@whiteslove/parsing-lexicon 0.7.5 → 0.7.7

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.7.5",
+  "version": "0.7.7",
   "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
   "repository": {
     "type": "git",

package/src/hiring-ats.d.ts

@@ -1,6 +1,6 @@
 import type { HiringSeniority, DegreeLevel, DegreeField } from './hiring-requirements.js';
 export type HiringAtsProfile = Readonly<{ raw: string; skills: Set<string>; skillEvidence: Map<string, number>; terms?: Set<string>; experienceYears?: number; seniority?: HiringSeniority; degreeLevel?: DegreeLevel; degreeFields: Set<DegreeField>; requiresUsSponsorship?: boolean }>;
-export type HiringAtsJob = Readonly<{ title: string; description?: string; tags?: string[]; skills?: string[]; niceToHave?: string[]; experienceMinYears?: number; seniority?: string | null; education?: string; country?: string; location?: string }>;
+export type HiringAtsJob = Readonly<{ title: string; description?: string; tags?: string[]; skills?: string[]; niceToHave?: string[]; experienceMinYears?: number; seniority?: string | null; education?: string; country?: string; location?: string; sponsorshipEvidence?: string[] }>;
 export function buildHiringAtsProfile(cvText: unknown, options?: { fuzzySkills?: boolean; referenceDate?: Date }): HiringAtsProfile;
 export function scoreHiringAts(profileOrCv: HiringAtsProfile | string, job: HiringAtsJob, options?: { fuzzySkills?: boolean; referenceDate?: Date }): Readonly<{ score: number; fitScore: number; eligible: boolean; blockers: readonly { code: string; label: string; critical: boolean }[]; breakdown: Readonly<{ skills: number; experience: number; seniority: number; scope: number; education: number; relevance: number }>; matched: readonly string[]; missing: readonly string[] }>;
 export function hiringAtsScoreColor(score: number): string;

package/src/hiring-ats.js

@@ -137,7 +137,10 @@ export function scoreHiringAts(profileOrCv, job, options = {}) {
   const relevance = extractTerms(keywordSource).size ? Math.round(coverage(profile.terms || extractTerms(profile.raw), extractTerms(keywordSource)) * 100) : 60;
   const breakdown = Object.freeze({ skills, experience: experience.score, seniority: seniority.score, scope: scope.score, education: education.score, relevance });
   let fitScore = Math.round(skills * .30 + experience.score * .20 + seniority.score * .20 + scope.score * .15 + education.score * .10 + relevance * .05);
-  const usText = `${job?.location || ''} ${title} ${description}`;
+  // Source-specific ATS ingestion can supply the wording that established a
+  // sponsorship policy separately from the rendered description.  It remains
+  // input evidence, not consumer-side parsing logic.
+  const usText = `${job?.location || ''} ${title} ${description} ${tagText} ${(job?.sponsorshipEvidence || []).join(' ')}`;
   const visaBlocked = (job?.country || '').toUpperCase() === 'US' || detectCountryCodeFromText(usText) === 'US'
     ? profile.requiresUsSponsorship === true && isNoSponsorshipRequirement(usText) : false;
   const blockers = visaBlocked ? [Object.freeze({ code: 'visa_sponsorship', label: 'Visa sponsorship unavailable', critical: true })] : [];

