@whiteslove/parsing-lexicon 0.3.2 → 0.3.3

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.3.2",
+  "version": "0.3.3",
   "description": "Shared deterministic multilingual parsing lexicon for Whiteslove housing and hiring services",
   "repository": {
     "type": "git",

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

@@ -21,12 +21,14 @@ export interface SourceRoleNormalization {
   canonical: string;
   label: string;
   re: RegExp;
+  labels: readonly string[];
 }
 
 export const SOURCE_PROFESSION_ALIASES: readonly SourceProfessionAlias[];
 export const SOURCE_ROLE_NORMALIZATION_RULES: readonly SourceRoleNormalization[];
 export const SOURCE_CANDIDATE_INTENT_ALIASES: readonly string[];
 export function normalizeSourceRole(value: unknown): Readonly<{ canonical: string; label: string }> | null;
+export function normalizeSourceRoleKeys(value: unknown): readonly string[] | null;
 export function matchesSourceCandidateIntent(value: unknown): boolean;
 export function professionDisplayLabel(canonical: string): string;
 export function matchExtendedProfessions(value: unknown, options?: { limit?: number; allowWeak?: boolean }): readonly ExtendedProfessionMatch[];

package/src/hiring-source-aliases.js

@@ -48,7 +48,9 @@ export const SOURCE_PROFESSION_ALIASES = Object.freeze([
   sourceRole('welder', 'Welder', ['сварщиком', 'сварщица', 'зварювальником'], 'construction'),
 ]);
 
-const sourceRoleNormalization = (canonical, label, re) => Object.freeze({ canonical, label, re });
+const sourceRoleNormalization = (canonical, label, re, labels) => Object.freeze({
+  canonical, label, re, labels: Object.freeze(labels || [label]),
+});
 
