shell

:(fas fa-info-circle fa-fw): This is a mix of shell, linux, and macOS commands. Comments are welcome with any corrections or suggestions.

CLI Usage🔗

PowerShell & Bash Comparison🔗

Brevity🔗

See all aliases with Get-Alias and to expedite your cli usage you could use a gist like this: Aliaser.ps1

Note that PowerShell eschews brevity for clarity, but you can alias anything you like to be nearly as succint as bash commands.

IMO readability/brevity trumps succintness. However for interactive terminal usage aliasing can be a great tool. Use VSCode to auto-expand aliases into fully qualified functions if you decide to turn your adhoc work into a script file.

Using pushd in a PowerShell session actually aliases to Push-Location. The difference is in addition to the path string pushd manages, you get a .NET object back with [System.Management.Automation.PathInfo] information including: Drive, Path, Provider, ProviderPath.

PowerShell Works With Native Tooling🔗

I've included the similar PowerShell command to help those jumping between multiple shells.

Please note that unlike Python, PowerShell works as a terminal with native tools + scripting language.

You can use pwsh in almost every case in Linux & macOS and use the same tools you prefer, while being able to execute PowerShell commands as well.

For example something like aws cli returning json could be automatically unmarshaled into an object instead of using jq

& (aws ec2 describe-instances | ConvertFrom-Json).Instances.InstanceId

Another example is paths.

Prerequiresites for the PowerShell examples:

Install-Module Microsoft.PowerShell.ConsoleGuiTools -Scope CurrentUser -Force

Command	shell	pwsh
View history	`history`	`Get-History`
Execute Line from History	`!Number`	`Invoke-Expression (Get-History \\| Out-ConsoleGridView -OutputMode Single).CommandLine`
Execute Last Command But With Sudo	`sudo !!`
Test file exists	`test -f ./filename`	`Test-Path $filename -PathType Leaf` or using .NET `[io.file]::exists($filename)`

Installation🔗

Common App Installs🔗

Application	Notes	Install Command
HomeBrew	Works on Linux and macOS now 👏.	`/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install.sh)"`

Ansible Initialization🔗

#!/usr/bin/env bash
sh -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install.sh)"
brew install python3 ansible

For docker/linux

sudo python3 -m pip install --upgrade pip
CFLAGS=-Qunused-arguments CPPFLAGS=-Qunused-arguments python3 -m pip install --user ansible
echo "======== Ansible Version Info ======== "
ansible-galaxy --version

A Few More Ansible Commands🔗

Command	Code
Run ansible playbook against a specific tag	`ansible-playbook main.yml --inventory inventory --ask-become-pass -t 'ui'`
Install requirements	`ansible-galaxy collection install community.general && ansible-galaxy install --role-file requirements.yml --force --ignore-errors`

Installing go-task🔗

This tool is great for cross-platform shell scripting as it runs all the commands in the Taskfile.yml using a built in go shell library that supports bash syntax (and others).

Quickly get up and running using the directions here: Install Task

Command	Code
Default Installation to local directory with debug logging enabled	`sh -c "$(curl -ssL https://taskfile.dev/install.sh)" -- -d`
Installation for user level access	`sh -c "$(curl -ssL https://taskfile.dev/install.sh)" -- -d -b /usr/local/bin`

Installing Brew Packages🔗

This eliminates any attempt to install if the package already exists. For quick adhoc installs, this is useful.

#!/usr/bin/env bash

# Minimize Homebrew updates for each run, speeding things up
export HOMEBREW_NO_AUTO_UPDATE=1

# if linux install script, might want to include this: export PATH="/home/linuxbrew/.linuxbrew/bin:$PATH"

# Example of installing with a tap
brew tap lucagrulla/tap
package=cw
brew list $package &>/dev/null || brew install $package

# git-delta needs an updated version, so make sure it's available
package=less
brew list $package &>/dev/null || brew install $package

package=git-delta
brew list $package &>/dev/null || brew install $package

Reduce Noise With Progress Bar🔗

Use unzip with a progress bar to display progress, rather than the thousands of lines of output. This is an example of installing the AWS CLI v2 in a Dockerfile, while not forcing the output of each line when unzipping.

