@whiteslove/parsing-lexicon 0.6.15 → 0.6.17

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.6.15",
+  "version": "0.6.17",
   "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.d.ts

@@ -4,6 +4,19 @@ export type HousingPriceParseResult = Readonly<{
   approximate: boolean;
 }>;
 
-export function parseHousingPrice(value: unknown, fallbackCurrency?: string): HousingPriceParseResult;
-export function parseHousingPricePerSqm(value: unknown, fallbackCurrency?: string): HousingPriceParseResult;
-export const parsePriceFromText: typeof parseHousingPrice;
+export type HousingMoneyParseContext = Readonly<{
+  country?: string;
+  currency?: string;
+  fallbackCurrency?: string;
+  dealType?: 'sale' | 'longRent' | 'shortRent' | string | null;
+}>;
+
+export function parseHousingPrice(
+  value: unknown,
+  context?: string | HousingMoneyParseContext,
+): HousingPriceParseResult;
+export function parseHousingPricePerSqm(
+  value: unknown,
+  context?: string | HousingMoneyParseContext,
+): HousingPriceParseResult;
+export const parsePriceFromText: typeof parseHousingPrice;

package/src/housing-money.js

@@ -9,6 +9,11 @@ import {
 } from './money-core.js';
 import { maskPhoneLikeSpans } from './contact.js';
 import { DEPOSIT_TERMS, SELLER_TERMS } from './housing.js';
+import { COUNTRIES, canonicalCountryCode, countryByCode } from './countries.js';
+import {
+  classifyHousingSingleMSpans,
+  HOUSING_NUMERIC_SPAN_TYPES,
+} from './housing-numeric-spans.js';
 
 const PRICE_KEYWORD = '(?:цена|ціна|нарх(?:и)?|narx(?:i)?|price|стоимост[ьи]|аренд(?:а|ная\\s+плата)?|rent)';
 // moneyCurrencyPattern() includes short codes (cad, ron, aed...) with no
@@ -30,7 +35,12 @@ function escapeRegex(value) {
   return String(value).replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
 }
 
+// One-letter `m/м` is deliberately excluded from the generic multiplier set.
+// Its meaning is resolved by housing-numeric-spans.js first, so area/distance/
+// microdistrict notation cannot leak into money parsing while explicit price
+// contexts can still use compact million shorthand.
 const SCALE_PATTERN = [...new Set(NUMBER_MULTIPLIERS.flatMap((entry) => aliasesOf(entry)).filter(Boolean))]
+  .filter((alias) => !/^[mм]$/iu.test(String(alias)))
   .sort((a, b) => String(b).length - String(a).length)
   .map(escapeRegex)
   .join('|');
@@ -51,6 +61,32 @@ const PER_SQM_BEFORE_AMOUNT_RE = new RegExp(
   'igu',
 );
 
+const COMMON_HOUSING_STRUCTURE_PATTERNS = Object.freeze([
+  /(?:^|[^\p{L}\p{N}_])\d{1,5}(?:[.,]\d{1,2})?\s*(?:м(?:2|²)|m(?:2|²)|sqm|sq\.?\s*m|м\s*кв\.?)\s*(?=$|[^\p{L}\p{N}_])/giu,
+]);
+
+function moneyParsingContext(value = '') {
+  if (value && typeof value === 'object' && !Array.isArray(value)) {
+    const country = canonicalCountryCode(value.country) || '';
+    const currency = String(value.currency || value.fallbackCurrency || countryByCode(country)?.currency || '')
+      .trim()
+      .toUpperCase();
+    const rawDealType = String(value.dealType || '').trim();
+    const dealType = ['sale', 'longRent', 'shortRent'].includes(rawDealType) ? rawDealType : null;
+    return { country, currency, dealType };
+  }
+
+  const currency = String(value || '').trim().toUpperCase();
+  const countryMatches = currency
+    ? COUNTRIES.filter((item) => item.currency === currency)
+    : [];
+  return {
+    country: countryMatches.length === 1 ? countryMatches[0].code : '',
+    currency,
+    dealType: null,
+  };
+}
+
 function parsedMoneyAmount(numberValue, scaleValue) {
   const amount = scaleValue
     ? parseScaledAmount(numberValue, scaleValue)
@@ -99,6 +135,22 @@ function maskRanges(text, ranges) {
   return masked;
 }
 