package/src/hiring-requirements.js

@@ -72,7 +72,7 @@ export function requiresUsSponsorship(value) {
 
 export function isNoSponsorshipRequirement(value) {
   const text = String(value || '');
-  return /(?:\bno\s+(?:visa\s+|immigration\s+|employment\s+)?sponsorship\b|\b(?:will\s+not|cannot|can't|unable\s+to|not\s+able\s+to)\s+sponsor\b|\bwithout\s+(?:the\s+need\s+for\s+)?(?:current\s+or\s+future\s+)?(?:employer\s+|visa\s+)?sponsorship\b|\bmust\s+(?:be\s+)?(?:legally\s+)?authoriz\w+\s+to\s+work[^.!?]{0,100}\bwithout\s+(?:current\s+or\s+future\s+)?sponsorship\b|\bsponsorship\s+(?:is\s+)?not\s+(?:available|provided|offered)\b|\bmay\s+not\s+be\s+able\s+to\b[^\n!?]{0,450}\b(?:sponsor|support|provide)\b[^\n!?]{0,180}\bsponsorship\b|\b(?:will|can|may)\s+not\b[^\n!?]{0,220}\b(?:sponsor|support|provide)\b[^\n!?]{0,160}\bsponsorship\b|\bnot\s+(?:currently\s+)?(?:able\s+to\s+)?(?:sponsor|support|provide)\b[^\n!?]{0,160}\bsponsorship\b)/i.test(text);
+  return /(?:\bno\s+(?:visa\s+|immigration\s+|employment\s+)?sponsorship\b|\b(?:will\s+not|cannot|can't|unable\s+to|not\s+able\s+to)\s+sponsor\b|\b(?:does|do)\s+not\s+(?:offer|provide|support)\s+(?:current\s+or\s+future\s+)?(?:visa\s+|employment\s+)?sponsorship\b|\bwithout\s+(?:the\s+need\s+for\s+)?(?:current\s+or\s+future\s+)?(?:employer\s+|visa\s+)?sponsorship\b|\bmust\s+(?:be\s+)?(?:legally\s+)?authoriz\w+\s+to\s+work[^.!?]{0,100}\bwithout\s+(?:current\s+or\s+future\s+)?sponsorship\b|\bsponsorship\s+(?:is\s+)?not\s+(?:available|provided|offered)\b|\bmay\s+not\s+be\s+able\s+to\b[^\n!?]{0,450}\b(?:sponsor|support|provide)\b[^\n!?]{0,180}\bsponsorship\b|\b(?:will|can|may)\s+not\b[^\n!?]{0,220}\b(?:sponsor|support|provide)\b[^\n!?]{0,160}\bsponsorship\b|\bnot\s+(?:currently\s+)?(?:able\s+to\s+)?(?:sponsor|support|provide)\b[^\n!?]{0,160}\bsponsorship\b)/i.test(text);
 }
 
 const REQUIRED_MARKER_RE = /\b(requirements?|qualifications?|minimum qualifications?|required skills?|must[- ]?have|you have|what (?:we|you) (?:are looking for|need|bring)|you(?:'|’)ll need|who you are|ideal candidate|what makes you a fit)\b|требован|квалификац|обязательн|необходим(?:о|ые|ый)|что мы (?:жд[её]м|ожидаем)|кого мы ищем|вимог|кваліфікац|обов['’]?язков|необхідн|кого ми шукаємо/i;