This shows how to use the pv command line tool to help display progress in both a count fashion, and also by just using as a timer.

RUN apt-get -yqq update --fix-missing && apt-get -yqq install pv \
    && mkdir -p ./tmpinstall && curl --silent "https://awscli.amazonaws.com/awscli-exe-linux-x86_64.zip" -o "./tmpinstall/awscliv2.zip" \
    && COUNT=`unzip -q -l "./tmpinstall/awscliv2.zip" | wc -l` \
    && mkdir -p ./tmpinstall/aws \
    && unzip "./tmpinstall/awscliv2.zip" -d "./tmpinstall/"  | pv -l -s $COUNT >/dev/null \
    && ./tmpinstall/aws/install --update | (pv --timer --name "🤖 awscli")  \
    && rm -rf ./tmpinstall/ \
    && apt-get clean -y && rm -rf /var/lib/apt/lists/* /tmp/library-scripts

Check for And Install Tooling🔗

This can help give you an example of how to double check that some installed tools are available as part of a setup script.

if command -v ghq &>/dev/null; then
    echo '✔️ ghq installed'
else
    warning "❌ failed to find ghq, attempting to setup via source"
    go install github.com/x-motemen/ghq@latest || echo "✅ installed ghq"
fi
if command -v gum &>/dev/null; then
    echo '✔️ gum installed'
else
    warning "❌ failed to find gum, attempting to setup via source"
    go install github.com/charmbracelet/gum@latest || echo "✅ installed gum"
fi
if ! command -v gum &/dev/null; then
  echo 'might need go binaries on path, trying now..., try adding the line to your .zshrc'
  export PATH="$(go env GOPATH)/bin:${PATH}"
fi

Conditional🔗

Only Proceed If First Condition Returns Nothing

package=cw
brew list $package &>/dev/null || brew install $package

On error do this:

test -f nonexistentfile || echo "😢 boo. file does not exist"

On success do the next command:

test -f ~/.bashrc && echo "✅ congrats, you have a bashrc file"

Web Requests🔗

Fetch A GitHub Release🔗

This contains a few things, including curl, jq parsing, and movement commands.

This provides a shell script example of using those to get the latest release from GitHub, parse the json, then move this to target path. This release doesn't wrap in a tar file; it's just a binary.

This might fail due to anonymous API hits on GitHub api are rate limited aggressively.

#!/usr/bin/env bash

echo "Grabbing latest release of fetch (a github cli for release downloads)"
USER=gruntwork-io
REPO=fetch
TAG=latest
ASSET=fetch_linux_amd64
FILE=fetch
curl --silent "https://api.github.com/repos/$USER/$REPO/releases/latest" \
| jq -r ".assets[] | select(.name | test(\"${ASSET}\")) | .browser_download_url" \
| wget -qi - --output-document=$FILE --progress=bar:force

echo "setting as executable and moving to /usr/local/bin"
chmod +x $FILE
sudo mv fetch /usr/local/bin
echo "Downloaded $(fetch --version) successfully"

Fetch a GitHub Release That Requires Extraction🔗

This is more of a Linux focused shell script example for grabbing a release and extracting the tar file.

#!/usr/bin/env bash
sudo apt -qqy update
sudo apt -qqy -o Dpkg::Progress-Fancy=true install wget

curl -s https://api.github.com/repos/GitTools/GitVersion/releases/latest \
| grep "browser_download_url.*gitversion\-debian.*\-x64.*\.tar\.gz" \
| cut -d ":" -f 2,3 \
| tr -d \" \
| wget -qi -

tarball="$(find . -name "gitversion-debian*.tar.gz")"
tar -xzf $tarball

sudo chmod +x gitversion
sudo mv gitversion /usr/local/bin

sudo rm $tarball
echo ">>>> gitversion version: $(~/gitversion /version)"
echo "Trying to install dotnet tools version"
dotnet tool update --global GitVersion.Tool

Concepts🔗

Shebang🔗

A common pattern is just #!/bin/bash.