 /**
  * Anchored or strongly contextualized raw-role repairs. These rules are kept
@@ -65,7 +67,25 @@ export const SOURCE_ROLE_NORMALIZATION_RULES = Object.freeze([
   sourceRoleNormalization('sales_manager', 'Sales Manager', /^(?:sales\s+executive(?:\s+ind)?|роп(?:,?\s*sales\s+executive)?)$/iu),
   sourceRoleNormalization('operative_officer', 'Operative Officer', /^(?:оперативник|оперуполномоченн\p{L}*|оперативный\s+уполномоченн\p{L}*)$/iu),
   sourceRoleNormalization('water_supply_specialist', 'Water Supply Specialist', /^(?:suv\s+ta['’ʻʼ‘`]?minoti|водоснабжение)$/iu),
-  sourceRoleNormalization('any_role', 'Any Role', /^(?:onlayn(?:\s+ish(?:chi)?)?|online(?:\s+ish(?:chi)?)?|онлайн|удал[её]нно|remote(?:\s+work)?|boshqa\s+ishlar?|farqi\s+(?:yo['’ʻʼ‘`]?q|yuq)|tungi|bilmaym\p{L}*)$/iu),
+  sourceRoleNormalization('any_role', 'Any Role', /^(?:ищу\s+(?:работу|подработку)(?:\s+(?:онлайн|удал[её]нно))?|удал[её]нн\p{L}*\s+подработк\p{L}*(?:\s+за\s+компьютером)?(?:\s+для\s+студентов)?|работа\s+студентам|подработка|работа|работу|любая\s+работа|любая\s+занятость|не\s*важно|без\s+разницы|нет\s+разницы|farqi\s+yo['’ʻʼ‘`]?q|farqi\s+yuq|boshqa\s+ishlar?|ish|ish\s+kerak|ish\s+qidir(?:yapman|aman)|ish\s+izlayapman|onlayn(?:\s+ish(?:chi)?)?|online(?:\s+ish(?:chi)?)?|онлайн|удал[её]нно|remote(?:\s+work)?|tungi|uyda|ofisda|bilmaym\p{L}*|noma['’ʻʼ‘`]?lum(?:\s+\p{L}+)?|ba|va)$/iu),
+  sourceRoleNormalization('lawyer_teacher', 'Lawyer', /huquqshunos[^\n,;]*(?:,|\/|\s)+(?:pedagog|o['’ʻʼ‘`]?qituvchi)|pedagog[^\n,;]*(?:,|\/|\s)+huquqshunos/iu, ['Lawyer', 'Teacher']),
+  sourceRoleNormalization('erp_administrator_analyst', 'ERP Administrator', /^erp\s+administrator\p{L}*\s*(?:&|,|\/|va)\s*data\s+tahlilchi$/iu, ['ERP Administrator', 'Analyst']),
+  sourceRoleNormalization('translator_operator', 'Translator', /tarjimon\p{L}*[^\n]*(?:operator|data\s+otish)|operator\p{L}*[^\n]*tarjimon/iu, ['Translator', 'Operator']),
+  sourceRoleNormalization('lawyer', 'Lawyer', /^(?:yurisprudensiya\s+)?huquq(?:shunos)?[^\n]*|^yur(?:isprudensiya|ist)[^\n]*$/iu),
+  sourceRoleNormalization('teacher', 'Teacher', /^(?:matematika\s+)?o['’ʻʼ‘`]?qituvchi(?:lik)?$/iu),
+  sourceRoleNormalization('finance_banking_specialist', 'Finance / Banking Specialist', /^kredit\s+bo['’ʻʼ‘`]?yicha\s+mutaxa(?:s|ss)is$/iu),
+  sourceRoleNormalization('finance_banking_specialist', 'Finance / Banking Specialist', /^(?:финансы?\s*[,/&+]\s*банки?|банки?\s*[,/&+]\s*финансы?|finance\s*[,/&+]\s*banking)$/iu),
+  sourceRoleNormalization('welder', 'Welder', /^(?:svarchik|svarshik)$/iu),
+  sourceRoleNormalization('factory_worker', 'Factory Worker', /^(?:jizzax\s+)?kia\s+zavodidan\s+ish\s+kerak$|^zavod\s+ishlari(?:\s+.*)?$/iu),
+  sourceRoleNormalization('it_specialist', 'IT Specialist', /^(?:kompyuter\s+(?:sohasida|xizmatlari\s+bo['’ʻʼ‘`]?yicha\s+ish\s+kerak)|it\s+kompyuter)$/iu),
+  sourceRoleNormalization('restaurant_manager', 'Restaurant Manager', /^(?:restoran|restaurant)[^\n]*(?:boshqaruv|manager|menejer)/iu),
+  sourceRoleNormalization('administrator', 'Administrator', /^virtual\s+asistent$/iu),
+  sourceRoleNormalization('engineer', 'Engineer', /^(?:texnolog|technolog)\s+(?:injener|инженер)|^(?:injener|инженер)\s+(?:texnolog|technolog)/iu),
+  sourceRoleNormalization('electrician', 'Electrician', /^elektrik$/iu),
+  sourceRoleNormalization('construction_worker', 'Construction Worker', /^(?:yo['’ʻʼ‘`]?l|йул|йўл)\s+qurilish|^(?:yo['’ʻʼ‘`]?l|йул|йўл)\s+курилиш/iu),
+  sourceRoleNormalization('barista', 'Barista', /^(?:koffe|coffee)\s+ledy$/iu),
+  sourceRoleNormalization('chief_accountant', 'Chief Accountant', /^bosh\s+b(?:u(?:x|h)?|o)?galter$/iu),
+  sourceRoleNormalization('accountant', 'Accountant', /word[^\n]*excel[^\n]*hisob\s+kitob|^помо(?:ш|щ)ник\s+бухгалт/iu),
   sourceRoleNormalization('restaurant_cafe_worker', 'Restaurant / Cafe Worker', /^ищу\s+работу\s+(?:в\s+)?(?:кафе|ресторанах?|кафе\s+или\s+ресторанах)$/iu),
   sourceRoleNormalization('driver', 'Driver', /^(?:xaydovchilik|haydovchilik|shafyorlik|shofyorlik)/iu),
   sourceRoleNormalization('retail_worker', 'Retail Worker', /^do['’ʻʼ‘`]?kon$/iu),
@@ -79,7 +99,7 @@ export const SOURCE_ROLE_NORMALIZATION_RULES = Object.freeze([
   sourceRoleNormalization('hvac_technician', 'HVAC Technician', /^(?:konditsaner|kanditsaner|konditsioner)/iu),
   sourceRoleNormalization('mobile_content_creator', 'Mobile Content Creator', /^mobilografiya(?:\s+bo['’ʻʼ‘`]?yicha)?$/iu),
   sourceRoleNormalization('cctv_intercom_technician', 'CCTV / Intercom Technician', /kamera\s+(?:dama?fon|domofon)|domofon\s+xizmat/iu),
-  sourceRoleNormalization('internal_control_specialist', 'Internal Control Specialist', /^ichki\s+nazoratchi$/iu),
+  sourceRoleNormalization('internal_control_specialist', 'Internal Control Specialist', /^(?:ichki\s+nazoratchi|внутренний\s+аудит)$/iu),
   sourceRoleNormalization('brand_ambassador', 'Brand Ambassador', /^(?:бренд\s+фейс|brand\s+face)$/iu),
   sourceRoleNormalization('insurance_specialist', 'Insurance Specialist', /^sug['’ʻʼ‘`]?urta$/iu),
   sourceRoleNormalization('bank_operations_specialist', 'Bank Operations Specialist', /^стаж[её]р\s+операционист|^операционист$/iu),
@@ -97,6 +117,15 @@ export function normalizeSourceRole(value) {
   return match ? Object.freeze({ canonical: match.canonical, label: match.label }) : null;
 }
 
