Executable scripts

Table of Contents

1. Presentation

This file will present all the executable scripts I wrote. It is also their original source code, all the following code snippets are exported and tangled from this file to the actual executables.

2. Autostart

Because I sometimes switch from window manager to window manager, creating a script that handles by itself autostarting things for me is way easier than rewriting every time the autostart part of my configuration. As you can every instance will be launched asynchronously, and only if there is no other instance of said command running.

Command Arguments Run once?
pactl load-module module-switch-on-connect  
mpc stop no
xrdb -merge "$XDG_CONFIG_HOME"/X11/Xresources no
picom --experimental-backends yes
numlockx on yes
pumopm   yes
xfce-polkit   yes
nm-applet   yes
xwallpaper --zoom "$(cat "$HOME"/.cache/wallpaper)" no
xss-lock plock yes
/usr/lib/kdeconnectd   yes
dunst   yes

I also have an external sound card, a Scarlet 2i2 G3, that I would like to use as my default audio output. However, it might not be always connected, hence the following code:

SOUNDCARD=$(pactl list short sinks | grep "Focusrite")
if [[ -n $SOUNDCARD ]]; then
    pactl set-default-sink "$(echo "$SOUNDCARD" | awk '{print $2}')"

3. cli utilities

3.1. Backup

backup is a very simple, oneliner script that will create a local copy of a file and add the date at which it was copied in the filename. You can see its source code here:

cp -r $argv[1] $argv[1].bak.(date +"%Y%m%d%H%M%S")

3.2. CPU Scaling

As I am using a laptop, maximum performance isn’t always what I want. Sometimes, it’s just better to have not so great but less battery-demanding performance. It is possible to achieve this by modifying the CPU governor with cpupower. The Arch Wiki has, as usual, some really good documentation on this.

The following script asks the user through rofi which governor to apply, and it relies on askpass to retrieve the user’s password.

governors=("performance" "powersave" "userspace" "ondemand" "conservative" "schedutil")
governor=$(printf "%s\n" "${governors[@]}" | rofi -dmenu)
sudo -A cpupower frequency-set -g "$governor"

3.3. Development

3.3.1. Cnew

cnew is a small utility script similar to but simpler than cppnew that creates a CMake template C project from the template that already exists in ~/dev/templateC. If no argument was passed, display an error message and exit.

if ! count $argv > /dev/null
    echo "Missing argument: PROJECT" && return -1

Pass the first argument to a switch statement.

switch "$argv[1]"

If the argument is -h or --help, then display the help message and exit the script normally.

case -h --help
    man ~/dev/fishfunctions/cnew.man
    exit 0

Else, the argument is the name of the project the user wants to create.

case '*'
    set -g project_name $argv[1]

Let’s close the switch statement.


Now, let’s copy the template where the user is executing cnew from, give it the name of the project and move to the project.

cp -r ~/dev/templateC $argv[1]
cd $argv[1]

The default files have a placeholder for the name of the project. Let’s replace these placeholders with the project’s name.

sed -i "s/PROJECTNAME/$argv[1]/g" CMakeLists.txt
sed -i "s/PROJECTNAME/$argv[1]/g" README.org
sed -i "s/CPROJECTNAME/$argv[1]/g" doc/Doxyfile

Now, let’s create a git repository and initialize it.

git init
git add .
git commit -m "initial commit"

And we’re done!

3.3.2. Dart Language Server

Spacemacs' recommendations on how to use Dart with LSP is outdated, since dart_language_server is obsolete. As recommended by the repo owner, we should launch instead the following code:

/usr/bin/dart $DART_SDK/snapshots/analysis_server.dart.snapshot --lsp

So, instead of using the obsolete executable, instead we will be calling the analysis server as requested.

3.3.3. UpdateFlutter

This is a simple utility to be ran when the flutter package is updated.

sudo chown -R :flutterusers /opt/flutter
sudo chmod -R g+w /opt/flutter
sudo chmod a+rw /opt/flutter/version
sudo chown $USER:(id -g $USER) /opt/flutter/bin/cache

3.4. mu-unread

mu-unread is a very simple utility that simply returns the amount of unread emails I have through the use of mu.

As you can see, the output string contains two font switchers for StumpWM so I can switch from the main font to Siji for the caracter contained between them: U+E072 (an email icon).

UNREAD=$(mu find "flag:unread AND (maildir:/Inbox OR maildir:/Junk)" | wc -l)
printf "^f2^f0%s" "$UNREAD"

3.5. Post scrot script

It is possible to call a script on the resulting image of a scrot command. Not only do I want to move them to a specific directory, I also want to be able to see them in nsxiv (a replacement for sxiv) in case I want to edit the image, copy it or simply delete it (sometimes I take screenshots by accident).

mv "$@" ~/Pictures/Screenshots/
nsxiv -abfs f "$HOME/Pictures/Screenshots/$*"

3.6. sshbind

Something that I did not know for quite some time but that is actually crazy useful about SSH is its ability to bind locally the port of a remote machine, and vice versa. The syntax is actually very simple, but I prefer a more intuitive way of writing it. Its usage is sshbind PORT FROMHOST TOHOST.

