Notes
The required endpoints are implemented. Bad requests get a 400 repsponse with JSON error data. The schema is in data/schema.sql.
I only had a couple of days to work on this and I really wanted to emphasise thorough testing and things which I view as best practices. Using small subpackages and handy CLI tools are two of those.
I chose sqlite3 as my datastore for the sake of simplicity and portability. I chose gin as my router because I've used it before. It is fast, really simple, and uses a 0-allocation router. I used ginkgo & gomega for testing because I like the easily nested structures that BDD frameworks give. I used gorm for the in-app database access because it makes common tasks really easy. For the data loading and for any complicated queries I just used raw SQL. I chose toml for a config file format because it's quick and simple and I like to have a command line option to generate a default config file. No package management.
Run it
-
Setup
go get ./... -
Test
ginkgo -r
-
Build
go build -
Generate your config file
./RestApiProject generate-config > config.toml -
Initialize the database
./RestApiProject init-db -
Run the server
./RestApiProject api-server
Bonus points
These features are accounted for in partial implementations, but not actually implemented. I chose to focus my time on building thorough tests.
Authenticate users
For full success it would need:
- an authentication middleware which would check for either a username and password combo or a valid auth token.
- a create user endpoint
- a change password endpoint
- a user token retreival endpoint. This would take username and password, verify the hash (already implemented using bcrypt), and return a time-limited JWT token, cryptographically signed with asynchronous keys
Coordinates
My intent here was to eventually allow users to "visit" merely by posting their current coordinates. The haversine formula can be used to find nerest points or every point within a radius. The only problem is that sqlite doesn't have mathematical functions. So I pre-calculate the sin and cos for each lat and lon. Using trigonometric identities and some algebra, it should be possible to query the database directly.