@whiteslove/parsing-lexicon 0.2.0 → 0.2.2

package/src/housing-money.js

@@ -1,102 +1,102 @@
-import { aliasesOf, findCanonical } from './normalization.js';
-import { CURRENCY_TERMS, NUMBER_MULTIPLIERS } from './money-lexicon.js';
-import {
-  MONEY_NUMBER_PATTERN,
-  moneyCurrencyFromText,
-  moneyCurrencyPattern,
-  parseNumericAmount,
-  parseScaledAmount,
-} from './money-core.js';
-import { maskPhoneLikeSpans } from './contact.js';
-
-const PRICE_KEYWORD = '(?:цена|ціна|нарх(?:и)?|narx|price|стоимост[ьи]|аренд(?:а|ная\\s+плата)?|rent)';
-const PRICE_CURRENCY = `(?:${moneyCurrencyPattern()})`;
-
-function escapeRegex(value) {
-  return String(value).replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
-}
-
-export function parseHousingPrice(value, fallbackCurrency = '') {
-  const original = String(value || '');
-  if (!original) return Object.freeze({ price: null, currency: fallbackCurrency || '' });
-
-  // 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);
-  let currency = moneyCurrencyFromText(text, fallbackCurrency || '') || '';
-  const explicit = Boolean(findCanonical(text, CURRENCY_TERMS, { partial: true }));
-  let price = null;
-
-  // Common Ukrainian/Russian classifieds shorthand: "10 т грн" / "10 т гр".
-  // Keep this housing-specific instead of adding globally ambiguous aliases
-  // `т` (tonne) and `гр` (gram) to the shared money lexicon. A nearby price
-  // keyword plus an explicit hryvnia shorthand makes the intent unambiguous.
-  const compactThousandUah = text.match(new RegExp(
-    `${PRICE_KEYWORD}[^\\r\\n]{0,48}?(\\d{1,6}(?:[.,]\\d{1,2})?)\\s*т(?:ыс\\.?)?\\s*(?:гр(?:н)?|₴|uah)(?=$|[^\\p{L}\\p{N}_])`,
-    'iu',
-  ));
-  if (compactThousandUah) {
-    const amount = parseNumericAmount(compactThousandUah[1]);
-    if (amount != null && amount >= 1 && amount <= 5_000_000) {
-      price = Math.round(amount * 1000);
-      currency = 'UAH';
-    }
-  }
-
-  const labelled = text.match(new RegExp(`${PRICE_KEYWORD}\\s*[:\\-–—]?\\s*(${MONEY_NUMBER_PATTERN})`, 'i'));
-  if (price == null && labelled) {
-    const amount = parseNumericAmount(labelled[1]);
-    if (amount != null && amount >= 50 && amount <= 5_000_000_000) price = amount;
-  }
-
-  if (price == null) {
-    let tagged = null;
-    const reNumSym = new RegExp(`(${MONEY_NUMBER_PATTERN})\\s*${PRICE_CURRENCY}`, 'ig');
-    const reSymNum = new RegExp(`${PRICE_CURRENCY}\\s*(${MONEY_NUMBER_PATTERN})`, 'ig');
-    for (const regex of [reNumSym, reSymNum]) {
-      let match;
-      while ((match = regex.exec(text)) !== null) {
-        const amount = parseNumericAmount(match[1]);
-        if (amount != null && amount >= 50 && amount <= 5_000_000_000 && (tagged == null || amount > tagged)) tagged = amount;
-      }
-    }
-    price = tagged;
-  }
-
-  if (price == null) {
-    const scalePattern = [...new Set(NUMBER_MULTIPLIERS.flatMap((entry) => aliasesOf(entry)).filter(Boolean))]
-      .sort((a, b) => String(b).length - String(a).length)
-      .map(escapeRegex)
-      .join('|');
-    // 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 = text.match(new RegExp(`(\\d+(?:[.,]\\d+)?)\\s*(${scalePattern})(?=$|[^\\p{L}\\p{N}_])`, 'iu'));
-    if (match) {
-      const amount = parseScaledAmount(match[1], match[2]);
-      if (amount != null && amount >= 1000 && amount <= 5_000_000_000) price = Math.round(amount);
-    }
-  }
-
-  if (price == null) {
-    const matches = text.match(/\d{1,3}(?:[ \u00A0.,]\d{3})+|\d{4,}/g) || [];
-    let best = null;
-    for (const raw of matches) {
-      const digits = raw.replace(/[\s.,]/g, '');
-      if (digits[0] === '0') continue;
-      const amount = parseNumericAmount(raw);
-      if (amount != null && amount >= 1000 && amount <= 5_000_000_000 && (best == null || amount > best)) best = amount;
-    }
-    price = best;
-  }
-
-  if (!explicit && fallbackCurrency === 'UZS' && price != null) {
-    const dailyUzbek = /(?:kunlik|sutkaga|kecha[- ]?kunduz|посуточн|суточн)/i.test(text);
-    currency = price >= 1_000_000 || (dailyUzbek && price >= 10_000) ? 'UZS' : 'USD';
-  }
-
-  return Object.freeze({ price, currency });
-}
-
-// Compatibility name for callers migrating from Flat Finder's local parser.
