One of the things that the Engineering team at Freestar works on is software that runs ad auctions, either in the browser or server-side. The goal of this challenge is to write a program that will run a simple auction, while enforcing data validity.
Freestar manages ads on many different websites. Each site has a different configuration: which bidders are permitted to bid on ads on that site, and an auction 'floor' that bids must meet or exceed.
Each bidder also has a configuration. For this challenge we have significantly reduced the configuration to consist of an adjustment to be applied to each bid to account for the difference between how much the bidder claims that they'll pay, and how much they will actually pay, based on historical data.
On start-up, your program should load the config file (config.json) which lists
all valid sites and bidders, and their respective configurations.
The program should then load the input (JSON) from standard input that contains a list of auctions to run. Each auction lists the site, which ad units are being bid on, and a list of bids that have been requested on your behalf.
For each auction, you should find the highest valid bidder for each ad unit, after applying the adjustment factor. An adjustment factor of -0.01 means that bids are reduced by 1%; an adjustment of 0.05 would increase them by 5%. (Positive adjustments are rare in real life, but possible.) For example, a bid of $0.95 and an adjustment factor of 0.05 (adjusted to $0.9975) will beat a bid of $1.10 with an adjustment factor of -0.1 (adjusted to $0.99). When reporting the winners, use the bid amounts provided by the bidder, rather than the adjusted values.
It is possible that a bidder will submit multiple bids for the same ad unit in the same auction.
A bid is invalid and should be ignored if the bidder is not permitted to bid on this site, the bid is for an ad unit not involved in this auction, the bidder is unknown, or if the adjusted bid value is less than the site's floor.
An auction is invalid if the site is unrecognized, or there are no valid bids. In the case of an invalid auction, just return an empty list of bids.
The output of your program should be a JSON list of auction results printed to stdout.
The result of each auction is a list of winning bids. The contents of output.json
are an example of valid output. Ensure that when run as above, the only output
of your program on stdout is the auction results.
Ensuring that inputs are well-formed (e.g. all fields are present and are of the expected types) is important, but also uninteresting. You may therefore assume that all inputs will be well-formed. All numeric values will be 64-bit floats (as is default for JSON).
Sample config, inputs and expected output are included in this repo to help you test your submission.
Also included are some sample Dockerfiles for various languages, to be used for
building and testing your submission. You may use any other language you are
comfortable with, though keep in mind that the code will be reviewed by a developer.
Rename the appropriate template to Dockerfile, or create your own.
By using Docker, you can ensure that we are building and running your
submission with the same toolchain as you.
Your code will be first evaluated mechanically, running it through various test cases to check its correctness. Your code will be built and run using these commands:
$ docker build -t challenge .
$ docker run -i -v /path/to/test/config.json:/auction/config.json challenge < /path/to/test/input.jsonPlease ensure that your submission can be run this way.
If your build process requires additional steps, please note them near the top of your submission's README.md, but the end result must be a Docker image that can be run using the command above.
After running our tests, the code will be reviewed by a developer. This is an opportunity for you to demonstrate your knowledge of best practices, as well as your familiarity with your chosen language, and its library ecosystem.