@whiteslove/parsing-lexicon 0.6.15 → 0.6.16

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.6.15",
+  "version": "0.6.16",
   "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
   "repository": {
     "type": "git",

package/src/housing-address.js

@@ -39,6 +39,12 @@ function compactStreet(value) {
     // "проживания" -> "оживания").
     .replace(new RegExp(`^${PREFIX_STREET_MARKER}(?!\\p{L})\\s*`, 'iu'), '')
     .replace(new RegExp(`\\s+${POSTFIX_STREET_MARKER}$`, 'iu'), '')
+    // OCR frequently substitutes “оя” for the Ukrainian/Russian “ля” in
+    // “шлях”; correct the street-token typo before canonical lookup.
+    .replace(/(?<!\p{L})шоях(?!\p{L})/giu, 'шлях')
+    // Listings commonly abbreviate this Kharkiv street to an initial and
+    // surname; expand it so address-only consumers do not retain just “Л”.
+    .replace(/(?<!\p{L})л\s*\.\s*малой(?!\p{L})/giu, 'Любови Малой')
     .replace(/[\s,;:.\-–—]+$/gu, '')
     .trim() || null;
 }

package/src/housing-card-fields.js

@@ -51,7 +51,7 @@ function enumeratedNearby(text) {
   const items = [];
   for (const item of blocks.join(', ').split(/[,;•·|]+/).map((part) => part
     .replace(/\s{2,}/g, ' ')
-    .replace(/^[\s\-–—.]+|[\s\-–—.!]+$/g, '')
+    .replace(/^[\s\-–—."'«»“”„]+|[\s\-–—.!"'«»“”„]+$/g, '')
     .trim())) {
     if (item.length < 3 || item.length > 45 || NEARBY_NOISE_RE.test(item) || NEARBY_FILLER_RE.test(item)) continue;
     if (!/\p{L}{3,}/u.test(item)) continue;

package/src/housing-listing-enrichment.d.ts

@@ -71,4 +71,7 @@ export function parseHousingCommissionAmount(value: unknown): HousingCommissionA
 export function parseHousingPerPersonPrice(value: unknown, options?: { country?: string }): HousingPerPersonPrice | null;
 export function parseHousingTransitRoutes(value: unknown): readonly string[];
 export function parseHousingObservedAmenities(value: unknown): readonly string[];
-export function parseHousingListingEnrichment(value: unknown, options?: { country?: string }): Readonly<HousingListingEnrichment>;
+export function parseHousingListingEnrichment(
+  value: unknown,
+  options?: { country?: string; dealType?: 'sale' | 'longRent' | 'shortRent' | null },
+): Readonly<HousingListingEnrichment>;

package/src/housing-listing-enrichment.js

@@ -9,6 +9,7 @@ import { parseHousingListingFields } from './housing-listing-fields.js';
 import { parseHousingResidentialComplex } from './housing-text.js';
 import { parseHousingAddress } from './housing-address.js';
 import { HOUSING_LANDMARK_EXTENSIONS, HOUSING_POI_EXTENSIONS } from './housing-poi-extensions.js';
+import { resolveHousingIntent } from './housing-intent.js';
 
 const GENERIC_CATEGORY = Object.freeze({
   Park: 'park', Metro: 'metro', 'Bus stop': 'transport', 'Public transport': 'transport', 'Main road': 'transport',
@@ -229,10 +230,11 @@ function walkMinutes(text) {
   return Number.isInteger(value) && value > 0 && value <= 180 ? value : null;
 }
 
-export function parseHousingListingEnrichment(value, { country = '' } = {}) {
+export function parseHousingListingEnrichment(value, { country = '', dealType = null } = {}) {
   const text = normalizeUnicode(value ?? '');
   if (!text) return deepFreeze({});
-  const listingFields = parseHousingListingFields(text, { country });
+  const resolvedDealType = dealType || resolveHousingIntent(text)?.dealType || null;
+  const listingFields = parseHousingListingFields(text, { country, dealType: resolvedDealType });
   const floor = parseHousingFloor(text);
   const areas = parseHousingAreas(text);
   const audience = parseHousingAudience(text);
@@ -271,7 +273,7 @@ export function parseHousingListingEnrichment(value, { country = '' } = {}) {
     euroLayout: listingFields.euroLayout ?? null,
     gas: listingFields.gas ?? null,
     newBuilding: listingFields.newBuilding ?? null,
-    communalSeparated: listingFields.communalSeparated ?? null,
+    communalSeparated: resolvedDealType === 'sale' ? null : (listingFields.communalSeparated ?? null),
     parking: listingFields.parking ?? null,
     elevator: listingFields.elevator ?? null,
     heating: listingFields.heating ?? null,
@@ -284,7 +286,7 @@ export function parseHousingListingEnrichment(value, { country = '' } = {}) {
     furnished: listingFields.furnished ?? null,
     deposit: listingFields.depositRequired ?? null,
     firstRental: firstRental(text, listingFields),
-    utilitiesAmount: listingFields.utilitiesAmount ?? null,
+    utilitiesAmount: resolvedDealType === 'sale' ? null : (listingFields.utilitiesAmount ?? null),
     commission,
     commissionPercent: commission === false ? 0 : null,
     commissionAmount,

