@whiteslove/parsing-lexicon 0.6.2 → 0.6.5

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (42) hide show
  1. package/package.json +5 -1
  2. package/src/central-asia-locations.js +87 -3
  3. package/src/central-asia.js +26 -2
  4. package/src/geo.js +7 -7
  5. package/src/geography-display.js +1 -1
  6. package/src/geography-zone-display.js +1 -1
  7. package/src/geography.js +13 -4
  8. package/src/hiring-requirements.js +5 -5
  9. package/src/hiring-salary-context.js +6 -2
  10. package/src/hiring-temporal.js +4 -0
  11. package/src/housing-address.js +1 -1
  12. package/src/housing-card-fields.js +20 -4
  13. package/src/housing-display.js +14 -1
  14. package/src/housing-listing-enrichment.js +53 -10
  15. package/src/housing-money.js +1 -1
  16. package/src/housing-poi-extensions.js +6 -0
  17. package/src/housing-source-aliases.js +3 -0
  18. package/src/housing-structured.js +18 -1
  19. package/src/index.js +2 -0
  20. package/src/kg-location-extensions.js +91 -11
  21. package/src/kz-district-extensions.js +36 -0
  22. package/src/kz-location-extensions.js +3 -3
  23. package/src/locations-runtime.js +10 -0
  24. package/src/locations.js +127 -39
  25. package/src/money-core.js +1 -1
  26. package/src/money-lexicon.js +1 -1
  27. package/src/money.js +16 -1
  28. package/src/normalization.js +39 -4
  29. package/src/tashkent-colloquial.js +8 -1
  30. package/src/tashkent-gastronomic-streets.js +108 -0
  31. package/src/tashkent-pois.js +38 -18
  32. package/src/ua-city-location-expansions.js +96 -0
  33. package/src/ua-geo-coordinates.d.ts +14 -0
  34. package/src/ua-geo-coordinates.js +209 -0
  35. package/src/ua-geo-resolver.d.ts +31 -0
  36. package/src/ua-geo-resolver.js +171 -0
  37. package/src/ua-location-extensions-metro.js +42 -1
  38. package/src/ua-location-extensions-regional.js +2 -2
  39. package/src/ua-secondary-cities.js +6 -4
  40. package/src/uz-geo-coordinates.d.ts +14 -0
  41. package/src/uz-geo-coordinates.js +166 -0
  42. package/src/uz-tashkent-context-extensions.js +37 -0
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@whiteslove/parsing-lexicon",
3
- "version": "0.6.2",
3
+ "version": "0.6.5",
4
4
  "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
5
5
  "repository": {
6
6
  "type": "git",
@@ -37,6 +37,9 @@
37
37
  },
38
38
  "./central-asia": "./src/central-asia.js",
39
39
  "./ukraine": "./src/ukraine.js",
40
+ "./ua-geo-coordinates": { "types": "./src/ua-geo-coordinates.d.ts", "import": "./src/ua-geo-coordinates.js" },
41
+ "./ua-geo-resolver": { "types": "./src/ua-geo-resolver.d.ts", "import": "./src/ua-geo-resolver.js" },
42
+ "./uz-geo-coordinates": { "types": "./src/uz-geo-coordinates.d.ts", "import": "./src/uz-geo-coordinates.js" },
40
43
  "./location-merge": "./src/location-merge.js",
41
44
  "./locations": "./src/locations-runtime.js",
42
45
  "./kg-location-extensions": "./src/kg-location-extensions.js",
@@ -50,6 +53,7 @@
50
53
  "./tashkent-colloquial": "./src/tashkent-colloquial.js",
51
54
  "./tashkent-residential-complexes": "./src/tashkent-residential-complexes.js",
52
55
  "./tashkent-pois": "./src/tashkent-pois.js",
56
+ "./tashkent-gastronomic-streets": "./src/tashkent-gastronomic-streets.js",
53
57
  "./tashkent-housing-geography": {
54
58
  "types": "./src/tashkent-housing-geography.d.ts",
55
59
  "import": "./src/tashkent-housing-geography.js"
@@ -1,4 +1,4 @@
1
- import { LOCATION_DICTIONARIES } from './locations.js';
1
+ import { LOCATION_DICTIONARIES } from './locations-runtime.js';
2
2
  import { LOCATION_LIST_KEYS } from './location-merge.js';
3
3
  import { CITIES_BY_COUNTRY, canonicalCity } from './geography.js';
4
4
  import { aliasesOf, aliasesToRegex, normalizeForMatch } from './normalization.js';
@@ -6,7 +6,7 @@ import { KZ_AMBIGUOUS_LOCAL_NAMES, KZ_SEARCH_CLUSTERS } from './kz-location-exte
6
6
  import { UZ_AMBIGUOUS_LOCAL_NAMES } from './uz-location-extensions.js';
7
7
 
8
8
  // Compatibility names retained for consumers. The canonical dictionaries are
9
- // assembled once in locations.js; matcher layers must consume, not re-merge.
9
+ // assembled once in the public runtime registry; matcher layers must consume, not re-merge.
10
10
  export const KZ_EXPANDED_LOCATION_DICTIONARIES = LOCATION_DICTIONARIES.KZ || Object.freeze({});
11
11
  export const UZ_EXPANDED_LOCATION_DICTIONARIES = LOCATION_DICTIONARIES.UZ || Object.freeze({});
12
12
 
@@ -139,6 +139,86 @@ function semanticBounds(value, match) {
139
139
  return { start, end };
140
140
  }
141
141
 