ssh -L $argv[1]:$argv[3]:$argv[1] $argv[2] -N

3.7. Nsxiv key handler

One thing I like with nsxiv is you can trigger different behaviors based on keypresses. For instance, with my current nsxiv configuration, if I press the space key followed by a character, it can delete to the trashcan, delete definitively, or open the current image in GIMP. All of that is done through one script file stored in $HOME/.config/nsxiv/exec/key-handler. The fact it reacts to first the spacebar instead of Ctrl-x is because I use a custom version of nsxiv I first modified to fit the bépo layout, and then I decided to change the prefix to fit how I use Emacs and StumpWM. You can read the PKGBUILD and my nsxiv patch in my dotfiles repo.

while read file
    set -g FILES "$file" $FILES

switch "$argv[1]"
    case "d"
	trash $FILES
    case "D"
	rm $FILES
    case "g"
	gimp $FILES
    case "j"
	for f in $FILES
	    set basename (echo "$f" | sed 's/\.[^.]*$//')
	    convert "$f" "$basename.jpg"
    case "r"
	for f in $FILES
	    convert -rotate 90 "$f" "$f"
    case "R"
	for f in $FILES
	    convert -rotate 270 "$f" "$f"
    case "y"
	set FILE "$FILES[1]"
	set TYPE (file -i "$FILE" | sed -r 's|.*(image/[a-z]+);.*|\1|')
	xclip -sel clip -t "$TYPE" -i "$FILE"

Here is a step by step explanation of the source code. First, we want to store in the variable FILES all the files marked in nsxiv. This is done with a while loop and the read command.

while read file
    set -g FILES "$file" $FILES

We can then read from the first member of argv which key the user pressed. Depending on it, we can chose what to execute.

switch "$argv[1]"
    case "d"
	trash $FILES
    case "D"
	rm $FILES
    case "g"
	gimp $FILES
    case "j"
	for f in $FILES
	    set basename (echo "$f" | sed 's/\.[^.]*$//')
	    convert "$f" "$basename.jpg"
    case "r"
	for f in $FILES
	    convert -rotate 90 "$f" "$f"
    case "R"
	for f in $FILES
	    convert -rotate 270 "$f" "$f"
    case "y"
	set FILE "$FILES[1]"
	set TYPE (file -i "$FILE" | sed -r 's|.*(image/[a-z]+);.*|\1|')
	xclip -sel clip -t "$TYPE" -i "$FILE"

The first option with the letter d is to move the file to the trash can. For this, we use the trash command from trash-cli.

case "d"
    trash $FILES

In case we want to definitively delete a file without using the trash can, we can use rm instead when we press the D key.

case "D"
    rm $FILES

It’s not rare I want to modify an image I have open in nsxiv, especially since screenshots are automatically open in this image viewer aften they are taken. In that case, a simple command will do.

case "g"
    gimp $FILES

Often, I use nsxiv to convert an image to a jpeg file, because my internet connection is not that great and jpeg screenshots are faster to upload than png screenshots. So what I do is for each file selected, I take the base name of the file (i.e. remove its extension), and then I use the convert command from imagemagik to convert it from its original format to a jpg format — imagemagik detects the formats based on the extension.

case "j"
    for f in $FILES
	set basename (echo "$f" | sed 's/\.[^.]*$//')
	convert "$f" "$basename.jpg"

I have two commands to rotate my files, and both only differ by the angle of rotation. Once again I rely on convert in both cases. To rotate clockwise, this is the code needed.

case "r"
    for f in $FILES
	convert -rotate 90 "$f" "$f"

On the other hand, to rotate counter-clockwise, we need this code:

case "R"
    for f in $FILES
	convert -rotate 270 "$f" "$f"

Lastly, when I want to copy a file, I just hit the y key for “yank” (that’s a term from Emacs). For that, I rely on file to tell me the mimetype of the image, then I can pass it to xclip along with the filename to copy it to the clipboard. In this case, we only copy the first of the selected files since the clipboard cannot hold several files at once.

case "y"
    set FILE "$FILES[1]"
    set TYPE (file -i "$FILE" | sed -r 's|.*(image/[a-z]+);.*|\1|')
    xclip -sel clip -t "$TYPE" -i "$FILE"

3.8. Starwars

This is a one-liner that allows you to watch Star Wars episode 4 in ASCII art in your terminal. Here is the code:

telnet towel.blinkenlights.nl

3.9. Toggle touchpad tapping

For some reasons, my firmware does not recognize the function key for toggling the touchpad. I’m not going to really complain about it since it lets me program it like I want. Since I often don’t need to completely deactivate the touchpad, I’ll instead toggle whether tapping is enabled or not when pressing XF86TouchpadToggle. And for that, I need this small script that will actually toggle it, and it will be used in my window manager configuration.

First let’s declare some variables to make this script more personal. With my current computer (a Gazelle by System76), the name of my touchpad is the following:

set TPNAME "ELAN0412:00 04F3:3162 Touchpad"

Let’s now get the identifier of the touchpad for xinput:

set TPID (xinput list | grep $TPNAME | awk '{print $6}' | sed 's|id=\(.*\)|\1|g')

Now, let’s detect the current status of the touchpad:

set TPSTATUS (xinput list-props $TPID | grep "Tapping Enabled" | \
	      grep -v "Default" | awk '{print $5}')

This will set TPSTATUS either to 0, meaning tapping is disabled, or to 1, meaning it’s enabled. I will consider any other value as being disabled.

test [[ $TPSTATUS = "1" ]] && set NEWTPSTATUS 0 || set NEWTPSTATUS 1

Finally, let’s update the touchpad’s options:

xinput set-prop $TPNAME "libinput Tapping Enabled" $NEWTPSTATUS

3.10. Wacom setup

I made a small and quick utility to set up my Wacom tablet so it is only bound to one screen. This is quite easy, we simply need to find the Wacom stylus’ ID and assign it to the display we want.

