nesting command substitutions

My goal is to use dmenu to browse a small set of applications. A list of such applications is in \~/prj/dmenus/favorites/a.txt. If I invoke $(cat ~/prj/dmenus/favorites/a.txt | dmenu)

I get just what I'm after. If I invoke

$(cat ~/prj/dmenus/favorites/a.txt | dmenu -fn 'Droid Sans Mono-18')

I get a output that is nicer to read. Next step, I would like to put the formatting options in a file. I can access that file and read it into a variable by another command substitution.

Example:x=$(<~/.config/dmenu/layout.txt); echo $x yields -fn 'Droid Sans Mono-18'

That is as far as I get. Can't seem to execute in the out command substitution.

$(cat ~/prj/dmenus/favorites/a.txt | dmenu $x)

usage: dmenu [-bfiv] [-l lines] [-p prompt] [-fn font] [-m monitor]

[-nb color] [-nf color] [-sb color] [-sf color] [-w windowid]

Not what I want Similarly, if I use

$(cat ~/prj/dmenus/favorites/a.txt | dmenu $(<~/.config/dmenu/layout.txt))

usage: dmenu [-bfiv] [-l lines] [-p prompt] [-fn font] [-m monitor]

[-nb color] [-nf color] [-sb color] [-sf color] [-w windowid]

Same failure. I bet the solution is really simple, and will enlighten me immensely.

I am using ubuntu 24.04 with fluxbox.

Thanks

Ti











https://redd.it/1ih44b2
@r_bash

Is there a way to get History without <enter>?

Hi, I'd like to get a past command of history for example !1900 but without enter, so I can rewrite that command for this instance and then manually I will do then <enter> for this new changed command?

Regards!

https://redd.it/1ih5bpo
@r_bash

Sed/awk help

Hi, I have text files that contain lists of numbers. Each number is on a separate line. Some of the numbers have forward slashes in the middle (eg 11152/3), some of them don't (eg 11276), which is fine.

However due to the way I collected the data, there are some lines that just have an assortment of slashes and spaces on them and nothing else.

Is there any way I can use sed or awk to get rid of the unwanted slashes whilst keeping the wanted ones?

https://redd.it/1ihd4yv
@r_bash

looking for a way to have a yes or no option at the end of a noscript to start another noscript or exit.

I have a simple backup noscript that creates archives of data. At the end of the noscript it encrypts and then uploads to a cloud server.

I'd like to make this into two noscripts with an option at the end of the first to run the second noscript or exit. i.e, I don't always want to encrypt and upload.

Any ideas?

https://redd.it/1ihisxo
@r_bash

Window Tiling Script w/ xdotool and wmctrl not persistent after switch workspaces

So I'm running xfce4 as my DE (w/ xfwm4 as the WM) and the window tiling that comes default is awesome \*if\* you're using a monitor(s) with normal dimensions. I got a nice ultrawide monitor a while back and the tiling, while still effective, doesn't quite tile the way it would on a standard monitor. I've been meaning to write a noscript to tile more effectively on an ultrawide monitor for a while now and I **finally got around to it** about a week ago.

Things are going great and it works exactly as expected (sort of). I pasted the code below (but I must warn you I'm still pretty new to bash noscripting so there might be an simpler way to accomplish the same thing). I basically find what window is active with xdotool, figure out what monitor that window is on (with some wizardry I found on StackExchange), and tile the window with wmctrl based on the argument passed to the noscript. Then I just programmed each of the variations with different arguments to different keyboard shortcuts and \*chef's kiss\*

**Here's the problem:** every time I change workspaces and change back, one or more of the windows I've tiled with my noscript move around to a different position and size. Is there any reason this could be happening with my noscript or could it be something else in the window manager overriding things?



