@whiteslove/parsing-lexicon 0.2.0

package/src/hiring-skills.js

@@ -0,0 +1,315 @@
+// These canonical labels are ordinary words or one-letter tokens. Matching the
+// label itself would create noisy results; only their explicit aliases are safe.
+const AMBIGUOUS_CANONICALS = new Set(['C', 'Go', 'Make', 'REST', 'Spring'])
+
+const group = (category, subcategory, entries) =>
+  entries.map(([name, aliases = []]) => ({
+    name,
+    category,
+    subcategory,
+    aliases: [...(AMBIGUOUS_CANONICALS.has(name) ? [] : [name]), ...aliases],
+  }))
+
+// Shared by Nitro enrichment and the browser-only ATS scorer. Keep canonical
+// labels readable; aliases hold abbreviations, spelling variants and local terms.
+export const SKILL_CATALOG = [
+  ...group('IT', 'Frontend', [
+    ['HTML', ['html5']], ['CSS', ['css3']], ['Sass', ['scss']], ['Less', ['less css']],
+    ['JavaScript', ['ecmascript', 'es6', 'js developer', 'js framework']], ['TypeScript', ['type script']],
+    ['React', ['react.js', 'reactjs']], ['React Native', ['react-native']],
+    ['Vue', ['vue.js', 'vuejs']], ['Nuxt', ['nuxt.js', 'nuxtjs']],
+    ['Next.js', ['nextjs', 'next js']], ['Angular', ['angular.js', 'angularjs']],
+    ['Svelte'], ['SvelteKit', ['svelte kit']], ['Astro', ['astro.js']], ['jQuery'],
+    ['Knockout', ['knockout.js', 'knockout js']], ['AlpineJS', ['alpine.js', 'alpine js']],
+    ['HTMX'], ['Web Components', ['custom elements']], ['Tailwind', ['tailwindcss', 'tailwind css']],
+    ['Bootstrap'], ['Material UI', ['mui', '@mui/material']], ['Ant Design', ['antd']],
+    ['Vuetify'], ['Nuxt UI'], ['Quasar'], ['PrimeVue', ['prime vue']], ['Storybook'],
+    ['Redux'], ['Redux Toolkit', ['rtk', '@reduxjs/toolkit']], ['Zustand'], ['MobX'],
+    ['Pinia'], ['Vuex'], ['RxJS'], ['React Query', ['tanstack query', '@tanstack/react-query']],
+    ['Axios'], ['Fetch API', ['window.fetch']], ['Webpack'], ['Vite'], ['Rollup'], ['Babel'],
+    ['ESLint'], ['Prettier'], ['Stylelint'], ['npm'], ['Yarn'], ['pnpm'],
+    ['Freemarker'], ['Velocity'], ['Liquid'], ['Nunjucks'],
+    ['Responsive Design', ['responsive web design', 'responsive layout', 'адаптивная верстка', 'адаптивний дизайн']],
+    ['Cross-browser Development', ['cross-browser', 'cross browser', 'кроссбраузерность']],
+    ['Accessibility', ['web accessibility', 'wcag', 'a11y']], ['BEM', ['бэм']],
+    ['SEO', ['search engine optimization', 'поисковая оптимизация']],
+    ['Technical SEO', ['техническое seo']], ['i18n', ['internationalization', 'localization', 'локализация']],
+  ]),
+  ...group('IT', 'Backend', [
+    ['Node.js', ['nodejs', 'node js']], ['Express', ['express.js', 'expressjs']],
+    ['NestJS', ['nest.js', 'nest js']], ['Fastify'], ['Python', ['python3']], ['Django'],
+    ['Flask'], ['FastAPI', ['fast api']], ['PHP'], ['Laravel'], ['Symfony'], ['WordPress'],
+    ['Java', ['java ee', 'jakarta ee']], ['Spring', ['spring framework']], ['Spring Boot', ['springboot']],
+    ['Hibernate'], ['Maven'], ['Gradle'], ['Kotlin'], ['Ktor'], ['C#', ['c sharp', 'csharp']],
+    ['.NET', ['dotnet', 'dot net']], ['ASP.NET', ['asp net', 'asp.net core']],
+    ['Entity Framework', ['entityframework', 'ef core']], ['Tomcat'], ['Go', ['golang', 'go language']],
+    ['Rust', ['rustlang']], ['C', ['c programming', 'c language']], ['C++', ['cpp', 'c plus plus']],
+    ['Ruby'], ['Ruby on Rails', ['rails framework']], ['Scala'], ['Elixir'], ['Erlang'],
+    ['OOP', ['object-oriented programming', 'object oriented programming']], ['SOLID', ['solid principles']],
+  ]),
+  ...group('IT', 'API & Architecture', [
+    ['REST', ['rest api', 'restful api', 'rest services']], ['GraphQL'],
+    ['Apollo GraphQL', ['apollo client', 'apollo server']], ['gRPC'], ['WebSocket', ['websockets', 'web socket']],
+    ['Socket.IO', ['socketio']], ['API Integration', ['api integrations', 'third-party integration', 'интеграция api']],
+    ['Webhooks', ['webhook']], ['SOAP', ['soap api', 'soap web service']], ['Microservices', ['micro services']],
+    ['PWA', ['progressive web app']], ['SPA', ['single-page application']],
+    ['SSR', ['server-side rendering']], ['SSG', ['static site generation']], ['JSON'], ['XML'], ['YAML', ['yml']],
+  ]),
+  ...group('IT', 'Mobile & Desktop', [
+    ['Android'], ['Android Studio'], ['Jetpack Compose'], ['Android SDK'], ['iOS'], ['Swift'], ['Objective-C', ['objective c']],
+    ['SwiftUI'], ['UIKit'], ['Xcode'], ['Flutter'], ['Dart'], ['Electron', ['electron.js', 'electronjs']],
