Add Docusaurus plugin for client-side semantic search

[vaibhaviitk]

Overview

Add docusaurus-plugin-altor-vec package to enable client-side semantic search for Docusaurus sites using the altor-vec WASM vector search engine.

Motivation

Docusaurus users need semantic search capabilities without:

• Server-side infrastructure
• Third-party services (Algolia costs $0.50/1K searches)
• API keys for basic functionality
• Per-query costs

This plugin provides:

• 🚀 Zero-config semantic search
• ⚡ Client-side execution (privacy-first)
• 🎯 Semantic understanding with Transformers.js or OpenAI
• 📦 Automatic index building during build
• 🔒 No data sent to external servers

Implementation

Package Structure

• Location: packages/docusaurus-plugin-altor-vec/
• Language: TypeScript
• Architecture: Modular with SOLID principles

Features Implemented

• ✅ Content extraction from markdown files
• ✅ Embedding generation (Transformers.js & OpenAI)
• ✅ HNSW index building using altor-vec
• ✅ React search UI component
• ✅ Web Worker integration (off-main-thread)
• ✅ Full configuration system with validation
• ✅ Comprehensive error handling
• ✅ Structured logging
• ✅ i18n support
• ✅ Security: Path traversal protection, file size limits, fetch validation

Security Fixes

• Path traversal validation in ContentExtractor
• File size limits (10MB per file, 100MB index)
• Fetch response validation in Web Worker
• API key sanitization in logs

Documentation

• Comprehensive README with quick start
• Security best practices section
• Browser compatibility notes
• Troubleshooting guide
• Updated main README with plugin reference
• Updated CONTRIBUTING.md for monorepo

Testing

• ✅ TypeScript compilation successful
• ✅ All Rust tests passing (31 tests)
• ✅ Cargo clippy with no warnings
• ✅ Code properly formatted
• 🔄 Integration testing to be done post-merge

Breaking Changes

None - this is a new package addition.

Checklist

• Code follows project style guidelines
• Tests pass (Rust: 31/31)
• Documentation updated
• Security vulnerabilities addressed
• No hardcoded values
• LICENSE files consistent
• Integration testing (post-merge)

Related

• Closes #[issue-number-will-be-here]