`# This noscript is meant to tile windows into smaller regions`

`# than what is available by default in xfce4.`

`#`

`# Ultra-wide monitors are effective as a seamless dual monitor,`

`# but window tiling acts different. This is a fix for that issue.`

`# Don't bother using this noscript on a standard monitor. It will`

`# work, but the windows will be unusable.`

`#`

`# This noscript will separate the monitor into 8 regions, 4 on`

`# the top half of the screen and 4 on the bottom, with each`

`# given a letter signifier representing a physical mapping of`

`# a keyboard, like so:`

`#`

`# -----------------`

`# | Q | W | E | R |`

`# |---------------|`

`# | A | S | D | F |`

`# -----------------`

`#`

`# Additionally, there will be 4 more regions with 100% height,`

`# from left to right:`

`#`

`# -----------------`

`# | | | | |`

`# | H | J | K | L |`

`# | | | | |`

`# -----------------`

`#`

`# This gives a total of 12 tiling variations available that`

`# mimic default tiling on a standard monitor. Simply pass`

`# the letter designation of the region you wish to tile your`

`# focused window to as the only argument.`

`#`

`# For example:`

`# 'window-tile.sh -Q' tiles the active window to the top-left`

`# region.`

`#`

`# Each variation can be tied to keyboard shortcuts for easy tiling.`

`# I used <ctrl>+<super>+<letter>`







`# Get active window as decimal using xdotool`

`FOCUSED=$(xdotool getactivewindow)`



`# Convert decimal value to hex for use with wmctrl`

`FOCUSED=$( echo "obase=16; $FOCUSED" | bc )`

`FOCUSED=$( echo "0x0$FOCUSED" | awk '{print tolower($0)}' )`



`# Thanks to terdon from the PowerUser StackExchange for this`

`# next section to determine the current monitor.`



`## Get screen info`

`screen1=($(xrandr | grep -w connected | awk -F'[ +]' '{print $1,$3,$4}' |`

`head -n 1))`

`screen2=($(xrandr | grep -w connected | awk -F'[ +]' '{print $1,$3,$4}' |`

`tail -n 1))`



`## Figure out which screen is to the right of which`

`if [ ${screen1[2]} -eq 0 ]`

`then`

`right=(${screen2[@]});`

`left=(${screen1[@]});`

`else`

`right=(${screen1[@]});`

`left=(${screen2[@]});`



`fi`



`## Get window position`

`pos=$(xwininfo -id $(xdotool getactivewindow) | grep "Absolute upper-left X" |`

`awk '{print $NF}')`



`## Which screen is this window displayed in? If $pos`

`## is greater than the offset of the rightmost screen,`

`## then the window is on the right hand one`



`# Parse resolution of current monitor and assign to`

`# $WIDTH and $HEIGHT`

`if [ "$pos" -gt "${right[2]}" ]`

`then`

`# echo

"${right[0]} : ${right[1]}"`

`IFS=x read -r WIDTH HEIGHT <<< ${right[1]}`

`else`

`# echo "${left[0]} : ${left[1]}"`

`IFS=x read -r WIDTH HEIGHT <<< ${left[1]}`

`fi`



`# Tile the focused window based on argument passed.`

`# Position and size is determined by the resolution of the current moniter:`

`# if $HEIGHT=1440 and I want the window to equal half the height of the`

`# screen, I would use $(( $HEIGHT / 2 )). Enter 'man wmctrl' in your`

`# terminal prompt to get more information on the wmctrl command.`



`if [ $1 = '-Q' ]`

`then`

`wmctrl -ir $FOCUSED -e 0,0,0,$(( $WIDTH / 4 )),$(((( $HEIGHT / 2 )) - 1))`

`elif [ $1 = '-W' ]`

`then`

`wmctrl -ir $FOCUSED -e 0,$(( $WIDTH / 4 )),0,$(( $WIDTH / 4 )),$(((( $HEIGHT / 2 )) - 1))`

`elif [ $1 = '-E' ]`

`then`

`wmctrl -ir $FOCUSED -e 0,$(( 2 * (( $WIDTH / 4 )))),0,$(( $WIDTH / 4 )),$(((( $HEIGHT / 2 )) - 1))`

`elif [ $1 = '-R' ]`

`then`

`wmctrl -ir $FOCUSED -e 0,$(( 3 * (( $WIDTH / 4 )))),0,$(( $WIDTH / 4 )),$(((( $HEIGHT / 2 )) - 1 ))`

`elif [ $1 = '-A' ]`

`then`

`wmctrl -ir $FOCUSED -e 0,0,$(( $HEIGHT / 2)),$(( $WIDTH / 4 )),$(( $HEIGHT / 2 ))`

`elif [ $1 = '-S' ]`

`then`

`wmctrl -ir $FOCUSED -e 0,$(( $WIDTH / 4 )),$(( $HEIGHT / 2 )),$(( $WIDTH / 4 )),$(( $HEIGHT / 2 ))`

`elif [ $1 = '-D' ]`

`then`

`wmctrl -ir $FOCUSED -e 0,$(( 2 * (( $WIDTH / 4 )))),$(( $HEIGHT / 2 )),$(( $WIDTH / 4 )),$(( $HEIGHT / 2 ))`