+    ['Tauri'], ['Qt'], ['WPF'], ['WinForms'], ['Unity', ['unity3d']], ['Unreal Engine', ['ue4', 'ue5']], ['Godot'],
+  ]),
+  ...group('IT', 'Databases', [
+    ['SQL'], ['NoSQL', ['no sql', 'no-sql']], ['PostgreSQL', ['postgres', 'psql']], ['MySQL'], ['MariaDB', ['maria db']], ['SQLite'],
+    ['Microsoft SQL Server', ['mssql', 'sql server']], ['Oracle Database', ['oracle db']],
+    ['MongoDB', ['mongo db', 'mongo']], ['Redis'], ['Elasticsearch', ['elastic search']], ['OpenSearch'],
+    ['ClickHouse', ['click house']], ['DynamoDB'], ['Firestore'], ['Firebase'], ['Neo4j'],
+    ['Supabase'], ['Snowflake'], ['BigQuery'], ['Prisma', ['prisma orm']], ['TypeORM'], ['Sequelize'], ['SQLAlchemy'],
+  ]),
+  ...group('IT', 'DevOps & Cloud', [
+    ['Git'], ['GitHub'], ['GitLab'], ['Bitbucket'], ['DevOps'], ['Docker'], ['Docker Compose', ['docker-compose']],
+    ['Kubernetes', ['k8s']], ['Helm'], ['Terraform'], ['Ansible'], ['Jenkins'],
+    ['GitHub Actions'], ['GitLab CI', ['gitlab ci/cd']], ['CircleCI'], ['TeamCity'], ['ArgoCD', ['argo cd']],
+    ['CI/CD', ['ci cd', 'cicd', 'continuous integration', 'continuous delivery']],
+    ['AWS', ['amazon web services']], ['Azure', ['microsoft azure']], ['Google Cloud', ['gcp', 'google cloud platform']],
+    ['Cloudflare'], ['Vercel'], ['Netlify'], ['Nginx'], ['Apache'], ['Linux'], ['Ubuntu'], ['Windows Server'],
+    ['Bash', ['shell scripting']], ['PowerShell'], ['SSH'], ['Active Directory'], ['VMware'], ['Grafana'],
+    ['Prometheus'], ['Zabbix'], ['Sentry'], ['ELK', ['elastic stack']], ['Kibana'],
+  ]),
+  ...group('IT', 'QA & Security', [
+    ['Jest'], ['Vitest'], ['Cypress'], ['Playwright'], ['Selenium'], ['Appium'], ['Postman'],
+    ['Swagger', ['openapi']], ['TestRail'], ['TDD', ['test-driven development']], ['BDD', ['behavior-driven development']],
+    ['Manual Testing', ['ручное тестирование']],
+    ['Test Automation', ['automation testing', 'автоматизация тестирования']],
+    ['Regression Testing', ['регрессионное тестирование']], ['Unit Testing', ['unit tests']],
+    ['Integration Testing'], ['E2E Testing', ['end-to-end testing']], ['Cybersecurity', ['information security']],
+    ['SIEM'], ['Splunk'], ['OAuth', ['oauth2']], ['JWT'], ['OWASP'], ['Penetration Testing', ['pentest', 'пентест']],
+  ]),
+  ...group('Data', 'Analytics & AI', [
+    ['Data Analysis', ['analytics', 'data analytics', 'анализ данных']], ['Business Analytics'], ['Commercial Analytics'],
+    ['Pandas'], ['NumPy'], ['Jupyter'], ['Power BI', ['powerbi']], ['Tableau'], ['Looker'], ['Qlik'],
+    ['Apache Spark', ['pyspark']], ['Hadoop'], ['Airflow'], ['Kafka'], ['RabbitMQ'], ['ETL'],
+    ['Data Warehouse'], ['Data Science'], ['Machine Learning', ['машинное обучение']], ['Deep Learning'], ['TensorFlow'],
+    ['PyTorch'], ['Scikit-learn', ['sklearn']], ['NLP'], ['Computer Vision'], ['LLM', ['large language models']],
+    ['Generative AI', ['genai', 'генеративный ии']], ['OpenAI'], ['LangChain'], ['RAG'], ['AI Tools', ['ai tools']],
+  ]),
+  ...group('Office', 'Productivity', [
+    ['Microsoft Office', ['ms office', 'мс офис']], ['Microsoft 365', ['office 365', 'm365']],
+    ['Microsoft Word', ['ms word', 'ворд']],
+    ['Microsoft PowerPoint', ['ms powerpoint', 'powerpoint']], ['Microsoft Outlook', ['ms outlook']],
+    ['Microsoft Access', ['ms access']], ['Microsoft Visio', ['ms visio']], ['Microsoft Project', ['ms project']],
+    ['Microsoft Teams', ['ms teams']], ['SharePoint'], ['OneDrive'], ['Google Workspace', ['g suite']],
+    ['Google Sheets'], ['Google Docs'], ['Google Slides'], ['LibreOffice'],
+  ]),
+  ...group('Office', 'Spreadsheets', [
+    ['Microsoft Excel', ['ms excel', 'excel', 'эксель']], ['Excel Pivot Tables', ['pivot tables', 'сводные таблицы']],
+    ['VLOOKUP', ['впр excel']], ['XLOOKUP'],
+    ['Power Query', ['powerquery']], ['Power Pivot', ['powerpivot']], ['Excel Macros', ['макросы excel']], ['VBA'],
+  ]),
+  ...group('Design', 'UI & UX', [
+    ['Figma'], ['Adobe Photoshop', ['photoshop']], ['Adobe Illustrator', ['illustrator']],
+    ['Adobe XD'], ['Sketch', ['sketch app']], ['UI Design'], ['UX Design'], ['Design Systems'], ['Prototyping'],
+  ]),
+  ...group('Business', 'Management', [
+    ['Business Strategy'], ['Strategic Planning'], ['Corporate Planning'], ['Business Planning'],
+    ['Corporate Strategy'], ['Commercial Strategy'], ['Management Consulting'], ['M&A', ['mergers and acquisitions']],
+    ['Project Management'], ['Portfolio Management'],