-export const parsePriceFromText = parseHousingPrice;
+import { aliasesOf, findCanonical } from './normalization.js';
+import { CURRENCY_TERMS, NUMBER_MULTIPLIERS } from './money-lexicon.js';
+import {
+  MONEY_NUMBER_PATTERN,
+  moneyCurrencyFromText,
+  moneyCurrencyPattern,
+  parseNumericAmount,
+  parseScaledAmount,
+} from './money-core.js';
+import { maskPhoneLikeSpans } from './contact.js';
+
+const PRICE_KEYWORD = '(?:цена|ціна|нарх(?:и)?|narx|price|стоимост[ьи]|аренд(?:а|ная\\s+плата)?|rent)';
+const PRICE_CURRENCY = `(?:${moneyCurrencyPattern()})`;
+
+function escapeRegex(value) {
+  return String(value).replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
+}
+
+export function parseHousingPrice(value, fallbackCurrency = '') {
+  const original = String(value || '');
+  if (!original) return Object.freeze({ price: null, currency: fallbackCurrency || '' });
+
+  // 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);
+  let currency = moneyCurrencyFromText(text, fallbackCurrency || '') || '';
+  const explicit = Boolean(findCanonical(text, CURRENCY_TERMS, { partial: true }));
+  let price = null;
+
+  // Common Ukrainian/Russian classifieds shorthand: "10 т грн" / "10 т гр".
+  // Keep this housing-specific instead of adding globally ambiguous aliases
+  // `т` (tonne) and `гр` (gram) to the shared money lexicon. A nearby price
+  // keyword plus an explicit hryvnia shorthand makes the intent unambiguous.
+  const compactThousandUah = text.match(new RegExp(
+    `${PRICE_KEYWORD}[^\\r\\n]{0,48}?(\\d{1,6}(?:[.,]\\d{1,2})?)\\s*т(?:ыс\\.?)?\\s*(?:гр(?:н)?|₴|uah)(?=$|[^\\p{L}\\p{N}_])`,
+    'iu',
+  ));
+  if (compactThousandUah) {
+    const amount = parseNumericAmount(compactThousandUah[1]);
+    if (amount != null && amount >= 1 && amount <= 5_000_000) {
+      price = Math.round(amount * 1000);
+      currency = 'UAH';
+    }
+  }
+
+  const labelled = text.match(new RegExp(`${PRICE_KEYWORD}\\s*[:\\-–—]?\\s*(${MONEY_NUMBER_PATTERN})`, 'i'));
+  if (price == null && labelled) {
+    const amount = parseNumericAmount(labelled[1]);
+    if (amount != null && amount >= 50 && amount <= 5_000_000_000) price = amount;
+  }
+
+  if (price == null) {
+    let tagged = null;
+    const reNumSym = new RegExp(`(${MONEY_NUMBER_PATTERN})\\s*${PRICE_CURRENCY}`, 'ig');
+    const reSymNum = new RegExp(`${PRICE_CURRENCY}\\s*(${MONEY_NUMBER_PATTERN})`, 'ig');
+    for (const regex of [reNumSym, reSymNum]) {
+      let match;
+      while ((match = regex.exec(text)) !== null) {
+        const amount = parseNumericAmount(match[1]);
+        if (amount != null && amount >= 50 && amount <= 5_000_000_000 && (tagged == null || amount > tagged)) tagged = amount;
+      }
+    }
+    price = tagged;
+  }
+
+  if (price == null) {
+    const scalePattern = [...new Set(NUMBER_MULTIPLIERS.flatMap((entry) => aliasesOf(entry)).filter(Boolean))]
+      .sort((a, b) => String(b).length - String(a).length)
+      .map(escapeRegex)
+      .join('|');
+    // 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 = text.match(new RegExp(`(\\d+(?:[.,]\\d+)?)\\s*(${scalePattern})(?=$|[^\\p{L}\\p{N}_])`, 'iu'));
+    if (match) {
+      const amount = parseScaledAmount(match[1], match[2]);
