/interviewTask

Primary LanguageJava

Isabel KUBE interview task

For this task we are developing a backend application which is used by a web application to search for available flights.

The SearchFlightService is called with the following parameters: from, to and date. Our backend uses several SearchFlightService to fetch available fligths which conform to the criteria and aggregates the results. The results are ordered from best to worst.

Feel free to create new classes, refactor existing ones and introduce libraries as you see fit.

Tasks:

  1. Implement fromTwoClients which creates a SearchFlightService by combining the results from two SearchFlightClient. For example, imagine we fetch flight data from Swissair and lastminute.com. Please order the aggregated flights using the "best" ordering strategy (see SearchResult).

  2. Several clients can return data for the same flight. For example, if we combine data from British Airways and lastminute.com, lastminute.com may include flights from British Airways. Update fromTwoClients so that if we get two or more flights with the same flightId, SearchFlightService selects the flight with the lowest unitPrice and discards the other ones.

  3. A client may occasionally throw an exception. fromTwoClients should handle the error gracefully, for example log a message and ignore the error. In other words, fromTwoClients should consider that a client which throws an exception is the same as a client which returns an empty list.

  4. Implement fromClients which behaves like fromTwoClients but it takes a list of SearchFlightClient. Note: You can use a recursion/loop/reduce to call all the clients and combine their results. Note: We can assume clients to contain less than 100 elements.

  5. Each client's search request is executed sequentially - one after another. Currently, we only send the request to client 2 after we have received the response from client 1. Instead, it would be better to send both search requests concurrently

  6. Bonus: Implement the concurrent approach for fromClients. fromClients waits for the results from every single client. This means that if one client is extremely slow, it will slow down the overall request. Implement a timeout per client so that the service doesn't spend more than 500 milliseconds per client.

There is a test implementation available for the SearchFlightClient.