+    ['Corporate Governance', ['group governance']], ['Business Process'], ['Process Improvement', ['process improvements']],
+    ['Operational Efficiency'], ['Operations'], ['Stakeholder Management'], ['Negotiation'],
+    ['Cross-functional Collaboration', ['cross-functional', 'cross functionally']], ['KPI', ['key performance indicators']],
+    ['Dashboards'], ['Forecasting'], ['Risk Management'], ['Resource Planning'], ['Gantt Charts'],
+    ['Agile'], ['Scrum'], ['Kanban'], ['Jira'], ['Confluence'], ['SDLC'],
+  ]),
+  ...group('Finance', 'Accounting Software', [
+    ['1C', ['1с', '1c accounting', '1с бухгалтерия']], ['SAP', ['sap erp', 'sap fico']],
+    ['SONO', ['соно']], ['M.E.Doc', ['medoc', 'медок']], ['Didox'], ['MySoliq', ['my soliq']],
+    ['ProZorro', ['прозорро']], ['SAGA', ['saga accounting', 'saga software']],
+  ]),
+  ...group('Finance', 'Accounting & Banking', [
+    ['Accounting', ['бухгалтерский учет', 'бухгалтерський облік']],
+    ['Financial Analysis'], ['P&L', ['profit and loss']], ['Budgeting'], ['Management Accounting'],
+    ['Tax Accounting'], ['Payroll'], ['IFRS', ['мсфо']], ['GAAP'], ['Audit'], ['Treasury'],
+    ['Investment Banking'], ['Credit Analysis'], ['AML', ['anti-money laundering']], ['KYC', ['know your customer']],
+  ]),
+  ...group('Business Systems', 'CRM & ERP', [
+    ['Microsoft Dynamics', ['dynamics 365']], ['HubSpot', ['hubspot crm']],
+    ['Bitrix24', ['битрикс24']], ['amoCRM', ['амо срм']], ['Smartup', ['smart up']], ['Odoo'], ['Oracle ERP'],
+  ]),
+  ...group('Sales', 'Sales & Customer Service', [
+    ['Sales', ['продажи', 'сотув', 'savdo']], ['Sales Pipeline'], ['Lead Generation', ['лидогенерация']],
+    ['Cold Calling', ['холодные звонки']], ['Account Management'], ['Key Account Management'],
+    ['CRM'], ['Customer Service', ['обслуживание клиентов']], ['Customer Support', ['поддержка клиентов']],
+    ['Call Center', ['колл-центр']], ['Client Communication', ['переписка с клиентами', 'мижозлар билан ёзишмалар']],
+    ['Complaint Handling'], ['Onboarding', ['employee onboarding', 'courier onboarding']], ['Conversion Funnel'],
+  ]),
+  ...group('Sales', 'CRM', [
+    ['Salesforce', ['salesforce crm']],
+  ]),
+  ...group('Marketing', 'Digital Marketing', [
+    ['Digital Marketing'], ['Content Marketing'], ['Social Media Marketing', ['smm']], ['Email Marketing'],
+    ['Google Ads', ['google adwords']], ['Meta Ads', ['facebook ads']], ['Google Analytics', ['ga4']],
+    ['Google Tag Manager', ['gtm']], ['Marketing Automation'], ['Copywriting', ['копирайтинг']],
+    ['Market Research'], ['A/B Testing', ['ab testing']], ['Conversion Rate Optimization', ['cro']],
+  ]),
+  ...group('HR', 'Recruiting & People', [
+    ['Recruitment', ['recruiting', 'подбор персонала']], ['Candidate Sourcing'], ['Boolean Search'],
+    ['LinkedIn Recruiter'], ['Interviewing', ['проведение собеседований']], ['HR Administration'],
+    ['HRIS'], ['Workday HCM'], ['Greenhouse ATS'], ['BambooHR'], ['SAP SuccessFactors'],
+  ]),
+  ...group('Legal', 'Legal & Compliance', [
+    ['Legal Research', ['правовой анализ']], ['Contract Drafting', ['составление договоров']],
+    ['Contract Management', ['договорная работа']], ['Due Diligence'], ['Corporate Law'], ['Commercial Law'],
+    ['Labor Law', ['трудовое право']], ['Litigation'], ['Compliance', ['комплаенс']], ['GDPR'],
+  ]),
+  ...group('Administration', 'Office Work', [
+    ['Document Management', ['документооборот']], ['Electronic Document Management', ['эдо']],
+    ['Records Management', ['делопроизводство']], ['Business Correspondence', ['деловая переписка']],
+    ['Data Entry', ['ввод данных']], ['Office Administration'], ['Calendar Management'], ['Meeting Coordination'],
+  ]),
+  ...group('Logistics', 'Supply Chain & Transport', [
+    ['Logistics', ['логистика']], ['Supply Chain', ['управление цепями поставок']], ['Procurement', ['закупки']],
+    ['Strategic Sourcing'], ['Tendering', ['тендеры']], ['Vendor Management'], ['Inventory Management'],
+    ['Warehouse Management'], ['WMS'], ['TMS'], ['Incoterms'], ['Customs Clearance'], ['Import/Export'],
+    ['Freight Forwarding'], ['Route Planning', ['маршрутизация']], ['Last-mile Logistics'],
+    ['Driving License B', ['права категории b']], ['Driving License C', ['права категории c']],
+    ['Forklift', ['водитель погрузчика']], ['Barcode Scanner', ['тсд', 'терминал сбора данных']],
+  ]),
+  ...group('Retail & Hospitality', 'Retail, POS & HoReCa', [