+function maskPatternMatches(text, patterns) {
+  let masked = text;
+  for (const pattern of patterns) {
+    pattern.lastIndex = 0;
+    masked = masked.replace(pattern, (match) => ' '.repeat(match.length));
+  }
+  return masked;
+}
+
+function maskHousingStructuralSpans(text, context) {
+  const strictMasked = maskPatternMatches(text, COMMON_HOUSING_STRUCTURE_PATTERNS);
+  const nonMoneySpans = classifyHousingSingleMSpans(strictMasked, context)
+    .filter((span) => span.type !== HOUSING_NUMERIC_SPAN_TYPES.MONEY);
+  return maskRanges(strictMasked, nonMoneySpans);
+}
+
 function isPaymentScopedAmount(text, start, end) {
   const left = text
     .slice(Math.max(0, start - 56), start)
@@ -108,16 +160,31 @@ function isPaymentScopedAmount(text, start, end) {
     .slice(end, Math.min(text.length, end + 28))
     .split(/[\r\n.;!?]/u)[0] || '';
 
-  // "deposit 500$" / "commission 100$" are payment details, not the
-  // listing price. For the reverse form ("500$ deposit"), only suppress the
-  // amount when there is no explicit price/rent label immediately to its left;
-  // this preserves text such as "rent 800$, deposit 500$".
+  // Prefix payment labels bind to the amount on their right: "deposit 500$".
   if (findCanonical(left, PAYMENT_AMOUNT_TERMS, { partial: true })) return true;
+
+  // A suffix label binds only when it appears before another numeric token.
+  // This prevents the listing price in "6000грн+комуналка+6000(залог)" from
+  // being suppressed merely because the later deposit keyword is nearby, while
+  // still excluding direct suffix forms such as "500$ deposit".
+  const nextDigit = right.search(/\d/u);
+  const suffixScope = nextDigit === -1 ? right : right.slice(0, nextDigit);
   return !PRICE_KEYWORD_RE.test(left)
-    && Boolean(findCanonical(right, PAYMENT_AMOUNT_TERMS, { partial: true }));
+    && Boolean(findCanonical(suffixScope, PAYMENT_AMOUNT_TERMS, { partial: true }));
+}
+
+function parseContextualSingleLetterMillion(text, context) {
+  for (const span of classifyHousingSingleMSpans(text, context)) {
+    if (span.type !== HOUSING_NUMERIC_SPAN_TYPES.MONEY) continue;
+    if (isPaymentScopedAmount(text, span.start, span.end)) continue;
+    const amount = span.amount * 1_000_000;
+    if (amount >= 1_000_000 && amount <= 5_000_000_000) return Math.round(amount);
+  }
+  return null;
 }
 
-export function parseHousingPricePerSqm(value, fallbackCurrency = '') {
+export function parseHousingPricePerSqm(value, context = '') {
+  const { currency: fallbackCurrency } = moneyParsingContext(context);
   const original = String(value || '');
   if (!original) return Object.freeze({ amount: null, currency: fallbackCurrency || '', approximate: false });
 
@@ -135,18 +202,21 @@ export function parseHousingPricePerSqm(value, fallbackCurrency = '') {
   });
 }
 
-export function parseHousingPrice(value, fallbackCurrency = '') {
+export function parseHousingPrice(value, context = '') {
+  const { country, currency: fallbackCurrency, dealType } = moneyParsingContext(context);
   const original = String(value || '');
   if (!original) return Object.freeze({ amount: null, currency: fallbackCurrency || '', approximate: false });
 