ID=$(xinput | grep -oPi "wacom.+stylus.+id=\K([0-9]+)")
SCREEN=$(xrandr -q --current | \
             grep -iPo '^([^ ])+(?= connected)' | \
             rofi -dmenu -i -p 'Select your display' | \
             tr -d '\n')
xinput map-to-output "$ID" "$SCREEN"

4. Emacs stuff

4.1. Dired

emacsclient -c -a emacs -e "(dired \"$*\")"

4.2. Emacsmail

This short script is used in my ~/.local/share/applications/mu4e.desktop file in order to send to Emacs any mailto: requests made in my system.

emacsclient -c -n -a emacs -e "(browse-url-mail \"$*\")"

5. Media

5.1. mp42webm

This function allows me to convert easily an mp4 video to the webm format. Nothing too fancy here.

ffmpeg -i $argv[1] -c:v libvpx -crf 10 -b:v 1M -c:a libvorbis $argv[1].webm

5.2. youtube-dl wrappers

5.2.1. ytplay

ytplay is a simple script I’ve written that allows me to play in mpv any YouTube video at the desired resolution. The script relies on dmenu (or rofi in dmenu-mode), youtube-dl and of course mpv itself.

set URL (rofi -dmenu -i -p "Video URL")
if test -n "$URL"
    set FORMAT                              \
	(youtube-dl --list-formats "$URL" | \
	egrep "webm.*[0-9]+x[0-9]+"       | \
	awk '{print $3 " " $1}'           | \
	sort -gu                          | \
	rofi -dmenu -i -p "Resolution"    | \
	string split " ")
    set FCODE $FORMAT[2]
    mpv --ytdl-format=$FCODE+bestaudio/best "$URL"

I’ll even add a .desktop entry for this script:

[Desktop Entry]
Name=ytplay (YouTube in mpv)
Comment=Play YouTube videos in mpv

5.2.2. ytdl - a youtube-dl wrapper

This script is a wrapper around youtube-dl which I use mainly for archiving YouTube videos on my NAS (at the time I’m writing this, I have already 2.1TB worth of videos archived). The principle behind this script is quite simple: I want to avoid as much as possible to redownload any video already downloaded in order to avoid pinging too much YouTube’s servers, 429 Too Many Requests errors are really annoying, and it comes really early when you have only a couple of new videos to download among the few 14k videos already downloaded.

Be aware this script was written for the Fish shell (3.1.0 and above), and makes use of youtube-dl 2020.03.24 and above, Fish getopts and ripgrep. Setting default values

Some variables in this script will have default values, we do not want to have a mile-long command each time we wish to download a single video. We’ll also set some global variables that won’t change:

Variable Name Default Value String?
YTDL_SHARED_DIR $HOME/.local/share/ytdl no
FORMAT_DEFAULT %(uploader)s/%(upload_date)s ~- ~%(title)s.%(ext)s yes
PREFFERED_FORMAT bestvideo[ext=mp4]+bestaudio[ext=m4a]/bestvideo+bestaudio yes
VERSION 0.3 yes

There is one more default variable pointing to ytdl’s root directory which depends on whether the videos directory has a French or English name:

if test -d "$HOME/Vidéos"
    set -g ROOTDIR_DEFAULT "$HOME/Vidéos" # French name
    set -g ROOTDIR_DEFAULT "$HOME/Videos" # English name
set -g YTDL_SHARED_DIR  "$HOME/.local/share/ytdl"
set -g FORMAT_DEFAULT   "%(uploader)s/%(upload_date)s - %(title)s.%(ext)s"
set -g ERRFILE_DEFAULT  "$YTDL_SHARED_DIR/video-errors"
set -g PREFFERED_FORMAT "bestvideo[ext=mp4]+bestaudio[ext=m4a]/bestvideo+bestaudio"
set -g VERSION          "0.3"
set -g YTDL_SHARED_DIR  "$HOME/.local/share/ytdl"
set -g FORMAT_DEFAULT   "%(uploader)s/%(upload_date)s - %(title)s.%(ext)s"
set -g ERRFILE_DEFAULT  "$YTDL_SHARED_DIR/video-errors"
set -g PREFFERED_FORMAT "bestvideo[ext=mp4]+bestaudio[ext=m4a]/bestvideo+bestaudio"
set -g VERSION          "0.3"
if test -d "$HOME/Vidéos"
    set -g ROOTDIR_DEFAULT "$HOME/Vidéos" # French name
    set -g ROOTDIR_DEFAULT "$HOME/Videos" # English name

We’ll also create the directory pointed at by YTDL_SHARED_DIR if it doesn’t exist already:

mkdir -p $YTDL_SHARED_DIR Help message

The next step is displaying the help message for the script. For that, just a long string echo’d will do, wrapped in the function _ytdl_help.

function _ytdl_help
    echo "Usage: ytdl [OPTION]... URL [URL]...

    -4, --ipv4
	Download with forced IPv4
	Default: no

    -6, --ipv6
	Download with forced IPv6
	Default: no

    -a, --batch-file <file>
	File containing URLs to download, one URL per line. Lines starting with
	'#', ';' or ']' are considered as comments and ignored.
	Default: None

    -c, --id-cache <file>
	File containing the video IDs that were already downloaded, one ID per

    -d, --directory <dir>
	Root directory in which to download videos.