+    ['POS', ['point of sale']], ['R-Keeper', ['rkeeper']], ['iiko', ['айко ресторан']],
+    ['Poster POS', ['poster restaurant']], ['MICROS', ['oracle micros']], ['Cash Register', ['работа с кассой']],
+    ['Merchandising', ['мерчандайзинг']], ['Stocktaking', ['инвентаризация']], ['HACCP', ['хассп']],
+    ['Hotel Management'], ['Opera PMS', ['oracle hospitality opera']], ['Restaurant Management'],
+  ]),
+  ...group('Engineering', 'CAD', [
+    ['AutoCAD', ['autocad civil 3d']], ['SolidWorks'], ['Revit'], ['ArchiCAD'], ['SketchUp'],
+  ]),
+  ...group('Engineering', 'Manufacturing', [
+    ['CNC', ['чпу']], ['PLC', ['плк']], ['SCADA', ['асу тп']], ['TIA Portal'], ['Siemens S7'],
+    ['Electrical Engineering'], ['Mechanical Engineering'], ['Mechatronics'], ['Welding', ['сварка']],
+    ['Lean Manufacturing', ['бережливое производство']], ['Six Sigma'], ['Kaizen'], ['5S'],
+    ['ISO 9001', ['iso9001']], ['ISO 14001'], ['ISO 45001'], ['Quality Management'], ['HSE', ['охрана труда']],
+  ]),
+  ...group('Construction', 'Construction & Design', [
+    ['Construction Management'], ['Quantity Surveying'], ['Cost Estimation', ['сметное дело', 'составление смет']],
+    ['Project Documentation'], ['Technical Drawings', ['технические чертежи']], ['BIM'],
+  ]),
+  ...group('Healthcare', 'Medical & Pharma', [
+    ['EMR', ['electronic medical record']], ['EHR', ['electronic health record']],
+    ['Medical Records', ['медицинская документация']], ['GMP'], ['GLP'],
+    ['Good Clinical Practice', ['gcp clinical']], ['PCR', ['пцр']], ['ELISA', ['ифа анализ']],
+    ['Laboratory Equipment'], ['Pharmacovigilance'], ['Clinical Research'], ['Patient Care'],
+  ]),
+  ...group('Education', 'Teaching & Learning', [
+    ['Teaching', ['преподавание', 'викладання']], ['Curriculum Development'], ['Lesson Planning'],
+    ['Classroom Management'], ['Moodle'], ['Google Classroom'], ['LMS', ['learning management system']],
+  ]),
+  ...group('Aviation', 'Aviation Operations', [
+    ['Airport Operations'], ['Flight Operations'], ['Ground Handling'], ['Aviation Safety'],
+    ['IATA'], ['ICAO'], ['Amadeus'], ['Sabre'],
+  ]),
+  ...group('E-commerce', 'Marketplaces', [
+    ['E-commerce', ['ecommerce', 'электронная коммерция', 'онлайн савдо']], ['Shopify'], ['Magento'],
+    ['OpenCart'], ['Amazon Seller Central'], ['Wildberries', ['вайлдберриз']], ['Ozon'],
+    ['Kaspi', ['kaspi marketplace', 'kaspi магазин']], ['Uzum', ['uzum market']],
+    ['Marketplace Management', ['ведение маркетплейсов']], ['Product Cards', ['карточки товаров']],
+  ]),
+  ...group('Automation', 'Low Code', [
+    ['Automation'], ['Zapier'], ['Make', ['make.com', 'integromat']], ['n8n'], ['Power Automate'], ['Power Apps'],
+    ['UiPath'], ['Automation Anywhere'], ['RPA', ['robotic process automation']],
+  ]),
+  ...group('Professional', 'Soft Skills', [
+    ['Communication', ['communication skills', 'коммуникация', 'коммуникабельность', 'коммуникация кўникмалари']],
+    ['Teamwork', ['team player', 'работа в команде', 'командная работа', 'жамоа билан']],
+    ['Leadership'], ['Time Management'], ['Problem Solving', ['решение проблем']], ['Critical Thinking'],
+    ['Analytical Thinking'], ['Multitasking', ['многозадачность']], ['Attention to Detail'],
+    ['Presentation Skills'], ['Responsibility', ['ответственность', 'масъулиятлилик']],
+    ['Adaptability', ['адаптивность', 'мослашувчан']], ['Fast Learner', ['быстрая обучаемость', 'тез ўрганувчи']],
+    ['Goal Orientation', ['стремление к цели', 'мақсадларга эришиш']],
+  ]),
+]
+
+export const SKILL_KEYWORDS = Object.freeze(
+  Object.fromEntries(SKILL_CATALOG.map(({ name, aliases }) => [name, Object.freeze([...new Set(aliases)])])),
+)
+
+export const SKILL_META = Object.freeze(
+  Object.fromEntries(SKILL_CATALOG.map(({ name, category, subcategory }) => [name, { category, subcategory }])),
+)
+
+export function normalizeSkillText(value) {
+  return value
+    .normalize('NFKC')
+    .toLocaleLowerCase('en')
+    .replace(/[‘’`´]/g, "'")
+    .replace(/[‐‑‒–—]/g, '-')
+    .replace(/\s+/g, ' ')
+    .trim()
+}
+
+export function escapeRegex(value) {
+  return value.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')
+}
+
+export function buildSkillRegex(alias) {
+  const normalized = normalizeSkillText(alias)
+  const pattern = escapeRegex(normalized).replace(/\s+/g, '\\s+')
+  // Unicode-aware boundaries prevent `react` matching `reactive`, `reaction`
+  // or `interaction`, while still handling C++, C#, .NET, 1C and Vue.js.
