utf9k utf9k

My zsh config

Table of contents

Setting up PATHs

Universal system folders

These paths generally exist on most every system so we’ll set them seperately from other PATH additions.

      $(brew --prefix)/bin
      "/Applications/Sublime Text.app/Contents/SharedSupport/bin"
export PATH


This section consists of helpers functions and global variables used by various applications.

A few of the helper functions are intended to make sure my configuration acts mostly identical across all machines and OSes without any extra configuration.

Whether that statement holds true is… debatable :)

Determining the current OS

In order to save having to remember how to use uname and all that, I just have my own little configuration within my shell that I can reference

if [[ $(uname -r) =~ 'microsoft' ]]; then
  export OPSYS="windows"
  export OPSYS=${(L)$(uname -s)}

Windows is a bit of a misnomer here because what I’m really checking for is whether the shell is running inside of Windows Subsystem for Linux (WSL)

Functionally, I can treat WSL and Linux the same (and I do) but there are some minor alterations I make use of, such as pointing the DISPLAY environment variable at an X display server on my host system

It’s worth noting that the value of $OPSYS on macOS is darwin. I could change it to be clearer but Darwin is technically the correct name for the base operating system

Setting my workspace

All of my development occurs in $HOME/Code regardless of what machine I’m on. One day I might change it though hence the variable.

export WORKSPACE="$HOME/Code"

Setting various global constants

export CONFIG_FILE="$HOME/.zshrc"
export CONFIG_SRC="$(chezmoi source-path)/zshrc.md"
export EDITOR="lvim"
export GPG_TTY=$(tty)
export LANGUAGE="en_NZ:en"
export LAST_MODIFIED="$(date)"

if [[ $TERM_PROGRAM == "iTerm.app" ]]; then
  export PROMPT=' ' # Installing iTerm helpers adds an arrow prompt
  export PROMPT='%B%F{green}>%f%b ' # I'd like a prompt in every other terminal

Setting some Windows / WSL specific constants

if [[ $OPSYS == "windows" ]]; then
  export DISPLAY=$(cat /etc/resolv.conf | grep nameserver | awk '{print $2; exit;}'):0.0
  export BROWSER="/mnt/c/Windows/explorer.exe"
  1. If I’m running on a Windows machine, I run Emacs by starting a daemon inside my terminal and connecting with emacsclient. Doing so spawns a new frame using the X display server running on the Windows host itself

  2. While I don’t believe this actually works, I attempt to override the BROWSER environment variable to open links on the Windows host from within Emacs

Module autoloading

Currently, this is just used for kubectl shell completion

autoload -Uz compinit



The version manager to rule them all

It wraps a number of existing language version managers into plugins that can be managed through one unified CLI tool

export ASDF_DIR=$HOME/.asdf
if [[ -f $ASDF_DIR/asdf.sh ]]; then
  . $ASDF_DIR/asdf.sh
export PATH=$(asdf where nodejs)/.npm/bin:$PATH
export PATH=$(asdf where python)/bin:$PATH


A fuzzy finder which comes with some autocompletions

if [[ $(command -v fzf) ]]; then
  [ -f ~/.fzf.zsh ] && source ~/.fzf.zsh


To save me having to set up each machine, I just set my Git identifiers each time

git config --global user.name "{{ .name }}"
git config --global user.email "{{ .email }}"


Kubectl comes with some shell completions for zsh

source <(kubectl completion zsh)


Less is great by default but it’d be even nicer with syntax highlighting!

if [[ $(which src-hilite-lesspipe.sh) ]]; then
  LESSPIPE=`which src-hilite-lesspipe.sh`
  export LESSOPEN="| ${LESSPIPE} %s"
  export LESS=' -R -X -F '


I don’t use it yet but Home Manager is promising

Setup is:

  • sh <(curl -L https://nixos.org/nix/install) --darwin-use-unencrypted-nix-store-volume
  • nix-channel --add https://github.com/nix-community/home-manager/archive/master.tar.gz home-manager
  • nix-channel --update
  • nix-shell '<home-manager>' -A install
export NIX_SSL_CERT_FILE=/etc/ssl/cert.pem
if [[ $(command -v nix) ]]; then
  export NIX_PATH=$HOME/.nix-defexpr/channels${NIX_PATH:+:}$NIX_PATH



Whenever I compile erlang (using asdf), I always use the same flags so it’s easier to just set them within my shell

export KERL_CONFIGURE_OPTIONS="--disable-debug --without-javac"
export KERL_BUILD_DOCS="yes"


I don’t explicitly set GOROOT as it is defined by asdf generally.

export PATH="$GOPATH/bin:$PATH"
export GO111MODULE="on"

Global packages

There’s no native functionality for keeping globally installed packages in sync, to my knowledge, so this is going to be a hack for that!

