@whiteslove/parsing-lexicon 0.5.9 → 0.5.12

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.5.9",
+  "version": "0.5.12",
   "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
   "repository": {
     "type": "git",

package/src/hiring-advanced.js

@@ -130,6 +130,7 @@ export function parseExperience(value) {
 
   const range = text.match(RANGE_RE);
   if (range) {
+    if (/(?:возраст|вік|age|yosh|ёш|vârsta|varsta)/iu.test(range[0]) && !/(?:опыт|experience|досвід|tajriba|тәжірибе)/iu.test(range[0])) return null;
     const minYears = Number(range[1].replace(',', '.'));
     const maxYears = Number(range[2].replace(',', '.'));
     return Object.freeze({ requirement: modifier || 'required', minYears: Math.min(minYears, maxYears), maxYears: Math.max(minYears, maxYears) });
@@ -137,6 +138,7 @@ export function parseExperience(value) {
 
   const single = text.match(MIN_RE);
   if (single) {
+    if (/(?:возраст|вік|age|yosh|ёш|vârsta|varsta)/iu.test(single[0]) && !/(?:опыт|experience|досвід|tajriba|тәжірибе)/iu.test(single[0])) return null;
     const years = Number(single[1].replace(',', '.'));
     const isExact = !/(?:от|не\s+менее|minimum|min\.?|at\s+least|від|kamida|кемінде|\+)/iu.test(single[0]);
     return Object.freeze({ requirement: modifier || 'required', minYears: years, maxYears: isExact ? years : null });

package/src/hiring-context.js

@@ -194,7 +194,7 @@ export const HIRING_SECTION_MARKERS = Object.freeze({
 export const HIRING_NON_CONTENT_TERMS = Object.freeze([
   group('vacancyDigest', { ru: ['дайджест вакансий', 'подборка вакансий', 'вакансии недели'], en: ['job digest', 'vacancy digest', 'jobs roundup'], uk: ['добірка вакансій'], ro: ['selecție de joburi'], uzLatn: ['vakansiyalar toplami'], uzCyrl: ['вакансиялар тўплами'], kk: ['вакансиялар топтамасы'] }),
   group('recruitmentAd', { ru: ['поможем подобрать сотрудника', 'услуги рекрутера', 'закроем вакансию'], en: ['recruitment services', 'we find candidates', 'staffing services'], uk: ['послуги рекрутера'], ro: ['servicii de recrutare'], uzLatn: ['rekruting xizmati'], uzCyrl: ['рекрутинг хизмати'], kk: ['рекрутинг қызметтері'] }),
-  group('jobService', { ru: ['помогу найти работу', 'подбор вакансий', 'сервис поиска работы'], en: ['job search service', 'we help you find a job'], uk: ['допоможемо знайти роботу'], ro: ['serviciu de căutare job'], uzLatn: ['ish topishga yordam'], uzCyrl: ['иш топишга ёрдам'], kk: ['жұмыс табуға көмектесеміз'] }),
+  group('jobService', { ru: ['помогу найти работу', 'подбор вакансий', 'сервис поиска работы', 'каналы с вакансиями', 'выбирай свою сферу и подписывайся'], en: ['job search service', 'we help you find a job'], uk: ['допоможемо знайти роботу'], ro: ['serviciu de căutare job'], uzLatn: ['ish topishga yordam'], uzCyrl: ['иш топишга ёрдам'], kk: ['жұмыс табуға көмектесеміз'] }),
   group('spam', { ru: ['подпишись на канал', 'заработок без вложений', 'доход без вложений', 'легкий заработок'], en: ['subscribe to our channel', 'earn money without investment', 'easy money'], uk: ['підпишись на канал', 'заробіток без вкладень'], ro: ['abonează-te la canal', 'câștig fără investiții'], uzLatn: ['kanalga obuna boling', 'sarmoyasiz daromad'], uzCyrl: ['каналга обуна бўлинг'], kk: ['арнаға жазылыңыз', 'инвестициясыз табыс'] }),
 ]);
 
