JournalAI — Personal Journal with Local AI

A self-hosted journaling app with local AI-powered theme extraction, weekly reflections, and handwriting OCR. Built with FastAPI + vanilla JS. All data stays on your machine.

Features

• Markdown-based daily journal entries with auto-save
• AI-powered theme extraction (Ollama, llama3.1:8b)
• Handwriting OCR via camera/scan upload (Ollama, minicpm-v)
• Full-text search across all entries
• Weekly AI-generated reflections
• Entry statistics and theme tracking
• Multi-user support with cookie-based session auth
• File-based storage (no database needed)
• Export entries as markdown
• Fully self-hosted and private

Prerequisites

1. Python 3.8+
2. Ollama — local AI inference engine

   # Install Ollama (see https://ollama.ai)
   curl https://ollama.ai/install.sh | sh
   
   # Pull required models
   ollama pull llama3.1:8b      # text (themes, reflections)
   ollama pull minicpm-v         # vision (handwriting OCR)
   
   # Start Ollama (if not already running)
   ollama serve

Installation

1. Clone the repo:

   git clone git@github.com:Singheskan/JournalAI.git
   cd JournalAI

2. Create a virtual environment and install dependencies:

   python3 -m venv venv
   source venv/bin/activate
   pip install -r requirements.txt

Running

1. Ensure Ollama is running:

   ollama serve

2. Start the server:

   cd backend
   python main.py

3. Open http://localhost:8000 in your browser.

Usage

Writing Entries

• The app opens to today's date
• Start typing — entries auto-save after 2 seconds of inactivity
• AI automatically extracts 3-5 themes from your entry

Navigation

• Today: click "Go to Today" in the sidebar
• Browse: click any date in "Recent Entries"
• Prev/Next: arrow buttons in the editor header
• Search: full-text search across all entries

Handwriting OCR (Scan)

1. Click "Scan" in the header
2. Pick a target date and upload a photo of handwritten text
3. Preview the transcribed text
4. Click "Add to Entry" to append it to the selected date's entry

Weekly Reflection

Click "Weekly Reflection" — AI analyzes your last 7 days and generates reflection questions.

Statistics

Click "Stats" for entry count, unique themes, and most common themes.

Project Structure

JournalAI/
├── backend/
│   ├── main.py            # FastAPI app, all API endpoints
│   ├── ai_service.py      # Ollama integration (themes, reflections, OCR)
│   ├── auth.py            # Session-based authentication
│   ├── storage.py         # File-based entry storage
│   ├── models.py          # Pydantic models
│   ├── process_entries.py # Entry processing utilities
│   └── test_reflection.py # Reflection tests
├── frontend/
│   └── index.html         # Single-page app (HTML + CSS + JS)
├── data/                  # Created at runtime, gitignored
│   └── users/<name>/      # Per-user entries and metadata
├── requirements.txt
├── .gitignore
└── README.md

API Endpoints

Method	Endpoint	Description
POST	/api/register	Register a new user
POST	/api/login	Log in
POST	/api/logout	Log out
GET	/api/entry/{date}	Fetch entry for a date
POST	/api/entry/{date}	Save entry (triggers theme extraction)
GET	/api/entries	List all entry dates with themes
GET	/api/search?q={query}	Full-text search
GET	/api/themes	Unique themes with counts
GET	/api/reflection/weekly	Generate weekly reflection
GET	/api/stats	Entry statistics
POST	/api/ocr	Transcribe handwritten text from image
GET	/api/export	Export all entries as markdown

Troubleshooting

Ollama Connection Issues

1. Check if Ollama is running: ps aux | grep ollama
2. Start it: ollama serve
3. Verify models: ollama list
4. Test: ollama run llama3.1:8b "Hello"

Port Already in Use

Change the port in backend/main.py:

uvicorn.run(app, host="0.0.0.0", port=8001)

License

MIT