`elif [ $1 = '-F' ]`

`then`

`wmctrl -ir $FOCUSED -e 0,$(( 3 * (( $WIDTH / 4 )))),$(( $HEIGHT / 2 )),$(( $WIDTH / 4 )),$(( $HEIGHT / 2 ))`

`elif [ $1 = '-H' ]`

`then`

`wmctrl -ir $FOCUSED -e 0,0,0,$(( $WIDTH / 4 )),$(( $HEIGHT ))`

`elif [ $1 = '-J' ]`

`then`

`wmctrl -ir $FOCUSED -e 0,$(( $WIDTH / 4 )),0,$(( $WIDTH / 4 )),$(( $HEIGHT ))`

`elif [ $1 = '-K' ]`

`then`

`wmctrl -ir $FOCUSED -e 0,$(( 2 * (( $WIDTH / 4 )))),0,$(( $WIDTH / 4 )),$(( $HEIGHT ))`

`elif [ $1 = '-L' ]`

`then`

`wmctrl -ir $FOCUSED -e 0,$(( 3 * (( $WIDTH / 4 )))),0,$(( $WIDTH / 4 )),$(( $HEIGHT ))`

`else`

`echo "Argument required"`

`fi`



https://redd.it/1iirxja
@r_bash

help in named pipes

Hi everyone,

I have a question, I was studying a Linux privilege escalation course, and I came across a systemctl abuse https://gtfobins.github.io/gtfobins/systemctl/#sudo

and then I ask myself why not to do it but get interactive shell, using two named pipes, example:

f1=/tmp/infifo
f2=/tmp/outfifo
mkfifo $f1 $f2
sf=mktemp.service
echo -e "Service\nExecStart=eval \"/bin/bash < $f1 > $f2 &\"\nInstall\nWantedBy=multi-user.target" > $sf
sudo systemctl link $sf
sudo systemctl enable $sf --now
cat $f2 &
cat > $f1

but it did not work, but if I tried it without systemctl, am I using pipes incorrect?
and can you help me understanding named pipes and how to use it?

https://redd.it/1ij1agl
@r_bash

Authorized one way but unauthorized the other

When I run curl commands independently in iTerm2, I am able to get a file back that I need with no issue. However, when I run those same commands in a bash noscript, I get unauthorized. The tokens are the same both ways. I can give more info if needed, but why would this be happening?

https://redd.it/1ij5qpj
@r_bash

is anything like "rm all except this, this2, this3"

Hi, I should remove some files.jpg (from 20 +/-) except 3 of them

rm all except DSC1011.jpg Dsc1015.jpg Dsc1020.jpg

what will be the command?

and of course for your GIANT HELPING ALWAYS GENIUSES

https://redd.it/1ijda50
@r_bash

Archstrap - Bootsrap Arch Linux from an existing Linux distro
https://github.com/wick3dr0se/archstrap

https://redd.it/1ikanhw
@r_bash

Depth first or breadth first after learning Linux terminal & Shell noscripting?


Few things on my mind but unsure if I am eligible to learn it..

- gitlab (Mastering gitlab 12)

- Elasticsearch(Not sure of the text to read)

- DNS & BIND (Pro DNS & BIND 10)

- DBA (Pro MySQL)


(Obviously looking into devops route)

https://redd.it/1ikkw26
@r_bash

line buffering vs block buffering

Hi, after trying appending to a file with awk some weird occurrence happened

awk -i inplace '{print $0} ENDFILE{print "endoffile"}' somefile

the next command in terminal finish immediately and throws an error with exit status 1:

cat -A
cat: -: input file is output file

Now the `grep` (which has `--line-buffered` as a possible flag) does fine

grep -

So, my suspicion was `awk -i inplace` has done something wrong, and the [inplace extension manual](https://www.gnu.org/software/gawk/manual/htmlnode/Extension-Sample-Inplace.html) does suggest so

>redirect gawk's standard output to /dev/null

Slightly different from suggested, but this works

awk -i inplace '{print $0} ENDFILE{print "endoffile"}' somefile &>/dev/null

also `sed --in-place` has no problem at all

sed -i '$r /dev/stdin' somefile <<< "endoffile"

So what is the cause of this, and is the manual slightly wrong? It doesn't seems awk -i inplace is like sed -i emulation, like suggested. Also, is &>/dev/null mandatory to follow inplace extension?