@@ -207,8 +207,15 @@ const SECTION_PATTERNS = Object.freeze(Object.fromEntries(
   Object.entries(HIRING_SECTION_MARKERS).map(([key, entry]) => [key, sectionPattern(entry)]),
 ));
 
+const INLINE_SECTION_LABELS = Object.values(HIRING_SECTION_MARKERS)
+  .flatMap((entry) => aliasesOf(entry))
+  .sort((a, b) => b.length - a.length)
+  .map((value) => escapeRegex(value).replace(/\s+/g, '\\s+'))
+  .join('|');
+const INLINE_SECTION_MARKER_RE = new RegExp(`([^\\p{L}\\p{N}\\s])(?=(?:${INLINE_SECTION_LABELS})\\s*[:：—-])`, 'giu');
+
 export function splitHiringSections(value) {
-  const text = String(value || '').replace(/\r/g, '');
+  const text = String(value || '').replace(/\r/g, '').replace(INLINE_SECTION_MARKER_RE, '\n$1');
   const out = {};
   let current = null;
   for (const rawLine of text.split('\n')) {

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

@@ -6,6 +6,13 @@ export type VisaSponsorshipWording = 'offered' | 'notOffered' | null;
 export function cleanHiringSourceText(value: unknown): string;
 export function extractCandidateStructuredField(value: unknown, key: string, maxLength?: number): string | null;
 export function extractJobStructuredField(value: unknown, key: string, maxLength?: number): string | null;
+export type HiringVacancyEntry = Readonly<{ title: string; text: string }>;
+export type SplitHiringVacancyPost = Readonly<{
+  prefix: string;
+  suffix: string;
+  entries: readonly HiringVacancyEntry[];
+}>;
+export function splitHiringVacancyEntries(value: unknown): SplitHiringVacancyPost;
 export function extractCandidateDisplayName(value: unknown): string;
 export function extractCandidateExperienceMentions(value: unknown): readonly CandidateExperienceMention[];
 export function parseHiringSourceSalary(value: unknown): SalaryParseResult | null;

package/src/hiring-source-semantics.js

@@ -21,7 +21,7 @@ const FIELD_EXTRA_ALIASES = Object.freeze({
   }),
 });
 
-const APPLICATION_MODAL_RE = /(?:отправить\s+резюме\s+)?пол\s+мужской\s+женский\s+образование[\s\S]{0,500}?выберите\s+вакансию[\s\S]*/iu;
+const APPLICATION_MODAL_RE = /(?:отправить\s*резюме\s*)+(?:\[telegram\]\s*)?пол\s*мужской\s*женский\s*образование[\s\S]{0,500}?выберите\s*вакансию[\s\S]*/iu;
 const LEAKED_STYLESHEET_START_RE = /:root\s*\{\s*--[\w-]+\s*:/iu;
 const LEAKED_CSS_RULE_RE = /\s*(?:[.#][\w\\-]+|:root)(?:[^{}\n]{0,300})\{\s*(?:(?:--)?[\w-]+\s*:\s*[^;{}]+;?\s*)+\}\s*/giu;
 
@@ -62,6 +62,39 @@ export function extractJobStructuredField(value, key, maxLength = 220) {
   return extractField(value, JOB_FIELD_TERMS[key], FIELD_EXTRA_ALIASES.job[key] || [], maxLength);
 }
 
