This project is a demo full-text-search application that compares the results from SQLite FT5 and Algolia Search platform.
Conceived as an experimental venture, this project serves as a demonstration of an unconventional monolith tech stack. It features an interactive front-end, using a mix of traditional Server Side Rendering (SSR) declarative web framework with zero custom JS:
- FastAPI the server framework;
- Jinja for the SSR templating;
- HTMX to enable front-end interactivity declaratively directly in the HTML.
Live demo (the server goes to sleep, give at least 30s for a cold start)
app-preview.mp4
The outcome of this project is something very simple and minimal. The served content is tiny and fast. There's no initial loading, everything is pre-rendered on the server, and each API request renders HTML that is injected into the DOM - no need for Hydration, Resumability nor even data serialization. It is compatible with most browsers, all the way back to IE11, where it struggles a little with style, but works.
The application works by serving a full rendered Jinja HTML template when the user navigates to a
Front-End route.
These templates are composed via smaller reusable templates (using include).
And then the templates (components) are also served, de-coupled from the whole page in the
HTML API (/html-api/...).
HTMX handles the rest, listens to DOM events and updates it when when necessary.
-
Create a
.envfile based on the.env.template.
You will need an Algolia account, more info in their Quick start guide.
More information about the env vars can be found in./text_search_app/config.py -
Setup a local environment with
make setup-venv, activate it withsource ./venv/bin/activate(or with your favourite tool). -
Install dependencies:
make install-dev. -
Start the development server:
make dev.
For deployment one would use the ./Dockerfile and set the required environment variables.
For running locally a production like build, install the dependencies with make install
and run the application with make start. That's it.
To keep this repository clean, no sample data is provided.
You can upload your own data via the /index/upload-csv endpoint.
Information about the supported CSV columns available in the Swagger UI (/docs).
Recommended datasets (adapt column names when needed):
MIT