Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/sebsheep/la-tournee-technical-test


https://github.com/sebsheep/la-tournee-technical-test

Last synced: 24 days ago
JSON representation

Awesome Lists containing this project

README 

        
# La Tournée - Dispatch Order Exercise



To run the app, just run:

```

docker-compose -f docker-compose.yml up

```

You can explore the API at http://localhost:8888/.

(on the first build, the migration can be launched before 

the SQL server is ready so you'd need to relaunch the 

command if this happens -- yeah that's not ideal, sorry for

that).



It will spin up server, database and test database.



To run the tests, when the above server is running:

```

docker exec -it la-tournee-technical-test_web_1 pytest

```

(you may have to rename `la-tournee-technical-test_web_1` accordingly to the

actual name of the `web` container if you renamed the cloned repo).



To format and sort imports:

```

docker exec -it la-tournee-technical-test_web_1 black .

docker exec -it la-tournee-technical-test_web_1 isort .

```



## Interesting files



### [`app/tests/test_orders.py`](app/tests/test_orders.py)



The tests are decoupled into 2 functions:

* the first one just ensure the received JSON is well 

  processed by the API is well understood.

* the second ensure the complicated logic behaves well



### [`app/models.py`](app/models.py)



The product table structure is slightly different from the

json we have in the example.



Indeed, to compute the order dispatch, we only need to

know if we can use supplier crates and in this case

how many slots there are in those crates. That's why

`packing` is an optional integer.



We also need to know the size of the product. The deposit price 

should be derived from this size (and will probably change

over time), on the contrary of what is suggested in the

store JSON. By the way, it would be safer expressing this

price in cents in order to only compute with integer,

since floating point arithmetic can be problematic for

price computations.



The `load_product_from_json` is called at server startup,

it would be maybe better having some command to launch

this but I didn't have time to search how to do this

properly with FastAPI.



### [`app/api/endpoints/orders.py`](app/api/endpoints/orders.py)



The main role of this endpoint is to ensure the requested

products exist and joining them with the Product table

to have the packing and the size of each item.



Once we have clean data, we can go through...



### [`app/core/dispatch.py`](app/core/dispatch.py)



This is the file encapsulating the complicated logic to dispatch the order. The main function is 

`to_dispatch_response` which is a "pure" function 

meaning we ca easily test it.



## Dev container



I tried to develop in VSCode dev container, the

configuration is in the `.devcontainer` folder.



This enables VSCode using the container which is used

to run the server, hence Pylance knows the libraries

used.



The experience was not entirely satisfactory: the startup

time is consequent and some features work randomly 

(e.g. "formatting on save" doesn't always work...).



## If I had more time



* Check types with MyPy

* Automating testing/formating before launching in production

  (probably in the CI)

* Write more tests for the dispatch logic, maybe some fuzzy

  tests checking we can always at least fit the products

  in the crates (maybe complicated?)

* Improve the dispatch algorithm (the tests have a sub 

  optimal case).

* Writing a separated docker architecture for production:

  the DB will probably be a managed one in AWS or so, thus

  we don't need this in the docker-compose.