Google Keep Vibe Search

A semantic search application for your Google Keep notes export. This app lets you search through your notes using natural language queries, finding the most relevant notes based on the "vibe" of your search.

Features

• Semantic Search: Find notes based on meaning, not just exact keyword matches
• AI Chatbot: Ask questions about your notes and get AI-generated answers powered by Ollama
• Clean UI: Simple, responsive interface inspired by Google Keep
• Self-contained: Easy to set up and run locally

Requirements

• Python 3.8+
• Google Keep export from Google Takeout
• Basic understanding of command-line operations
• Ollama running locally or accessible via network (for AI chat features)

Setup Instructions

1. Export your Google Keep notes

1. Go to Google Takeout
2. Select only "Keep" from the list of Google products
3. Click "Next step" and choose your delivery method
4. Create export and download the ZIP file
5. Extract the ZIP file to a location on your computer

2. Set up the Python environment

# Clone this repository or download and extract the ZIP file
git clone https://github.com/Harduex/google-keep-vibe-search.git
cd google-keep-vibe-search

# Create a virtual environment
python -m venv venv

# Activate the virtual environment
# On Windows:
venv\Scripts\activate
# On macOS/Linux:
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# For development, set up pre-commit hooks
pip install pre-commit
pre-commit install

3. Set up Ollama (required for chat features)

1. Install Ollama from the official website
2. Start the Ollama service
3. Pull your preferred model (e.g., ollama pull llama3)

4. Configure the application

Copy the provided .env.example file to create a .env file:

# On Windows
copy .env.example .env

# On macOS/Linux
cp .env.example .env

Then edit the .env file to set your Google Keep export path and Ollama settings:

# Path to Google Keep export
GOOGLE_KEEP_PATH=D:\\Takeout\\Keep  # On Windows
# Or
GOOGLE_KEEP_PATH=/home/user/Downloads/Takeout/Keep  # On macOS/Linux

# Ollama settings (for chatbot)
OLLAMA_API_URL=http://localhost:11434
LLM_MODEL=llama3  # Or any other model you've pulled with Ollama
CHAT_CONTEXT_NOTES=5  # Number of relevant notes to use for context

5. Run the backend API

# Make sure your virtual environment is activated
python -m app.main

The application should now be running at http://127.0.0.1:8000

6. Run the React client

# Navigate to the client directory
cd client

# Install dependencies (first time only)
npm install

# Start the development server
npm run dev

The React client should now be running at http://localhost:3000 and will automatically connect to the backend API.

7. Example Usage

You can use the Vibe Search to ask questions to your notes and receive very accurate results. Here is an example of how to use the search functionality:

1. Open your browser and go to http://127.0.0.1:8000.
2. In the search bar, enter a query such as "meeting notes from last week" or "ideas for project".
3. Click the "Search" button.
4. The results will display matching notes sorted by relevance based on the "vibe" of your search.

For example, if you have a note with the content "Discussed project milestones in the meeting last week", and you search for "What are my meeting notes from the last week?", the Vibe Search will find and display this note as a relevant result.

Using the AI Chat

1. Click on the "Chat" tab in the navigation bar
2. Type your question in the chat input (e.g., "What were the main topics discussed in last month's meetings?")
3. The AI will retrieve relevant notes from your collection and use them to generate a detailed answer
4. You can see which notes were used as context for the AI's response in the right panel

Feel free to experiment with different queries to see how well the Vibe Search can find the most relevant notes based on the meaning of your search terms.

Find Related Notes

Each note in the search results has a "Show related" button that helps you discover connections between your notes:

1. When you click the "Show related" button on a note, the system performs a new search using that note's content
2. This reveals other notes that share similar topics, ideas, or themes
3. It's a great way to rediscover forgotten notes and see connections between different ideas

Docker Setup

This application can be run using Docker containers, which simplifies setup and ensures consistent environments.

Prerequisites

• Docker
• Docker Compose
• Google Keep export from Google Takeout (as mentioned in the Setup Instructions)
• Ollama service (for chat features)

Running with Docker

1. Create a .env file from the example:

cp .env.example .env

2. Edit the .env file to set your Google Keep export path and Ollama settings:

GOOGLE_KEEP_PATH=/path/to/your/Takeout/Keep
OLLAMA_API_URL=http://host.docker.internal:11434  # Access host Ollama from container
LLM_MODEL=llama3

3. Build and start the containers:

docker compose up -d

4. Access the application at http://localhost

Troubleshooting

• If the backend can't find your notes, check that the volume mapping in docker-compose.yml is correct for your Google Keep export path.
• If you encounter permission issues with the cache directory, run chmod -R 777 cache on your host machine.
• If the chatbot isn't working, ensure Ollama is running and accessible from the container using the correct URL.

Development

Code Formatting

This project uses automatic code formatting to maintain consistent code style:

• Python code is formatted with Black and isort
• JavaScript/TypeScript code is formatted with Prettier

VS Code Setup

If you're using VS Code, we've included settings that will automatically format your code on save. Just install the recommended extensions when prompted.

Manual Formatting

You can manually format the code using:

# For Python files
black app/
isort app/

# For JavaScript/TypeScript files
cd client
npm run format

How it works

The application uses semantic search to find relevant notes:

Semantic Search: Uses the sentence-transformers model to convert your notes and queries into vector embeddings and finds similar content based on meaning.

AI Chatbot: Uses Ollama's LLM capabilities to generate responses based on your notes. The system:

1. Searches for notes relevant to your question
2. Formats them as context for the LLM
3. Sends both your question and the context to Ollama
4. Returns the generated response along with the source notes used

This approach allows you to find notes that are conceptually related to your query even if they don't contain the exact same words.

Customization

You can adjust the following settings in your .env file:

• MAX_RESULTS: Maximum number of results to return
• SEARCH_THRESHOLD: Minimum relevance score (0.0 - 1.0) for search results
  • 0.0: Show all results (default)
  • 0.5: Show moderately relevant results
  • 0.7: Show highly relevant results
  • 1.0: Show only perfect matches
• HOST and PORT: Server binding settings
• OLLAMA_API_URL: URL to your Ollama API service
• LLM_MODEL: The specific model to use with Ollama (must be available in your Ollama installation)
• CHAT_CONTEXT_NOTES: Number of relevant notes to include as context for the LLM

Troubleshooting

No notes are being loaded

• Check that the GOOGLE_KEEP_PATH in your .env file points to the correct location
• Ensure the folder contains .json files from your Google Keep export

App loads slow

• The first load might take longer as the model loads and computes embeddings
• Subsequent loads should be faster

Missing dependencies

• Ensure you're using the virtual environment and have installed all requirements

Chatbot not working

• Verify that Ollama is running (ollama serve)
• Check the Ollama API URL in your .env file
• Ensure the model specified in LLM_MODEL is available in your Ollama installation
• Look at the console logs for any API connection errors

License

MIT License