@whiteslove/parsing-lexicon 0.5.3 → 0.5.4

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.5.3",
+  "version": "0.5.4",
   "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
   "repository": {
     "type": "git",

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

@@ -3,6 +3,7 @@ import type { SalaryParseResult } from '../index.d.ts';
 export type CandidateExperienceMention = Readonly<{ years: number; context: string; approximate?: true }>;
 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 function extractCandidateDisplayName(value: unknown): string;

package/src/hiring-source-semantics.js

@@ -21,6 +21,18 @@ 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 LEAKED_CSS_RULE_RE = /\s*[.#][\w\\-]+(?:\s+[.#][\w\\-]+)?(?::[\w-]+)?\s*\{\s*(?:[\w-]+\s*:\s*[^;{}]+;?\s*)+\}\s*/gu;
+
+/** Remove source-page controls and stylesheet fragments copied into vacancy text. */
+export function cleanHiringSourceText(value) {
+  return String(value || '')
+    .replace(APPLICATION_MODAL_RE, '')
+    .replace(LEAKED_CSS_RULE_RE, ' ')
+    .replace(/[ \t]+(?=\n)/g, '')
+    .trim();
+}
+
 function entryAliases(entry, extras = []) {
   return [...new Set([entry?.canonical, ...aliasesOf(entry || {}), ...extras].filter(Boolean))];
 }

package/src/hiring-vacancy-fields.js

@@ -1,8 +1,13 @@
 import { detectDegreeLevel } from './hiring-requirements.js';
+import { cleanHiringSourceText } from './hiring-source-semantics.js';
 import { normalizeUnicode } from './normalization.js';
 
 export function detectHiringEducation(value) {
-  const text = String(value || '');
+  // Some vacancy pages append the fields of their application modal to the
+  // vacancy body. Degree choices in that UI describe the applicant form, not
+  // the vacancy requirement (for example: "Пол Мужской Женский Образование
+  // ... Высшее специалитет, магистратура Выберите вакансию").
+  const text = cleanHiringSourceText(value);
   // Vacancy requirements often state "Bachelor's or Master's degree". That is
   // an accepted-alternative list, not a Master's minimum; expose the least
   // restrictive accepted level so filters do not hide valid Bachelor holders.