142
+ const NUMBERED_LOCAL_AREA_NAME_RE = /(?:^|[-–—/\s])\d{1,2}[aа]?(?=$|[^\p{L}\p{N}_])/iu;
143
+ const NUMBERED_CHILD_SUFFIX_RE = /^\s*[-–—/]\s*\d{1,2}[aа]?(?=$|[^\p{L}\p{N}_])/iu;
144
+
145
+ function isUmbrellaAreaPrefixOfNumberedBlock(value, candidate) {
146
+ if (candidate.type !== 'local_area') return false;
147
+ if (NUMBERED_LOCAL_AREA_NAME_RE.test(String(candidate.name || ''))) return false;
148
+ return NUMBERED_CHILD_SUFFIX_RE.test(value.slice(candidate.end));
149
+ }
150
+
151
+ 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)`;
152
+ const residentialMarkedAliasCache = new WeakMap();
153
+
154
+ function stripResidentialMarker(value) {
155
+ return String(value || '')
156
+ .replace(new RegExp(`^\\s*${RESIDENTIAL_MARKER_TEXT}\\s*[:\\-–—]?\\s*`, 'iu'), '')
157
+ .replace(new RegExp(`\\s*${RESIDENTIAL_MARKER_TEXT}\\s*$`, 'iu'), '')
158
+ .trim();
159
+ }
160
+
161
+ function residentialMarkedAliasRegex(item) {
162
+ if (residentialMarkedAliasCache.has(item)) return residentialMarkedAliasCache.get(item);
163
+ const aliases = [...new Set([item?.name, ...(item?.aliases || [])]
164
+ .map(stripResidentialMarker)
165
+ .filter(Boolean))];
166
+ const re = aliases.length ? aliasesToRegex(aliases) : null;
167
+ residentialMarkedAliasCache.set(item, re);
168
+ return re;
169
+ }
170
+
171
+ function markedResidentialMatch(value, item) {
172
+ const aliasRe = residentialMarkedAliasRegex(item);
173
+ if (!aliasRe) return null;
174
+ const markerRe = new RegExp(`(?:^|[^\\p{L}\\p{N}_])${RESIDENTIAL_MARKER_TEXT}\\s*[:\\-–—]?\\s*`, 'giu');
175
+ for (const marker of value.matchAll(markerRe)) {
176
+ const tailStart = (marker.index ?? 0) + marker[0].length;
177
+ const tail = value.slice(tailStart, Math.min(value.length, tailStart + 96));
178
+ const aliasMatch = tail.match(aliasRe);
179
+ if (!aliasMatch) continue;
180
+ const offset = aliasMatch.index ?? 0;
181
+ if (offset > 0 && /\S/u.test(tail.slice(0, offset))) continue;
182
+ return { 0: aliasMatch[0], index: tailStart + offset };
183
+ }
184
+ return null;
185
+ }
186
+
187
+ const LOCATION_CLAUSE_BOUNDARY_RE = /[\n\r.!?;|]/u;
188
+ 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*)?)$/iu;
189
+ const NEARBY_LOCATION_PREFIX_RE = /(?:^|[^\p{L}\p{N}_])(?:до|около|возле|рядом(?:\s+(?:с|со))?|недалеко\s+от|неподал[её]ку\s+от|вблизи|напротив|ориентир(?:ом)?|near|nearby|close\s+to|next\s+to|opposite|landmark|yaqin(?:ida)?|yonida|mo['’ʻ]?ljal|яқин(?:ида)?|ёнида|мўлжал|жақын|жанында|маңында)(?=$|[^\p{L}\p{N}_])[\s\S]{0,96}$/iu;
190
+ const DISTANCE_LOCATION_PREFIX_RE = /(?:\d+(?:[.,]\d+)?\s*(?:мин(?:ут\p{L}*)?|min(?:ute)?s?|daqiqa|м|m|метр\p{L}*|км|km)\s*(?:от|до|from|to)\s+)[\s\S]{0,80}$/iu;
191
+ 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;
192
+
193
+ function clauseStart(value, start) {
194
+ for (let index = start - 1; index >= 0; index -= 1) {
195
+ if (LOCATION_CLAUSE_BOUNDARY_RE.test(value[index])) return index + 1;
196
+ }
197
+ return 0;
198
+ }
199
+
200
+ function clauseEnd(value, end) {
201
+ for (let index = end; index < value.length; index += 1) {
202
+ if (LOCATION_CLAUSE_BOUNDARY_RE.test(value[index])) return index;
203
+ }
204
+ return value.length;
205
+ }
206
+
207
+ function locationMentionRole(value, candidate) {
208
+ const start = clauseStart(value, candidate.start);
209
+ const end = clauseEnd(value, candidate.end);
210
+ const before = value.slice(start, candidate.start);
211
+ const after = value.slice(candidate.end, end);
212
+
213
+ // A direct subject marker later in the same clause wins over an unrelated
214
+ // earlier temporal "до": "до пятницы квартира в ЖК X" is still ЖК X itself.
215
+ if (PRIMARY_LOCATION_PREFIX_RE.test(before)) return 'primary';
216
+ if (NEARBY_LOCATION_PREFIX_RE.test(before) || DISTANCE_LOCATION_PREFIX_RE.test(before) || NEARBY_LOCATION_SUFFIX_RE.test(after)) {
217
+ return 'nearby';
218
+ }
219
+ return 'mentioned';
220
+ }
221
+
142
222
  function publicMatch(candidate) {
143
223
  const { start, end, matchedText, explicitContext, ...result } = candidate;
144
224
  return Object.freeze(result);
@@ -149,7 +229,8 @@ function findEntryMatches(text, cityName, data) {
149
229
  const raw = [];
150
230
  for (const key of LOCATION_LIST_KEYS) {
151
231
  for (const item of data?.[key] || []) {
152
- const match = value.match(item?.re);
232
+ const match = value.match(item?.re)
233
+ || (key === 'residentialComplexes' ? markedResidentialMatch(value, item) : null);
153
234
  if (!match) continue;
154
235
  const { start, end } = semanticBounds(value, match);
155
236
  const candidate = {
@@ -167,8 +248,11 @@ function findEntryMatches(text, cityName, data) {
167
248
  end,
168
249
  matchedText: value.slice(start, end),
169
250
  explicitContext: false,
251
+ role: 'mentioned',
170
252
  };
253
+ if (isUmbrellaAreaPrefixOfNumberedBlock(value, candidate)) continue;
171
254
  candidate.explicitContext = hasExplicitContext(value, candidate);
255
+ candidate.role = locationMentionRole(value, candidate);
172
256
  raw.push(candidate);
173
257
  }
174
258
  }
@@ -51,6 +51,9 @@ export const UZ_SEARCH_TARGETS = Object.freeze([
51
51
  city("Ellikqal'a", { uzLatn: ["Ellikqal'a", 'Ellikkala'], kaaLat: ['Elliqala'], ru: ['Элликкала'], en: ['Ellikkala'] }, { country: 'UZ', type: 'search_target', priority: 'P4', region: "Qoraqalpog'iston" }),
52
52
  ]);
53
53
 
54
+ /** Canonical Kyrgyzstan city catalog used by geography consumers. */
55
+ export const KG_CITY_CATALOG = CITIES_BY_COUNTRY.KG;
56
+
54
57
  export const KZ_LOCATION_TERMS = Object.freeze({
55
58
  microdistrict: Object.freeze(['микрорайон', 'микр.', 'микр', 'мкр.', 'мкр', 'м-н', 'мкр-н', 'ықшамаудан', 'ықш.ауд.', 'ықш']),
56
59
  residentialArea: Object.freeze(['жилой массив', 'жилмассив', 'ж/м', 'ж.м.', 'тұрғын алабы', 'тұрғын массиві', 'массив']),
@@ -66,6 +69,14 @@ export const UZ_LOCATION_TERMS = Object.freeze({
66
69
  residentialComplex: Object.freeze(['ЖК', 'жк', 'turar joy majmuasi', 'residential complex', 'residence']),
67
70
  });
68
71
 
72
+ export const KG_LOCATION_TERMS = Object.freeze({
73
+ city: Object.freeze(['шаар', 'шаары', 'город', 'г.', 'city']),
74
+ district: Object.freeze(['район', 'району', 'р-н', 'district']),
75
+ microdistrict: Object.freeze(['микрорайон', 'мкр', 'мкр.', 'кичирайон', 'кичи район']),
76
+ residentialArea: Object.freeze(['жилмассив', 'жилой массив', 'массив', 'конуш', 'жаңы конуш']),
77
+ residentialComplex: Object.freeze(['ЖК', 'жк', 'жилой комплекс', 'турак жай комплекси', 'residential complex', 'residence']),
78
+ });
79
+
69
80
  export function canonicalKazakhstanCity(value) {
70
81
  return canonicalCity(value, 'KZ');
71
82
  }
@@ -74,14 +85,27 @@ export function canonicalUzbekistanCity(value) {
74
85
  return canonicalCity(value, 'UZ');
75
86
  }
76
87
 