+  return new RegExp(`(?:^|[^\\p{L}\\p{N}_])${pattern}(?=$|[^\\p{L}\\p{N}_])`, 'iu')
+}
+
+const CONTEXTUAL_SKILL_PATTERNS = Object.freeze([
+  ['Corporate Governance', /\b(?:group|company|corporate) governance\b/i],
+  ['Data Analysis', /\banalys(?:e|is|ing)\b[^.;\n]{0,60}\b(?:data|metrics?|funnels?)\b|\b(?:data|metrics?|funnels?)\b[^.;\n]{0,60}\banalys(?:e|is|ing)\b/i],
+  ['Conversion Funnel', /\b(?:conversion|onboarding|registration)[- ](?:to[- ]\w+\s+)?funnel\b|\bonboarding funnel\b/i],
+  ['Cross-functional Collaboration', /\bcross[- ]function(?:al|ally)\b/i],
+])
+
+const COMPILED_SKILLS = SKILL_CATALOG.map((definition) => ({
+  definition,
+  patterns: [...new Set(definition.aliases.map(normalizeSkillText))].map(buildSkillRegex),
+}))
+
+const CANONICAL_BY_ALIAS = new Map()
+for (const { name, aliases } of SKILL_CATALOG) {
+  CANONICAL_BY_ALIAS.set(normalizeSkillText(name), name)
+  for (const alias of aliases) CANONICAL_BY_ALIAS.set(normalizeSkillText(alias), name)
+}
+
+export function canonicalSkillName(value) {
+  const normalized = normalizeSkillText(value)
+  return CANONICAL_BY_ALIAS.get(normalized) ?? extractSkillNames(normalized)[0]
+}
+
+export function extractSkillDetails(text) {
+  const normalized = normalizeSkillText(text)
+  const found = []
+  const names = new Set()
+  for (const { definition, patterns } of COMPILED_SKILLS) {
+    if (!patterns.some((pattern) => pattern.test(normalized))) continue
+    found.push({
+      name: definition.name,
+      category: definition.category,
+      subcategory: definition.subcategory,
+    })
+    names.add(definition.name)
+  }
+  for (const [name, pattern] of CONTEXTUAL_SKILL_PATTERNS) {
+    if (names.has(name) || !pattern.test(normalized)) continue
+    const meta = SKILL_META[name]
+    if (!meta) continue
+    found.push({ name, ...meta })
+    names.add(name)
+  }
+  return found
+}
+
+export function extractSkillNames(text) {
+  return extractSkillDetails(text).map(({ name }) => name)
+}
+
+export function getSkillMeta(name) {
+  const canonical = canonicalSkillName(name)
+  return canonical ? SKILL_META[canonical] : undefined
+}

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

@@ -0,0 +1,32 @@
+export interface ExtendedProfessionMatch {
+  canonical: string;
+  group: string;
+  family: string;
+  score: number;
+  strength: 'strong' | 'weak';
+  matched: string;
+  index: number;
+  label: string;
+}
+
+export interface SourceProfessionAlias {
+  canonical: string;
+  label: string;
+  group: string;
+  aliases: readonly string[];
+  re: RegExp;
+}
+
+export interface SourceRoleNormalization {
+  canonical: string;
+  label: string;
+  re: RegExp;
+}
+
+export const SOURCE_PROFESSION_ALIASES: readonly SourceProfessionAlias[];
+export const SOURCE_ROLE_NORMALIZATION_RULES: readonly SourceRoleNormalization[];
+export const SOURCE_CANDIDATE_INTENT_ALIASES: readonly string[];
+export function normalizeSourceRole(value: unknown): Readonly<{ canonical: string; label: string }> | null;
+export function matchesSourceCandidateIntent(value: unknown): boolean;
+export function professionDisplayLabel(canonical: string): string;
+export function matchExtendedProfessions(value: unknown, options?: { limit?: number; allowWeak?: boolean }): readonly ExtendedProfessionMatch[];

package/src/hiring-source-aliases.js

@@ -0,0 +1,186 @@
+import { aliasesToRegex } from './normalization.js';
+import { matchProfessions } from './hiring-professions.js';
+
+const sourceRole = (canonical, label, aliases, group = 'other') => Object.freeze({
+  canonical,
+  label,
+  group,
+  aliases: Object.freeze([...new Set(aliases)]),
+  re: aliasesToRegex([...new Set(aliases)]),
+});
+
+/**
+ * Source spellings seen in real CV feeds which are too source-specific or
+ * colloquial for the core profession rows. They still live in the shared
+ * package so consumers do not grow their own profession regex catalogs.