package/src/housing-listing-fields.d.ts

@@ -53,5 +53,5 @@ export type HousingListingFields = Readonly<{
  */
 export function parseHousingListingFields(
   value: unknown,
-  options?: { country?: string },
+  options?: { country?: string; dealType?: 'sale' | 'longRent' | 'shortRent' | null },
 ): Readonly<Partial<HousingListingFields>>;

package/src/housing-listing-fields.js

@@ -2,6 +2,7 @@ import { deepFreeze } from './lexicon-core.js';
 import { normalizeUnicode } from './normalization.js';
 import { parseHousingContext } from './housing-context.js';
 import { parseHousingFeatures } from './housing-features.js';
+import { resolveHousingIntent } from './housing-intent.js';
 
 const bool = (text, positive, negative = null) => {
   if (negative?.test(text)) return false;
@@ -69,6 +70,11 @@ function parseAvailableFrom(text) {
   const match = text.match(/(?:доступн\p{L}*\s+с|свободн\p{L}*\s+с|заселени\p{L}*\s+с|заезд\s+с|available\s+from|move[- ]?in\s+from)\s*[:\-]?\s*((?:\d{1,2}[./-]\d{1,2}(?:[./-]\d{2,4})?)|(?:\d{1,2}\s+[\p{L}]{3,12}(?:\s+\d{4})?))/iu);
   if (match?.[1]) return match[1].trim();
 
+  // Ukrainian listings commonly phrase this as “move-in is possible only
+  // from …”, rather than using the shorter “available from …” form.
+  const ukrainianMoveIn = text.match(/(?:перегляд\s+(?:і|та)\s+)?заселен\p{L}*\s+можлив\p{L}*(?:\s+лише)?\s+з\s*((?:\d{1,2}[./-]\d{1,2}(?:[./-]\d{2,4})?)|(?:\d{1,2}\s+[\p{L}]{3,12}(?:\s+\d{4})?))/iu);
+  if (ukrainianMoveIn?.[1]) return ukrainianMoveIn[1].trim();
+
   // Uzbek listings commonly attach the ablative suffix directly to the month:
   // "1-sentyabrdan beriladi" / "1-сентябрдан берилади".
   const uzbek = text.match(/(\d{1,2}\s*[-./]?\s*[\p{L}]{3,12})(?:dan|дан)(?=$|[^\p{L}\p{N}_])[^.\r\n]{0,32}(?:beriladi|берилади|bo['’]?sh|бўш)/iu);
@@ -76,7 +82,9 @@ function parseAvailableFrom(text) {
 }
 
 function parseUtilitiesAmount(text) {
-  const match = text.match(/(?:коммунальн\p{L}*|коммуналк\p{L}*|ком\.?\s*услуг\p{L}*|utilities?|bills?)\D{0,24}(?:около|примерно|~|≈)?\s*(\d{2,8}(?:[.,]\d{1,2})?)\s*(₴|грн|uah|\$|usd|€|eur|сум|uzs|тг|kzt|lei|ron|руб|rub)?/iu);
+  // Do not bridge a sentence boundary. Otherwise “plus utilities. Tel. 06…”
+  // turns the first phone digits into a fictitious utility bill.
+  const match = text.match(/(?:коммунальн\p{L}*|коммуналк\p{L}*|ком\.?\s*услуг\p{L}*|utilities?|bills?)[^\d.;!?…\r\n]{0,24}(?:около|примерно|~|≈)?\s*(\d{2,8}(?:[.,]\d{1,2})?)\s*(₴|грн|uah|\$|usd|€|eur|сум|uzs|тг|kzt|lei|ron|руб|rub)?/iu);
   if (!match) return null;
   const amount = Number(match[1].replace(',', '.'));
   if (!Number.isFinite(amount) || amount <= 0) return null;
@@ -103,11 +111,16 @@ function parseDepositRequired(text) {
   return null;
 }
 