88
+ export function canonicalKyrgyzstanCity(value) {
89
+ return canonicalCity(value, 'KG');
90
+ }
91
+
77
92
  export function canonicalCentralAsiaCity(value, country = null) {
78
93
  if (country === 'KZ') return canonicalKazakhstanCity(value);
79
94
  if (country === 'UZ') return canonicalUzbekistanCity(value);
80
- return canonicalKazakhstanCity(value) || canonicalUzbekistanCity(value);
95
+ if (country === 'KG') return canonicalKyrgyzstanCity(value);
96
+ return canonicalKazakhstanCity(value)
97
+ || canonicalUzbekistanCity(value)
98
+ || canonicalKyrgyzstanCity(value);
81
99
  }
82
100
 
83
101
  export function centralAsiaCityAliases(canonical, country) {
84
- const catalog = country === 'KZ' ? KZ_CITY_CATALOG : country === 'UZ' ? UZ_CITY_CATALOG : [...KZ_CITY_CATALOG, ...UZ_CITY_CATALOG];
102
+ const catalog = country === 'KZ'
103
+ ? KZ_CITY_CATALOG
104
+ : country === 'UZ'
105
+ ? UZ_CITY_CATALOG
106
+ : country === 'KG'
107
+ ? KG_CITY_CATALOG
108
+ : [...KZ_CITY_CATALOG, ...UZ_CITY_CATALOG, ...KG_CITY_CATALOG];
85
109
  const item = catalog.find((entry) => entry.canonical === canonical);
86
110
  return item ? Object.freeze([...new Set([item.canonical, ...aliasesOf(item)])]) : Object.freeze([]);
87
111
  }
package/src/geo.js CHANGED
@@ -48,7 +48,7 @@ export const TASHKENT_METRO = Object.freeze([
48
48
  station('Hamid Olimjon', 'Хамид Алимджан', 'Hamid Olimjon', 'chilonzor', ['Хамид Олимжон', 'Hamid Alimjan', 'Hamid Alimjon']),
49
49
  station('Amir Temur Xiyoboni', 'Амир Темур Хиёбони', 'Amir Temur Xiyoboni', 'chilonzor', ['Сквер', 'Amir Timur Square', 'Amir Temur']),
50
50
  station('Mustaqillik Maydoni', 'Мустакиллик майдони', 'Mustaqillik Maydoni', 'chilonzor', ['Мустақиллик майдони', 'Площадь Независимости', 'Independence Square']),
51
- station('Paxtakor', 'Пахтакор', 'Paxtakor', 'chilonzor', ['Pakhtakor']),
51
+ station('Paxtakor', 'Пахтакор', 'Paxtakor', 'chilonzor'),
52
52
  station('Xalqlar Dostligi', 'Халклар Дустлиги', 'Xalqlar Dostligi', 'chilonzor', ["Xalqlar Do'stligi", 'Xalqlar Do‘stligi', 'Xalqlar Doʻstligi', 'Халклар Дўстлиги', 'Халқлар Дўстлиги', 'Дружба народов', 'Bunyodkor', 'Бунёдкор']),
53
53
  station('Milliy Bog', 'Миллий Бог', 'Milliy Bog', 'chilonzor', ["Milliy Bog'", 'Milliy Bog‘', 'Milliy Bogʻ', 'Миллий Боғ', 'Yoshlik', 'Ёшлик']),
54
54
  station('Novza', 'Новза', 'Novza', 'chilonzor', ['Хамза', 'Hamza', 'Hamza metro', 'метро Хамза']),
@@ -123,9 +123,9 @@ export const TASHKENT_AREAS = Object.freeze({
123
123
  area('Karakamysh-2/4', ['каракамыш 2 4', 'қорақамиш 2 4', 'qoraqamish 2 4', 'karakamish 2 4']),
124
124
  area('Karakamysh-2/5', ['каракамыш 2 5', 'қорақамиш 2 5', 'qoraqamish 2 5', 'karakamish 2 5']),
125
125
  area('Olympia', ['олимпия', 'olimpiya', 'olympia']), area('Vuzgorodok', ['вузгородок', 'вуз городок', 'vuzgorodok']),
126
- area('Medgorodok', ['медгородок', 'мед городок', 'medgorodok']),
127
- area('Chimbay', ['чимбай', 'chimboy', 'chimbay']),
128
- area('Takhtapul', ['тахтапуль', 'тахтапул', 'taxtapul', 'takhtapul']),
126
+ area('Medgorodok', ['медгородок', 'мед городок', 'medgorodok', 'TashMI-2', 'TashMI 2', 'ToshMI-2', 'ToshMI 2', 'ТашМИ-2', 'ТашМИ 2', 'ТошМИ-2', 'ТошМИ 2', 'новый ташми', 'новый ташми-2', 'янги ташми', 'yangi tashmi']),
127
+ area('Chimbay', ['чимбай', 'чимбой', 'chimboy', 'chimbay']),
128
+ area('Taxtapul', ['тахтапуль', 'тахтапул', 'taxtapul', 'takhtapul', 'Takhtapul']),
129
129
  ]),
130
130
  Mirobod: Object.freeze([
131
131
  area('Hospitalny', ['госпитальный', 'госпиталка', 'hospitalny']),
@@ -140,13 +140,13 @@ export const TASHKENT_AREAS = Object.freeze({
140
140
  area('Traktorsozlar-2', ['тракторсозлар 2', 'traktorsozlar 2', 'ттз 2', 'ttz 2']),
141
141
  area('TTZ-3', ['ттз 3', 'ttz 3']),
142
142
  area('Traktorsozlar-4', ['тракторсозлар 4', 'traktorsozlar 4', 'ттз 4', 'ttz 4']),
143
- area('Yalangach', ['ялангач', 'yalangach', "yalang'och"]), area('Feruza', ['феруза', 'feruza']), area('Geofizika', ['геофизика', 'поселок геофизиков', 'geofizika']),
143
+ area('Yalangach', ['ялангач', 'ялангач массив', 'yalangach', "yalang'och", "yalang'och dahasi", 'высоковольтный', 'массив высоковольтный']), area('Feruza', ['феруза', 'feruza']), area('Geofizika', ['геофизика', 'поселок геофизиков', 'geofizika']),
144
144
  ]),
145
145
  Uchtepa: Object.freeze([
146
146
  area('Lolazor', ['лолазор', 'lolazor']),
147
147
  ]),
148
148
  Sergeli: Object.freeze([
149
- area('Yangi Sergeli', ['янги сергели', 'yangi sergeli']), area('Stroygorod', ['стройгород', 'stroygorod']),
149
+ area('Yangi Sergeli', ['янги сергели', 'yangi sergeli']),
150
150
  area('Uzgarish', ['узгарыш', 'ўзгариш', "o'zgarish", 'uzgarish']),
151
151
  ]),
152
152
  Chilanzar: Object.freeze([
@@ -176,4 +176,4 @@ export const TASHKENT_AREAS = Object.freeze({
176
176
  ...[1, 2, 3, 4].map((n) => area(`Tuzel-${n}`, [`тузель ${n}`, `tuzel ${n}`])),
177
177
  area('Tashselmash', ['ташсельмаш', 'tashselmash']), area('Alimkent', ['алимкент', 'olimkent', 'alimkent']), area('Shohimardon', ['шохимардон', 'shohimardon']),
178
178
  ]),
179
- });
179
+ });
@@ -1,6 +1,6 @@
1
1
  import { COUNTRIES, canonicalCountryCode, countryByCode } from './countries.js';
2
2
  import { CITIES, REGIONS } from './geography.js';
3
- import { dictionaryFor } from './locations.js';
3
+ import { dictionaryFor } from './locations-runtime.js';
4
4
  import { aliasesOf } from './normalization.js';
5
5
 
