This is a NextJS application that allows a user to create an Atlas Search Index and tune $search queries using a graphical interface.
To help application developers who are unfamiliar with search to add sophisticated text search experiences to their apps using Atlas Search.
Novice Atlas Search users and application developers.
The app uses the NextJS framework which is based on React and NodeJS. MongoDB Leafygreen-ui components have also been used, as well as NextJS API routes.
You can run the app in development mode:
$ cd ./tunafish
$ npm run dev
Or build it for deployment:
$ cd ./tunafish
$ npm build
$ npm start
The app will run on http://localhost:3000
You can run a pre-created image:
$ docker run -t -i -d -p 5050:3000 --name tunafish --restart unless-stopped johnunderwood197/tunafish:latest
Or you can build locally in docker
$ ./build.sh
The app will run on http://localhost:5050
You will need a valid connection string and network access to an Atlas Cluster to connect.
Once connected you can select an existing search index for your collection or create a new one.
When the user opts to create a new index the app uses the mongodb-schema package (as used by Compass) to retrieve 1000 samples and generate a schema analysis.With the schema analysis the application is able to make suggestions for how to index different fields for search. The user is guided through a 'placeholder' search application which provides visual cues to help them construct the search experience for their app.
The tutorial covers three sections: facets, autocomplete (search bar), and search fields (results). When the user clicks on a section they are presented with some helpful information, links to documention, and a dropdown of suggestd fields that they can select to create a new search index defintion.
Once the index defintion has been built up the user can click 'deploy'. This creates the new index definition. The App waits for the new index to be built and when it completes will retrieve some data to show what the user has configured. This is the same view that is shown when the user selects an already existing index.
When an existing index is selected (or after a new index is built) the Query Tuner allows the user to create search queries by picking fields to search across and giving them different 'weights'. This shows the developer how to create a search query and how to apply simple boosting rules. The tuner parses the index defintion to find out which fields can be suggested for 'string' and for 'autocomplete' queries.
The user can also select what facets to display and use them for filtering. Only string facets are currently supported.
The user can now copy the generated queries that are displayed in the UI and use it in their application or for further development in a tool such as Compass.