This installs a range of language servers in a very hacky way

function gsync() {
  yarn global add $global_npm_packages
  gem install $global_ruby_packages


Admittedly most of the git related stuff could live inside of a .gitconfig file but I never get around to moving it

That and I figure this will all eventually be superseded by nix anyway

You know… when I get around to doing that…

alias ae="deactivate &> /dev/null; source ./venv/bin/activate"
alias crush="pngcrush -ow"
alias de="deactivate &> /dev/null"
alias edit="$EDITOR $CONFIG_SRC"
alias gb="git branch -v"
alias gbm="git checkout master"
alias gcm="git commit -Si"
alias gr="git remote -v"
alias gs="git status"
alias gst="git status"
alias ipv4="dig @resolver4.opendns.com myip.opendns.com +short -4"
alias ipv6="dig @resolver1.ipv6-sandbox.opendns.com AAAA myip.opendns.com +short -6"
alias nvim="$EDITOR"
alias rebrew="brew bundle --file=$(chezmoi source-path)/Brewfile"
alias refresh="chezmoi apply && source $CONFIG_FILE"
alias tabcheck="/bin/cat -e -t -v"
alias utd="cd ~/utf9k && yarn start"
alias venv="python3 -m venv venv && ae"
alias vi="$EDITOR"
alias view="less $CONFIG_FILE"
alias vim="$EDITOR"
alias ws="cd $WORKSPACE"


These are some handy functions I use from time to time

Quick shortcuts to push and pull the current branch

While I can just do git pull, setting tracking branches is annoying because I always call them the same as their upstream branch.

These commands just push to and pull from the current branch explicitly.

function gpl { git branch | grep '*' | cut -c3- | xargs -I{} git pull origin {} }
function gps { git branch | grep '*' | cut -c3- | xargs -I{} git push origin {} }
function pap { git branch | grep '*' | cut -c3- | xargs -I{} git pull upstream {} && git push origin {} }

What application is listening on any given port?

function whomport() { lsof -nP -i4TCP:$1 | grep LISTEN }

I’d like to tangle a markdown file please

I have my own little Markdown tangling tool which you can read about here

function tangle-md() {
  if [[ $(command -v lugh) ]]; then
    lugh -f "$1"
    echo "lugh isn't installed. You can find it at https://github.com/marcus-crane/lugh"

I’d like to tangle an org file please

function tangle-file() {
  emacs --batch -l org $@ -f org-babel-tangle

What’s inside that JWT?

This function is used to display both encrypted and regular JWT tokens, as opposed to using an online service like https://jwt.io

It’s taken almost verbatim from this post except the original exit 0 would cause my terminal session to exit so I swapped it for a break instead.

To pretty print a JWT line, just use it like so: jwt <token>

If you’d like to use a JWT stored as a file, you can do that pretty easily too: jwt $(cat a_saved_jwt)

function jwt() {
  for part in 1 2; do
    b64="$(cut -f$part -d. <<< "$1" | tr '_-' '/+')"
    n=$((len % 4))
    if [[ 2 -eq n ]]; then
    elif [[ 3 -eq n ]]; then
    d="$(openssl enc -base64 -d -A <<< "$b64")"
    python -mjson.tool <<< "$d"
    # don't decode further if this is an encrypted JWT (JWE)
    if [[ 1 -eq part ]] && grep '"enc":' <<< "$d" >/dev/null ; then

I’d like to see all resources in any given namespace

Annoyingly, the kubectl get all command doesn’t actually do what it says on the tin.

Specifically, “all” in this case only includes a couple of different resources like pods and services.

As a workaround, it’s a bit slow but we can just enumerate through all of the supported resources and see what we get back.

function get-all-resources() {
    kubectl api-resources --verbs=list --namespaced -o name | xargs -n 1 kubectl get --show-kind --ignore-not-found -n $@

What functions have I defined?

Often I’ll forget what little shortcut functions I’ve made so here’s a quick cheatsheet

There is a built-in functions but it shows the actual source code rather than a list of names

function funcs() {
    functions | grep "()" | grep -v '"'

How’s that for disorientation? Enough “functions” for ya?

The extra grep is a bit of a hack because without it, the actual function body will match the search for grep "()"

It’s quite interesting in a way, that it would recursively search itself so I added in a second grep to remove any lines that feature a double quote.