+/** Like normalizeSourceRole, but returns every display label a rule maps to
+ * (some raw values name two simultaneous roles, e.g. "Lawyer, Teacher"). */
+export function normalizeSourceRoleKeys(value) {
+  const text = String(value || '').trim();
+  if (!text) return null;
+  const match = SOURCE_ROLE_NORMALIZATION_RULES.find((entry) => entry.re.test(text));
+  return match ? [...match.labels] : null;
+}
+
 export const SOURCE_CANDIDATE_INTENT_ALIASES = Object.freeze([
   'работу ищу',
   'роботу шукаю',

package/src/housing-context.js

@@ -18,7 +18,7 @@ export const PROPERTY_CONDITION_TERMS = Object.freeze([
 ]);
 
 export const LAYOUT_TERMS = Object.freeze([
-  group('separateRooms', { ru: ['раздельные комнаты', 'изолированные комнаты'], en: ['separate rooms', 'isolated rooms'], uk: ['роздільні кімнати', 'ізольовані кімнати'], ro: ['camere separate'], uzLatn: ['alohida xonalar'], uzCyrl: ['алоҳида хоналар'], kk: ['бөлек бөлмелер'] }),
+  group('separateRooms', { ru: ['раздельные комнаты', 'изолированные комнаты'], en: ['separate rooms', 'isolated rooms'], uk: ['роздільні кімнати', 'ізольовані кімнати'], ro: ['camere separate'], uzLatn: ['alohida xonalar', 'aloxida xonalar'], uzCyrl: ['алоҳида хоналар', 'алохида хоналар'], kk: ['бөлек бөлмелер'] }),
   group('adjacentRooms', { ru: ['смежные комнаты', 'проходные комнаты', 'проходная комната'], en: ['adjacent rooms', 'walk-through room'], uk: ['суміжні кімнати', 'прохідна кімната'], ro: ['camere decomandate parțial', 'cameră de trecere'], uzLatn: ['o‘tish xonasi'], uzCyrl: ['ўтиш хонаси'], kk: ['өтпелі бөлме'] }),
   group('kitchenStudio', { ru: ['кухня-студия', 'кухня студия'], en: ['kitchen studio', 'open kitchen living room'], uk: ['кухня-студія'], ro: ['bucătărie open space'], uzLatn: ['oshxona studiya'], uzCyrl: ['ошхона студия'], kk: ['асүй-студия'] }),
   group('euroTwoRoom', { ru: ['евродвушка', 'евро двушка'], en: ['euro two-room', 'euro one bedroom'], uk: ['євродвушка'], ro: ['apartament euro două camere'], uzLatn: ['evro ikki xonali'], uzCyrl: ['евро икки хонали'], kk: ['еуро екі бөлмелі'] }),
@@ -29,7 +29,7 @@ export const LAYOUT_TERMS = Object.freeze([
 ]);
 