    -e, --error-file <file>
	File containing the IDs of videos that failed to download, one ID per

    -f, --format <format>
	Format name for downloaded videos, including path relative to root

    -l, --logs <file>
	File in which to store logs.

    -V, --verbose
	Show verbose output
	Default: no

    -v, --version
	Show version of ytdl.

    -h, --help
	Shows this help message"

We also have the function _ytdl_version to display the current version of ytdl:

function _ytdl_version
    echo "ytdl 0.3, developped for fish 3.1.0 and youtube-dl 2020.03.24 or newer"
    echo "requires Fish getopts <https://github.com/jorgebucaran/fish-getopts>"
    echo "and ripgrep <https://github.com/BurntSushi/ripgrep>"
end Arguments Handling

The function _ytdl_parse_ops is a little bit trickier: we use getopts to parse the arguments passed to the script in order to get some preferences from the user. Here is a quick reference on what options are available and what they do:

Short Long Takes a value? Associated Variable Default Value What it does
4 ipv4 no IPV4 None Force IPv4
6 ipv6 no IPV6 None Force IPv6
a batch-file yes FILE None Batch file
c cache yes DOWNFILE $DOWNFILE_DEFAULT Cache file
d directory yes ROOTDIR $ROOTDIR_DEFAULT Root directory
e error-file yes ERRFILE $ERRFILE_DEFAULT Error logs
f format yes FORMAT $FORMAT_DEFAULT Filename format
V verbose no VERBOSE 1 Verbose output
v version command None None Script version
h help command None None Display this message

We can also pass individual YouTube URLs without any options or switches associated to them, they will be downloaded as part of a single queue.

case 4 ipv4
	set -g IPV4
case 6 ipv6
	set -g IPV6
case a batch-file
	set -g FILE $value
case c cache
	set -g DOWNFILE $value
case d directory
	set -g ROOTDIR $value
case e error-file
	set -g ERRFILE $value
case f format
	set -g FORMAT $value
case l logs
	set -g LOGFILE $value
case V verbose
	set -g VERBOSE
case v version
	_ytdl_version && exit
case h help
	_ytdl_help && exit

The following shows how getopts is used to catch the options and switches passed to the script:

getopts $argv | while read -l key value
    switch $key
	case _
	    for v in $value
		set -g VIDEOS $VIDEOS $v

Some values need to be set to their default, so let’s assign them their value if no user value was passed:


Both these code blocks are executed in _ytdl_parse_ops:

function _ytdl_parse_ops
    getopts $argv | while read -l key value
	switch $key
	    case _
		for v in $value
		    set -g VIDEOS $VIDEOS $v
end Logging

_ytdl_log is a very simple function used for logging information for the user in the file pointed to by LOGFILE. The first argument the function should receive is its log level. I generally use either "INFO" or "ERR". The second argument is the message to log.

function _ytdl_log
    set -l INFOLEVEL $argv[1]
    set -l MSG $argv[2]
    set -l LOG (printf "[%s] %s %s\n" $INFOLEVEL (date +"%F %T") $MSG)
    printf "%s\n" $LOG >> $LOGFILE
    if test $VERBOSE -eq 1
	echo $LOG
end Download a Single Video

In order to download a single video, a simple function has been written for this that will display when downloaded how far it is down the list of videos to be downloaded and it will add its ID to the file listing all videos downloaded. The script will also try to download the video according to the PREFFERED_FORMAT variable, but if the download fails it will download the default format selected by youtube-dl. If both downloads fail, the ID of the video will be added to the list of failed videos. If one of the downloads succeeds, it will remove the ID from the list of failed downloads.

The first argument of the function is the video ID from YouTube, the second argument is the position of the video in the queue, and the third argument is the queue length –can be the amount of videos in a whole YouTube channel, the amount of videos in a playlist, or simply the amount of YouTube URLs passed as arguments to the script.

function _ytdl_download_video
    set ID $argv[1]
    _ytdl_log "INFO" (printf "Downloading video with ID $ID (%4d/%4d)" $argv[2] $argv[3])
    if youtube-dl -f $PREFFERED_FORMAT -ciw -o $FORMAT "https://youtube.com/watch?v=$ID"
	echo $ID >> $DOWNFILE
    else if youtube-dl -ciw -o $FORMAT "https://youtube.com/watch?v=$ID"
	echo $ID >> $DOWNFILE
	_ytdl_log "ERR" "Could not download $VIDEO"
	echo $ID >> $ERRFILE

Note that this function is not meant to be called without any checks before. It is meant to be called by _ytdl_download_queue described below. Download a Queue of Videos

One of the main goals of this tool is to check if a video has already been downloaded. This is why, as you will see below, we use ripgrep to check if the ID of the video we want to download is already present in the list of downloaded videos. If not, it will then be downloaded though _ytdl_download_video described above.

function _ytdl_download_queue
    for i in (seq (count $argv))
	rg -- $argv[$i] $DOWNFILE 2&> /dev/null
	if test $status -ne 0
	    _ytdl_download_video $argv[$i] $i (count $argv)
end Download Videos From Arguments

The main aim of this function is to transform the URLs contained in the arguments passed to the script to a list of IDs usable later on by ytdl.

function _ytdl_download_arg_urls
    set -g IDs
    for VIDEO in $argv
	_ytdl_log "Info" "Getting video ID for $VIDEO"
	set -g IDs $IDs (youtube-dl --get-id $VIDEO)
    _ytdl_download_queue $IDs
end Download Videos From a Batch File

The final function to declare before the main body of the script is _ytdl_download_batch: it will look for each line, ignoring the ones beginning by #, ; and ] (just like youtube-dl) and will download them, assuming these are channel URLs or playlist URLs, however it should also work with direct video URLs.