6
6
  // Centralized presentation names for canonical geography values.
@@ -1,5 +1,5 @@
1
1
  import { geographyDisplayName } from './geography-display.js';
2
- import { dictionaryFor } from './locations.js';
2
+ import { dictionaryFor } from './locations-runtime.js';
3
3
  import { aliasesOf, findCanonical } from './normalization.js';
4
4
  import { canonicalCountryCode } from './countries.js';
5
5
 
package/src/geography.js CHANGED
@@ -57,10 +57,19 @@ const RO_CITIES_BASE = Object.freeze([
57
57
  export const RO_CITIES = Object.freeze([...RO_CITIES_BASE, ...RO_CITY_EXTENSIONS]);
58
58
 
59
59
  export const KG_CITIES = Object.freeze([
60
- entity('Bishkek', { ru: ['Бишкек'], en: ['Bishkek'], kk: ['Бішкек'], uzLatn: ['Bishkek'] }, { country: 'KG', type: 'city' }),
61
- entity('Osh', { ru: ['Ош'], en: ['Osh'], kk: ['Ош'], uzLatn: ['Osh'] }, { country: 'KG', type: 'city' }),
62
- entity('Karakol', { ru: ['Каракол'], en: ['Karakol'], kk: ['Қаракөл'], uzLatn: ['Karakol'] }, { country: 'KG', type: 'city' }),
63
-
60
+ entity('Bishkek', { ky: ['Бишкек'], ru: ['Бишкек'], en: ['Bishkek'], kk: ['Бішкек'], uzLatn: ['Bishkek'] }, { country: 'KG', type: 'city' }),
61
+ entity('Osh', { ky: ['Ош'], ru: ['Ош'], en: ['Osh'], kk: ['Ош'], uzLatn: ['Osh'] }, { country: 'KG', type: 'city' }),
62
+ entity('Jalal-Abad', { ky: ['Жалал-Абад', 'Жалал Абад'], ru: ['Джалал-Абад', 'Джалал Абад'], en: ['Jalal-Abad', 'Jalalabad'] }, { country: 'KG', type: 'city' }),
63
+ entity('Karakol', { ky: ['Каракол'], ru: ['Каракол', 'Пржевальск'], en: ['Karakol', 'Przhevalsk'], kk: ['Қаракөл'], uzLatn: ['Karakol'] }, { country: 'KG', type: 'city' }),
64
+ entity('Tokmok', { ky: ['Токмок'], ru: ['Токмок'], en: ['Tokmok'] }, { country: 'KG', type: 'city' }),
65
+ entity('Naryn', { ky: ['Нарын'], ru: ['Нарын'], en: ['Naryn'] }, { country: 'KG', type: 'city' }),
66
+ entity('Talas', { ky: ['Талас'], ru: ['Талас'], en: ['Talas'] }, { country: 'KG', type: 'city' }),
67
+ entity('Batken', { ky: ['Баткен'], ru: ['Баткен'], en: ['Batken'] }, { country: 'KG', type: 'city' }),
68
+ entity('Kara-Balta', { ky: ['Кара-Балта', 'Кара Балта'], ru: ['Кара-Балта', 'Кара Балта'], en: ['Kara-Balta', 'Kara Balta'] }, { country: 'KG', type: 'city' }),
69
+ entity('Balykchy', { ky: ['Балыкчы'], ru: ['Балыкчи'], en: ['Balykchy', 'Balykchi'] }, { country: 'KG', type: 'city' }),
70
+ entity('Kant', { ky: ['Кант'], ru: ['Кант'], en: ['Kant'] }, { country: 'KG', type: 'city' }),
71
+ entity('Uzgen', { ky: ['Өзгөн', 'Озгон'], ru: ['Узген', 'Озгон'], en: ['Uzgen', 'Özgön', 'Ozgon'] }, { country: 'KG', type: 'city' }),
72
+ entity('Kyzyl-Kiya', { ky: ['Кызыл-Кыя', 'Кызыл Кыя'], ru: ['Кызыл-Кия', 'Кызыл Кия'], en: ['Kyzyl-Kiya', 'Kyzyl Kiya'] }, { country: 'KG', type: 'city' }),
64
73
  ]);
65
74
 
66
75
  /** Canonical cities used across global hiring and housing ingestion. */
@@ -21,12 +21,12 @@ export function detectHiringSeniority(value) {
21
21
  ['director', /\b(?:engineering\s+)?director\b|директор/i],
22
22
  ['principal', /\bprincipal\b(?=[^,;\n]{0,60}\b(?:engineer|developer|architect)\b)/i],
23
23
  ['staff', /\bstaff\b(?=[^,;\n]{0,60}\b(?:engineer|developer|architect)\b)/i],
24
- ['lead', /\b(?:team\s*lead|tech\s*lead|lead\s+(?:engineer|developer|frontend|backend))\b|тимлид|техлид|ведущ\w*/i],
24
+ ['lead', /\b(?:team\s*lead|tech\s*lead|lead\s+(?:engineer|developer|frontend|backend))\b|тимлид|техлид|ведущ\w*|yetakchi\s+(?:dasturchi|mutaxassis|muhandis)|жетекші\s+(?:бағдарламашы|маман|инженер)/iu],
25
25
  ['manager', /\bengineering manager\b|руководител/i],
26
- ['senior', /\bsenior\b(?=[^,;\n]{0,60}\b(?:engineer|developer|frontend|backend|software)\b)|сеньор|старш\w*\s+(?:разработ|инженер)/i],
27
- ['middle', /\b(?:middle|mid[- ]?level)\b(?=[^,;\n]{0,60}\b(?:engineer|developer|frontend|backend|software)\b)|мидл/i],
28
- ['junior', /\bjunior\b(?=[^,;\n]{0,60}\b(?:engineer|developer|frontend|backend|software)\b)|джун\w*|младш\w*/i],
29
- ['intern', /\b(?:intern|trainee)\b|стаж[ёе]р|стажир/i],
26
+ ['senior', /\bsenior\b(?=[^,;\n]{0,60}\b(?:engineer|developer|frontend|backend|software)\b)|сеньор|старш\w*\s+(?:разработ|инженер)|katta\s+(?:dasturchi|mutaxassis|muhandis)|аға\s+(?:бағдарламашы|маман|инженер)/iu],
27
+ ['middle', /\b(?:middle|mid[- ]?level)\b(?=[^,;\n]{0,60}\b(?:engineer|developer|frontend|backend|software)\b)|мидл|o['’ʻ]?rta\s+(?:dasturchi|mutaxassis|muhandis)|орта\s+(?:бағдарламашы|маман|инженер)/iu],
28
+ ['junior', /\bjunior\b(?=[^,;\n]{0,60}\b(?:engineer|developer|frontend|backend|software)\b)|джун\w*|младш\w*|kichik\s+(?:dasturchi|mutaxassis|muhandis)|кіші\s+(?:бағдарламашы|маман|инженер)/iu],
29
+ ['intern', /\b(?:intern|trainee)\b|стаж[ёе]р|стажир|amaliyotchi|stajyor|тәжірибеден\s+өтуші|стажер/iu],
30
30
  ];
31
31
  return checks.find(([, pattern]) => pattern.test(text))?.[0] || null;
32
32
  }
@@ -111,7 +111,11 @@ export function parseHiringSalaryWithContext(value, options = {}) {
111
111
  return withContext(parseSalary(value), value, options);
112
112
  }
113
113
 