+      if (amount != null && amount >= 1000 && amount <= 5_000_000_000) price = Math.round(amount);
+    }
+  }
+
+  if (price == null) {
+    const matches = text.match(/\d{1,3}(?:[ \u00A0.,]\d{3})+|\d{4,}/g) || [];
+    let best = null;
+    for (const raw of matches) {
+      const digits = raw.replace(/[\s.,]/g, '');
+      if (digits[0] === '0') continue;
+      const amount = parseNumericAmount(raw);
+      if (amount != null && amount >= 1000 && amount <= 5_000_000_000 && (best == null || amount > best)) best = amount;
+    }
+    price = best;
+  }
+
+  if (!explicit && fallbackCurrency === 'UZS' && price != null) {
+    const dailyUzbek = /(?:kunlik|sutkaga|kecha[- ]?kunduz|посуточн|суточн)/i.test(text);
+    currency = price >= 1_000_000 || (dailyUzbek && price >= 10_000) ? 'UZS' : 'USD';
+  }
+
+  return Object.freeze({ price, currency });
+}
+
+// Compatibility name for callers migrating from Flat Finder's local parser.
+export const parsePriceFromText = parseHousingPrice;

package/src/housing-source-aliases.d.ts

@@ -1,8 +1,8 @@
-export const HOUSING_DEAL_TYPE_EXTENSIONS: readonly unknown[];
-export const HOUSING_ROOM_ONLY_EXTENSIONS: readonly unknown[];
-export function resolveExtendedHousingIntent(value: unknown): Readonly<{
-  action: string | null;
-  listingKind: string | null;
-  dealType: 'sale' | 'longRent' | 'shortRent';
-}> | null;
-export function isRoomOnlyHousing(value: unknown): boolean;
+export const HOUSING_DEAL_TYPE_EXTENSIONS: readonly unknown[];
+export const HOUSING_ROOM_ONLY_EXTENSIONS: readonly unknown[];
+export function resolveExtendedHousingIntent(value: unknown): Readonly<{
+  action: string | null;
+  listingKind: string | null;
+  dealType: 'sale' | 'longRent' | 'shortRent';
+}> | null;
+export function isRoomOnlyHousing(value: unknown): boolean;

package/src/housing-source-aliases.js

@@ -1,51 +1,51 @@
-import { lexiconEntity } from './lexicon-core.js';
-import { findCanonical } from './normalization.js';
-import { resolveHousingIntent } from './housing-intent.js';
-import { HOUSING_OCCUPANCY_TYPES, PROPERTY_TYPES } from './housing.js';
-
-const group = (canonical, aliases, extra = {}) => lexiconEntity(canonical, aliases, extra);
-
-/** Source/feed wording that is useful for normalization but too noisy for the generic core. */
-export const HOUSING_DEAL_TYPE_EXTENSIONS = Object.freeze([
-  group('longRent', {
-    ru: ['снять', 'в месяц', 'месяц'],
-    en: ['monthly'],
-    uzLatn: ['ijaraga', 'oyiga', 'beriladi'],
-    uzCyrl: ['ижарага', 'ойига', 'берилади'],
-    kk: ['айына'],
-  }),
-  group('shortRent', {
-    ru: ['сутки', 'суток', 'суточно'],
-    en: ['daily'],
-    uzLatn: ['kunlik', 'sutkalik'],
-    uzCyrl: ['кунлик', 'суткалик'],
-    kk: ['тәуліктік'],
-  }),
-]);
-
-export const HOUSING_ROOM_ONLY_EXTENSIONS = Object.freeze([
-  group('roomOnly', {
-    ru: ['шеринг'],
-    en: ['room only'],
-    uzLatn: ['student qizlarga', 'talaba qizlarga', 'opshijit dom', 'obshijit dom'],
-    uzCyrl: ['студент қизларга', 'талаба қизларга'],
-  }),
-]);
-
-export function resolveExtendedHousingIntent(value) {
-  const text = String(value || '');
-  const base = resolveHousingIntent(text);
-  if (base) return base;