What this function does is for each line, it will fetch the entierty of the video IDs found in a playlist or channel. Then, it will look each ID up the list of already downloaded videos and will add all new IDs to a queue of videos to be downloaded. It will then pass each new video ID to _ytdl_download_video directly. Beware that if you pass directly the URL of the channel, such as https://www.youtube.com/user/enyay if you want to download Tom Scott’s videos, it will download everything on the main page of their channel, which means it will even download videos from playlists they decided to put on their channel’s front page, even if it is not theirs. So in that case, we need to append /videos to any channel URL.

function _ytdl_download_batch
    set -q $FILE
    if test $status -eq 1
	set -g NEW
	set CHANNELS (cat $FILE | grep -vE "#|;|\]")
	for c in $CHANNELS
	    _ytdl_log "INFO" "Getting IDs for channel $c"
	    if test (egrep '\/c\/|user|channel' (echo $c |psub))
		set -g IDS (youtube-dl --get-id "$c/videos")
		set -g IDS (youtube-dl --get-id $c)
	    _ytdl_log "INFO" "Fetching new videos from channel"
	    for i in (seq (count $IDS))
		printf "\rsearching (%d/%d)" $IDn (count $IDS)
		rg -- $IDS[$i] $DOWNFILE 2&> /dev/null
		if test $status -ne 0
		    set -g NEW $IDS[$i] $NEW
	    printf "\n"
	for i in (seq (count $NEW))
	    _ytdl_download_video $NEW[$i] $i (count $NEW)
end Main Body

Now that we have all our functions declared, let’s call them! First, we need to parse our arguments. We’ll then download all files passed as arguments. Finally, we’ll download videos, playlists and channels specified from a batch file.

_ytdl_parse_ops $argv
_ytdl_download_arg_urls $VIDEOS

And that’s all! If you’re interested with a very simple interface for downloading one video once, I wrote a small rofi-ytdl script that calls the rofi utility to specify a single link and download it.

6. Plock

plock is a simple script that locks the screen with i3lock while setting as the background image of the locked screen a corrupted screenshot of the screen before it was locked.

set TMPBG /tmp/screen.png
scrot $TMPBG
corrupter -add 0 $TMPBG $TMPBG
i3lock -t -e -f -i $TMPBG

7. Rofi utilities

7.1. askpass

Askpass is a simple script that invokes rofi as a way to get from a GUI the user’s sudo password. It is inspired by this original tool, rewritten in fish and with rofi support instead of dmenu. As you can see, this is a oneliner if we ignore the initial shebang. This executable is pointed at by the

rofi -dmenu -password -no-fixed-num-lines -p (printf $argv[1] | sed s/://)

7.2. awiki

awiki is a simple script used with rofi that relies on the arch-wiki-docs package in order to provide the user a way to quickly find and display any English page from the Arch Wiki in a browser. The advantage of using this over the wiki-search utility from the arch-wiki-lite package is you get instant suggestion in rofi using fuzzy-search. The downside is rofi will only help you find pages by their title, and it will not help you find keywords in the content of said pages.

The first step is to create the list of all the pages that are currently stored on disk. arch-wiki-docs stores them in /usr/share/doc/arch-wiki/html/en. A simple ls piped in three sed will give us a list of page titles. We then pipe that into rofi in dmenu mode in order to choose the page we want to display. By the way, setting the location of the HTML files will come in handy later.

set WLOCATION /usr/share/doc/arch-wiki/html/en/
set WPAGE (/bin/ls $WLOCATION                              | \
    sed -e 's/_/ /g' -e 's/\.html$//' -e 's|.*/\(.*\)|\1|' | \
    rofi -dmenu -p "Arch Wiki" -i)
set WPAGE (echo $WPAGE | sed -r 's/\s+/_/g')

Now, all I need to do is to send this list into rofi and tell it to open the result with our favorite browser with xdg-open.

xdg-open $WLOCATION$WPAGE.html

7.3. dmenu

I wrote this very simple script in order to replace dmenu with rofi’s emulation of dmenu, since I prefer rofi’s appearance. It basically calls rofi’s dmenu emulation with the arguments initially passed to dmenu.

rofi -dmenu $argv

7.4. Emoji picker

The emoji picker is a simple fish script that uses rofi and ~/.config/emoji.txt to provide a small, local search for emojis. Once the emoji is selected, it is copied to the clipboard using xclipboard.

grep -v "#" ~/.config/emoji.txt | rofi -dmenu -p "Select emoji" -i | \
    awk '{print $1}' | tr -d '\n' | xclip -selection clipboard

Also, let’s send a notification telling the user the emoji has been copied!

EMOJI=$(xclip -o -selection clipboard | tr -d '\n')
test -z "$EMOJI" && notify-send "No emoji copied" -u low && exit
EMOJI="$EMOJI copied to clipboard"
notify-send -u low "$EMOJI"

It is inspired from this video from Luke Smith, rewritten in Fish.