To make your script more portable, by respecting the users env preferences try:

#!/usr/bin/env bash
#!/usr/bin/env zsh
#!/usr/bin/env sh

bash.cyberciti.biz reference

Some good info on this from Shebang

:(fas fa-code fa-fw): If you do not specify an interpreter line, the default is usually the /bin/sh

:(fas fa-code fa-fw): For a system boot script, use /bin/sh

:(fas fa-code fa-fw): The /usr/bin/env run a program such as a bash in a modified environment. It makes your bash script portable. The advantage of #!/usr/bin/env bash is that it will use whatever bash executable appears first in the running user's $PATH variable.

SSH🔗

Setup your permissions for ~/.ssh

echo "Setting full user permissions for ~/.ssh"
chmod -R u+rwX ~/.ssh
echo "Remove group access for ~/.ssh"
chmod go-rwx ~/.ssh
echo "now set any pem files to chmd 400 \$key to ensure read-only"
chmod 0600 ~/.ssh/id_rsa

For why 0600 see footnote.¹

Troubleshooting macOS permissions

I've had issues with macOS adding an @ with ACL issues on the ssh key's when downloaded.

To resolve this, just copy the contents of the ssh key to a new file and remove the original.

cat original_key.pem > key.pem

How To List Users In Linux

Search Contents of a File🔗

Using ripgrep you can search very quickly through file contents.

In this example, I found a text string in a PowerShell file that VSCode wasn't able to find after 1-2 mins due to the size of the directory.

rg -l -c "Start-ThreadJob" *.ps1

Benchmark

I ran a quick test to see how ripgrep performed compared to normal grep search. Grep wasn't optimized, and by default is single threaded. Ripgrep is multithreaded, automatically honors gitignore and more.

grep -rnw $HOME -e 'Start-ThreadJob'

Tool	Time
`ripgrep`	5m6s
`grep`	1h+timeout

Using yq to edit yaml files for Datadog service🔗

GitHub - mikefarah/yq: yq is a portable command-line YAML processor

I've use yq to edit yaml files programatically, such as datadog configuration files.

Here's a few samples on how to use this tool, using datadog agent config files as an example.

Quick Install of Datadog Service🔗

DD_HOST_TAGS="type:custom-server,stage:dev"
DD_HOSTNAME="custom-server"

DD_AGENT_MAJOR_VERSION=7 DD_API_KEY=FancyAPIKey DD_SITE="datadoghq.com" bash -c "$(curl -L https://s3.amazonaws.com/dd-agent/scripts/install_script.sh)"
sudo chmod -R 777 /etc/datadog-agent/

Start and stop the datadog services🔗

sudo systemctl stop datadog-agent
sudo systemctl start datadog-agent

Edit Default Datadog Config File🔗

Next, configure the main configuration with custom tags and host name, including additional ec2 tags, metadata, and a custom tag to show the specific load test this is capturing.

config=/etc/datadog-agent/datadog.yaml
nametag=mycustom-server
testname=bigloadtest

echo "set the basic config for app"
yq eval "
.hostname = \"$nametag\" |
.process_config.enabled = true |
.tags = [\"scope:loadtest\",\"testname:$testname\"] |
.env = \"dev\" |
.cloud_provider_metadata = [\"aws\"] |
.collect_ec2_tags = true" --inplace $config
yq eval ".hostname, .process_config.enabled, .tags, .env, .cloud_provider_metadata ,.collect_ec2_tags" $config

Enable Datadog Network Monitoring🔗

echo "set the process level config to search for ssh/sshd metrics"
sudo cp /etc/datadog-agent/system-probe.yaml.example /etc/datadog-agent/system-probe.yaml
netconfig=/etc/datadog-agent/system-probe.yaml
yq eval '.network_config.enabled' $netconfig
yq eval --inplace  '
.network_config.enabled = true
' $netconfig
yq eval '.network_config.enabled' $netconfig

Enable Datadog Process Level Tracking🔗

Enable process level tracking, with specific matching on ssh, sshd.