   // Contact spans are removed once, before every money branch. A phone can
   // therefore never win as a labelled, currency-tagged or fallback amount.
   const text = maskPhoneLikeSpans(original);
-  // Unit prices are not listing totals. Mask their exact spans before running
-  // the ordinary total-price parser so "От 13 млн за м2" cannot become a
-  // 13,000,000 UZS apartment price. The dedicated parser above still exposes
-  // the unit price to consumers that need it.
-  const priceText = maskRanges(text, perSquareMeterMatches(text, fallbackCurrency));
+  // Unit prices are not listing totals. Mask their exact spans first. Then the
+  // shared numeric-span classifier masks all N m/Nм spans that resolve to area,
+  // distance, microdistrict or unknown; only positively classified money spans
+  // remain visible to the price parser.
+  const withoutUnitPrices = maskRanges(text, perSquareMeterMatches(text, fallbackCurrency));
+  const numericContext = { country, dealType };
+  const priceText = maskHousingStructuralSpans(withoutUnitPrices, numericContext);
   let currency = moneyCurrencyFromText(priceText, fallbackCurrency || '')
     || moneyCurrencyFromText(text, fallbackCurrency || '')
     || '';
@@ -195,6 +265,17 @@ export function parseHousingPrice(value, fallbackCurrency = '') {
     }
   }
 
+  // Single-letter million shorthand is accepted only when the numeric-span
+  // classifier resolves the span as money. This supports explicit "price 2m"
+  // / "2m USD" everywhere and bare compact "800m" only for Uzbek sale context.
+  if (price == null) {
+    const contextualMillion = parseContextualSingleLetterMillion(priceText, numericContext);
+    if (contextualMillion != null) {
+      price = contextualMillion;
+      currency = moneyCurrencyFromText(priceText, fallbackCurrency || '') || fallbackCurrency || '';
+    }
+  }
+
   const labelled = priceText.match(new RegExp(
     `${PRICE_KEYWORD}\\s*[:\\-–—]?\\s*(${MONEY_NUMBER_PATTERN})(?:\\s*(${SCALE_PATTERN})(?=$|[^\\p{L}\\p{N}_]))?`,
     'iu',
@@ -237,9 +318,6 @@ export function parseHousingPrice(value, fallbackCurrency = '') {
   }
 
   if (price == null) {
-    // Multiplier aliases include useful one-letter forms such as `m` and `k`.
-    // Require the alias to end at a token boundary so measurements/words like
-    // `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) {
       const amount = parseScaledAmount(match[1], match[2]);

