Summary

DocuBrowser transforms local document collections into a private, AI-powered search engine. It targets users needing efficient, semantic search across diverse file types without cloud reliance, offering offline functionality and data privacy.

How It Works

It combines SQLite FTS5 for fast keyword search with local AI models (Ollama: nomic-embed-text for embeddings, dolphin3 for synopses) for semantic understanding. A default hybrid mode balances keyword and semantic relevance, enabling nuanced searches beyond exact matches.

Quick Start & Requirements

Distributed via native packages (RPM, DEB, tarball, zip, dmg) for easy installation. Requires Python 3.9+, Ollama (auto-installable), and Calibre for specific e-book formats. Minimum 8GB RAM, recommended 16GB RAM and 4GB+ vRAM for AI acceleration. GPU is optional but speeds up AI tasks. Detailed guides are in INSTALL.md.

Highlighted Details

• Hybrid Search: Combines FTS5 keyword and Ollama semantic search (70/30 default).
• AI Synopsis: On-demand, cached AI-generated summaries.
• Format Support: Indexes PDF, DOCX, PPTX, XLSX, ODT, ODS, ODP, EPUB, MOBI, AZW3, AZW, HTML, TXT, Markdown.
• PII Protection: Detects and blacklists documents with sensitive personal information.
• UI Features: Dark/light themes, pagination, alphabetic index, tag cloud, direct file opening.

Maintenance & Community

Maintained by a single developer (linuxrebel). No explicit community channels or sponsorships are detailed in the README.

Licensing & Compatibility

Licensed under GNU GPL-3.0-or-later. This copyleft license may restrict commercial use or integration with closed-source software.

Limitations & Caveats

DRM-encrypted e-books are metadata-only searchable. Scanned PDFs require external OCR. The system is English-only and lacks user authentication, designed strictly for local, single-user operation.