7.5. Partition mounting and unmounting

7.5.1. Rofi-mount

rofimount is a script inspired by this one, based on dmenu, which interactively asks the user what to mount, and where to mount it. What I did was replace dmenu with rofi, and fix a couple of bugs I encountered in the original script. Get the mountable elements

What the script does first is detect everything that can be mounted. Between a begin and end, let’s set LFS as a local variable. This si in order to get sane variables in the current block.

set -l LFS

Now, let’s detect the amount of mountable Android filesystems, and if any are detected, let’s read them into a global variable.

set -l a (math (jmtpfs -l | wc -l) - 2)
test $a -ge 0 && jmtpfs -l 2> /dev/null | tail -n $a | read -zg anddrives

We’ll do the same for external and internal drives and partitions that can be mounted here.

lsblk -rpo "name,type,size,mountpoint" | \
awk '$2=="part"&&$4==""{printf "%s (%s)\n",$1,$3}' | \
read -zg usbdrives

Finally, we look for any CD drive that could be mounted on our device.

blkid /dev/sr0 | awk '{print $1}' | sed 's/://' | read -z cddrives

And that’s the end of our first block!


Alright, we’ll save what kind on drives we can mount in a temporary file called /tmp/drives. We’ll make sure it’s blank by erasing it then creating it again with touch, like so. The -f flag on rm is here so we get no error if we try to delete a file that doesn’t exist (yet).

set -g TMPDRIVES /tmp/drives

Now, let’s write what type of drives we can mount in this temporary file.

test -n "$usbdrives" && echo "USB" >> $TMPDRIVES
test -n "$cddrives" && echo "CD" >> $TMPDRIVES
test -n "$anddrives" && echo "Android" >> $TMPDRIVES

Now, we want to declare where to look for mount directories. For now, we’ll only look in /media, but you can add more if you wish.

set -g basemount /media Get the mount point

Now, let’s declare a function that will allow us to chose the drive we want to mount.

function getmount

First, we want to get our mount point. We’ll run a find command on each of the directories listed in $basemount to look for folders on which our drive could be mounted. This list will be passed to rofi from which we will chose our mount point.

set -g mp (for d in $basemount
    find $d -maxdepth 5 -type d
end | rofi -dmenu -i -p 'Type in mount point.')

We should verify that something has been actually selected, otherwise we should abort the script.

if test -z $mp || test $mp = ""
    return 1

Now, if the selected mount point does not exist, we’ll ask the user whether the directory should be created. If no, the script will abort. If yes, an attempt will be made at creating the directory as the user; if that fails, a new attempt will be made as sudo.

if test ! -d $mp
    switch (printf "No\\nYes" | rofi -dmenu -i -p "$mp does not exist. Create it?")
	case 'Yes'
	    mkdir -p $mp || sudo -A mkdir -p $mp
	case '*'
	    return 1

Finally, let’s close the function

end Mount a USB drive, hard drive or partition

Alright, we want to mount a partition that answers by the name of /dev/sdXX, how do we do that? Let’s create first the function mountusb that will take care of it for us.

function mountusb

Now, the first thing we want to do is select the partition we want to mount. Remember, we stored those in $usbdrives earlier, so let’s pipe them into rofi so we can chose from it. Also, awk will get their path in /dev.

set -g chosen (echo $usbdrives | \
rofi -dmenu -i -p "Mount which drive?" | \
awk '{print $1}')

As usual after a user selection, let’s verify something has actually been selected. If not, let’s abort the script.

test -z $chosen && return 1

Now, let’s select the mount point of our partition. We’ll call the function getmount described in Get the mount point to select it.


Let’s verify the variable mp set in getmount is not empty, otherwise abort the script.

test -z $mp && return 1

Now, let’s mount it! We’ll use a switch which will detect the filesystem used so we know how to mount the partition.

switch (lsblk -no "fstype" $chosen)

We have two named case: vfat filesystems.

case "vfat"
    sudo -A mount -t vfat $chosen $mp -o rw,umask=0000

And ntfs filesystems.

case "ntfs"
    sudo -A mount -t ntfs $chosen $mp -o rw,umask=0000

Else, we’ll let mount determine which filesystem is used by the partition (generally ext4).

case '*'
    sudo -A mount $chosen $mp

We’ll also run a chown on this newly mounted filesystem so the user can access it without any issues.

sudo -A chown -R $USER:(id -g $USER) $mp

Let’s close the switch block and send a notification the partition has been mounted.

end && notify-send -a "dmount" "💻 USB mounting" "$chosen mounted to $mp."

And let’s close the function.

end Mount an Android device

The function that manages to mount Android filesystems is mountandroid. Let’s declare it.

function mountandroid -d "Mount an Android device"

We’ll select which Android we want to mount. We will be asked through rofi.

set chosen (echo $anddrives | rofi -dmenu -i -p "Which Android device?" | awk '{print $1 $2}' | sed 's/,$//')

Now, we need to get the bus of the Android device we want to mount. It will be useful later, after we authorized mounting from our device, to get the path to our partition.

set bus (echo $chosen | sed 's/,.*//')

Let’s temporarily mount our device.

jmtpfs -device=$chosen $mp

Now, we need to allow our computer to mount our Android device. Depending on the Android version it is running on, we either need to specify our device is USB connected in order to exchange files, or Android will explicitely ask us if it is OK for our computer to access it. Let’s inform the user of that.

echo "OK" | \
rofi -dmenu -i -p "Press (Allow) on your phone screen, or set your USB settings to allow file transfert"

Now, let’s get the actual path of our Android filesystem we wish to mount, and let’s unmount the previous temporary filesystem.

set newchosen (jmtpfs -l | grep $bus | awk '{print $1 $2}' | sed 's/,$//')
sudo -A umount $mp

Now we cam mount the new filesystem and send a notification if everything went well.

jmtpfs -device=$newchosen $mp && \
notify-send -a "dmount" "🤖 Android Mounting" "Android device mounted to $mp."