114
- const VACANCY_COMPENSATION_RE = /(?:salary|salary\s+range|base\s+pay|pay\s+range|annual\s+pay|compensation(?:\s+range)?|заработн\p{L}*\s+плат\p{L}*|зарплат\p{L}*|з\s*[\/\\.\-]?\s*п(?=$|[^\p{L}\p{N}_])|оклад\p{L}*|вилка\s+оплат\p{L}*|оплата\s+труда|компенсац\p{L}*|ставка|💵|💰)/giu;
114
+ // Include short local-board labels/markers used in real vacancy feeds: `З/п`,
115
+ // `ЗП`, and the money emoji that ish-bor-style cards use before a salary field.
116
+ // The marker is still required for long prose so unrelated large numbers are not
117
+ // promoted to compensation merely because the vacancy happens to be in UZ/KZ.
118
+ const VACANCY_COMPENSATION_RE = /(?:💵|💰|salary|salary\s+range|base\s+pay|pay\s+range|annual\s+pay|compensation(?:\s+range)?|з\s*[/\\.\-]?\s*п\b|заработн\p{L}*\s+плат\p{L}*|зарплат\p{L}*|оклад\p{L}*|вилка\s+оплат\p{L}*|оплата\s+труда|компенсац\p{L}*|ставка)/giu;
115
119
 
116
120
  // AI-recruiting/staffing postings (e.g. Mercor) routinely mention funding,
117
121
  // valuation or revenue figures in the same listing as the actual salary. A
@@ -141,4 +145,4 @@ export function parseHiringVacancySalary(value, options = {}) {
141
145
  if (parsed && (parsed.min != null || parsed.max != null)) return parsed;
142
146
  }
143
147
  return null;
144
- }
148
+ }
@@ -9,6 +9,10 @@ export const HIRING_MONTHS = Object.freeze({
9
9
  iulie: 6, august: 7, septembrie: 8, octombrie: 9, noiembrie: 10, decembrie: 11,
10
10
  січня: 0, лютого: 1, березня: 2, квітня: 3, травня: 4, червня: 5,
11
11
  липня: 6, серпня: 7, вересня: 8, жовтня: 9, листопада: 10, грудня: 11,
12
+ yanvar: 0, fevral: 1, mart: 2, aprel: 3, iyun: 5,
13
+ iyul: 6, avgust: 7, sentabr: 8, oktabr: 9, noyabr: 10, dekabr: 11,
14
+ қаңтар: 0, ақпан: 1, наурыз: 2, сәуір: 3, мамыр: 4, маусым: 5,
15
+ шілде: 6, тамыз: 7, қыркүйек: 8, қазан: 9, қараша: 10, желтоқсан: 11,
12
16
  });
13
17
 
14
18
  export const UNICODE_LEFT_BOUNDARY = '(?<![\\p{L}\\p{N}])';
@@ -8,7 +8,7 @@ const ADDRESS_LABEL_RE = /(?:адрес|адреса|адресація|адре
8
8
  const PREFIX_STREET_MARKER = String.raw`(?:(?:ул(?:ица)?|вул(?:иця)?|пр|просп(?:ект)?|пр-т|переул(?:ок)?|пров(?:улок)?|проезд|наб(?:ережная)?|шоссе|str(?:ada)?|street|st|avenue|ave|road|rd|көше)\.?)`;
9
9
  const POSTFIX_STREET_MARKER = String.raw`(?:ko['’ʼ\u02bc]?cha(?:si)?|кўча(?:си)?|коча(?:си)?|kocha(?:si)?)`;
10
10
  const POSTFIX_STREET_TYPE = String.raw`(?:вулиця|улица|провулок|переулок|проспект|бульвар|набережна|набережная|шосе|шоссе|площа|площадь|узвіз|спуск|алея|аллея|дорога|тупик)`;
11
- const HOUSE_MARKER = String.raw`(?:дом|д\.|будинок|буд\.|house|h\.|uy|уй|nr\.?|no\.?|№)`;
11
+ const HOUSE_MARKER = String.raw`(?:дом|д\.|будинок|буд\.|house|h\.|uy|уй|үй|nr\.?|no\.?|№)`;
12
12
  const BUILDING_MARKER = String.raw`(?:корп(?:ус)?\.?|к\.|строен(?:ие)?|стр\.|будова|секц(?:ия|ія)?|bloc|corp|building|bldg\.?|korpus)`;
13
13
  const NUMBER_TOKEN = String.raw`\d{1,5}(?:[-\/]?[\p{L}])?(?:[\/-]\d{1,4}(?:[-\/]?[\p{L}])?)?`;
14
14
  const STREET_WORD = String.raw`[\p{L}'’.-]{2,48}`;
@@ -39,6 +39,7 @@ const PLACE_KINDS = Object.freeze([
39
39
  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;
40
40
  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}*)$/iu;
41
41
  const NEARBY_FILLER_RE = /(?:инфраструктур\p{L}*|все\s+необходимо|вс[её]\s+рядом|шаговой\s+доступност|развитый\s+район)/iu;
42
+ const QUARTER_NEARBY_PREFIX_RE = /(?:^|[\s,;:—–-])(?:до|рядом\s+(?:с|со)|недалеко\s+от|возле|около|ориентир\s*[:—–-]?|near(?:by)?|close\s+to|lângă|aproape\s+de)\s*$/iu;
42
43
 
43
44
  function enumeratedNearby(text) {
44
45
  const blocks = [...text.matchAll(new RegExp(NEARBY_BLOCK_RE.source, `${NEARBY_BLOCK_RE.flags}g`))]
@@ -61,18 +62,33 @@ function enumeratedNearby(text) {
61
62
  return items;
62
63
  }
63
64
 
