DavidCardoso / my-ez-cli

CLI Tools using Docker + Shell Script.

Home Page:https://github.com/My-Tooling/my-ez-cli

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

My Ez CLI

Tools via Unix Command Line Interface with no installation and just using Docker + Shell Script.

Prerequisites

Setup

It adds aliases to your ~/.zshrc file and symbolic links to your /usr/local/bin/ folder:

./setup.sh
# --------------------------------------------------------------------------------
#                   My Ez CLI • Setup
# --------------------------------------------------------------------------------
#   Hope you enjoy it! :D
# --------------------------------------------------------------------------------
#   Note: Aliases may be created in '~/.zshrc' file...
# --------------------------------------------------------------------------------
#   Note: Symbolic links may be created in '/usr/local/bin/' folder...
# --------------------------------------------------------------------------------
#   Warning: Root access may be needed.
# --------------------------------------------------------------------------------
#   GitHub: https://github.com/DavidCardoso/my-ez-cli
# --------------------------------------------------------------------------------

# 1) ALL                6) node             11) docker-compose-viz
# 2) aws                7) yarn             12) playwright
# 3) terraform          8) yarn-berry       13) python
# 4) cdktf              9) serverless       14) EXIT
# 5) gcloud             10) speedtest
# Choose an option:

Usage examples

AWS CLI

See more.

# help
aws help

# list buckets
aws s3 ls --profile my-aws-profile

# download a file from a bucket
aws s3 cp s3://my-bucket/my-file /path/to/local/file --profile my-aws-profile

AWS Get Session Token

# authenticate using MFA
aws-get-session-token <MFA_DIGITS>

AWS SSO

# authenticate using SSO
aws-sso
#1) configure
#2) login
#3) logout
#Choose an option:

AWS SSO Get Credentials

Building the docker image.

If you need to get/know the SSO credentials being used, run:

aws-sso-cred $AWS_PROFILE
# or specify a profile of your choice
aws-sso-cred my-working-profile

Python

It is using version 3.12.4 as default.

# see version
python --version

# run interpreter
python

# run a script
python main.py

Using other Python versions

This script is using the same env var used by PyEnv.

So all you need to do is to declare the PYENV_VERSION before calling the python command.

# Export directly or add it to your profile configs (e.g.,`.zshrc`).
export PYENV_VERSION=3.9.19
python main.py

# or pass it inline
PYENV_VERSION=3.9.19 python main.py

# Note: the respective docker image will be downloaded if not found locally
# Unable to find image 'python:3.9.19' locally
# 3.9.19: Pulling from library/python
# 21988c13fd96: Download complete
# 42d758104bc9: Download complete
# 6d0099138f57: Download complete
# Digest: sha256:47d6f16aa0de11f2748c73e7af8d40eaf44146c6dc059b1d0aa1f917f8c5cc58
# Status: Downloaded newer image for python:3.9.19

Node

It is using Node 20 (current LTS version).

# see node version
node -v

# run node interpreter
node

# run a node script
node somefile.js

Using other Node versions

# just add the node version as a suffix
node14 -v
node16 -v
node18 -v

Yarn

It is using Node 20 (current LTS version).

# see yarn version
yarn -v

# start the package.json from a JS project
yarn init

# install a package as dev dependency
yarn add some-pkg --dev

# install a package globally
yarn global add another-pkg

Using Yarn with NPM token and custom ports

Use MEC_BIND_PORTS env var if you want to bind ports between the host and container:

MEC_BIND_PORTS="8080:80 9090:80" yarn

# or
export MEC_BIND_PORTS="8080:80 9090:80"
yarn

In order to be able to install NPM packages from a private repository, you might need to inform NPM_TOKEN env var.

NPM_TOKEN=your-token-here yarn

# or
export NPM_TOKEN=your-token-here
yarn

Hint: you can put this on your default shell config file.
Example for zsh: echo "export NPM_TOKEN=your-token-here" >> ~/.zshrc

Using Yarn with other Node versions

Some NPM packages aren't compatible with newer Node versions yet.

# just add the node version as a suffix
yarn14 -v
yarn16 -v
yarn18 -v

Yarn Berry (v2+)

# if you have replaced yarn
yarn --version # it should show 3.6+

# otherwise
yarn-berry --version # it should show 3.6+

Serverless Framework

It is ready to work with AWS.

See more about the docker image.

# see versions
serverless -v

# help
serverless --help

# Starting from a template
# note: replace "template-name" below with the folder name of the example you want to use

# method 1
serverless create \
  -u https://github.com/serverless/examples/tree/master/template-name \
  -n my-project-folder

# method 2 [recommended]
serverless init \
  template-name \
  -n my-project-folder

# Hint: if you get build errors, try it
cd my-project-folder && yarn

# Serverless.com account login
# note: It is also possible to use an access key to authenticate via serverless CLI
serverless login

# Deploy your project
serverless deploy

# Invoke a Lambda Function
serverless invoke -f hello

# Invoke and display lambda logs
serverless invoke -f hello --log

# Fetch lambda logs
serverless logs -f hello
serverless logs -f hello --tail

Serverless Getting Started docs.

Serverless templates.

Terraform

Important: ensure you are using the right provider credentials/roles/permissions before executing any command.