And now, we can close our function.

end Mount a CD drive

This part is way easier than the previous functions. As we will see, the function mountcd's body is only three lines long. First, let’s declare the function.

function mountcd

Now, let’s chose the CD drive we want to mount using rofi.

set chosen (echo $cddrives | rofi -dmenu -i -p "Which CD drive?")

We’ll also get the mountpoint thanks to the getmount function described earlier.


And finally, let’s mount it and send the notification everything went well.

sudo -A mount $chosen $mp && \
notify-send -a "dmount" "💿 CD mounting" "$chosen mounted."

Finally, let’s close our function.

end Ask what type of drive we want to mount

The first thing we will be asked if different types of drives are detected is which of these types the user wishes to mount. This is done with the function asktype which is declared below.

function asktype

We will use a switch statement which will use our anwser to rofi about what we wish to mount.

switch (cat $TMPDRIVES | rofi -dmenu -i -p "Mount which drive?")

If we chose the option "USB", we’ll mount a hard drive, partition or USB drive. In which case we’ll call the mountusb function.

case "USB"

If we chose the "Android" option, the mountandroid function is called.

case "Android"

Else if we chose the "CD" option, we’ll call the mountcd function.

case "CD"

If nothing is selected, the function will naturally exit. Now, let’s close our switch statement and our function.

end Launch the mounting functions

Now that we have declared our functions and set our variables, we’ll read the temporary file described in Get the mountable elements. The amount of lines is passed in a switch statement.

switch (wc -l < $TMPDRIVES)

If the file has no lines, i.e. it is empty, we have no mountable media. Let’s inform our user this is the case.

case 0
    notify-send "No USB drive or Android device or CD detected" -a "dmount"

If we only have one line, we have only one type of mountable media. We’ll pass this line to a second switch statement.

case 1
    switch (cat $TMPDRIVES)

This will allow the script to automatically detect what type of media it is, and mount the corresponding function.

case "USB"
case "Android"
case "CD"

Let’s close this nested switch case.


If we have more than one line, we’ll have to ask the user what type of media they want to mount.

case '*'

Now, let’s end our switch statement!


Finally, we’ll delete our temporary file.


And with that, this is the end of our script!

7.5.2. Rofi-umount

rofiumount is the counterpart of rofimount for unmounting our mounted partitions. Get the unmountable drives

First, we will need to list all the drives that can be safely unmounted. Let’s run this.

set -g drives (lsblk -nrpo "name,type,size,mountpoint" | \
awk '$2=="part"&&$4!~/\/boot|\/home$|SWAP/&&length($4)>1{printf "%s (%s)\n",$4,$3}')

Now, let’s get the android devices that are mounted.

set -g androids (awk '/jmtpfs/ {print $2}' /etc/mtab)

And let’s get the CD drives that are mounted.

set -g cds (awk '/sr0/ {print $2}' /etc/mtab)

We’ll store all of our information in a temporary file, /tmp/undrives.

set -g undrivefile /tmp/undrives

Let’s make sure we begin with a clean, empty file.

rm -f $undrivefile
touch $undrivefile

Depending on if the related variables are set, write the different types of mounted drives in the temporary file.

test -n "$drives" && echo "USB" >> $undrivefile
test -n "$cds" && echo "CD" >> $undrivefile
test -n "$androids" && echo "Android" >> $undrivefile Unmount disk partitions

The function unmountusb will take care of unmounting any drive we can safely unmount. First, let’s declare the function.

function unmountusb

Let’s chose the drive to unmount with rofi.

set chosen (echo $drives | \
rofi -dmenu -i -p "Unmount which drive?" | \
awk '{print $1}')

Let’s verify if the user actually selected any drive. If no, let’s abort the script.

test -z "$chosen" && exit 0

Now, let’s unmount the chosen drive and send a notification if it has been done.

sudo -A umount $chosen && \
notify-send "💻 USB unmounting" "$chosen unmounted." -a "dumount"

Now, let’s close the function.

end Unmount Android device

The function unmountandroid will take care of unmounting any mounted Android device. First, let’s declare our function.

function unmountandroid

Let the user choose which Android device to unmount.

set chosen (echo $androids | rofi -dmenu -i -p "Unmount which device?")

We’ll verify the user chose any device.

test -z "$chosen" && exit 0

If a device has been chosen, let’s unmount it and send a notification it has been successfuly unmounted.

sudo -A umount -l $chosen && \
notify-send "🤖 Android unmounting" "$chosen unmounted." -a "dumount"

Finally, let’s close the function.

end Unmount CD drive

unmountcd will take care of unmounting any mounted CD drive. Let’s declare this function.

function unmountcd

As before, let the user chose which CD drive to unmount.

set chosen (echo "$cds" | rofi -dmenu -i -p "Unmount which CD?")

We’ll verify the user chose any device.

test -z "$chosen" && exit 0

If a drive has been chosen, let’s unmount it and send a notification it has been successfuly unmounted.

sudo -A umount -l $chosen && \
notify-send "💿 CD unmounting" "$chosen unmounted." -a "dumount"

Now, let’s close the function.

end Ask what type of drive to unmount

If several types of unmountable drives are available, let’s ask the user which type to unmount based on the content of the temporary file declared in Get the unmountable drives. First, let’s declare the function.

function asktype

Let’s create a switch statement to which will be passed the selection of the user from rofi.

switch (cat $undrivefile | rofi -dmenu -i -p "Unmount which type of device?")

