This is a tutorial on how to design a simple server using flask,gevent and docker and how to handle its simple error cases.

• Structure
• Endpoints
• Error handlers
• Docker

main.py => The main file of the server containing the requests, functions and error handlers.

test_api.py => the test script for ci cd

Dockerfile => contain the serve in the format of Dockerfile Image

Requirements => Contain the libraries needed to be installed to run our server.

Alive[/alive]

This endpoint checks whether the server is alive

Request URL:

http://18.144.5.195/alive

Request Parameters

NA

Request Body

NA

Request Response

200 OK

Representation

{
  "status": "alive"
}

API test command

curl -X 'POST' \
  'http://18.144.5.195/alive' \
  -H 'accept: application/json' \
  -d ''

upload_image_to_predict[/upload-file_api]

This endpoint takes a text and transform it into its coorresponding asmo.

Request URL: http://18.144.5.195/upload-file_api/

Request Parameters

NA

Request Body

image file to predict format jpg

API test command

curl -X 'POST' \
  'http://18.144.5.195/upload-file_api/' \
  -H 'accept: application/json' \
  -H 'Content-Type: multipart/form-data' \
  -F '[email protected];type=image/jpeg'

string: This should be either an arabic text, an english text or english numbers.

Request Response

200 OK

Representation

{
  "Prediction Resut": "COVID"
}

404 In case of using a wrong endpoint.

Representation

"error":"wrong endpoint",
"Description":"you tried to access an undefined endpoint"

405 In case of using a sending unsupported data.

Representation

"error":"Unsupported input type",
"Description":"Description":"you tried to send an Unsupported input type"

500 In case of internal server error.

Representation

"error":"Internal server error",
"Description":"An internal server error has occured"

http://18.144.5.195/

• this is made by github actions

check ci_cd.png image

in this step we install the requirments in the repo then make tests wiht pytest for the apis post functions like checking server is alive, and prediction on dumy image on the image to check the model load and predictions

in this step we deploy to the aws server and setup our docker using docker-compuse then run the docker instance

action_script