Technically speaking, the second grep would potentially be filtering itself out recursively which I think is pretty interesting

It also makes my head hurt a little bit for what you’d think would be a pretty basic function!

What’s a quick way to archive backups?

In order to save on cloud storage space, while still keeping a home for rarely used backups, I like to store things in Backblaze B2

While I could use the web UI, it’s often just as fast to use rclone which is what this script is a wrapper around

function archive() {
    rclone copy "$(pwd)/$1" b2:long-term-backups --progress

Quick convert screen recording to a more suitable format

Often times, I find myself making screen recording with Quicktime but they export as .mov files. I much prefer having an mp4 file as it’s more universally accepted so this is a quick function to perform that convertion with ffmpeg.

function demov() {
  if [[ $(command -v "ffmpeg") ]]; then
      ffmpeg -i $1 -vcodec libx264 -acodec aac $(echo "$1" | rev | cut -f 2- -d '.' | rev).mp4
      print "It doesn't look like you have ffmpeg installed."

Quick convert h265 to 8 bit 264

function de265() {
  if [[ $(command -v "ffmpeg") ]]; then
      ffmpeg -i $1 -map 0 -c:v libx264 -crf 18 -vf format=yuv420p -c:a copy $(echo "$1" | rev | cut -f 2- -d '.' | rev).mp4
      print "It doesn't look like you have ffmpeg installed."

Extract emails from a webpage

I recently discovered html-xml-utils which has some handy functionality so this is a basic script to try and extract mailto links from a webpage

function emails() {
  wget --spider --recursive --level=2 --execute robots=off --user-agent="Mozilla/5.0 Firefox/4.0.1" $1 2>&1 |
    egrep -o 'https?://[^ ]+' |
    sed -e 's/^.*"\([^"]\+\)".*$/\1/g' |
    uniq |
    xargs curl -s |
    grep -s "mailto" |
    hxpipe |
    grep "mailto:" |
    cut -d ":" -f2 |
    sort |

Calculating nines

Often times, it can be useful to put service uptime into minutes and hours. Thankfully uptime.is is a handy tool for this plus it reserves JSON too!

function nines() {
  curl -s https://uptime.is/$1 | jq

Delete Git branches interactively with fzf

This function was quite shamelessly taken from this very good post by Sebastian Jambor.

It opens an interactive fzf window which shows a list of git branches, with their relevant history on the side as a preview pane.

You can press TAB to select multiple branches and ENTER to delete them.

If you decide to back out, you can press ESC to cancel.

function gbd() {
  git branch |
    grep --invert-match --extended-regexp 'master|main' |
    cut -c 3- |
    fzf --multi --preview="git log {} --" |
    xargs git branch --delete --force

Create an internet bookmark file

function bookmark() {
  local bookmarkName
  local bookmarkURL
  vared -p "Bookmark name: " bookmarkName
  vared -p "Bookmark URL: " bookmarkURL
  echo "[InternetShortcut]\nURL=$bookmarkURL\nIconIndex=0\n" > $HOME/Bookmarks/$bookmarkName.url

View and open internet bookmarks

function site() {
  fd . ~/Bookmarks |
    fzf --multi --preview="cat {} | grep URL | cut -c 5- | xargs curl --head --location --max-time 10" |
    sed 's/ /\\ /g' |
    xargs open

View unread Pinboard items

function pinboard() {
  if [[ ! $OP_SESSION_my ]]; then
    echo "Please log in using the op cli | eval (op signin my)"
    export pinboardToken=$(op get item Pinboard --fields "API Token")
    curl "https://api.pinboard.in/v1/posts/all?auth_token=$pinboardToken&format=json" |
      jq 'map(select(.toread == "yes")) | .[].href' |
      tr -d '"' |
      fzf --multi --preview="curl -L -I {}" |
      xargs open

View homebrew casks

I find that the Homebrew cask search doesn’t provide enough information to make an informed decision so I’m using fzf instead to help

function casks() {
  curl "https://formulae.brew.sh/api/cask.json" |
    jq '.[].token' |
    tr -d '"' |
    fzf --multi --preview="curl https://formulae.brew.sh/api/cask/{}.json | jq '.'" |
    xargs brew install --cask

View all ingress domain names found in a cluster

function ingresses() {
  kubectl get ingresses --all-namespaces -o json |
    jq -r '.items[].spec.rules[].host' | 
    fzf --preview="curl -I -L https://{}"

Regenerate a secret key that has the same length as the input

Something I commonly do is regenerate secret keys between environments when deploying software. These keys aren’t necessarily secret in themselves so much as they are just used to provide extra entropy.

