './location-merge.js';
-export * from './locations.js';
+export * from './locations-runtime.js';
+export * from './kg-location-extensions.js';
 export * from './kz-location-extensions.js';
 export * from './uz-location-extensions.js';
 export * from './central-asia-locations.js';

package/src/kg-location-extensions.js

@@ -0,0 +1,25 @@
+import { aliasesToRegex } from './normalization.js';
+
+const district = (name, aliases = []) => {
+  const all = Object.freeze([...new Set([name, ...aliases].filter(Boolean))]);
+  return Object.freeze({
+    canonical: name,
+    name,
+    type: 'district',
+    entityType: 'district',
+    country: 'KG',
+    aliases: all,
+    re: aliasesToRegex(all),
+  });
+};
+
+export const KG_LOCATION_EXTENSIONS = Object.freeze({
+  Bishkek: Object.freeze({
+    districts: Object.freeze([
+      district('Pervomaisky', ['Первомайский', 'Первомайский район', 'Биринчи Май', 'Биринчи Май району', 'Birinchi May', 'Birinchi May district']),
+      district('Leninsky', ['Ленинский', 'Ленинский район', 'Ленин району', 'Lenin district']),
+      district('Oktyabrsky', ['Октябрьский', 'Октябрьский район', 'Октябрь району', 'Oktyabr district']),
+      district('Sverdlovsky', ['Свердловский', 'Свердловский район', 'Свердлов району', 'Sverdlov district']),
+    ]),
+  }),
+});

package/src/locations-runtime.js

@@ -0,0 +1,52 @@
+import {
+  LOCATION_DICTIONARIES as BASE_LOCATION_DICTIONARIES,
+  UA_REGION_ENTRIES,
+  UA_SECONDARY_CITIES,
+  matchUkraineRegion,
+  matchUkraineSecondaryCity,
+} from './locations.js';
+import { LOCATION_LIST_KEYS, mergeLocationCountries } from './location-merge.js';
+import { KG_LOCATION_EXTENSIONS } from './kg-location-extensions.js';
+
+export { UA_REGION_ENTRIES, UA_SECONDARY_CITIES, matchUkraineRegion, matchUkraineSecondaryCity };
+
+export const LOCATION_DICTIONARIES = Object.freeze({
+  ...BASE_LOCATION_DICTIONARIES,
+  KG: mergeLocationCountries(
+    BASE_LOCATION_DICTIONARIES.KG || {},
+    KG_LOCATION_EXTENSIONS,
+  ),
+});
+
+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;
+}