-export function parseHousingListingFields(value, { country = '' } = {}) {
+export function parseHousingListingFields(value, { country = '', dealType = null } = {}) {
   const text = normalizeUnicode(value ?? '');
   const context = parseHousingContext(text);
   const features = parseHousingFeatures(text);
   if (!text) return deepFreeze({});
+  // Utility billing is a rental-only concept. A source may carry a generic
+  // payment-table value on every listing, including sales, so never let that
+  // value become a persisted housing attribute for a sale.
+  const resolvedDealType = dealType || resolveHousingIntent(text)?.dealType || null;
+  const isSale = resolvedDealType === 'sale';
 
   const gas = bool(text,
     /(?:^|[^\p{L}\p{N}_])(?:газ|gaz|gas)(?=$|[^\p{L}\p{N}_])|метан|aragaz|gaz\s+ta['’]?min/iu,
@@ -211,7 +224,7 @@ export function parseHousingListingFields(value, { country = '' } = {}) {
     ),
     gas,
     newBuilding: bool(text, /новостро|новобуд|новый\s+дом|novast(?:royka|iroyka)|navast(?:royka|iroyka)|new\s*build|newly\s+built|yangi\s+(?:bino|qurilgan|uy)|bloc\s+nou/iu),
-    communalSeparated: parseCommunalSeparated(text, country),
+    communalSeparated: isSale ? null : parseCommunalSeparated(text, country),
     parking: bool(
       text,
       /паркинг|парков|машино[- ]?мест|parking|avtoturargoh|mashina\s*joyi/iu,
@@ -239,6 +252,6 @@ export function parseHousingListingFields(value, { country = '' } = {}) {
     firstRent,
     minRentTerm: parseMinRentTerm(text),
     availableFrom: parseAvailableFrom(text),
-    utilitiesAmount: parseUtilitiesAmount(text),
+    utilitiesAmount: isSale ? null : parseUtilitiesAmount(text),
   });
 }

package/src/housing-money.js

@@ -242,8 +242,17 @@ export function parseHousingPrice(value, fallbackCurrency = '') {
     // `500 m2` and `5 minut` cannot be promoted to 500 million / 5 million.
     const match = priceText.match(new RegExp(`(\\d+(?:[.,]\\d+)?)\\s*(${SCALE_PATTERN})(?=$|[^\\p{L}\\p{N}_])`, 'iu'));
     if (match) {
+      // Ukrainian listings abbreviate a microdistrict as “531 м/р”. The
+      // one-letter million scale is otherwise valid, but this trailing
+      // “/р” makes the token an area identifier, never a price.
+      const tail = priceText.slice((match.index ?? 0) + match[0].length);
+      if (/^\s*\/\s*[рr](?=$|[^\p{L}\p{N}_])/iu.test(tail)) {
+        // Fall through to the ordinary bare-number pass, which can still
+        // recover a genuine rent amount later in the listing.
+      } else {
       const amount = parseScaledAmount(match[1], match[2]);
       if (amount != null && amount >= 1000 && amount <= 5_000_000_000) price = Math.round(amount);
+      }
     }
   }
 

package/src/housing-structured.js

@@ -430,6 +430,12 @@ export function parseHousingStructured(value, options = {}) {
   const country = String(options.country || '').trim();
   const fallbackCurrency = options.fallbackCurrency || countryCurrency(country) || '';
   const phoneCountry = options.phoneCountry || countryPhoneHint(country) || null;
+  const intent = resolveHousingIntent(text);
+  const listingFields = parseHousingListingFields(text, { country, dealType: intent?.dealType || null });
+  const parsedPayments = parseHousingPaymentDetails(text);
+  const payments = intent?.dealType === 'sale'
+    ? deepFreeze({ ...parsedPayments, utilities: null })
+    : parsedPayments;
 
   return deepFreeze({
     text,
@@ -437,7 +443,7 @@ export function parseHousingStructured(value, options = {}) {
       platform: sourcePost.source,
       contact: sourcePost.contact,
     },
-    intent: resolveHousingIntent(text),
+    intent,
     context: parseHousingContext(text),
     rooms: parseHousingRoomCount(text),
     floor: parseHousingFloor(text),
@@ -450,8 +456,8 @@ export function parseHousingStructured(value, options = {}) {
     }),
     residentialComplex: parseHousingResidentialComplex(text),
     amenities: parseHousingAmenities(text),
-    listingFields: parseHousingListingFields(text, { country }),
-    payments: parseHousingPaymentDetails(text),
+    listingFields,
+    payments,
     seller: parseHousingSeller(text),
     infrastructure: parseHousingInfrastructure(text),
     contacts: parseHousingContacts(text, { countryHint: phoneCountry, sourcePost }),

