defect-prediction

This module predicts the smells, anti-patterns, and bugs in IaC (Ansible) and TOSCA. It also recommends the fixes.

Prerequisites

This module depends on the SODALITE sub-project “semantic-reasoner”. Thus, first built it

The information about building semantic reasoner can be found at https://github.com/SODALITE-EU/semantic-reasoner

Build Process

Ansible Smell detection extends Ansible-Lint tool by adding custom detection rules. Thus, first need to install Ansible-Lint

pip install ansible-lint

To build and run Ansible Smells Tests:

Go to the directory “defect-prediction\ansible” (via cd command) and run

python -m unittest tests/Test*.py

This requires Python 3.X with unittest module.

To build TOSCA smell detection modules:

Go to directory “defect-prediction\tosca” and run

mvn clean install

This requires maven 3.x

Deployment

The built artifact for TOSCA is a web application (.war file) that can be deployed in any Web server.

The built artifact for Ansible is a Flask Web application.

Docker Image Building and Usage: Ansible Smell Detector

sudo docker build -t sodalite/ansiblesmells .
sudo docker run -p 5000:5000 -d --name=anssAPI sodalite/ansiblesmells
sudo docker start anssAPI
sudo docker logs anssAPI
sudo docker stop anssAPI
sudo docker rm  anssAPI

Docker Image Building and Usage: TOSCA Smell Detector

sudo docker build -t sodalite/toscasmells .
sudo docker run -p 8080:8080 -d --name=tossAPI sodalite/toscasmells
sudo docker start tossAPI
sudo docker logs tossAPI
sudo docker stop tossAPI
sudo docker rm  tossAPI

Run Docker Compose

sudo docker-compose up
sudo docker image ls

REST APIs

TOSCA Smells

POST http://ip:8080/bug-predictor-api/v0.1/bugs/tosca/jsonv2

A sample request

{
	"aadmid": "z",
	"repository": "x", # optional
	"server" : "y" # optional
}

Or, Send the TOSCA file as multipart/form-data (name:” file”, value: actual file)

POST http://{serverIP}:8080/bug-predictor-api/v0.1/bugs/tosca/file

Ansible Smells

http://{serverIP}:5000/bugs/ansible/file

Send the Ansible file as multipart/form-data (name:” file”, value: actual file)

Ansible Linguistic Anti-patterns

To retrain the deep learning models (optinal)

http://{serverIP}:5000/bugs/ansible/linguistic/train

To detect/predict linguistic anti-patterns, send the Ansible file as multipart/form-data (name:” file”, value: actual file)

http://{serverIP}:5000/bugs/ansible/linguistic/file

seifashraf1/defect-prediction