This project is carried out as part of Datascientest's Data Engineering training (Fev-Dec 2021). It takes part of the evaluation process for the module entitled "FastAPI" .
For this evaluation, we consider a company creating quizzes via a smart-phone or web browser apps. The quizzes take the form of Multiple Choice Question (Questions à Choix Multiple in French, hence the name OpenQCM). The idea is to propose a service for individuals/companies to test themselves/their team's knowledge while having fun :D.
To optimize the architecture, the company wants to set up an API with the purpose of unifying the way to query a database and return a series of questions. The objective then is to create this API.
The database is represented by a csv file that the api manipulate and stock at the current directory. The original version in available at :
https://dst-de.s3.eu-west-3.amazonaws.com/fastapi_fr/questions.csv
The database contains the following fields:
- question: the title of the question
- subject: the category of the question
- correct: the list of correct answers
- use: the type of Multiple Choice Question _(_MCQ) for which this question is used
- responseA: answer A
- responseB: answer B
- responseC: response C
- responseD: the answer D (if it exists).
The API, referred to as OpenQCM from now on, is a simple HTTP REST API for technical quizzes including a wide variety of topics such as: Automation, BDD, Classification, Data Science, Docker, Machine Learning, and lots more.
A user can search for questions using the following pattern:
The user choose a type of test (use) and one or more categories (subject). The application produce MCQs of either 5, 10 or 20 questions (to be specified in user request, 5 being the default value). The API returns this number of questions in JSON format, c.f. Output format.
As the API must be able to generate many MCQs, these are returned in a random order: thus, a request with the same parameters may return different questions.
Users are expected to have an account. Credentials verification is done using basic username:password authentication: the string containing Basic username:password is passed via the Authorization header (encoded/no)
OpenQCM has many endpoint, illustrated in the following:
| Parameters | Details | |
|---|---|---|
| Get /status | None |
Verify that the API is functional. |
| GET /users/me | None |
Get the current user |
| GET /subjets | None |
List available subjects in the database |
| GET /uses | None |
List available uses in the database |
| GET /qcm | Number (type :integer): must be either 5, 10 or 20 default value : 5 Use (type :string) : must be one of available uses from the database, e,g, 'Test de positionnement', 'Test de validation', 'Total Bootcamp' Default value : 'Test de positionnement' Subjects (type list[string]) could be one or more subjects available from database.` |
- Get a MCQ from database : (1) containing 5,10 or 20 questions (2) Related to one or more subjects (category) (3) Related to one use (e.g. Test de positionnement, Test de validation, etc) |
| POST /qcm/add | question, subject, correct,use, responseA, responseB,responseC,responseD |
- Create a new question (requires admin privileges). |
The following example will output 10 random questions.
- Curl
curl -X 'http://127.0.0.1:8000/qcm/?number=10&use=Test%20de\ %20positionnement&subjects=Data%20Science&subjects= Docker&subjects=Machine%20Learning&subjects=Syst%C3%A8mes %20distribu%C3%A9s'
-H 'accept: application/json'
- Request URL
This returns a JSON object with the results in an array you can iterate over.
{
"response_code": 0,
"results": {
"use": "Test de positionnement",
"subject": [
"Docker",
"Machine Learning",
"Systèmes distribués",
"Data Science"
],
"number": 10,
"results": [
"Le théorème CAP oppose",
"Dans Hadoop, les partitioners permettent",
"Hive permet",
"Spark se différencie de Hadoop par",
"Quels sont les trois éléments constitutifs de Hadoop ?",
"Lors de l'étape de Map d'un wordcount appliqué à la phrase \"cette phrase
est une phrase\", les valeurs émises sont:",
"Docker permet de persister des changements",
"DockerHub est",
"Des containers Docker peuvent communiquer entre eux grâce à",
"Docker est utilisé"
]
}
}
Response Headers
content-length: 509
content-type: application/json
date: Wed,28 Jul 2021 18:28:31 GMT
server: uvicorn