package/src/housing-numeric-spans.js

@@ -0,0 +1,93 @@
+import { canonicalCountryCode } from './countries.js';
+import { moneyCurrencyPattern, parseNumericAmount } from './money-core.js';
+
+export const HOUSING_NUMERIC_SPAN_TYPES = Object.freeze({
+  MONEY: 'money',
+  AREA: 'area',
+  DISTANCE: 'distance',
+  MICRODISTRICT: 'microdistrict',
+  UNKNOWN: 'unknown',
+});
+
+const PRICE_LABEL_BEFORE_RE = /(?:цена|ціна|нарх(?:и)?|narx(?:i)?|price|стоимост[ьи]|аренд(?:а|ная\s+плата)?|rent)[^\r\n]{0,16}$/iu;
+const CURRENCY_AFTER_RE = new RegExp(
+  `^\\s*(?:uzbek(?:istan)?\\s+)?(?:${moneyCurrencyPattern()})(?![\\p{L}\\p{N}_])`,
+  'iu',
+);
+const UZ_SALE_SIGNAL_RE = /(?:sotiladi|sotuv(?:da)?|sotaman|прода[её]тся|продам|продаж[аеи]|for\s+sale|sale)/iu;
+const UNICODE_TOKEN_END = String.raw`(?![\p{L}\p{N}_])`;
+
+const MICRODISTRICT_AFTER_RE = /^\s*\/\s*[rр](?=$|[^\p{L}\p{N}_])/iu;
+const AREA_BEFORE_RE = /(?:общ(?:ая)?\.?\s*пл(?:ощад[ьи])?\.?|загальн\p{L}*\s+площ\p{L}*|площа|площадь|umumiy\s+maydon|maydon|area|surface|suprafa(?:ță|ta))[^\r\n,;]{0,12}$/iu;
+const AREA_AFTER_RE = new RegExp(
+  `^[\\s,;:()\\-–—]*(?:общ(?:ая)?\\.?\\s*пл(?:ощад[ьи])?\\.?|загальн\\p{L}*\\s+площ\\p{L}*|площа|площадь|kv\\.?\\s*m|sqm|sq\\.?\\s*m|umumiy\\s+maydon|maydon|area|surface|suprafa(?:ță|ta))${UNICODE_TOKEN_END}`,
+  'iu',
+);
+
+const DISTANCE_OBJECT = String.raw`(?:metro(?:ga|dan|gacha)?|метро|maktab(?:ga|dan|gacha)?|school|школ\p{L}*|bozor(?:ga|dan|gacha)?|market|рынок\p{L}*|bekat(?:ga|dan|gacha)?|stop|station|остановк\p{L}*|park(?:ga|dan|gacha)?|парк\p{L}*|aeroport(?:ga|dan|gacha)?|airport|аэропорт\p{L}*|do['’ʻʼ]?kon(?:ga|dan|gacha)?|магазин\p{L}*)`;
+const DISTANCE_BEFORE_RE = new RegExp(
+  `(?:${DISTANCE_OBJECT}|masofa|distance|(?:до|от|to|from)\\s+${DISTANCE_OBJECT})[^\\r\\n,;]{0,14}$`,
+  'iu',
+);
+const DISTANCE_AFTER_RE = new RegExp(
+  `^[\\s,;:()\\-–—]*(?:(?:gacha|гача|masofa|distance|metr\\p{L}*|метр\\p{L}*)${UNICODE_TOKEN_END}|(?:до|от|to|from)\\s+${DISTANCE_OBJECT}${UNICODE_TOKEN_END}|${DISTANCE_OBJECT}${UNICODE_TOKEN_END}|(?:walk(?:ing)?|piyoda)\\s+(?:to\\s+)?${DISTANCE_OBJECT}${UNICODE_TOKEN_END})`,
+  'iu',
+);
+
+const SINGLE_M_CANDIDATE_RE = /(?<![\p{L}\p{N}_])(\d{1,4}(?:[.,]\d{1,2})?)([ \u00A0]*)([mм])(?![\p{L}\p{N}_])/giu;
+
+function normalizedDealType(value) {
+  const dealType = String(value || '').trim();
+  return ['sale', 'longRent', 'shortRent'].includes(dealType) ? dealType : null;
+}
+
+function classifySingleMSpan(text, match, { country, dealType }) {
+  const start = match.index ?? 0;
+  const end = SINGLE_M_CANDIDATE_RE.lastIndex;
+  const before = text.slice(Math.max(0, start - 64), start);
+  const after = text.slice(end, Math.min(text.length, end + 64));
+
+  if (MICRODISTRICT_AFTER_RE.test(after)) return HOUSING_NUMERIC_SPAN_TYPES.MICRODISTRICT;
+  if (AREA_BEFORE_RE.test(before) || AREA_AFTER_RE.test(after)) return HOUSING_NUMERIC_SPAN_TYPES.AREA;
+  if (DISTANCE_BEFORE_RE.test(before) || DISTANCE_AFTER_RE.test(after)) return HOUSING_NUMERIC_SPAN_TYPES.DISTANCE;
+
+  const explicitPrice = PRICE_LABEL_BEFORE_RE.test(before);
+  const explicitCurrency = CURRENCY_AFTER_RE.test(after);
+  if (explicitPrice || explicitCurrency) return HOUSING_NUMERIC_SPAN_TYPES.MONEY;
+
+  const compact = match[2].length === 0;
+  const localSaleSignal = country === 'UZ' && UZ_SALE_SIGNAL_RE.test(`${before} ${after}`);
+  if (country === 'UZ' && compact && (dealType === 'sale' || localSaleSignal)) {
+    return HOUSING_NUMERIC_SPAN_TYPES.MONEY;
+  }
+
+  return HOUSING_NUMERIC_SPAN_TYPES.UNKNOWN;
+}
+
+export function classifyHousingSingleMSpans(value, options = {}) {
+  const text = String(value || '');
+  if (!text) return Object.freeze([]);
+
+  const country = canonicalCountryCode(options.country) || String(options.country || '').trim().toUpperCase();
+  const dealType = normalizedDealType(options.dealType);
+  const spans = [];
+
+  SINGLE_M_CANDIDATE_RE.lastIndex = 0;
+  let match;
+  while ((match = SINGLE_M_CANDIDATE_RE.exec(text)) !== null) {
+    const amount = parseNumericAmount(match[1]);
+    if (amount == null) continue;
+    const start = match.index ?? 0;
+    const end = SINGLE_M_CANDIDATE_RE.lastIndex;
+    spans.push(Object.freeze({
+      start,
+      end,
+      raw: match[0],
+      amount,
+      compact: match[2].length === 0,
+      type: classifySingleMSpan(text, match, { country, dealType }),
+    }));
+  }
+
+  return Object.freeze(spans);
+}

