@whiteslove/parsing-lexicon 0.5.3 → 0.5.5

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@whiteslove/parsing-lexicon",
-  "version": "0.5.3",
+  "version": "0.5.5",
   "description": "Shared deterministic multilingual parsing lexicon for WhitesLove housing and hiring services",
   "repository": {
     "type": "git",
@@ -176,6 +176,10 @@
       "import": "./src/housing-money.js"
     },
     "./hiring-professions": "./src/hiring-professions.js",
+    "./hiring-profession-display": {
+      "types": "./src/hiring-profession-display.d.ts",
+      "import": "./src/hiring-profession-display.js"
+    },
     "./hiring-skills": {
       "types": "./src/hiring-skills.d.ts",
       "import": "./src/hiring-skills.js"

package/src/hiring-profession-display.d.ts

@@ -0,0 +1,2 @@
+export function professionDisplayName(value: unknown): string;
+export function professionCanonicalDisplayName(canonical: unknown): string;

package/src/hiring-profession-display.js

@@ -0,0 +1,51 @@
+import { matchProfession, professionByCanonical } from './hiring-professions.js';
+
+const PROFESSION_ACRONYMS = Object.freeze(new Map([
+  ['ai', 'AI'],
+  ['api', 'API'],
+  ['bi', 'BI'],
+  ['crm', 'CRM'],
+  ['dba', 'DBA'],
+  ['erp', 'ERP'],
+  ['grc', 'GRC'],
+  ['hr', 'HR'],
+  ['ios', 'iOS'],
+  ['ml', 'ML'],
+  ['pmo', 'PMO'],
+  ['qa', 'QA'],
+  ['seo', 'SEO'],
+  ['sre', 'SRE'],
+  ['ui', 'UI'],
+  ['ux', 'UX'],
+]));
+
+function titleCaseCanonical(canonical) {
+  return String(canonical || '')
+    .split('_')
+    .filter(Boolean)
+    .map((part) => PROFESSION_ACRONYMS.get(part) || `${part.charAt(0).toUpperCase()}${part.slice(1)}`)
+    .join(' ');
+}
+
+/**
+ * Return one deterministic display label for a profession owned by this package.
+ *
+ * This is intentionally not a search transliteration API. Search folding and
+ * alias matching remain separate concerns. Unknown values are returned trimmed
+ * instead of being locally transliterated or guessed by consumers.
+ */
+export function professionDisplayName(value) {
+  const raw = String(value ?? '').trim();
+  if (!raw) return '';
+
+  const exact = professionByCanonical(raw);
+  if (exact) return titleCaseCanonical(exact.canonical);
+
+  const match = matchProfession(raw, { allowWeak: false });
+  return match ? titleCaseCanonical(match.canonical) : raw;
+}
+
+export function professionCanonicalDisplayName(canonical) {
+  const entry = professionByCanonical(String(canonical ?? '').trim());
+  return entry ? titleCaseCanonical(entry.canonical) : '';
+}

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.