+ */
+export const SOURCE_PROFESSION_ALIASES = Object.freeze([
+  sourceRole('chief_executive_officer', 'Chief Executive Officer', ['ceo', 'chief executive officer', 'генеральный директор', 'гендиректор', 'виконавчий директор'], 'management'),
+  sourceRole('chief_technology_officer', 'Chief Technology Officer', ['cto', 'chief technology officer', 'технический директор', 'технічний директор'], 'management'),
+  sourceRole('sales_manager', 'Sales Manager', ['sales executive', 'head of sales', 'менеджер экспортных продаж', 'менеджер по экспортным продажам', 'роп', 'sotuv menejer'], 'sales'),
+  sourceRole('driver', 'Driver', ['се категория', 'се категория буйича', 'ce категория', 'ce category', 'shafyor'], 'transport'),
+  sourceRole('accountant', 'Accountant', ['bugalteriya', 'buxgalteriya', 'buhgalteriya'], 'finance'),
+  sourceRole('cashier', 'Cashier', ['kassa xodimi', 'kassa mudiri'], 'retail'),
+  sourceRole('notary', 'Notary', ['notary', 'notarius', 'нотариус'], 'legal'),
+  sourceRole('metrology_specialist', 'Metrology Specialist', ['metrologiya', 'metrology specialist', 'метролог', 'standartlashtirish'], 'manufacturing'),
+  sourceRole('security_guard', 'Security Guard', ['xavfsizlik', 'qoriqlash', 'qo‘riqlash', "qo'riqlash", 'охорона'], 'security'),
+  sourceRole('finance_banking_specialist', 'Finance / Banking Specialist', ['finance specialist', 'banking specialist', 'moliya', 'soliq', 'bank'], 'finance'),
+  sourceRole('teacher', 'Teacher', ['tyutorlik', 'тьютор', 'titur', "o'qituvchilik", 'o‘qituvchilik', 'oʻqituvchilik', 'ustoz'], 'education'),
+  sourceRole('english_teacher', 'English Teacher', ['ingliz tili ustoziman', 'ingliz tili ustoz', 'ingliz tili oqituvchi', "ingliz tili o'qituvchi"], 'education'),
+  sourceRole('it_specialist', 'IT Specialist', ['it specialist', 'itishnik', 'it ishnik', 'kompyuter boyicha ish', "kompyuter bo'yicha ish", 'kompyuter bo‘yicha ish'], 'infrastructure'),
+  sourceRole('biotechnologist', 'Biotechnologist', ['biotechnologist', 'биотехнолог', 'biotexnolog'], 'medicine'),
+  sourceRole('laboratory_technician', 'Laboratory Technician', ['laborant'], 'medicine'),
+  sourceRole('media_specialist', 'Media Specialist', ['media specialist', 'специалист по сми', 'matbuot'], 'media_content'),
+  sourceRole('engineer', 'Engineer', ['engineer', 'инженер', 'інженер', 'muhandis', 'injiner'], 'engineering'),
+  sourceRole('chat_operator', 'Chat Operator', ['chat operatori'], 'customer_support'),
+  sourceRole('call_center_operator', 'Call Center Operator', ['koll-markaz operatori', 'koll markaz operatori', 'call-markaz operatori'], 'customer_support'),
+  sourceRole('consultant', 'Consultant', ['consultant', 'консультант', 'консультантка'], 'commercial'),
+  sourceRole('supervisor', 'Supervisor', ['supervisor', 'супервайзер', 'начальник отряд', 'начальник отряда'], 'management'),
+  sourceRole('quality_inspector', 'Quality Inspector', ['quality inspector', 'инспектор по качеству', 'інспектор з якості'], 'manufacturing'),
+  sourceRole('oil_gas_worker', 'Oil & Gas Worker', ['oil and gas', 'oil & gas', 'нефть и газ', 'нефтегаз', 'neft va gaz', 'neft vagaz', 'neft vagaz sohasida', 'neft va gaz sohasida'], 'manufacturing'),
+  sourceRole('cybersecurity_specialist', 'Cybersecurity Specialist', ['cybersecurity specialist', 'ciso', 'руководитель по информационной безопасности', 'руководитель информационной безопасности'], 'information_security'),
+  sourceRole('architect', 'Architect', ['arxitektor loyihachi', 'arxitektor'], 'construction'),
+  sourceRole('economist', 'Economist', ['iqtisodchi', 'iqtsodchi', 'iqtisodiy'], 'finance'),
+  sourceRole('logistics_manager', 'Logistics Specialist', ['logist', 'logistics specialist', 'логист'], 'logistics'),
+  sourceRole('frontend_developer', 'Frontend Developer', ['frontet', 'frontet developer', 'frontet dasturchi'], 'software_development'),
+  sourceRole('nanny', 'Nanny', ['bolalarga qarash', 'bolaga qarash'], 'education'),
+  sourceRole('loader', 'Loader', ['грузчиком', 'грузчика', 'грузчик'], 'logistics'),
+  sourceRole('welder', 'Welder', ['сварщиком', 'сварщица', 'зварювальником'], 'construction'),
+]);
+
+const sourceRoleNormalization = (canonical, label, re) => Object.freeze({ canonical, label, re });
+
+/**
+ * Anchored or strongly contextualized raw-role repairs. These rules are kept
+ * separate from SOURCE_PROFESSION_ALIASES so generic values such as `model`,
+ * `online`, or `bank` never become profession matches in arbitrary CV prose.