package/src/housing-structured.d.ts

@@ -65,6 +65,7 @@ export type HousingStructuredOptions = Readonly<{
   country?: string;
   fallbackCurrency?: string;
   phoneCountry?: string;
+  dealType?: 'sale' | 'longRent' | 'shortRent' | string | null;
   source?: string;
   knownStreet?: string | null;
   allowBareAddress?: boolean;
@@ -92,7 +93,10 @@ export type HousingStructuredResult = Readonly<{
 
 export function parseHousingRoomCount(value: unknown): number | null;
 export function parseHousingFloor(value: unknown): Readonly<{ floor: number | null; totalFloors: number | null }>;
-export function parseHousingAreas(value: unknown): HousingAreaDetails;
+export function parseHousingAreas(
+  value: unknown,
+  options?: Pick<HousingStructuredOptions, 'country' | 'dealType'>,
+): HousingAreaDetails;
 export function parseHousingPayments(value: unknown): HousingPaymentContext;
 export function parseHousingSeller(value: unknown): Readonly<{ type: 'owner' | 'agency' | null; confidence: number }>;
 export function parseHousingInfrastructure(value: unknown): readonly HousingInfrastructureMatch[];

package/src/housing-structured.js

@@ -13,6 +13,10 @@ import { parseHousingListingFields } from './housing-listing-fields.js';
 import { parseHousingPrice } from './housing-money.js';
 import { parseHousingAmenities, parseHousingResidentialComplex } from './housing-text.js';
 import { parseHousingSourcePost } from './housing-source-aliases.js';
+import {
+  classifyHousingSingleMSpans,
+  HOUSING_NUMERIC_SPAN_TYPES,
+} from './housing-numeric-spans.js';
 
 const NUMBER_WORDS = Object.freeze([
   [/(?<![\p{L}\p{N}_])(?:однушк\p{L}*|однокомнатн\p{L}*|bir\s+xona(?:li)?|бир\s+хона(?:ли|лик)?|1\s*(?:-\s*)?к(?:омн\p{L}*)?|1\s*(?:-\s*)?xona(?:li)?|1\s*(?:-\s*)?хона(?:лик|ли)?|1\s*бөлмелі|one[- ]bedroom|one[- ]room)(?![\p{L}\p{N}_])/iu, 1],
@@ -63,11 +67,6 @@ function parseCompactSpec(text) {
 }
 
 function currencyNear(text) {
-  // findCanonical's partial match requires the alias to be its own
-  // space-delimited token, so a currency symbol glued directly to its
-  // number with no space ("500.$", "350$") never matches. Fall back to
-  // moneyCurrencyFromText, which detects symbols and codes independent of
-  // surrounding whitespace (used by parseHousingPrice for the same reason).
   return findCanonical(text, CURRENCY_TERMS, { partial: true })?.canonical
     || moneyCurrencyFromText(text) || null;
 }
@@ -76,11 +75,6 @@ export function parseHousingRoomCount(value) {
   const text = normalizeUnicode(value ?? '');
   if (!text) return null;
 
-  // Uzbek market shorthand: a unit formally registered as N rooms that was
-  // physically converted into a different count ("1 xonali ... 2 xona
-  // qilingan" — nominally 1-room, made into 2). The converted count is the
-  // one that matters and must win over the nominal count mentioned earlier
-  // in the same text.
   const converted = text.match(/(\d{1,2})\s*(?:ta\s*)?xona(?:ga)?\s+(?:qilingan|aylantirilgan|bo['’ʻʼ‘`]?lingan)(?=$|[^\p{L}\p{N}_])/iu);
   if (converted) {
     const rooms = toNumber(converted[1]);
@@ -99,14 +93,10 @@ export function parseHousingRoomCount(value) {
     if (rooms != null && rooms >= 1 && rooms <= 20) return rooms;
   }
 
