Features:
- Embeddings and Similarity search within pdf, images, docx etc.
- A nice UI (Ctrl + Space) like Spotlight Search to search from anywhere.
- Settings page to run indexing, set source folder, index folder, similarity score etc.
Demo.mov
Models used:
- PyPDF2
- pytesseract
- beautifulsoup4
- python-docx
There are two portions - backend which does heavy lifting creating embeddings and frontend which communicates using rest APIs.
./install.sh
./start-frontend.sh
./start-backend.sh
The following build process creates a .dmg file that can be installed.
./build.sh
outputs into dist_electron/ folder.
Run/install it!
And then, Ctrl + Space or Cmd + Space shows/hides the search!
- Backspace to go back to input
- Esc to close
- Top/Bottom arrows to navigate and Enter to open folder