+ */
+export const SOURCE_ROLE_NORMALIZATION_RULES = Object.freeze([
+  sourceRoleNormalization('economist', 'Economist', /^iqt(?:i)?sodchi$/iu),
+  sourceRoleNormalization('economist', 'Economist', /^iqtisodiy$/iu),
+  sourceRoleNormalization('logistics_manager', 'Logistics Specialist', /^logist(?:ika)?(?:\s+updater)?$/iu),
+  sourceRoleNormalization('english_teacher', 'English Teacher', /^ingliz\s+tili\s+ustoz(?:iman)?$/iu),
+  sourceRoleNormalization('frontend_developer', 'Frontend Developer', /mobilagraf[\s\S]*itishnik[\s\S]*front(?:et|ent|end)/iu),
+  sourceRoleNormalization('nanny', 'Nanny', /farqi\s+yo[\s\S]*bolalarga\s+qarash/iu),
+  sourceRoleNormalization('sales_manager', 'Sales Manager', /^(?:sales\s+executive(?:\s+ind)?|роп(?:,?\s*sales\s+executive)?)$/iu),
+  sourceRoleNormalization('operative_officer', 'Operative Officer', /^(?:оперативник|оперуполномоченн\p{L}*|оперативный\s+уполномоченн\p{L}*)$/iu),
+  sourceRoleNormalization('water_supply_specialist', 'Water Supply Specialist', /^(?:suv\s+ta['’ʻʼ‘`]?minoti|водоснабжение)$/iu),
+  sourceRoleNormalization('any_role', 'Any Role', /^(?:onlayn(?:\s+ish(?:chi)?)?|online(?:\s+ish(?:chi)?)?|онлайн|удал[её]нно|remote(?:\s+work)?|boshqa\s+ishlar?|farqi\s+(?:yo['’ʻʼ‘`]?q|yuq)|tungi|bilmaym\p{L}*)$/iu),
+  sourceRoleNormalization('restaurant_cafe_worker', 'Restaurant / Cafe Worker', /^ищу\s+работу\s+(?:в\s+)?(?:кафе|ресторанах?|кафе\s+или\s+ресторанах)$/iu),
+  sourceRoleNormalization('driver', 'Driver', /^(?:xaydovchilik|haydovchilik|shafyorlik|shofyorlik)/iu),
+  sourceRoleNormalization('retail_worker', 'Retail Worker', /^do['’ʻʼ‘`]?kon$/iu),
+  sourceRoleNormalization('salesperson', 'Salesperson', /^(?:savdo|sotuvchi)$/iu),
+  sourceRoleNormalization('pharmacist', 'Pharmacist', /^(?:dorishunos|farmatsevt)$/iu),
+  sourceRoleNormalization('notary_assistant', 'Notary Assistant', /^(?:natarus|notarius)\s+yordamchisi/iu),
+  sourceRoleNormalization('librarian', 'Librarian', /^kutubxonachi$/iu),
+  sourceRoleNormalization('singer_vocalist', 'Singer / Vocalist', /^(?:vokal\s*:\s*)?xonanda$/iu),
+  sourceRoleNormalization('model', 'Model', /^model$/iu),
+  sourceRoleNormalization('flight_attendant', 'Flight Attendant', /^bortprovodnik$|^бортпроводник$/iu),
+  sourceRoleNormalization('hvac_technician', 'HVAC Technician', /^(?:konditsaner|kanditsaner|konditsioner)/iu),
+  sourceRoleNormalization('mobile_content_creator', 'Mobile Content Creator', /^mobilografiya(?:\s+bo['’ʻʼ‘`]?yicha)?$/iu),
+  sourceRoleNormalization('cctv_intercom_technician', 'CCTV / Intercom Technician', /kamera\s+(?:dama?fon|domofon)|domofon\s+xizmat/iu),
+  sourceRoleNormalization('internal_control_specialist', 'Internal Control Specialist', /^ichki\s+nazoratchi$/iu),
+  sourceRoleNormalization('brand_ambassador', 'Brand Ambassador', /^(?:бренд\s+фейс|brand\s+face)$/iu),
+  sourceRoleNormalization('insurance_specialist', 'Insurance Specialist', /^sug['’ʻʼ‘`]?urta$/iu),
+  sourceRoleNormalization('bank_operations_specialist', 'Bank Operations Specialist', /^стаж[её]р\s+операционист|^операционист$/iu),
+  sourceRoleNormalization('commercial_director', 'Commercial Director', /^коммерческ\p{L}*\s+директор|\bchief\s+commercial\s+officer\b|\bCCO\b/iu),
+  sourceRoleNormalization('security_specialist', 'Security Specialist', /^по\s+безопасност\p{L}*\s+объекта$/iu),
+  sourceRoleNormalization('healthcare_specialist', 'Healthcare Specialist', /^mededsina$|^meditsina$|^медицина$/iu),
+  sourceRoleNormalization('tourism_hospitality_specialist', 'Tourism / Hospitality Specialist', /^mehmonxona[^\n]*turfirma|^turfirma[^\n]*mehmonxona/iu),
+  sourceRoleNormalization('confectioner', 'Confectioner', /qandolat|qandolatchi/iu),
+]);
+
+export function normalizeSourceRole(value) {
+  const text = String(value || '').trim();
+  if (!text) return null;
+  const match = SOURCE_ROLE_NORMALIZATION_RULES.find((entry) => entry.re.test(text));
+  return match ? Object.freeze({ canonical: match.canonical, label: match.label }) : null;
+}
+
+export const SOURCE_CANDIDATE_INTENT_ALIASES = Object.freeze([
+  'работу ищу',
+  'роботу шукаю',
+  'срочно работу ищу',
+  'терміново роботу шукаю',
+  'могу работать',
+  'можу працювати',
+]);
+
+const SOURCE_CANDIDATE_INTENT_RE = aliasesToRegex(SOURCE_CANDIDATE_INTENT_ALIASES);
+
+export function matchesSourceCandidateIntent(value) {
+  return SOURCE_CANDIDATE_INTENT_RE.test(String(value || ''));
+}
+
+const LABELS = new Map(SOURCE_PROFESSION_ALIASES.map((entry) => [entry.canonical, entry.label]));