package/src/housing-address.js

@@ -23,6 +23,7 @@ const ADDRESS_FIELD_STOP_RE = /\s+(?:цена|ціна|нарх(?:и)?|narx|pric
 const PROPERTY_AREA_LINE_RE = /(?:^|[^\p{L}\p{N}_])(?:(?:общая|жилая|полезная|кухонная)\s+площадь|площадь\s+(?:квартиры|дома|комнаты))(?=$|[^\p{L}\p{N}_])/iu;
 const NON_ADDRESS_BARE_RE = /^(?:(?:(?:перш(?:ий|ому)|перв(?:ый|ом)|друг(?:ий|ому)|втор(?:ой|ом)|трет(?:ій|ьем|ий)|\d{1,3}(?:-?й)?)\s+(?:поверх|этаж|floor|qavat|қабат))|(?:поверх|этаж|floor|qavat|қабат)(?:\s|$)|(?:район|р-н|рн|мікрорайон|микрорайон|мкр\.?|жк|ж\.к\.|жилой\s+комплекс|житловий\s+комплекс|residential\s+complex)(?:\s|$)|(?:недалеко|поруч|рядом|біля|около|возле)(?=$|[^\p{L}\p{N}_])|(?:зупинка|остановка|станція|станция)(?:\s|$))/iu;
 const DELIMITED_STREET_REJECT_RE = /(?:^|\s)(?:город|місто|city|район|р-н|рн|мікрорайон|микрорайон|мкр|жк|метро|поверх|этаж|floor|qavat|кімнат\p{L}*|комнат\p{L}*|квартира|квартири|квартиры|оренда|аренда|продаж\p{L}*|цена|ціна|площад\p{L}*|площа|зупинка|остановка|ориентир\p{L}*|ор[-–—]?р\.?)(?:\s|$)/iu;
+const LOCATION_RELATION_RE = /(?:yonida|yaqin(?:ida)?|ro['’ʻʼ`]?parasida|near(?:by)?|close\s+to|next\s+to|рядом|возле|около|недалеко|поруч|біля|lângă|aproape)/iu;
 const UNIT_COMPONENT_PATTERNS = Object.freeze([
   String.raw`(?:^|[\s,;])(?:кв\.?|кв-ра)(?!\p{L})\s*(?:№|#)?\s*(${SECONDARY_TOKEN})(?=$|[^\p{L}\p{N}])`,
   String.raw`(?:^|[\s,;])квартира\s*(?:№|#)\s*(${SECONDARY_TOKEN})(?=$|[^\p{L}\p{N}])`,
@@ -188,9 +189,12 @@ function result(address, street = null, houseNumber = null, building = null, con
   const normalizedHouseNumber = normalizeNumber(houseNumber);
   const compactBuilding = normalizedHouseNumber?.match(/^(\d{1,5})(?:к|k)(\d{1,4})$/iu);
   const normalizedBuilding = normalizeNumber(building) || compactBuilding?.[2] || null;
-  const canonicalAddress = normalizedStreet && normalizedHouseNumber
+  // The public address is always a verified street component, optionally
+  // followed by its house/building. Never preserve a source label's whole
+  // location sentence as an address just because it was parsed on that path.
+  const canonicalAddress = normalizedStreet
     ? composeHousingAddress({ street: normalizedStreet, houseNumber: compactBuilding ? compactBuilding[1] : normalizedHouseNumber, building: normalizedBuilding })
-    : clean(address) || null;
+    : null;
   return Object.freeze({
     address: canonicalAddress,
     street: normalizedStreet,
@@ -536,7 +540,22 @@ function labelledAddress(text, rawValue) {
 
   const line = clean(rawLine).slice(0, 140);
   if (!line) return null;
-  return parseHousingAddress(line, { allowBare: true });
+  const structured = parseHousingAddress(line, { allowDelimitedBare: true });
+  if (structured.street || structured.houseNumber) return structured;
+
+  // A label alone is not evidence that arbitrary prose is a street.  Accept
+  // an unmarked form only when it is a short street-plus-house expression;
+  // landmark lists such as "Yakkasaroy, Magic City yonida" remain geo data.
+  const tail = splitAddressTail(line);
+  const candidateStreet = compactStreet(tail?.street);
+  const safeBare = Boolean(
+    tail?.houseNumber
+    && candidateStreet
+    && !LOCATION_RELATION_RE.test(candidateStreet)
+    && candidateStreet.split(/\s+/u).length <= 5
+    && candidateStreet.split(/\s+/u).every((word) => /^[\p{L}'’.-]{2,48}$/u.test(word)),
+  );
+  return safeBare ? parseHousingAddress(line, { allowBare: true }) : structured;
 }
 
 function plausibleDelimitedStreet(value) {

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

@@ -1,4 +1,4 @@
-export type HousingAudience = 'women' | 'men' | 'family';
+export type HousingAudience = 'women' | 'men' | 'family' | 'students';
 export interface HousingPerPersonPrice { amount: number; currency: string | null; approximate: boolean; scope: 'person' }
 export interface HousingCommissionAmount { amount: number; currency: string | null; approximate: boolean }
 export interface HousingListingEnrichment {