65
+ function isNearbyQuarterReference(text, match) {
66
+ const start = match?.index ?? -1;
67
+ if (start < 0) return false;
68
+ const before = text.slice(Math.max(0, start - 64), start);
69
+ return QUARTER_NEARBY_PREFIX_RE.test(before);
70
+ }
71
+
72
+ function firstPrimaryQuarterMatch(text, pattern) {
73
+ const flags = pattern.flags.includes('g') ? pattern.flags : `${pattern.flags}g`;
74
+ for (const match of text.matchAll(new RegExp(pattern.source, flags))) {
75
+ if (!isNearbyQuarterReference(text, match)) return match;
76
+ }
77
+ return null;
78
+ }
79
+
64
80
  export function parseHousingQuarterLabel(value) {
65
81
  const text = String(value || '');
66
82
  if (!text) return null;
67
83
  if (/(?:^|[^\p{L}\p{N}_])(?:глинк[аи]?|glinka)(?:$|[^\p{L}\p{N}_])/iu.test(text)) return 'Glinka';
68
- 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)
69
- || text.match(/(?:квартал|kvartal(?:i)?|квартил|kvartil|мкр|микрорайон|массив|massiv|daha(?:si|dan)?|hudud|худуд)\s*[-№#]?\s*(\d{1,3})/iu)
70
- || text.match(/(?:чиланзар|chilonzor|chilanzar)\s*[-№#]?\s*(\d{1,2})(?!\d)/iu);
84
+ 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)
85
+ || firstPrimaryQuarterMatch(text, /(?:квартал|kvartal(?:i)?|квартил|kvartil|мкр|микрорайон|массив|massiv|daha(?:si|dan)?|hudud|худуд)\s*[-№#]?\s*(\d{1,3})/iu)
86
+ || firstPrimaryQuarterMatch(text, /(?:чиланзар|chilonzor|chilanzar)\s*[-№#]?\s*(\d{1,2})(?!\d)/iu);
71
87
  if (match) {
72
88
  const label = `${Number(match[1])} kvartal`;
73
89
  return STREET_LIKE_PLACE_RE.test(label) ? null : label;
74
90
  }
75
- const centralBlock = text.match(/(?:^|[^\p{L}\p{N}_])(?:ц|c)\s*[-–]?\s*(\d{1,2})(?:$|[^\p{L}\p{N}_])/iu);
91
+ const centralBlock = firstPrimaryQuarterMatch(text, /(?:^|[^\p{L}\p{N}_])(?:ц|c)\s*[-–]?\s*(\d{1,2})(?:$|[^\p{L}\p{N}_])/iu);
76
92
  return centralBlock ? `C-${Number(centralBlock[1])}` : null;
77
93
  }
78
94
 
@@ -24,10 +24,23 @@ const DISPLAY = Object.freeze({
24
24
  'Bobur Park': 'Парк Бобура', 'Independence Square': 'Площадь Независимости', 'Mega Planet': 'Mega Planet',
25
25
  'Bek Baraka': 'Бек-Барака', Khadra: 'Хадра',
26
26
  }),
27
+ uk: Object.freeze({
28
+ Park: 'Парк', Metro: 'Метро', 'Bus stop': 'Автобусна зупинка', 'Public transport': 'Громадський транспорт',
29
+ 'Main road': 'Головна дорога', Clinic: 'Клініка', 'Maternity hospital': 'Пологовий будинок', Hospital: 'Лікарня',
30
+ School: 'Школа', Kindergarten: 'Дитячий садок', Childcare: 'Дитячі заклади', University: 'Університет',
31
+ 'Shopping center': 'Торговий центр', Shop: 'Магазин', Korzinka: 'Korzinka', Supermarket: 'Супермаркет',
32
+ Market: 'Ринок', Cafe: 'Кафе', Restaurant: 'Ресторан', Playground: 'Дитячий майданчик', Pharmacy: 'Аптека',
33
+ Mosque: 'Мечеть', Church: 'Церква', 'Railway station': 'Залізничний вокзал', Airport: 'Аеропорт',
34
+ 'Bobur Park': 'Парк Бабура', 'Independence Square': 'Площа Незалежності', 'Mega Planet': 'Mega Planet',
35
+ 'Bek Baraka': 'Бек-Барака', Khadra: 'Хадра',
36
+ }),
27
37
  });
28
38
 
29
39
  function languageKey(locale) {
30
- return String(locale || 'en').toLowerCase().startsWith('ru') ? 'ru' : 'en';
40
+ const key = String(locale || 'en').toLowerCase();
41
+ if (key.startsWith('uk')) return 'uk';
42
+ if (key.startsWith('ru')) return 'ru';
43
+ return 'en';
31
44
  }
32
45
 
33
46
  function exactPoi(value) {
@@ -3,6 +3,7 @@ import { findAllCanonical, normalizeUnicode } from './normalization.js';
3
3
  import { GENERIC_LANDMARK_TERMS } from './landmarks.js';
4
4
  import { matchTashkentHousingDistrict, matchTashkentHousingMetro, matchTashkentHousingQuarter } from './tashkent-housing-geography.js';
5
5
  import { TASHKENT_LANDMARKS } from './tashkent-pois.js';
6
+ import { TASHKENT_RESIDENTIAL_COMPLEXES, matchTashkentResidentialComplex } from './tashkent-residential-complexes.js';
6
7
  import { parseHousingRoomCount, parseHousingFloor, parseHousingAreas } from './housing-structured.js';
7
8
  import { parseHousingListingFields } from './housing-listing-fields.js';
8
9
  import { parseHousingResidentialComplex } from './housing-text.js';
@@ -26,18 +27,21 @@ const APPLIANCE_PATTERNS = Object.freeze([
26
27
  ]);
27
28
 