https://redd.it/1ikudxt
@r_bash

Learning bash, trying to get it to do something stupid

I'm writing a noscript to handle my code projects, and something stupid I want to add is an ffmpeg command to play every mp3 in a folder after it opens my project in the IDE. Me & GPT (good idea for a romance novel, you're welcome) got this far:

for i in *.mp3; do

ffplay -nodisp -autoexit "/home/scottishcomedian/Music/bash_bullshit/$i"

done

And when I run it, it just hits me with the blank console. What am I doing wrong, oh wise elders?

https://redd.it/1invmde
@r_bash

illegal number problem

Hey, I struggle with some noscript.

var="nef892na9s1p9asn2aJs71nIsm"

for counter in {1..40}
do
var=$(echo $var | base64)
if $counter -eq 35
then
echo $var | WC -c
fi
done

It always give me: illegal number: {1..40}
Can someone help?

https://redd.it/1iogfzm
@r_bash

How to parse a nested JSON file in an old unix version

Hi, I'm trying to split a JSON file by transaction_id wherein 1 transaction_id=1 record. However, I feel like my shell noscript is failing due to the fact that it cannot read the JSON file and it won't proceed to processing on what I want it to do. You may see the snippet of my code below.


# Extract all transaction_id values using a regular expression

echo "$content" | sed -n 's/.*"transaction_id":\s*"\([^"]*\)".*/\1/p' | while read transaction_id; do

# Debugging: Show the current transaction_id being processed and log it

echo "Processing transaction_id: $transaction_id" | tee -a "$BATCH_LOG"



# Get the last character of the transaction_id

last_char="${transaction_id: -1}"



# Debugging: Show the last character of the transaction_id and log it

echo "Last character of '$transaction_id': $last_char" | tee -a "$BATCH_LOG"



# Check the last character and categorize

if [[ "$last_char" =~ [0-4] ]]; then

echo "$transaction_id" >> "${file%.json}_01.json"

# Debugging: Log which file the transaction_id is being saved to

echo "Saved to: ${file%.json}_01.json" | tee -a "$BATCH_LOG"

elif [[ "$last_char" =~ [5-9] ]]; then

echo "$transaction_id" >> "${file%.json}_02.json"

# Debugging: Log which file the transaction_id is being saved to

echo "Saved to: ${file%.json}_02.json" | tee -a "$BATCH_LOG"

elif [[ "$last_char" =~ [a-l] ]]; then

echo "$transaction_id" >> "${file%.json}_03.json"

# Debugging: Log which file the transaction_id is being saved to

echo "Saved to: ${file%.json}_03.json" | tee -a "$BATCH_LOG"

elif [[ "$last_char" =~ [m-z] ]]; then

echo "$transaction_id" >> "${file%.json}_04.json"

# Debugging: Log which file the transaction_id is being saved to

echo "Saved to: ${file%.json}_04.json" | tee -a "$BATCH_LOG"

else

# Debugging: Log unexpected last characters

echo "Unexpected last character '$last_char' for transaction_id: $transaction_id" | tee -a "$BATCH_LOG"

fi

done


I hope someone can help I've been losing my mind over this.

https://redd.it/1ip6rpi
@r_bash

Transposing args in noscript, including quotes

I'm trying to create a noscript to interact with my docker containers without having to shell in and run commands manually. It's a very simple noscript:

#!/bin/bash

ALLARGS="$@"
docker compose exec api ash -c "cd ../ && alembic ${ALLARGS}"

I tried a few things (${ALL_ARGS//\"/\\\"}, sed, others), but finally noticed that "$@" simply doesn't contain the double quotes. Is there a way to transpose the args as is?

https://redd.it/1iow04v
@r_bash

Check if number of arguments is one after all the flag

I have a noscript who can take more than one flag.

./noscript -a list is the same than ./noscript list all but list can have other parameter than all so what i want is ./noscript -a list somethingHere give a error.


So what i have test is if $3 is empty when -a is given.

But if the user type ./noscript -a -s list this give a error because $3 is no longer empty but the exeption behavior is to work.

if aflag = 1 and (after 'list' is empty)
do something
else
error

So my idea is this on pseudo code. But i don't know how to check dynamicly if the $n+1 after list ( $n) is empty

https://redd.it/1ip7zyv
@r_bash

Unexpected curl command behaviour ?