package/src/housing-listing-enrichment.js

@@ -30,12 +30,12 @@ const APPLIANCE_PATTERNS = Object.freeze([
 
 const FIRST_RENT_UZ_RE = /(?:hali\s+hech\s+kim\s+(?:yashamagan|turmagan)|ҳали\s+ҳеч\s+ким\s+(?:яшамаган|турмаган))/iu;
 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;
-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;
+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|(?:oila|oyla)(?:ga|lar|li)?\s+yoki\s+\d{1,2}\s+ta\s+bola(?:lar)?(?:ga)?\s+(?:ijara(?:ga)?\s+)?(?:beril|topshiril))/iu;
 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;
 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;
 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;
 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;
-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;
+const FAMILY_RE = /(?:семь\p{L}*|family|oila(?:ga|lar|li)?|oyla(?: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;
 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;
 const AIR_CONDITIONER_RE = /(?:кондицион|air\s*con|konditsioner|kandit(?:s|c)?aner|kanditsaner|кандитсанер)/iu;
 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;
@@ -138,7 +138,10 @@ export function parseHousingAudience(value) {
   const family = FAMILY_RE.test(text);
   const women = WOMEN_RE.test(text);
   const men = MEN_RE.test(text);
+  const students = STUDENT_RE.test(text);
+  if (family && students) return deepFreeze({ primary: 'family', alternatives: ['family', 'students'] });
   if (family && women) return deepFreeze({ primary: 'family', alternatives: ['family', 'women'] });
+  if (students) return deepFreeze({ primary: 'students', alternatives: ['students'] });
   if (family) return deepFreeze({ primary: 'family', alternatives: ['family'] });
   if (women) return deepFreeze({ primary: 'women', alternatives: ['women'] });
   if (men) return deepFreeze({ primary: 'men', alternatives: ['men'] });

package/src/housing-poi-relations.js

@@ -18,11 +18,16 @@ const TYPE_MARKERS = Object.freeze([
 ]);
 
 const RELATION_RE = /(?<relation>рядом\s+(?:с|со)|возле|около|недалеко\s+от|напротив|за|перед|near(?:by)?|close\s+to|next\s+to|opposite|behind|in\s+front\s+of|yaqin(?:ida)?|yonida|ro['’ʻʼ`]?parasida|орналасқан\s+жерде|поруч|біля|поблизу|lângă|aproape\s+de)\s+(?<target>[^,;.!?\r\n]{2,96})/giu;
+// Uzbek usually places the relation after the landmark: "Magic City yonida"
+// rather than "yonida Magic City". Keep its target span separate so it never
+// leaks into an address/street field.
+const POSTFIX_RELATION_RE = /(?<target>[^,;.!?\r\n]{2,96}?)\s+(?<relation>yonida|yaqin(?:ida)?|ro['’ʻʼ`]?parasida)(?=$|[,;.!?\r\n])/giu;
 const DISTANCE_RE = /(?<amount>\d{1,3}(?:[.,]\d+)?)\s*(?<unit>km|км|min(?:ute)?s?|мин(?:ут(?:ы|а|ах)?)?|дақиқа|daqiqa|метр(?:а|ов)?|m)\s*(?<mode>пешком|пішки|walking?|yayov|piyoda|на\s+машине|by\s+car)?\s*(?:до|от|from|to|до\s+станции)\s+(?<target>[^,;.!?\r\n]{2,96})/giu;
 
 function cleanTarget(value) {
   return String(value || '')
     .replace(/\b(?:на\s+машине|пешком|пішки|walking?|piyoda|yayov)\b/giu, ' ')
+    .replace(/(?<!\p{L})(\p{L}{3,})(?:ga|qa|ka)(?!\p{L})/giu, '$1')
     .replace(/\s+/g, ' ').trim();
 }
 
@@ -91,7 +96,7 @@ export function extractHousingPoiRelations(value, { country = '', city = '', res
   if (!text || !normalizedCountry || typeof resolveGeoCandidates !== 'function') return deepFreeze([]);
   const context = { country: normalizedCountry, city, resolveGeoCandidates };
   const relations = [];
-  for (const pattern of [RELATION_RE, DISTANCE_RE]) {
+  for (const pattern of [RELATION_RE, POSTFIX_RELATION_RE, DISTANCE_RE]) {
     for (const match of text.matchAll(pattern)) {
       const groups = match.groups || {};
       const target = groups.target || '';