Demonstration project with SpringBoot and Elasticsearch / OpenSearch to provide an autocomplete REST endpoint from a dataset of DB stations.
- Docker with Compose V2 installed. For Compose V1 use "docker-compose" instead.
Start the development environment by running:
docker compose up -d --buildThis should build two containers, springboot and opensearch.
Requests to the autocomplete endpoint can be sent e.g. with cURL:
curl --location 'http://localhost:8080/api/v1/auto-complete/aac'In the rare case that the Spring Boot container starts before OpenSearch is ready, simply restart Spring Boot:
docker compose up -dThis could only happen after the very first build of the images, when OpenSearch needs a minute or two to load, so a second start of the container should fix the problem.
Due to the limited amount of time, some (additional) features are not fully tested and implemented:
- TODO items are left intentionally in the code, as they document improvements or possible future features:
- The Spring Boot image is quite large, which should be improved with a JDK/JRE that is better suited for this purpose
- With the permutations, it would also be possible to search for the EVA_NR and DS100 fields
- The weighting and score building of indexing and querying OpenSearch documents would improve search results
- Additional SuggestStrategies (see FVStationSuggestionStrategy) could provide autocomplete features only to e.g. "RV" stations
- There is only one unit test and no integration tests yet
- A deployment to AWS using a GitHub CI/CS pipeline was planned, but not implemented in time due to the incomplete documentation of the OpenSearch Java Client
- The basic API key check is disabled, as a deployment to AWS did not take place