-  // Structured "Label - Value" reposts (Telegram channel copies of OLX ads)
-  // put the label before the number: "Комнат - 4" instead of "4 комнаты".
   const reversed = text.match(/(?:комнат\p{L}*|xona(?:lar)?(?:i)?|хона(?:лар)?(?:и|лик|ли)?|rooms?)\s*[-:–—]\s*(\d{1,2})(?=$|[^\p{L}\p{N}])/iu);
   const reversedRooms = toNumber(reversed?.[1]);
   if (reversedRooms != null && reversedRooms >= 1 && reversedRooms <= 20) return reversedRooms;
 
-  // Listings that only carry the compact "1/5/12-50m" shorthand state the room
-  // count in its first number.
   return parseCompactSpec(text)?.rooms ?? null;
 }
 
@@ -114,17 +104,11 @@ export function parseHousingFloor(value) {
   const text = normalizeUnicode(value ?? '');
   if (!text) return deepFreeze({ floor: null, totalFloors: null });
 
-  // The rooms/floor/total shorthand must be read before the generic
-  // floor/total fraction parser or its first two numbers would incorrectly
-  // become floor=2, totalFloors=10 for "2/10/16".
   const compactSpec = parseCompactSpec(text);
   if (compactSpec) {
     return deepFreeze({ floor: compactSpec.floor, totalFloors: compactSpec.totalFloors });
   }
 
-  // Skip a pair that is merely the head of a rooms/floor/total triple the
-  // shorthand parser above rejected: there the floor and the total floors are
-  // the second and third numbers, never the first two.
   const fraction = text.match(/(?:^|[^\d])(\d{1,3})\s*[\/\\]\s*(\d{1,3})(?!\s*[\/\\]\s*\d)(?=$|[^\d])/u);
   if (fraction) {
     const floor = toNumber(fraction[1]);
@@ -143,9 +127,6 @@ export function parseHousingFloor(value) {
     }
   }
 
-  // Prefer the common "7 этаж" form before trying "этаж 7". Otherwise the
-  // marker-first parser can consume the next unrelated number (for example
-  // "7 этаж 44м²") and report floor 44.
   const beforeMarker = text.match(/(?:^|[^\d])(\d{1,3})\s*-?\s*(?:(?:chi|чи)\s*)?(?:этаж(?:да)?|поверх|floor|etaj|qavat(?:i(?:da(?:gi)?)?|da)?|қабат(?:ы(?:нда(?:ғы)?)?|та)?|кават(?:и(?:да(?:ги)?)?|да)?|қават(?:и(?:да(?:ги)?)?|да)?)(?=$|[^\p{L}\p{N}_])/iu);
   let floor = toNumber(beforeMarker?.[1]);
   let totalFloors = null;
@@ -164,7 +145,6 @@ export function parseHousingFloor(value) {
   }
 
   if (totalFloors == null) {
-    // Structured "Label - Value" reposts: "Этажность - 5" instead of "5 этажей".
     const reversedTotal = text.match(/(?:этажность|поверховість|qavatlilik|қабаттылық)\s*[-:–—]\s*(\d{1,3})(?=$|[^\p{L}\p{N}])/iu);
     totalFloors = toNumber(reversedTotal?.[1]);
   }
@@ -180,17 +160,18 @@ const AREA_LABELS = Object.freeze([
   ['kitchen', /(?:площадь\s+кухни|кухня|kitchen(?:\s+area)?|bucătărie|bucatarie|oshxona|асүй)/iu],
   ['balcony', /(?:балкон|лоджия|balcony|loggia|balcon|balkon)/iu],
   ['terrace', /(?:терраса|terrace|teras(?:ă|a)|terrasa)/iu],
-  ['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],
+  ['total', /(?:общ(?:ая)?\.?\s*пл(?:ощад[ьи])?\.?|общая\s+площадь|загальн\p{L}*\s+площ\p{L}*|площа|площадь|total\s+area|surface\s+area|suprafa(?:ță|ta)(?:\s+total(?:ă|a))?|umumiy\s+maydon|жалпы\s+аудан)/iu],
 ]);
 
 const AREA_UNIT_RE = String.raw`(?:м²|м2|m²|m2|sqm|sq\.?\s*m|mp|кв\.?\s*м|кв2|квадрат\p{L}*)`;
+const AREA_UNIT_AFTER_LABEL_RE = String.raw`(?:${AREA_UNIT_RE}|[mм])`;
 
 function areaAfterLabel(text, labelRe) {
-  const re = new RegExp(`${labelRe.source}\\s*[:=-]?\\s*(\\d{1,4}(?:[.,]\\d{1,2})?)\\s*${AREA_UNIT_RE}`, 'iu');
+  const re = new RegExp(`${labelRe.source}\\s*[:=-]?\\s*(\\d{1,4}(?:[.,]\\d{1,2})?)\\s*${AREA_UNIT_AFTER_LABEL_RE}(?=$|[^\\p{L}\\p{N}_])`, 'iu');
   return toNumber(text.match(re)?.[1]);
 }
 