-  const deal = findCanonical(text, HOUSING_DEAL_TYPE_EXTENSIONS, { partial: true });
-  if (!deal?.canonical) return null;
-  return Object.freeze({ action: null, listingKind: null, dealType: deal.canonical });
-}
-
-export function isRoomOnlyHousing(value) {
-  const text = String(value || '');
-  if (!text.trim()) return false;
-  const occupancy = findCanonical(text, HOUSING_OCCUPANCY_TYPES, { partial: true })?.canonical;
-  if (occupancy && ['room', 'sharedRoom', 'bedSpace'].includes(occupancy)) return true;
-  if (findCanonical(text, PROPERTY_TYPES, { partial: true })?.canonical === 'dormitory') return true;
-  return Boolean(findCanonical(text, HOUSING_ROOM_ONLY_EXTENSIONS, { partial: true }));
-}
+import { lexiconEntity } from './lexicon-core.js';
+import { findCanonical } from './normalization.js';
+import { resolveHousingIntent } from './housing-intent.js';
+import { HOUSING_OCCUPANCY_TYPES, PROPERTY_TYPES } from './housing.js';
+
+const group = (canonical, aliases, extra = {}) => lexiconEntity(canonical, aliases, extra);
+
+/** Source/feed wording that is useful for normalization but too noisy for the generic core. */
+export const HOUSING_DEAL_TYPE_EXTENSIONS = Object.freeze([
+  group('longRent', {
+    ru: ['снять', 'в месяц', 'месяц'],
+    en: ['monthly'],
+    uzLatn: ['ijaraga', 'oyiga', 'beriladi'],
+    uzCyrl: ['ижарага', 'ойига', 'берилади'],
+    kk: ['айына'],
+  }),
+  group('shortRent', {
+    ru: ['сутки', 'суток', 'суточно'],
+    en: ['daily'],
+    uzLatn: ['kunlik', 'sutkalik'],
+    uzCyrl: ['кунлик', 'суткалик'],
+    kk: ['тәуліктік'],
+  }),
+]);
+
+export const HOUSING_ROOM_ONLY_EXTENSIONS = Object.freeze([
+  group('roomOnly', {
+    ru: ['шеринг'],
+    en: ['room only'],
+    uzLatn: ['student qizlarga', 'talaba qizlarga', 'opshijit dom', 'obshijit dom'],
+    uzCyrl: ['студент қизларга', 'талаба қизларга'],
+  }),
+]);
+
+export function resolveExtendedHousingIntent(value) {
+  const text = String(value || '');
+  const base = resolveHousingIntent(text);
+  if (base) return base;
+  const deal = findCanonical(text, HOUSING_DEAL_TYPE_EXTENSIONS, { partial: true });
+  if (!deal?.canonical) return null;
+  return Object.freeze({ action: null, listingKind: null, dealType: deal.canonical });
+}
+
+export function isRoomOnlyHousing(value) {
+  const text = String(value || '');
+  if (!text.trim()) return false;
+  const occupancy = findCanonical(text, HOUSING_OCCUPANCY_TYPES, { partial: true })?.canonical;
+  if (occupancy && ['room', 'sharedRoom', 'bedSpace'].includes(occupancy)) return true;
+  if (findCanonical(text, PROPERTY_TYPES, { partial: true })?.canonical === 'dormitory') return true;
+  return Boolean(findCanonical(text, HOUSING_ROOM_ONLY_EXTENSIONS, { partial: true }));
+}

package/src/housing-structured.d.ts

@@ -1,55 +1,55 @@
-export type HousingAreaDetails = Readonly<{
-  total: number | null;
-  living: number | null;
-  kitchen: number | null;
-  balcony: number | null;
-  terrace: number | null;
-}>;
-
-export type HousingPaymentContext = Readonly<{
-  deposit: Readonly<{
-    required: boolean | null;
-    kind: string | null;
-    amount: number | null;
-    currency: string | null;
-  }>;
-  prepaymentMonths: number | null;