Three types of values can be returned: "USB", "CD", or "Android". These values will be used to launch their corresponding function.

case 'USB'
case 'CD'
case 'Android'

Let’s close the switch statement.


Let’s now close the function.

end Launch the unmounting functions

Now back to the body of our script, let’s input in a switch case the number of lines contained in our temporary file.

switch (wc -l < $undrivefile)

If the file containes no lines. i.e. it is empty, nothing is to be unmounted. Let’s inform the user of that.

case 0
    notify-send "No USB drive or Android device or CD to unmount" -a "dumount"

Else, if there is only one type of drive, we’ll automatically let our script choose based on the content of this sole line.

case 1
    switch (cat $undrivefile)
	case 'USB'
	case 'CD'
	case 'Android'

And if there are more types than one, let’s ask the user.

case '*'

Let’s close our main switch statement.


And finally, let’s delete our temporary file.

rm -f $undrivefile

7.6. Rofi-pass

rofi-pass is a simple utility that gets a password stored in the pass password manager with rofi as its interface, and then stores the password in the clipboard.

Let’s parse all the arguments passed to the script. If one of them is --type, -t or type, the script will attempt to type the password to the text area already selected without pasting the password to the clipboard.

for arg in $argv
    switch $arg
	case '--type' '-t' 'type'
	    set -g TYPE "yes"
	case '*'
	    printf 'Unknown argument: %s\n.' $arg
	    exit 1

Now, let’s get the list of the passwords that exist in our pass repository.

set passwords (find $HOME/.password-store -type f -name "*.gpg" | \
string replace -r ".*.password-store/" "" | \
string replace -r ".gpg" "" | sort)

Let the user choose which password they wish to select.

set password (for elem in $passwords
    echo $elem
end | rofi -dmenu -i -p "Select your password")

Let’s verify we actually selected a password and not just exited. If no password was selected, let’s simply exit the script.

if test -z $password

Depending on the arguments passed earlier, we might want some different behavior.

if test $TYPE = "yes"
    set -l IFS
    set pass (pass show $password | string split -n \n)[1]
    printf %s $pass | xvkbd -file -
    pass show -c $password 2> /dev/null

The default behavior is to copy the password to the clipboard for 45 seconds, so let’s do that.

pass show -c $password 2> /dev/null

Else, if we passed --type, -t or type as an argument of the script, we want it to attempt to type our password in the currently selected text input. Let’s do that.

set -l IFS
set pass (pass show $password | string split -n \n)[1]
printf %s $pass | xvkbd -file -

To correctly get the password from pass, we need to parse the output and only get the first line, hence the following command.

set pass (pass show $password | string split -n \n)[1]

7.7. rofi-ytdl

This is just a simple wrapper around ytdl so I can easily download a video from rofi, which we’ll use first to retrieve the URL of the video we want to download, be it from YouTube or other website supported by youtube-dl.

URL=$(echo "Video to download:" | rofi -dmenu -i -p "Video to download:")

Now, if the variable URL is not empty (i.e. the user specified a link and did not abort the operation), we’ll proceed to teh download. Before it begins, we’ll send a notification saying the download is about to begin. When the ytdl process ends, we’ll also send a notification notifying the user on the success or failure of the download.

if [ -n "$URL" ]; then
    notify-send -u normal "YTDL" "Starting downloading\n$URL"
    if [[ $(ytdl "$URL") ]]
        notify-send -u normal "YTDL" "Finished downloading!"
        notify-send -u critical "YTDL" "Failed downloading\n$URL"

8. Wallpaper utilities

8.1. pape-update

This little tool sets a random wallpaper using xwallpaper.

PAPE=$(find "$PAPESDIR" -type f | sort -R | tail -1)
set-pape "$PAPE"

8.2. Select wallpaper

This script is based on what nsxiv can do as an image viewer as well as xwallpaper.

PAPE=$(nsxiv -orbft ~/Pictures/Wallpapers/*)
set-pape "$PAPE"

8.3. Set a wallpaper

This utility is not meant to be called by the user directly, but rather by scripts that may be written by the user. Its role is simple: check if the provided wallpaper exists and if it is an image. If both requirements are met, the path to this image is then stored in $XDG_CACHE_HOME/wallpaper, or if this variable is empty in $HOME/.cache/wallpaper.

CACHEFILE=$([ -n "$XDG_CACHE_HOME" ] && echo "$XDG_CACHE_HOME/wallpaper" || echo "$HOME/.cache/wallpaper")
[[ -f $1 ]] &&                              \
    grep image <(file -b --mime-type "$1") && \
    echo "$1" > "$CACHEFILE"                    \
    && xwallpaper --zoom "$1"

9. Weather

A quick and useful script I often use is a curl request to v2.wttr.in to get a weather forecast in the terminal. By default, I want the request to be about the city I live in, but it is also possible for the script to accept as its arguments a search inquiry.

if count $argv > /dev/null
    set -l SEARCH (string join '+' $argv)
    curl http://v2.wttr.in/~$SEARCH
    curl http://v2.wttr.in/Aubervilliers

Author: Lucien Cartier-Tilet

Email: lucien@phundrak.com

Created: 2022-08-24 Wed 18:09