+const INLINE_VACANCY_HEADING_RE = /✔️?\s*([A-ZА-ЯЁЎҚҒҲІЇЄ][A-ZА-ЯЁЎҚҒҲІЇЄ\s()\/-]{2,100}?)(?=\s*(?:[•⏳✔📍✈📞]|$))/gu;
+const SHARED_HIRING_CONTACT_RE = /(?:✈️?\s*Telegram\s*:|📞\s*Контакт\s*:)/iu;
+
+/** Split a multi-vacancy source post into role-local blocks and shared context. */
+export function splitHiringVacancyEntries(value) {
+  const text = cleanHiringSourceText(value);
+  if (!text) return Object.freeze({ prefix: '', suffix: '', entries: Object.freeze([]) });
+
+  const headings = [...text.matchAll(INLINE_VACANCY_HEADING_RE)]
+    .map((match) => ({ start: match.index ?? 0, title: match[1].replace(/\s+/g, ' ').trim() }));
+  if (headings.length < 2) return Object.freeze({ prefix: text, suffix: '', entries: Object.freeze([]) });
+
+  let suffix = '';
+  let contentEnd = text.length;
+  const lastStart = headings.at(-1).start;
+  const contact = SHARED_HIRING_CONTACT_RE.exec(text.slice(lastStart));
+  if (contact?.index != null) {
+    contentEnd = lastStart + contact.index;
+    suffix = text.slice(contentEnd).trim();
+  }
+
+  const entries = headings.map((heading, index) => {
+    const end = Math.min(headings[index + 1]?.start ?? contentEnd, contentEnd);
+    return Object.freeze({ title: heading.title, text: text.slice(heading.start, end).trim() });
+  }).filter(({ text: entryText }) => Boolean(entryText));
+
+  return Object.freeze({
+    prefix: text.slice(0, headings[0].start).trim(),
+    suffix,
+    entries: Object.freeze(entries),
+  });
+}
+
 export function extractCandidateDisplayName(value) {
   const text = String(value || '');
   const structured = extractCandidateName(text);

package/src/hiring-work-semantics.d.ts

@@ -26,9 +26,11 @@ export type HiringWorkSchedule =
 
 export type HiringProbationKind = 'probation' | 'noProbation' | 'paidProbation' | 'unpaidProbation';
 export type HiringExperienceRequirement = 'noExperience' | 'experienceRequired';
+export type HiringWorkTimeRange = Readonly<{ start: string; end: string }>;
 
 export function detectEmploymentTypes(value: unknown): readonly HiringEmploymentType[];
 export function detectWorkModes(value: unknown): readonly HiringWorkMode[];
 export function detectWorkSchedules(value: unknown): readonly HiringWorkSchedule[];
+export function extractWorkTimeRanges(value: unknown): readonly HiringWorkTimeRange[];
 export function detectProbation(value: unknown): HiringProbationKind | null;
 export function detectExperienceRequirement(value: unknown): HiringExperienceRequirement | null;

package/src/hiring-work-semantics.js

@@ -66,6 +66,24 @@ export function detectWorkSchedules(value) {
   return Object.freeze(collectCanonical(String(value || ''), [...SCHEDULE_TERMS, ...WORK_SCHEDULE_EXTENSIONS]));
 }
 
+const WORK_TIME_RANGE_RE = /(?:^|[^\p{L}\p{N}])(?:с|from|de\s+la)?\s*([01]?\d|2[0-3])[:.]([0-5]\d)\s*(?:[-–—]|до|to|până\s+la|pana\s+la|gacha)\s*([01]?\d|2[0-3])[:.]([0-5]\d)(?!\d)/giu;
+
+/** Extract explicit working-time intervals without confusing them with contact hours. */
+export function extractWorkTimeRanges(value) {
+  const text = String(value || '');
+  const ranges = [];
+  const seen = new Set();
+  for (const match of text.matchAll(WORK_TIME_RANGE_RE)) {
+    const start = `${match[1].padStart(2, '0')}:${match[2]}`;
+    const end = `${match[3].padStart(2, '0')}:${match[4]}`;
+    const key = `${start}-${end}`;
+    if (seen.has(key)) continue;
+    seen.add(key);
+    ranges.push(Object.freeze({ start, end }));
+  }
+  return Object.freeze(ranges);
+}
+
 export function detectProbation(value) {
   const text = String(value || '');
   for (const entry of PROBATION_MATCH_ORDER) {

package/src/money.js

@@ -101,6 +101,10 @@ function rangeSearchText(text) {
   for (const symbol of Object.keys(CURRENCY_SYMBOL_CANDIDATES)) {
     normalized = normalized.split(symbol).join(' '.repeat(symbol.length));
   }
+  // Some job boards render salary parts as separate UI tokens and join their
+  // accessible text with middle dots: "2 500 · - 3 000 · €". Treat those
+  // decorative separators as whitespace while preserving source offsets.
+  normalized = normalized.replace(/[·•]/gu, ' ');
   return normalized;
 }