echo "set the process level config to search for ssh/sshd metrics"
sudo cp /etc/datadog-agent/conf.d/process.d/conf.yaml.example  /etc/datadog-agent/conf.d/process.d/conf.yaml
processconfig=/etc/datadog-agent/conf.d/process.d/conf.yaml
yq eval '.instances' $processconfig
yq eval --inplace  '
.instances[0].name = "ssh" |
.instances[0].search_string = ["ssh","sshd"]
' $processconfig
yq eval --inplace  '
.instances[1].name = "myprocess" |
.instances[1].search_string = ["myprocess"]
' $processconfig
yq eval '.instances' $processconfig

You can do a lot with yq.

Parse Kubernetes Secrets Using JQ🔗

Using jq, you can parse out secrets from base64 encoded values for some quick scripting.

NOTE: This uses sttr but you can modify to whatever your platform provides (zsh base64 -decode or pwsh [System.Convert]::FromBase64String($Base64String))) If you have Go installed (everyone should! 😀) then run go install github.com/abhimanyu003/sttr@latest.

This example parses an encoded json string to help registry an Azure Container Registry from a Kubernetes stored secret.

namespace="mynamespace"
secretname="mysecretname"

kubectl config set-context --current --namespace=$namespace
configEncoded=$(kubectl get secret $secretname -o jsonpath='{.data.\.dockerconfigjson}')
configDecoded=$(sttr base64-decode $config)
registry=$(echo $configDecoded | jq '.auths | keys[]')
echo -e "👉 registry: $registry"
creds=$(echo $configDecoded | jq .auths.${registry}.auth --raw-output)
echo -e "👉 username:password: $( sttr base64-decode $creds )"

GitHub CLI🔗

View The Logs Of A Prior Run🔗

View the logs of the last run (or toggle to error logs with the switch).

gh run view --log $(gh run list -L1 --json 'databaseId' --jq '.[].DatabaseId')
gh run view $(gh run list --limit 1 --json databaseId --jq '.[0].databaseId' ) --log

This can be chained together with other commands to quickly iterate on testing. When appropriate, you might avoid this by running act but I've had limited success with it due to various restrictions.

git commit -am 'ci: get github release working' && \
  git push && \
  gh workflow run release && \
  sleep 5 && \
  gh run watch -i1 || gh run view --log --job $(gh run list -L1 --json 'workflowDatabaseId' --jq '.[].workflowDatabaseId')

Use To Configure Settings on Many Repos At Once🔗

This example uses [gum]² to filter. Use tab when selecting in the multi-entry option.

org=$(gum input --prompt 'enter GitHub org: ')
originallist=$( gh repo list $org --json 'name' --jq '.[].name' |  tr ' ' '\n' )
repos="$( echo $originallist | gum filter --no-limit )"

for repo in $( echo $repos | tr '\n' ' ') ;
do
    printf "processing %s ... " "${repo}"
    gh api \
        --method PATCH \
        -H "Accept: application/vnd.github+json" \
        /repos/$org/$repo \
        -F use_squash_pr_title_as_default=true \
        -F squash_merge_commit_title=PR_TITLE \
        -F squash_merge_commit_message=PR_BODY \
        --silent
        printf "✔️\n"
    # return # for testing
done

Clone All The Desired🔗

Uses gum[^gum-repo] & [ghq]³. See setup directions.

Configure ghq🔗

To configure ghq defaults run:

git config --global ghq.vcs git
git config --global ghq.root $(gum input -prompt 'base git directory for repos: (recommend ~/git):  ' )

Clone All Repos Selected🔗

org=$(gum input --prompt 'enter GitHub org: ')
originallist=$( gh repo list $org --json 'name' --jq '.[].name' |  tr ' ' '\n' )
echo 'select repos (use tab to select, and type to filter)'
repos="$( echo $originallist | gum filter --no-limit )"

for repo in $( echo $repos | tr '\n' ' ') ;
do
    printf "processing %s ... " "${repo}"
    ghq get "https://github.com/${org}/${repo}" &> /dev/null
    printf "✔️\n"
done