 export const BUILDING_TYPE_TERMS = Object.freeze([
-  group('brick', { ru: ['кирпичный дом', 'кирпич'], en: ['brick building', 'brick house'], uk: ['цегляний будинок'], ro: ['clădire din cărămidă'], uzLatn: ['g‘ishtli uy', "g'ishtli uy"], uzCyrl: ['ғиштли уй'], kk: ['кірпіш үй'] }),
+  group('brick', { ru: ['кирпичный дом', 'кирпич'], en: ['brick building', 'brick house'], uk: ['цегляний будинок'], ro: ['clădire din cărămidă'], uzLatn: ['g‘ishtli uy', "g'ishtli uy", 'gishtli uy', 'g‘ishtli', "g'ishtli", 'gishtli'], uzCyrl: ['ғиштли уй', 'гиштли уй', 'ғиштли', 'гиштли'], kk: ['кірпіш үй'] }),
   group('panel', { ru: ['панельный дом', 'панелька'], en: ['panel building', 'prefab panel'], uk: ['панельний будинок'], ro: ['bloc din panouri'], uzLatn: ['panelli uy'], uzCyrl: ['панелли уй'], kk: ['панельді үй'] }),
   group('monolithic', { ru: ['монолитный дом', 'монолит'], en: ['monolithic building', 'monolith'], uk: ['монолітний будинок'], ro: ['clădire monolit'], uzLatn: ['monolit uy'], uzCyrl: ['монолит уй'], kk: ['монолитті үй'] }),
   group('monolithicBrick', { ru: ['монолитно-кирпичный', 'монолит кирпич'], en: ['monolithic brick'], uk: ['монолітно-цегляний'], ro: ['monolit-cărămidă'], uzLatn: ['monolit g‘isht'], uzCyrl: ['монолит ғишт'], kk: ['монолит кірпіш'] }),
@@ -116,7 +116,7 @@ export const FINANCING_TERMS = Object.freeze([
 ]);
 
 export const LOCATION_RELATIONS = Object.freeze([
-  group('near', { ru: ['рядом с', 'возле', 'около'], en: ['near', 'next to', 'close to'], uk: ['поруч', 'біля'], ro: ['lângă', 'aproape de'], uzLatn: ['yaqinida', 'yonida', 'yaqin'], uzCyrl: ['яқинида', 'ёнида', 'яқин'], kk: ['жанында', 'маңында'] }),
+  group('near', { ru: ['рядом', 'рядом с', 'возле', 'около'], en: ['near', 'next to', 'close to'], uk: ['поруч', 'біля'], ro: ['lângă', 'aproape de'], uzLatn: ['yaqinida', 'yonida', 'yaqin'], uzCyrl: ['яқинида', 'ёнида', 'яқин'], kk: ['жанында', 'маңында'] }),
   group('opposite', { ru: ['напротив', 'через дорогу'], en: ['opposite', 'across the road'], uk: ['навпроти', 'через дорогу'], ro: ['vizavi', 'peste drum'], uzLatn: ['ro‘parasida'], uzCyrl: ['рўпарасида'], kk: ['қарсысында'] }),
   group('behind', { ru: ['за парком', 'за рынком', 'за базаром'], en: ['behind'], uk: ['за парком'], ro: ['în spatele'], uzLatn: ['orqasida'], uzCyrl: ['орқасида'], kk: ['артында'] }),
   group('intersection', { ru: ['на пересечении', 'на перекрёстке', 'перекрёсток'], en: ['at the intersection', 'crossroads'], uk: ['на перехресті'], ro: ['la intersecție'], uzLatn: ['chorrahada'], uzCyrl: ['чорраҳада'], kk: ['қиылыста'] }),