-export function parseHousingAreas(value) {
+export function parseHousingAreas(value, options = {}) {
   const text = normalizeUnicode(value ?? '');
   const result = { total: null, living: null, kitchen: null, balcony: null, terrace: null };
   if (!text) return deepFreeze(result);
@@ -200,8 +181,11 @@ export function parseHousingAreas(value) {
     const generic = text.match(new RegExp(`(?:^|[^\\d])(\\d{1,4}(?:[.,]\\d{1,2})?)\\s*${AREA_UNIT_RE}(?=$|[^\\p{L}\\p{N}])`, 'iu'));
     result.total = toNumber(generic?.[1]);
   }
-  // The compact "1/5/12-50m" shorthand glues the area to the floor block, where
-  // the unit is often a bare "m" the generic area matcher cannot trust alone.
+  if (result.total == null) {
+    const typedArea = classifyHousingSingleMSpans(text, options)
+      .find((span) => span.type === HOUSING_NUMERIC_SPAN_TYPES.AREA);
+    result.total = typedArea?.amount ?? null;
+  }
   if (result.total == null) result.total = parseCompactSpec(text)?.areaSqm ?? null;
   for (const key of Object.keys(result)) {
     const number = result[key];
@@ -319,7 +303,7 @@ function distanceFromWindow(window, entityOffset = 0) {
       : /(?:на\s+машине|by\s+car|drive|mashinada|көлікпен)/iu.test(local) ? 'drive' : null;
     candidates.push({ distance: Math.abs(index - entityOffset), value: Number(match[1]), unit: 'minute', mode });
   }
-  const metricRe = /(\d{1,4}(?:[.,]\d+)?)\s*(км|km|километр\p{L}*|м|meter(?:s)?|метр\p{L}*)/giu;
+  const metricRe = /(\d{1,4}(?:[.,]\d+)?)\s*(км|km|километр\p{L}*|м|m(?![\p{L}\p{N}_])|meter(?:s)?|метр\p{L}*)/giu;
   for (const match of window.matchAll(metricRe)) {
     const index = match.index ?? 0;
     const rawUnit = match[2].toLocaleLowerCase();
@@ -430,6 +414,17 @@ 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 sourceDealType = ['sale', 'longRent', 'shortRent'].includes(String(options.dealType || ''))
+    ? String(options.dealType)
+    : null;
+  const effectiveDealType = intent?.dealType || sourceDealType || null;
+  const listingFields = parseHousingListingFields(text, { country, dealType: effectiveDealType });
+  const parsedPayments = parseHousingPaymentDetails(text);
+  const payments = effectiveDealType === 'sale'
+    ? deepFreeze({ ...parsedPayments, utilities: null })
+    : parsedPayments;
+  const numericOptions = { country, dealType: effectiveDealType };
 
   return deepFreeze({
     text,
@@ -437,12 +432,16 @@ export function parseHousingStructured(value, options = {}) {
       platform: sourcePost.source,
       contact: sourcePost.contact,
     },
-    intent: resolveHousingIntent(text),
+    intent,
     context: parseHousingContext(text),
     rooms: parseHousingRoomCount(text),
     floor: parseHousingFloor(text),
-    area: parseHousingAreas(text),
-    price: parseHousingPrice(text, fallbackCurrency),
+    area: parseHousingAreas(text, numericOptions),
+    price: parseHousingPrice(text, {
+      country,
+      currency: fallbackCurrency,
+      dealType: effectiveDealType,
+    }),
     address: parseHousingAddress(text, {
       knownStreet: options.knownStreet || null,
       allowDelimitedBare: options.allowDelimitedBareAddress === true,
@@ -450,8 +449,8 @@ 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'],
     ]),