28
29
  const FIRST_RENT_UZ_RE = /(?:hali\s+hech\s+kim\s+(?:yashamagan|turmagan)|ҳали\s+ҳеч\s+ким\s+(?:яшамаган|турмаган))/iu;
29
- const LANDLORD_PRESENT_RE = /(?:xozaykali|hojaykali|xo['’]?jaykali|с\s+хозяйк(?:ой|ой\s+в\s+квартире)|хозяйк\p{L}*\s+(?:жив[её]т|прожива\p{L}*))/iu;
30
- const STUDENT_RE = /(?:studentlar\s+uchun|talabalar\s+uchun|студент(?:ам|ы|ок|ов)?\s+(?:можно|для)|для\s+студент)/iu;
31
- const NO_BROKER_RE = /(?:bez\s+makler|maklersiz|vositachisiz|без\s+(?:маклер|посредник|риелтор|риэлтор|комисси))/iu;
32
- const BROKER_RE = /(?:makler|vositachi|макл(?:ер[а-яё]*)?|ри[еэ]лтор[а-яё]*|агентств[а-яё]*|комисси[а-яё]*|broker|realtor|commission)/iu;
33
- const MEN_RE = /(?:o['’ʻʼ‘`]?g['’ʻʼ‘`]?il\s+bola(?:lar)?(?:ga)?|ogil\s+bola(?:lar)?(?:ga)?|sherik\s+bola|эркак(?:лар)?|erkak(?:lar)?(?:ga)?|только\s+(?:мужчин|парн))/iu;
34
- const WOMEN_RE = /(?:qiz(?:lar)?(?:ga)?|ayol(?:lar)?(?:ga|ni)?|киз(?:лар)?(?:га)?|аёл(?:лар)?(?:га|ни)?|девушк\p{L}*|женщин\p{L}*|girls?\s+only|women\s+only)/iu;
35
- const FAMILY_RE = /(?:семь\p{L}*|family|oila(?:ga|lar|li)?|oila\s+uchun|оилага|оелага|оилавий|oelaga)/iu;
36
- const ROOM_SHARE_RE = /(?:sherik(?:ka|lik|likga)?|шерик(?:ка|лик)?|roommate|flatmate|подселени|койко[-\s]?мест|место\s+в\s+(?:комнат|квартир)|birga\s+yashash(?:ga)?|kvartira(?:ga|da)?[^\r\n.!?]{0,36}(?:\d+|bitta|1)\s*(?:ta\s*)?(?:qiz|ayol)[^\r\n.!?]{0,20}(?:ijarachi\s*)?(?:kerak|kere))/iu;
30
+ const LANDLORD_PRESENT_RE = /(?:xozaykali|hojaykali|xo['’]?jaykali|с\s+хозяйк(?:ой|ой\s+в\s+квартире)|хозяйк\p{L}*\s+(?:жив[её]т|прожива\p{L}*)|with\s+(?:the\s+)?(?:landlord|owner)\s+(?:present|living\s+in)|cu\s+proprietar(?:ul)?\s+în\s+cas(?:ă|a)|үй\s*иесі\s+(?:тұрады|бірге\s+тұрады))/iu;
31
+ const STUDENT_RE = /(?:studentlar\s+uchun|talabalar\s+uchun|студент(?:ам|ы|ок|ов)?\s+(?:можно|для)|для\s+студент|students?\s+(?:only|welcome)|for\s+students|pentru\s+studen[țt]i|studen[țt]i(?:lor)?|студенттерге|студенттер\s+үшін)/iu;
32
+ const NO_BROKER_RE = /(?:bez\s+makler|maklersiz|vositachisiz|без\s+(?:маклер|посредник|риелтор|риэлтор|комисси)|no\s+(?:broker|agent|commission|agency\s+fee)|f[ăa]r[ăa]\s+(?:comision|agen[țt]ie|intermediari)|делдалсыз|комиссиясыз)/iu;
33
+ const BROKER_RE = /(?:makler|vositachi|макл(?:ер[а-яё]*)?|ри[еэ]лтор[а-яё]*|агентств[а-яё]*|комисси[а-яё]*|broker|realtor|commission|comision(?:ul)?|agen[țt]ie|delda[lл]\p{L}*|делдал\p{L}*)/iu;
34
+ const MEN_RE = /(?:o['’ʻʼ‘`]?g['’ʻʼ‘`]?il\s+bola(?:lar)?(?:ga)?|ogil\s+bola(?:lar)?(?:ga)?|sherik\s+bola|эркак(?:лар)?|erkak(?:lar)?(?:ga)?|только\s+(?:мужчин|парн)|\bmen\s+only\b|b[ăa]rba[țt]i(?:lor)?|b[ăa]ie[țt]i(?:lor)?|жігіттерге|жігіттер(?:ге)?|хлопц(?:ям|і|ів)?|чоловік(?:ам|и)?)/iu;
35
+ const WOMEN_RE = /(?:qiz(?:lar)?(?:ga)?|ayol(?:lar)?(?:ga|ni)?|киз(?:лар)?(?:га)?|аёл(?:лар)?(?:га|ни)?|девушк\p{L}*|женщин\p{L}*|girls?\s+only|women\s+only|fete(?:lor)?|femei(?:lor)?|қыздарға|қыздар(?:ға)?|дівчат(?:ам|а|ок)?|жінк(?:ам|и)?)/iu;
36
+ const FAMILY_RE = /(?:семь\p{L}*|family|oila(?:ga|lar|li)?|oila\s+uchun|оилага|оелага|оилавий|oelaga|famil(?:ie|ia)|cuplu(?:ri)?|отбасына|отбасылы|жанұяға|сім['’ʼ]?[яїі](?:ям|ям[иі])?|сімейн\p{L}*)/iu;
37
+ const ROOM_SHARE_RE = /(?:sherik(?:ka|lik|likga)?|шерик(?:ка|лик)?|roommate|flatmate|подселени|койко[-\s]?мест|место\s+в\s+(?:комнат|квартир)|birga\s+yashash(?:ga)?|kvartira(?:ga|da)?[^\r\n.!?]{0,36}(?:\d+|bitta|1)\s*(?:ta\s*)?(?:qiz|ayol)[^\r\n.!?]{0,20}(?:ijarachi\s*)?(?:kerak|kere)|coleg\s+de\s+(?:apartament|camer[ăa])|bed\s+space|бөлмелес(?:\s+керек)?|көрші\s+керек)/iu;
37
38
  const AIR_CONDITIONER_RE = /(?:кондицион|air\s*con|konditsioner|kandit(?:s|c)?aner|kanditsaner|кандитсанер)/iu;
38
39
  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;
39
40
  const WALK_MINUTES_RE = /(?:yayov|piyoda|пешком)\s*(\d{1,2})\s*(?:daqiqa|min(?:ute)?s?|минут)/iu;
40
41
  const TRANSIT_ROUTES_RE = /(?:aftobuslar|avtobuslar|автобуслар|автобусы)[^\r\n\d]{0,24}((?:\d{1,4}[\s,;/]*){1,10})/iu;
42
+ 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;
43
+ 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;
44
+ 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;
41
45
 
42
46
  function unique(values) {
43
47
  return [...new Set(values.filter(Boolean))];
@@ -47,6 +51,42 @@ function categoryOf(entry) {
47
51
  return entry?.category || GENERIC_CATEGORY[entry?.canonical] || 'landmark';
48
52
  }
49
53
 
54
+ function nearbyReferenceRanges(text) {
55
+ const ranges = [];
56
+ for (const pattern of [NEARBY_RELATION_TAIL_RE, NEARBY_TRAVEL_TAIL_RE]) {
57
+ const regex = new RegExp(pattern.source, pattern.flags);
58
+ for (const match of text.matchAll(regex)) {
59
+ const start = match.index ?? 0;
60
+ ranges.push({ start, end: start + match[0].length });
61
+ }
62
+ }
63
+ return ranges;
64
+ }
65
+
66
+ function insideNearbyReference(match, ranges) {
67
+ return ranges.some((range) => match.start >= range.start && match.end <= range.end);
68
+ }
69
+
70
+ function withoutNearbyLocationReferences(text) {
71
+ return String(text || '')
72
+ .replace(NEARBY_RELATION_TAIL_RE, ' ')
73
+ .replace(NEARBY_TRAVEL_TAIL_RE, ' ');
74
+ }
75
+
76
+ function hasResidentialContext(text, match) {
77
+ if (!match.entry?.ambiguous) return true;
78
+ const context = text.slice(Math.max(0, match.start - 36), Math.min(text.length, match.end + 36));
79
+ return RESIDENTIAL_CONTEXT_RE.test(context);
80
+ }
81
+
82
+ function residentialComplexNearbyMatches(text) {
83
+ const ranges = nearbyReferenceRanges(text);
84
+ if (!ranges.length) return [];
85
+ return findAllCanonical(text, TASHKENT_RESIDENTIAL_COMPLEXES)
86
+ .filter((match) => insideNearbyReference(match, ranges) && hasResidentialContext(text, match))
87
+ .map((match) => ({ canonical: match.canonical, category: 'residential_complex', start: match.start }));
88
+ }
89
+
50
90
  function extensionMatches(text) {
51
91
  return findAllCanonical(text, HOUSING_POI_EXTENSIONS).map((match) => ({
52
92
  canonical: match.canonical,
@@ -83,7 +123,7 @@ function genericMatches(text) {
83
123
  export function parseHousingNearby(value) {
84
124
  const text = normalizeUnicode(value ?? '');
85
125
  if (!text) return deepFreeze([]);
86
- const specific = [...extensionMatches(text), ...catalogPoiMatches(text)].sort((a, b) => a.start - b.start);
126
+ const specific = [...extensionMatches(text), ...catalogPoiMatches(text), ...residentialComplexNearbyMatches(text)].sort((a, b) => a.start - b.start);
87
127
  const specificCategories = new Set(specific.map((item) => item.category));
88
128
  const generic = genericMatches(text)
89
129
  .filter((item) => !specificCategories.has(item.category))
@@ -201,7 +241,10 @@ export function parseHousingListingEnrichment(value, { country = '' } = {}) {
201
241
  const quarter = matchTashkentHousingQuarter(text);
202
242
  const district = matchTashkentHousingDistrict(text)?.name || quarter?.district || null;
203
243
  const metro = matchTashkentHousingMetro(text)?.name || null;
204
- const parsedRc = specificResidentialComplex(text) || parseHousingResidentialComplex(text);
244
+ const primaryResidentialText = withoutNearbyLocationReferences(text);
245
+ const parsedRc = specificResidentialComplex(primaryResidentialText)
246
+ || matchTashkentResidentialComplex(primaryResidentialText)?.name
247
+ || parseHousingResidentialComplex(primaryResidentialText);
205
248
  const commission = parseHousingCommission(text);
206
249
  const commissionAmount = parseHousingCommissionAmount(text);
207
250
  const address = parseHousingAddress(text);
@@ -271,4 +271,4 @@ export function parseHousingPrice(value, fallbackCurrency = '') {
271
271
  }
272
272
 