package/src/housing-features.js

@@ -21,7 +21,7 @@ export function parseHousingFeatures(value) {
 
   const internet = bool(
     text,
-    tokenRe(String.raw`интернет|wi[ -]?fi|we[ -]?fi|вай[ -]?фай|vayfay|роутер|router|інтернет`),
+    tokenRe(String.raw`интернет|wi[ -]?fi|we[ -]?fi|wf|вай[ -]?фай|vayfay|роутер|router|інтернет`),
     /(?:без\s+(?:интернет|wi[ -]?fi|вай[ -]?фай)|нет\s+(?:интернет\p{L}*|wi[ -]?fi)|(?:интернет|wi[ -]?fi)\s+нет|internet\s+(?:yo['’]?q|mavjud\s+emas)|интернет\s+йўқ)/iu,
   );
   const courtyard = bool(

package/src/housing-listing-enrichment.js

@@ -22,7 +22,7 @@ const APPLIANCE_PATTERNS = Object.freeze([
   ['Microwave', /(?:микроволнов|microwave|mikrovolnov|mikravolnof)/iu],
   ['Vacuum cleaner', /(?:пылесос|vacuum\s+cleaner|pilesos|pedesos)/iu],
   ['Oven', /(?:духовк|oven|duxovka|duhofka|dxofka)/iu],
-  ['Kitchen', /(?:кухн|kitchen|oshxona)/iu],
+  ['Kitchen', /(?:кухн|kitchen|oshxona|kuxn\p{L}*)/iu],
 ]);
 
 const FIRST_RENT_UZ_RE = /(?:hali\s+hech\s+kim\s+(?:yashamagan|turmagan)|ҳали\s+ҳеч\s+ким\s+(?:яшамаган|турмаган))/iu;

package/src/housing-money.js

@@ -8,6 +8,7 @@ import {
   parseScaledAmount,
 } from './money-core.js';
 import { maskPhoneLikeSpans } from './contact.js';
+import { DEPOSIT_TERMS, SELLER_TERMS } from './housing.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
@@ -19,11 +20,34 @@ const PRICE_KEYWORD = '(?:цена|ціна|нарх(?:и)?|narx(?:i)?|price|с
 const CURRENCY_ALT = `(?:${moneyCurrencyPattern()})`;
 const PRICE_CURRENCY_AFTER_NUMBER = `(?:${CURRENCY_ALT}(?![\\p{L}\\p{N}_]))`;
 const PRICE_CURRENCY_BEFORE_NUMBER = `(?:(?<![\\p{L}\\p{N}_])${CURRENCY_ALT})`;
+const PAYMENT_AMOUNT_TERMS = Object.freeze([
+  DEPOSIT_TERMS.deposit,
+  SELLER_TERMS.commission,
+].filter(Boolean));
+const PRICE_KEYWORD_RE = new RegExp(PRICE_KEYWORD, 'iu');
 
 function escapeRegex(value) {
   return String(value).replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
 }
 
+function isPaymentScopedAmount(text, start, end) {
+  const left = text
+    .slice(Math.max(0, start - 56), start)
+    .split(/[\r\n.;!?]/u)
+    .pop() || '';
+  const right = text
+    .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$, deposit 500$".
+  if (findCanonical(left, PAYMENT_AMOUNT_TERMS, { partial: true })) return true;
+  return !PRICE_KEYWORD_RE.test(left)
+    && Boolean(findCanonical(right, PAYMENT_AMOUNT_TERMS, { partial: true }));
+}
+
 const APPROXIMATE_RE = /около|примерно|~|≈/iu;
 