+const ACRONYMS = new Map([
+  ['qa', 'QA'], ['hr', 'HR'], ['ui', 'UI'], ['ux', 'UX'], ['ai', 'AI'], ['ml', 'ML'],
+  ['seo', 'SEO'], ['sre', 'SRE'], ['dba', 'DBA'], ['crm', 'CRM'], ['erp', 'ERP'], ['pmo', 'PMO'], ['it', 'IT'],
+]);
+
+export function professionDisplayLabel(canonical) {
+  if (!canonical) return '';
+  const explicit = LABELS.get(canonical);
+  if (explicit) return explicit;
+  return String(canonical).split('_').map((part) => ACRONYMS.get(part) || `${part.charAt(0).toUpperCase()}${part.slice(1)}`).join(' ');
+}
+
+function sourceMatches(text) {
+  const matches = [];
+  for (const entry of SOURCE_PROFESSION_ALIASES) {
+    const match = String(text || '').match(entry.re);
+    if (!match) continue;
+    matches.push({
+      canonical: entry.canonical,
+      group: entry.group,
+      family: entry.group,
+      score: 1,
+      strength: 'strong',
+      matched: match[0].trim(),
+      index: match.index ?? 0,
+      label: entry.label,
+    });
+  }
+  return matches;
+}
+
+/**
+ * Combine the core taxonomy with real-source aliases. More specific matches
+ * suppress contained generic matches ("Chief Accountant" must not also become
+ * "Accountant"), while separate spans remain separate professions.
+ */
+export function matchExtendedProfessions(value, { limit = 5, allowWeak = true } = {}) {
+  const text = String(value || '');
+  if (!text) return [];
+  const base = matchProfessions(text, { limit: Math.max(limit * 4, 24), allowWeak }).map((match) => {
+    const index = text.toLocaleLowerCase().indexOf(String(match.matched || '').toLocaleLowerCase());
+    return { ...match, index: index < 0 ? 0 : index, label: professionDisplayLabel(match.canonical) };
+  });
+  const matches = [...base, ...sourceMatches(text)]
+    .sort((a, b) => b.score - a.score || b.matched.length - a.matched.length || a.index - b.index);
+
+  const selected = [];
+  const canonicals = new Set();
+  for (const match of matches) {
+    if (canonicals.has(match.canonical)) continue;
+    const start = match.index;
+    const end = start + match.matched.length;
+    const contained = selected.some((chosen) => {
+      const chosenStart = chosen.index;
+      const chosenEnd = chosenStart + chosen.matched.length;
+      return chosen.score >= match.score && chosenStart <= start && chosenEnd >= end && chosen.matched.length > match.matched.length;
+    });
+    if (contained) continue;
+    canonicals.add(match.canonical);
+    selected.push({ ...match });
+  }
+
+  // A concrete software role outranks a generic "IT specialist" mention in
+  // the same role/title text, even when the two aliases occupy separate spans.
+  const hasSoftwareRole = selected.some((match) => match.group === 'software_development');
+  const filtered = hasSoftwareRole
+    ? selected.filter((match) => match.canonical !== 'it_specialist')
+    : selected;
+
+  return Object.freeze(filtered.slice(0, limit).map((match) => Object.freeze(match)));
+}

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

@@ -0,0 +1,31 @@
+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 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;
+export function extractCandidateExperienceMentions(value: unknown): readonly CandidateExperienceMention[];
+export function parseHiringSourceSalary(value: unknown): SalaryParseResult | null;
+export function parseCandidateSalary(value: unknown, country?: string): SalaryParseResult | null;
+export function detectUsLocation(value: unknown): boolean;
+export function detectVisaSponsorshipWording(value: unknown): VisaSponsorshipWording;
+export const TEMPORARY_WORK_AUTH_RE: RegExp;
+export function detectRecruitmentAgency(value: unknown): boolean;
+export function extractNiceToHaveContext(value: unknown, maxLength?: number): string;
+export type CandidatePostSignals = Readonly<{
+  candidateForm: boolean;
+  cvMarker: boolean;
+  firstPerson: boolean;
+  personalProfile: boolean;
+  contact: boolean;
+  emptyRecommendation: boolean;
+  sectionCount: number;
+}>;
+export function detectCandidatePostSignals(value: unknown): CandidatePostSignals;
+export function extractCandidateStructuredBlock(value: unknown, key: string, maxLength?: number): string | null;
+export function defaultHiringCurrency(country: unknown): string | null;
+export function isHiringCharityAppeal(value: unknown): boolean;
+export function isHiringRecruitingOpportunity(value: unknown): boolean;
+export function sameHiringProfessionFamily(a: unknown, b: unknown): boolean;