The following command reads the exchange rate information for the EUR/USD currency pair from HTML page and prints it.

page=$(curl -s https://www.widgets.investing.com/live-currency-cross-rates?theme=darkTheme&pairs=1); echo "$page" | pup 'div.pid-1-bid text{}'

But why doesn't the following command work, instead it prints the entire page?

curl -s https://www.widgets.investing.com/live-currency-cross-rates?theme=darkTheme&pairs=1 | pup 'div.pid-1-bid text{}'

https://redd.it/1imsp0s
@r_bash

WHAT IS BASH DOING?

**UPDATE**

So it looks like FFPMEG is interacting with the shell in some way... so adding this to the FFPMEG line seems to have resolved the issue.

</dev/null >/dev/null 2>&1



I am doing something dumb... I guess? But I can't figure out what in the heck, when using the EVAL statement, previous variables are stripping off a character for every other loop? Sound confusing? I am confused...

I am using FFMPEG and writing a quick little bash wrapper to automatically detect silences and split apart an audio file.

Let me see if I can show what is going on... This is WITHOUT the eval command...

IFS='\r\n'
while read -r line1; do
IFS= read -r line2
echo "Start: $line1"
echo "End: $line2"
echo "Prev: $PREV"
START="${PREV}"
END="${line1}"

echo "/usr/bin/ffmpeg -hidebanner -loglevel error -i ./${INPUT} -ss ${PREV} -to ${line1} output${COUNT}.wav"
COMMAND='/usr/bin/ffmpeg -hidebanner -loglevel error -i '
COMMAND+="./${INPUT} -ss ${START} -to ${END} output${COUNT}.wav"

echo "${COMMAND}"
# eval ${COMMAND}

COUNT=$(( COUNT + 1 ))
PREV=$line2
echo ''

done <<< $SILENCES

This outputs exactly what I would expect...

Start: 6.04
End: 6.30
Prev: 0
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 0 -to 6.04 output0.wav
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 0 -to 6.04 output0.wav
Start: 21.72
End: 21.98
Prev: 6.30
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 6.30 -to 21.72 output1.wav
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 6.30 -to 21.72 output1.wav
Start: 24.18
End: 24.53
Prev: 21.98
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 21.98 -to 24.18 output2.wav
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 21.98 -to 24.18 output2.wav
Start: 43.34
End: 43.58
Prev: 24.53
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 24.53 -to 43.34 output3.wav
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 24.53 -to 43.34 output3.wav

SO then I uncomment the eval command. That is the only change. I have tried with and without " ", using and not using { } to see if I am interpretting the string differently.

`eval ${COMMAND}`

SOOOO.... Here is the output

Start: 6.04
End: 6.30
Prev: 0
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 0 -to 6.04 output0.wav
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 0 -to 6.04 output0.wav

Start: 1.72
End: 21.98
Prev: 6.30
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 6.30 -to 1.72 output1.wav
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 6.30 -to 1.72 output1.wav
-to value smaller than -ss; aborting.

Start: 24.18
End: 24.53
Prev: 21.98
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 21.98 -to 24.18 output2.wav
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 21.98 -to 24.18 output2.wav

Start: 3.34
End: 43.58
Prev: 24.53
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 24.53 -to 3.34 output3.wav
/usr/bin/ffmpeg -hidebanner -loglevel error -i ./audio.wav -ss 24.53 -to 3.34 output3.wav
-to value smaller than -ss; aborting.

SO Every other iteration... the ${PREV} variable has the first digit/character stripped. So for the second iteration:

21.72 -> 1.72

BUT this ONLY happens when I have the EVAL command AFTER the echo commands. So somehow the eval command is affecting that variable, but I can't see how. Thanks!

https://redd.it/1im9sik
@r_bash

How can I combine these two awk commands?

git ls-remote -t `https://github.com/PowerShell/PowerShell.git` | awk -e '$2 ~ /v[0-9].[0-9].[0-9]$/ {sub(/refs\/tags\/v/,"");print $2}' | awk 'END {print}'


This command gives me the latest version of Powershell. I just want the result from the last line. How can I add the END rule to the first awk command?

https://redd.it/1ik223z
@r_bash

Does anyone know of an interactive BASH command builder that is built with BASH noscripts?

I'd love to have an interactive shell environment that helps students new to BASH navigates its opacity and easily survey the breadth of options. I would also like it for myself, since one can always learn more.

This is especially useful when BASH is required for a subject like Practical Control Theory with Python but is not the subject of the class. Think of it as a nice gateway drug for BASH.

And while I'm asking for ponies, I think it would be best to make this run natively inside terminal emulators. So writing it as a BASH noscript would be great.

https://redd.it/1iq37l3
@r_bash