package/src/housing-text.js

@@ -33,6 +33,7 @@ const RC_ATTRIBUTE_TOKEN_RE = /^\d|^[\d/]+$|^[a-zа-яё]?\d+(?:[а-яёa-z]{1,4
 const RC_LATIN_NOISE_RE = /^(?:vip|lux|luxe|elite|premium|euro|evro|new|top|super|best|hot|urgent|srochno|arenda|ijara|sotiladi|sale|rent|for|home|house|flat|apartment|apartments|kvartira|tashkent|toshkent|wifi|wi|fi|tv|ac|internet|telegram|yandex|google|instagram|whatsapp|iphone|samsung|lg|bosch|artel|km|km2|m2|sqm|usd|uzs|eur)$/iu;
 const RC_LATIN_WORD_RE = /^[A-Za-z][A-Za-z'’\-]{1,}$/u;
 const RC_DISTRICT_MARKER_RE = /^(?:район|районе|районы|туман|тумани|tumani|district)$/iu;
+const RC_ADDRESS_INTRO_RE = /^(?:за|адрес(?:ою|у|ом)?|вул(?:иця)?|улица)$/iu;
 const RC_MARKER_RE = /(?:жк|жм|ж\/к|residential complex|ansamblu(?: rezidential)?|turar[- ]?joy majmuasi)\s*/i;
 
 function leadingLatinName(text) {
@@ -65,7 +66,7 @@ export function parseHousingResidentialComplex(value) {
   const parsed = String(candidate).trim().replace(/([\p{Ll}\d])(\p{Lu})/gu, '$1 $2').replace(/\s{2,}/g, ' ').split(/\s+/).reduce((words, token) => {
     if (words.stopped || words.list.length >= 4) return { ...words, stopped: true };
     const clean = token.replace(/^[«»"'„“]+|[«»"'„“!|,]+$/g, '');
-    if (!clean || RC_STOP_WORD_RE.test(clean) || RC_ATTRIBUTE_TOKEN_RE.test(clean)) return { ...words, stopped: true, by: clean };
+    if (!clean || RC_STOP_WORD_RE.test(clean) || RC_ADDRESS_INTRO_RE.test(clean) || RC_ATTRIBUTE_TOKEN_RE.test(clean)) return { ...words, stopped: true, by: clean };
     return { list: [...words.list, clean], stopped: false, by: words.by };
   }, { list: [], stopped: false, by: '' });
   const words = RC_DISTRICT_MARKER_RE.test(parsed.by) ? parsed.list.slice(0, -1) : parsed.list;

package/src/ua-kharkiv-location-translations.js

@@ -39,6 +39,8 @@ export const UA_KHARKIV_LOCATION_TRANSLATIONS = Object.freeze({
     ['Horizont', 'Обрій', 'Обрий', 'Obrii', 'Горизонт'],
   ]),
   streets: locationEntries([
+    ['Poltavskyi Shliakh Street', 'Полтавський шлях', 'Полтавский шлях', 'Полтавський шоях', 'Полтавский шоях'],
+    ['Liubovi Maloi Street', 'вулиця Любові Малої', 'вул. Любові Малої', 'Любові Малої', 'улица Любови Малой', 'ул. Любови Малой', 'Любови Малой', 'Л. Малой', 'Л.Малой', 'ул. Л. Малой', 'ул.Л.Малой', 'улица Л. Малой', 'улица Л.Малой'],
     ['Hvardiitsiv-Shyronintsiv Street', 'вулиця Гвардійців-Широнінців', 'вул. Гвардійців-Широнінців', 'Гвардійців-Широнінців', 'улица Гвардейцев-Широнинцев', 'ул. Гвардейцев-Широнинцев', 'Гвардейцев-Широнинцев', 'Гвардейцев Широнинцев'],
     ['Yuvileinyi Avenue', 'проспект Ювілейний', 'пр-т Ювілейний', 'Ювілейний проспект', 'проспект Юбилейный', 'пр-т Юбилейный', 'Юбилейный проспект', 'проспект 50-річчя ВЛКСМ', 'проспект 50-летия ВЛКСМ'],
     ['Traktorobudivnykiv Avenue', 'проспект Тракторобудівників', 'пр-т Тракторобудівників', 'Тракторобудівників проспект', 'проспект Тракторостроителей', 'пр-т Тракторостроителей', 'Тракторостроителей проспект'],

package/src/ua-kharkiv-microdistrict-extensions.js

@@ -3,6 +3,7 @@ import { locationEntries } from './location-merge.js';
 export const UA_KHARKIV_MICRODISTRICT_EXTENSIONS = Object.freeze({
   Kharkiv: Object.freeze({
     microdistricts: locationEntries([
+      ['Kholodna Hora', 'Хол. Гора', 'Хол Гора'],
       ['337 microdistrict', '337-й мікрорайон', '337 мікрорайон', '337-й микрорайон', '337 микрорайон', '337 мкр', '337 мкр.', '337-й мкр', '337 м/р', '337 мрн'],
       ['339 microdistrict', '339-й мікрорайон', '339 мікрорайон', '339-й микрорайон', '339 микрорайон', '339 мкр', '339 мкр.', '339-й мкр', '339 м/р', '339 мрн'],
     ]),