@whiteslove/parsing-lexicon 0.10.8 → 0.10.9

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.10.8",
+  "version": "0.10.9",
   "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
   "repository": {
     "type": "git",

package/src/hiring-skills.js

@@ -368,8 +368,11 @@ export function canonicalSkillName(value) {
   return CANONICAL_BY_ALIAS.get(normalized) ?? extractSkillNames(normalized)[0]
 }
 
+const PAY_BONUS_KPI_RE = /(?:\+|plus|плюс)\s*(?:%\s*\/?\s*)?kpi(?![\p{L}\p{N}_])/giu
+
 export function extractSkillDetails(text) {
-  const normalized = normalizeSkillText(cleanHiringSourceText(text))
+  // "+ KPI" / "plus KPI" after a pay figure is a bonus scheme, not a skill.
+  const normalized = normalizeSkillText(cleanHiringSourceText(text).replace(PAY_BONUS_KPI_RE, ' '))
   const found = []
   const names = new Set()
   for (const { definition, patterns } of COMPILED_SKILLS) {

package/src/hiring-source-semantics.js

@@ -76,8 +76,8 @@ 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;
+const INLINE_VACANCY_HEADING_RE = /(?:✔️?|▶️?)\s*([A-ZА-ЯЁЎҚҒҲІЇЄ][A-ZА-ЯЁЎҚҒҲІЇЄ\s()\/-]{2,100}?(?:\s*\([^)\n]{1,30}\))?)(?=\s*(?:[:•⏳✔▶📍✈📞💰]|$))/gu;
+const SHARED_HIRING_CONTACT_RE = /(?:✈️?\s*Telegram\s*:|📞\s*Контакт\s*:|👍\s*ГРАФИК\s+РАБОТЫ)/iu;
 
 /** Split a multi-vacancy source post into role-local blocks and shared context. */
 export function splitHiringVacancyEntries(value) {

package/src/money-core.js

@@ -22,7 +22,10 @@ export const MONEY_SCALE_PATTERN = 'k|к|тыс\\.?|тысяч(?:а|и)?|тис\
 // present at all) would stop matching, since "до" doesn't satisfy the
 // boundary either.
 export const MONEY_RANGE_RE = new RegExp(`(${MONEY_NUMBER_PATTERN})\\s*(?:(${MONEY_SCALE_PATTERN})(?=$|[^\\p{L}\\p{N}_]))?\\s*(?:-|–|—|до|to|bis|dan\\s+gacha)\\s*(${MONEY_NUMBER_PATTERN})\\s*(?:(${MONEY_SCALE_PATTERN})(?=$|[^\\p{L}\\p{N}_]))?`, 'iu');
-export const MONEY_SINGLE_RE = new RegExp(`(${MONEY_NUMBER_PATTERN})\\s*(${MONEY_SCALE_PATTERN})?(?=$|[^\\p{L}\\p{N}_])`, 'giu');
+// Boards often glue the currency to the number ("10 000 000сум"); without this
+// the boundary check fails on the full number and backtracks to "10 000".
+const ATTACHED_CURRENCY = "сум|сўм|so['’ʻʼ`]?m|som|uzs|usd|eur|rub|руб|грн|kzt|kgs";
+export const MONEY_SINGLE_RE = new RegExp(`(${MONEY_NUMBER_PATTERN})\\s*(${MONEY_SCALE_PATTERN})?(?=$|[^\\p{L}\\p{N}_]|(?:${ATTACHED_CURRENCY})(?![\\p{L}\\p{N}_]))`, 'giu');
 
 export function parseNumericAmount(raw) {
   let value = String(raw || '').replace(/\u00a0/g, ' ').trim();

package/src/money.js

@@ -1,4 +1,4 @@
-import { findPhoneLikeSpans } from './contact.js';
+import { findPhoneLikeSpans, parsePhoneNumbers } from './contact.js';
 import { aliasesOf, findCanonical, normalizeUnicode } from './normalization.js';
 import {
   CURRENCY_SYMBOL_CANDIDATES,
@@ -56,7 +56,15 @@ function hasModifier(text, key) {
 function protectedPhoneSpans(text) {
   return findPhoneLikeSpans(text).filter((span) => {
     const prefix = text.slice(Math.max(0, span.start - 32), span.start);
-    return CONTACT_MARKER_RE.test(prefix);
+    if (CONTACT_MARKER_RE.test(prefix)) return true;
+    // A label-less international number ("998 953 506 090", typically scraped
+    // into a pay field) is a contact, not a 12-digit salary. Require a valid
+    // E.164 number of 11+ digits so ordinary grouped salaries stay money.
+    // A "998 95… - 998 95…" pair is merged into one span, so test each part.
+    return span.raw.split(/\s*[-–—]\s*/u).some((part) => {
+      const digits = part.replace(/\D/g, '');
+      return digits.length >= 11 && parsePhoneNumbers(`+${digits}`).length > 0;
+    });
   });
 }