PDF Content Search

AI-powered PDF search with hybrid semantic capabilities using Elasticsearch 9.3 vector search and Ollama embeddings.

Features

• 🧠 AI Hybrid Search - Combines keyword matching with semantic understanding (RRF algorithm)
• 📄 Page-level PDF search with Elasticsearch 9.3 vector search
• 🔍 Multiple search modes: Hybrid AI, Exact match, Prefix match
• 🌍 AI-powered PDF translation (Ollama qwen2.5:3b, ~52s/page on CPU)
• 🔄 Async job processing with Symfony Messenger
• 📊 Analytics Dashboard - Real-time search metrics: trends, click position distribution, CTR, CSV/JSON export
• 📝 OCR for scanned PDFs - Automatic text layer via ocrmypdf (enables search & highlighting)
• 📱 Responsive Vue.js frontend with in-PDF highlighting
• 🧪 Full test suite — PHPUnit (93% PHP coverage) + Vitest (89% JS/Vue coverage, 172 tests)

Quick Start

# 1. Clone and start
git clone https://github.com/josego85/pdf-content-search.git
cd pdf-content-search
make dev

# 2. Add PDFs and index them
cp your-pdfs/*.pdf public/pdfs/
docker compose -p pdf-content-search exec php php bin/console app:index-pdfs

# 3. Open: http://localhost

What make dev does automatically:

• ✅ Installs dependencies (Composer + NPM)
• ✅ Runs database migrations
• ✅ Creates Elasticsearch index structure
• ✅ Builds frontend assets

Prerequisites: Ollama must be installed and running natively on the host before make dev. See Getting Started for setup.

Note: .env is committed with safe defaults (Symfony standard).

Common Commands

make help          # Show all available commands
make dev           # Start development (http://localhost)
make prod          # Start production (http://localhost:8080)
make down          # Stop environment
make logs          # View logs (add SERVICE=php for specific service)
make shell         # Open shell in PHP container
make test          # Run PHPUnit tests (93% PHP coverage)
make status        # Show all environments status

# Translation monitoring (helper scripts)
./bin/monitor-jobs.sh --watch   # Real-time job tracking
./bin/worker-logs.sh -f         # Worker logs

# Translation monitoring (full commands)
docker compose -p pdf-content-search exec php php bin/console app:translation:monitor        # Active jobs
docker compose -p pdf-content-search exec php php bin/console app:translation:monitor --all  # All jobs
docker compose -p pdf-content-search exec php php bin/console app:translation:monitor --watch # Watch mode

Stack

• Backend: PHP 8.4, Symfony 7.4, PostgreSQL 16
• Search: Elasticsearch 9.3 (vector search, HNSW)
• Frontend: Vue.js 3.5, Tailwind CSS 3.4, PDF.js 5.4, ApexCharts
• AI: Ollama native (qwen2.5:3b translations, nomic-embed-text embeddings)
• Queue: Symfony Messenger (3 workers)
• Analytics: PostgreSQL 16 (metrics storage), Vue.js dashboard
• Testing: PHPUnit (PHP), Vitest + happy-dom (JS/Vue)

Documentation

Getting Started

• Getting Started - Complete setup in 5 minutes
• Configuration - Environment variables & advanced settings
• Production - Deploy, optimization & security
• Testing - PHPUnit (93% PHP) + Vitest (89% JS/Vue, 172 tests)
• Troubleshooting - Common issues & solutions

Features

• Analytics Dashboard - Search metrics & KPIs (http://localhost/analytics)
• REST API - API reference & endpoints
• PDF Translation - Ollama translation & job tracking

Reference

• Frontend Architecture - Webpack, Vue.js, Tailwind

License

Licensed under GNU General Public License v3.0.