-  utilities: string | null;
-  commission: Readonly<{
-    required: boolean | null;
-    percent: number | null;
-  }>;
-}>;
-
-export type HousingInfrastructureDistance = Readonly<{
-  value: number | null;
-  unit: 'minute' | 'meter' | 'kilometer';
-  mode: 'walk' | 'drive' | null;
-}>;
-
-export type HousingInfrastructureMatch = Readonly<{
-  poi: string;
-  relation: string | null;
-  distance: HousingInfrastructureDistance | null;
-  start: number;
-  end: number;
-}>;
-
-export type HousingStructuredResult = Readonly<{
-  intent: Readonly<{ action: string | null; listingKind: string | null; dealType: string }> | null;
-  context: Readonly<Record<string, unknown>>;
-  rooms: number | null;
-  floor: Readonly<{ floor: number | null; totalFloors: number | null }>;
-  area: HousingAreaDetails;
-  payments: HousingPaymentContext;
-  seller: Readonly<{ type: 'owner' | 'agency' | null; confidence: number }>;
-  infrastructure: readonly HousingInfrastructureMatch[];
-}>;
-
-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 parseHousingPayments(value: unknown): HousingPaymentContext;
-export function parseHousingSeller(value: unknown): Readonly<{ type: 'owner' | 'agency' | null; confidence: number }>;
-export function parseHousingInfrastructure(value: unknown): readonly HousingInfrastructureMatch[];
-export function parseHousingStructured(value: unknown): HousingStructuredResult;
+export type HousingAreaDetails = Readonly<{
+  total: number | null;
+  living: number | null;
+  kitchen: number | null;
+  balcony: number | null;
+  terrace: number | null;
+}>;
+
+export type HousingPaymentContext = Readonly<{
+  deposit: Readonly<{
+    required: boolean | null;
+    kind: string | null;
+    amount: number | null;
+    currency: string | null;
+  }>;
+  prepaymentMonths: number | null;
+  utilities: string | null;
+  commission: Readonly<{
+    required: boolean | null;
+    percent: number | null;
+  }>;
+}>;
+
+export type HousingInfrastructureDistance = Readonly<{
+  value: number | null;
+  unit: 'minute' | 'meter' | 'kilometer';
+  mode: 'walk' | 'drive' | null;
+}>;
+
+export type HousingInfrastructureMatch = Readonly<{
+  poi: string;
+  relation: string | null;
+  distance: HousingInfrastructureDistance | null;
+  start: number;
+  end: number;
+}>;
+
+export type HousingStructuredResult = Readonly<{
+  intent: Readonly<{ action: string | null; listingKind: string | null; dealType: string }> | null;
+  context: Readonly<Record<string, unknown>>;
+  rooms: number | null;
+  floor: Readonly<{ floor: number | null; totalFloors: number | null }>;
+  area: HousingAreaDetails;
+  payments: HousingPaymentContext;
+  seller: Readonly<{ type: 'owner' | 'agency' | null; confidence: number }>;
+  infrastructure: readonly HousingInfrastructureMatch[];
+}>;
+
+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 parseHousingPayments(value: unknown): HousingPaymentContext;
+export function parseHousingSeller(value: unknown): Readonly<{ type: 'owner' | 'agency' | null; confidence: number }>;
+export function parseHousingInfrastructure(value: unknown): readonly HousingInfrastructureMatch[];
+export function parseHousingStructured(value: unknown): HousingStructuredResult;