docman/README.md

DocMan - Document Management System

AI-powered document scanning, OCR, classification, and search.

Quick Start

# Run directly (dev mode)
cd ~/dev/docman
make dev

# Or run the installed binary
~/bin/docman -port 8200

Open http://localhost:8200

Features

• Auto-processing: Drop PDFs/images into ~/documents/inbox/ → auto-classified and indexed
• PDF Preview: Built-in PDF viewer
• Full-text Search: FTS5 + semantic search
• Categories: taxes, expenses, bills, medical, contacts, legal, insurance, banking, receipts
• Expense Tracking: Filter by date, export to CSV, track tax-deductible items
• Markdown Records: Each document gets a searchable markdown file

Scanner Setup

1. Set scanner to save to SMB share:

   • Share: \\192.168.1.16\documents\inbox (or wherever james is)
   • Or use your scanner's app to save to ~/documents/inbox/

2. Workflow:

   • Scan document → lands in inbox
   • DocMan auto-processes (OCR → classify → store)
   • View/search in web UI

Directory Structure

~/documents/
├── inbox/      # Drop scans here (auto-processed)
├── store/      # PDF storage (by checksum)
├── records/    # Markdown records by category
│   ├── taxes/
│   ├── expenses/
│   ├── bills/
│   └── ...
└── index/      # SQLite database

Configuration

Environment Variables

FIREWORKS_API_KEY=fw_xxx    # Required for AI classification

Command Line Options

-port        HTTP port (default: 8200)
-data        Data directory (default: ~/documents)
-ai-endpoint AI API endpoint (default: Fireworks)
-ai-key      AI API key
-ai-model    Classification model
-embed-model Embedding model
-watch       Only watch inbox, don't start web server

Systemd Service

# Edit to add your Fireworks API key
nano ~/.config/systemd/user/docman.service

# Enable and start
systemctl --user daemon-reload
systemctl --user enable docman
systemctl --user start docman

# Check status
systemctl --user status docman
journalctl --user -u docman -f

API Endpoints

Method	Path	Description
GET	/	Dashboard
GET	/browse	Browse documents
GET	/doc/:id	View document
GET	/search	Search page
GET	/expenses	Expenses tracker
GET	/upload	Upload page
POST	/api/upload	Upload document
GET	/api/documents/:id	Get document JSON
PATCH	/api/documents/:id	Update document
DELETE	/api/documents/:id	Delete document
GET	/api/search?q=	Search API
GET	/api/expenses/export	Export CSV
GET	/api/stats	Dashboard stats

Dependencies

System packages (already installed):

• poppler-utils (pdftotext, pdfinfo, pdftoppm)
• tesseract-ocr (OCR for scanned images)

Notes

• Without FIREWORKS_API_KEY, documents will be categorized as "uncategorized"
• The inbox watcher runs continuously, processing new files automatically
• Markdown files are searchable even without embeddings