Take a look at Terraform AWS modules public registry and usage examples.

# help
terraform -help

# start the terraform in a project
mkdir my-terraform-project
cd my-terraform-project
terraform init

# set the right environment
# useful for multiple environments
# hint: avoid using default environment
terraform workspace list
terraform workspace new ${ENVIRONMENT}
terraform workspace select ${ENVIRONMENT}

# validate terraform files
terraform validate

# see  changes
terraform plan
# save changes to an output file (recommended)
terraform plan -out=tfplan

# apply changes to the providers (aws, gcp, azure, etc)
terraform apply
# apply changes using tfplan output file (recommended)
terraform apply tfplan

# destroy created resources on the providers
# warning: do not run it in production! ;D
terraform destroy

CONTEXT variable

By default, the parent directory is mounted on the container. This allows files inside parent folder to be referenced in the Terraform files.

For instance, if you need to use a Terraform module that is located two levels up in the filesystem, you can use CONTEXT variable before the terraform command to define the absolute path to that module (or another folder).

# option 1
CONTEXT=$(cd "$PWD/../../" && pwd) terraform --version
CONTEXT=$(cd "$PWD/../../" && pwd) terraform init

# option 2
CONTEXT=$(cd "$PWD/../../" && pwd)
CONTEXT=$CONTEXT terraform --version
CONTEXT=$CONTEXT terraform init

# option 3
export CONTEXT=$(cd "$PWD/../../" && pwd)
terraform --version
terraform init

DOTENV_FILE variable

All variables in DOTENV_FILE file will be available inside the container.

By default, the terraform container will use ${PWD}/.env file.

Inform a different value if you want to point to another one.

export DOTENV_FILE=local.env
terraform init
terraform plan

# or
DOTENV_FILE=local.env terraform init
DOTENV_FILE=local.env terraform plan

TF_RC_FILE variable

This is used for Terraform Cloud login.

By default, the terraform container will use ${HOME}/.terraformrc file.

Inform a different value if you want to point to another one.

export TF_RC_FILE=/another/path/to/terraform-credentials/file
terraform init
# it should recognize the backend config pointing to your TF Cloud workspace(s)

AWS_CREDENTIALS_FOLDER variable

This is used for AWS CLI authentication.

By default, the terraform container will use ${HOME}/.aws folder.

Inform a different value if you want to point to another one.

export AWS_PROFILE=your-aws-profile
export AWS_CREDENTIALS_FOLDER=/another/path/to/credentials/folder/
terraform init
terraform plan
terraform apply
# it should be able to deploy to your aws account based on the credentials used

See more about AWS auth configs.

GCLOUD_CREDENTIALS_FOLDER and GOOGLE_APPLICATION_CREDENTIALS variables

This is used for GCP CLI authentication.

By default, the terraform container will use ${HOME}/.config/gcloud folder, and /root/.config/gcloud/application_default_credentials.json file, respectively.

GOOGLE_APPLICATION_CREDENTIALS path starts with /root/ because this is the default user inside the container. Therefore you should not change it to your local user.

Inform different values if you want to point to another one.

export GCLOUD_CREDENTIALS_FOLDER=/another/path/to/credentials/folder/
export GOOGLE_APPLICATION_CREDENTIALS=/root/another/path/to/credentials/file
terraform init
terraform plan
terraform apply
# it should be able to deploy to your cloud account based on the credentials used

Cloud Development Kit for Terraform (CDKTF)

It is ready to work with Python.

Building the docker image for Python.

mkdir /my/folder/learn-cdktf
cd /my/folder/learn-cdktf

cdktf --help

# starts a new project from a template
cdktf init --template="python" --providers="aws@~>4.0"

Ookla Speedtest CLI

Building the docker image.

# help
speedtest --help

# run a speed test
speedtest

Google Cloud CLI

# If are not logged in, run the command below and follow the steps:
# 1. Copy/paste the provided URL in your browser
# 2. Authorize using your Google account
# 3. Copy/paste the generated auth code back in your terminal
gcloud-login

# If your current project is [None] or you wanna change it, set one.
gcloud config set project <PROJECT_ID>

# Test if it is working...
gcloud version
gcloud help
gcloud storage ls

gcloud CLI overview.

gcloud auth login.

Graph Viz for docker compose

This will create a dependency graph in display only, dot, or image formats based on a docker-compose YAML file (defaults to ./docker-compose.yml).

For more info, please check its official documentation.

# navigate to the directory where the docker-compose YAML file is
cd /my/project/with/docker-compose-file/

# using just default options
docker-compose-viz render

# using a custom docker compose file
docker-compose-viz render ./my-custom-docker-compose.yml

# dot output format
docker-compose-viz render --output-format=dot

# image output format
docker-compose-viz render --output-format=image

# setting the path/name of the output file
docker-compose-viz render --output-format=image --output-file=graph.png

Playwright

playwright # it will open the /bin/bash inside the container
# then you can run the other test related commands
npx playwright install chromium
npm run test
# etc...

For more info, please check its official documentation.

Author

David Cardoso

Contributors

Feel free to become a contributor! ;D

About

CLI Tools using Docker + Shell Script.

https://github.com/My-Tooling/my-ez-cli

License:MIT License


Languages

Language:Shell 88.0%Language:Dockerfile 12.0%