 export function parseHousingPrice(value, fallbackCurrency = '') {
@@ -78,11 +102,15 @@ export function parseHousingPrice(value, fallbackCurrency = '') {
     // 'u' is required for the \p{L}/\p{N} boundary escapes to work as
     // Unicode property classes — without it they silently match nothing,
     // which had made the boundary guard a no-op.
-    const reNumSym = new RegExp(`(${MONEY_NUMBER_PATTERN})\\s*${PRICE_CURRENCY_AFTER_NUMBER}`, 'igu');
+    // Source ads sometimes insert a stray period before the currency symbol:
+    // "500.$". Treat that punctuation as a separator, not as part of the
+    // numeric amount.
+    const reNumSym = new RegExp(`(${MONEY_NUMBER_PATTERN})\\s*[.]?\\s*${PRICE_CURRENCY_AFTER_NUMBER}`, 'igu');
     const reSymNum = new RegExp(`${PRICE_CURRENCY_BEFORE_NUMBER}\\s*(${MONEY_NUMBER_PATTERN})`, 'igu');
     for (const regex of [reNumSym, reSymNum]) {
       let match;
       while ((match = regex.exec(text)) !== null) {
+        if (isPaymentScopedAmount(text, match.index ?? 0, regex.lastIndex)) continue;
         const amount = parseNumericAmount(match[1]);
         if (amount != null && amount >= 50 && amount <= 5_000_000_000 && (tagged == null || amount > tagged)) tagged = amount;
       }
@@ -118,9 +146,12 @@ export function parseHousingPrice(value, fallbackCurrency = '') {
   }
 
   if (price == null) {
-    const matches = text.match(/\d{1,3}(?:[ \u00A0.,]\d{3})+|\d{4,}/g) || [];
+    const bareAmountRe = /\d{1,3}(?:[ \u00A0.,]\d{3})+|\d{4,}/g;
     let best = null;
-    for (const raw of matches) {
+    for (const match of text.matchAll(bareAmountRe)) {
+      const raw = match[0];
+      const start = match.index ?? 0;
+      if (isPaymentScopedAmount(text, start, start + raw.length)) continue;
       const digits = raw.replace(/[\s.,]/g, '');
       if (digits[0] === '0') continue;
       const amount = parseNumericAmount(raw);
@@ -138,4 +169,4 @@ export function parseHousingPrice(value, fallbackCurrency = '') {
 }
 
 // Compatibility name for callers migrating from Flat Finder's local parser.
-export const parsePriceFromText = parseHousingPrice;
+export const parsePriceFromText = parseHousingPrice;

package/src/housing-poi-extensions.js

@@ -23,7 +23,7 @@ export const HOUSING_LANDMARK_EXTENSIONS = Object.freeze([
     ru: ['детские сады', 'детские садики', 'детский садик', 'детский садики'],
     en: ['kindergartens', 'nurseries'],
     uzLatn: ["bog'chalar", 'bog‘chalar', 'bogʻchalar'],
-    uzCyrl: ['боғчалар'],
+    uzCyrl: ['боғчалар', 'богча', 'богчалар'],
   }, { category: 'kindergarten' }),
   group('Bus stop', {
     ru: ['автобусной остановке', 'автобусная остановка рядом', 'остановка рядом'],
@@ -59,7 +59,7 @@ export const HOUSING_LANDMARK_EXTENSIONS = Object.freeze([
     ru: ['парки', 'рядом парки'],
     en: ['parks'],
     uzLatn: ["bog'lar", 'bog‘lar', 'bogʻlar'],
-    uzCyrl: ['боғлар'],
+    uzCyrl: ['боғлар', 'сайлгоҳлар', 'сайлгох', 'сайлгохлар'],
   }, { category: 'park' }),
   group('Playground', {
     ru: ['детская площадка', 'детские площадки'],

package/src/housing-structured.js

@@ -95,7 +95,7 @@ export function parseHousingFloor(value) {
   let totalFloors = toNumber(explicit?.[2]);
 
   if (floor == null) {
-    const beforeMarker = text.match(/(?:^|[^\d])(\d{1,3})\s*-?\s*(?:этаж(?:да)?|поверх|floor|etaj|qavat(?:da)?|қабат(?:та)?|кават(?:да)?|қават(?:да)?)(?=$|[^\p{L}\p{N}_])/iu);
+    const beforeMarker = text.match(/(?:^|[^\d])(\d{1,3})\s*-?\s*(?:(?:chi|чи)\s*)?(?:этаж(?:да)?|поверх|floor|etaj|qavat(?:i(?:da(?:gi)?)?|da)?|қабат(?:ы(?:нда(?:ғы)?)?|та)?|кават(?:и(?:да(?:ги)?)?|да)?|қават(?:и(?:да(?:ги)?)?|да)?)(?=$|[^\p{L}\p{N}_])/iu);
     floor = toNumber(beforeMarker?.[1]);
   }