273
273
  // Compatibility name for callers migrating from Flat Finder's local parser.
274
- export const parsePriceFromText = parseHousingPrice;
274
+ export const parsePriceFromText = parseHousingPrice;
@@ -108,6 +108,12 @@ export const HOUSING_POI_EXTENSIONS = Object.freeze([
108
108
  uzLatn: ['Assalom Jomiy', 'Assalom Jomiy turar joy majmuasi'],
109
109
  uzCyrl: ['Ассалом Жомий'],
110
110
  }, { category: 'residential_complex', display: { ru: 'ЖК «Ассалом Жомий»', en: 'Assalom Jomiy' } }),
111
+ group('Assalom Sohil', {
112
+ ru: ['Ассалом Сохил', 'ЖК Ассалом Сохил', 'жилой комплекс Ассалом Сохил'],
113
+ en: ['Assalom Sohil', 'Assalom Sohil residential complex'],
114
+ uzLatn: ['Assalom Sohil', 'Assalom Sohil turar joy majmuasi'],
115
+ uzCyrl: ['Ассалом Сохил'],
116
+ }, { category: 'residential_complex', display: { ru: 'ЖК «Ассалом Сохил»', en: 'Assalom Sohil' } }),
111
117
  group('Jomiy Roundabout', {
112
118
  ru: ['круг Жомий', 'круг жомий', 'кольцо Жомий'],
113
119
  en: ['Jomiy Roundabout', 'Jomiy circle'],
@@ -10,6 +10,7 @@ export const HOUSING_DEAL_TYPE_EXTENSIONS = Object.freeze([
10
10
  group('longRent', {
11
11
  ru: ['снять', 'в месяц', 'месяц'],
12
12
  en: ['monthly'],
13
+ uk: ['зняти', 'на місяць', 'місяць'],
13
14
  uzLatn: ['ijaraga', 'oyiga', 'beriladi'],
14
15
  uzCyrl: ['ижарага', 'ойига', 'берилади'],
15
16
  kk: ['айына'],
@@ -17,6 +18,7 @@ export const HOUSING_DEAL_TYPE_EXTENSIONS = Object.freeze([
17
18
  group('shortRent', {
18
19
  ru: ['сутки', 'суток', 'суточно'],
19
20
  en: ['daily'],
21
+ uk: ['доба', 'діб', 'подобово'],
20
22
  uzLatn: ['kunlik', 'sutkalik'],
21
23
  uzCyrl: ['кунлик', 'суткалик'],
22
24
  kk: ['тәуліктік'],
@@ -27,6 +29,7 @@ export const HOUSING_ROOM_ONLY_EXTENSIONS = Object.freeze([
27
29
  group('roomOnly', {
28
30
  ru: ['шеринг'],
29
31
  en: ['room only'],
32
+ uk: ['шерінг'],
30
33
  uzLatn: ['student qizlarga', 'talaba qizlarga', 'opshijit dom', 'obshijit dom'],
31
34
  uzCyrl: ['студент қизларга', 'талаба қизларга'],
32
35
  }),
@@ -84,6 +84,23 @@ export function parseHousingFloor(value) {
84
84
  const text = normalizeUnicode(value ?? '');
85
85
  if (!text) return deepFreeze({ floor: null, totalFloors: null });
86
86
 
87
+ // Common classifieds shorthand is rooms/floor/total floors, e.g. 2/10/16.
88
+ // It must be checked before the generic floor/total fraction parser or the
89
+ // first two numbers would incorrectly become floor=2, totalFloors=10.
90
+ // Requiring the triple to occupy its own line avoids treating dates as this
91
+ // housing-specific shorthand inside ordinary prose.
92
+ const compactTriple = text.match(/(?:^|[\r\n])\s*(\d{1,2})\s*[\/\\]\s*(\d{1,3})\s*[\/\\]\s*(\d{1,3})\s*(?=$|[\r\n])/u);
93
+ if (compactTriple) {
94
+ const rooms = toNumber(compactTriple[1]);
95
+ const floor = toNumber(compactTriple[2]);
96
+ const totalFloors = toNumber(compactTriple[3]);
97
+ if (rooms != null && rooms >= 1 && rooms <= 20
98
+ && floor != null && totalFloors != null
99
+ && floor >= 1 && floor <= totalFloors && totalFloors <= 200) {
100
+ return deepFreeze({ floor, totalFloors });
101
+ }
102
+ }
103
+
87
104
  const fraction = text.match(/(?:^|[^\d])(\d{1,3})\s*[\/\\]\s*(\d{1,3})(?=$|[^\d])/u);
88
105
  if (fraction) {
89
106
  const floor = toNumber(fraction[1]);
@@ -142,7 +159,7 @@ const AREA_LABELS = Object.freeze([
142
159
  ['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],
143
160
  ]);
144
161
 
145
- const AREA_UNIT_RE = String.raw`(?:м²|м2|m²|m2|sqm|sq\.?\s*m|mp|кв\.?\s*м|квадрат\p{L}*)`;
162
+ const AREA_UNIT_RE = String.raw`(?:м²|м2|m²|m2|sqm|sq\.?\s*m|mp|кв\.?\s*м|кв2|квадрат\p{L}*)`;
146
163
 
147
164
  function areaAfterLabel(text, labelRe) {
148
165
  const re = new RegExp(`${labelRe.source}\\s*[:=-]?\\s*(\\d{1,4}(?:[.,]\\d{1,2})?)\\s*${AREA_UNIT_RE}`, 'iu');