I can never remember which keys require a specific length so this is a short function to take a key and regenerate a key that is the exact same length.

function secretregen() {
  local SECRET_LENGTH=$(echo -n $1 | wc -m | awk '{$1=$1};1')
  LC_ALL=C tr -dc A-Za-z0-9 </dev/urandom | head -c $SECRET_LENGTH ; echo ''

Decode URLs with percentage decoded values

Often times, it can be more useful to inspect the API calls made by a web application, than using the API documentation supplies but this can get a little annoying when you need to decode HTML entities.

As a result, this little function will decode a URL parameter like team_ids%5B%5D=ABC123 into team_ids[]=ABC123.

There are other types of HTML encoding of course but I only ever seem to run into percentage decoding on a day to day basis.

Remember to quote your input so that & symbols and the like aren’t interpreted as shell commands.

function percentdecode() {
  echo $1 | python3 -c 'import sys,urllib.parse as ul; print(ul.unquote_plus(sys.stdin.read()),end="")'

Create a new blog post for my site

Hugo archetypes are the way to do this but I’m not sure if I have my folders configured properly.

function newpost() {
  local SLUG=$(echo $1 | tr '[:upper:]' '[:lower:]' | tr ' ' '-')
  mkdir -p ~/utf9k/content/blog/20xx--$SLUG
  cp ~/utf9k/archetypes/blog.md ~/utf9k/content/blog/20xx--$SLUG/index.md
  sed -i '' -e "s/<TITLE>/$1/g" ~/utf9k/content/blog/20xx--$SLUG/index.md
  sed -i '' -e "s/<SLUG>/$SLUG/g" ~/utf9k/content/blog/20xx--$SLUG/index.md
  echo "Created a new post at ~/utf9k/content/blog/20xx--$SLUG/index.md"
  echo "Would you like to start writing?"
  select ynd in "Yes" "No" "Delete"; do
    case $ynd in
      Yes ) $EDITOR ~/utf9k/content/blog/20xx--$SLUG/index.md; echo "Nice work!"; break;;
      No ) break;;
      Delete ) rm -rf ~/utf9k/content/blog/20xx--$SLUG; echo "Deleted"; break;;


A small helper function for sourcing the contents of .env files into my shell

envy() {
  if [ -f ".env" ]; then
    set -o allexport
    source .env
    set +o allexport
    echo "No env file located"
    return 1

fly.io logs

I find myself checking fly logs (and sshing into them) a lot since some of my personal projects live there.

We can use fzf to make this easier, and faster without too much hassle.

There’s some data munging due to the CLI output being a little non-standard but nothing impossible

flogs() {
  fly apps list |
    tail -r |
    tail -n +2 |
    tail -r |
    tail -n +2 |
    awk '{ print $1 }' |
    fzf --preview="fly logs -a {}" --preview-window=80%,follow |
    xargs fly ssh console -a

Pretty print PATH

path() {
  echo -e "${PATH//:/\\n}"

Kumamon on demand

I like Kumamon but I don’t watch Kumamon videos enough so this is a small function that opens a random Kumamon YouTube video using mpv

kumamon() {
  channel=${channels[$(( $RANDOM % ${#channels[@]} +1 ))]}
  mpv $channel --shuffle --geometry=100%:0% --autofit=20% --ytdl-format="bestvideo[height<=480]+bestaudio/best[height<=480]" --ontop

defaults plist viewer

This is probably my weightiest command to date

viewdefaults() {
  defaults domains |
    sed 's/$/, NSGlobalDomain/' |
    tr -d ',' |
    tr ' ' '\n' |
    fzf --preview="defaults export {} - | python3 -c \"import sys,plistlib,pprint; pprint.pprint(plistlib.loads(sys.stdin.read().encode('utf-8')))\"" |
    xargs -n1 -I{} sh -c 'defaults export $1 - > $1.plist' -- {}

Pretty print URL params

Using the previously defined percentdecode function, this makes it easy to visualise request params in a URL

params() {
  percentdecode $1 |
    tr "?" "\n" |
    tr "&" "\n"

master to main

From time to time, I’ll update a repo’s branch for consistency and forget the steps to update my local

master2main() {
  git branch -m master main
  git fetch origin
  git branch -u origin/main main
  git remote set-head origin -a

iTerm 2 integration

I used iTerm 2 on my various devices as a terminal and so, there are some shell integrations that are handy to use

if [[ -f "$HOME/.iterm2_shell_integration.zsh" ]]; then
  . $HOME/.iterm2_shell_integration.zsh