Apropos an ongoing project, looking at sed, and realising:
I've only been using sed for, oh, 40 years.
Cute little script is cool! I wrote a little bash script to wrap around a thesaurus .txt file I found in a unix stack exchange answer (link in script).
It started out as a one-liner, but I adapted it after the results were not completely terrible, though I'm always looking for better ones.
Why are all thesauruses (thesauri? thesauruseseseses?) online and why aren't there any decent local offerings for linux?
Either way, enjoy this little standalone bash script - I'm p proud of the ranking algorithm implementation in the big piped one-liner bit at the end.
Today I learned from the flock(1) man page (on Debian) that you can put this line at the top of a shell script to prevent more than one copy of it from ever running at a time:
[ "${FLOCKER}" != "$0" ] && exec env FLOCKER="$0" flock -en "$0" "$0" "$@" || :
This is an extremely clever hack and I don't know how I've never stumbled upon it in 38 years of writing shell scripts.
#unix #linux #scripting #shellScripting
Update: derp, just use readlink, derp. XD
In the #iDontKnowWhyIdidntWriteThisYearsAgo category:
~ $ type deref
deref is a function
deref ()
{
stat -c "%N" "$*" | cut -f4 -d\'
}
~ $ which gcc
/usr/bin/gcc
~ $ deref $(which gcc)
gcc-14
~ $ deref $(which $(deref $(which gcc)))
x86_64-linux-gnu-gcc-14
Revolutionizing Shell Scripting: Meet CMDS-Framework for Zsh
The CMDS-Framework, or simply cmds, is set to transform the way developers write shell scripts by introducing robust argument parsing and completion features. This innovative zsh framework aims to str...
https://news.lavx.hu/article/revolutionizing-shell-scripting-meet-cmds-framework-for-zsh
I've probably tooted about this before, but I don't know why this isn't standard.
It's just so obvious, at least to me. ;)
~ $ type mcd
mcd is a function
mcd ()
{
[[ -n $1 ]] && mkdir "$1" && cd "$1"
}
I think I won't bother anymore with writing (bash) shell scripts that are longer than a few lines. I find the syntax too unintuitive and there are better programming languages like PHP that produce more readable code.
In the last two hours, I translated a bash shell script with ~250 lines to a PHP CLI script. The latter is nearly 400 lines long but definetly more readable and it also has more user-friendly output.
1/2
Interesting #ShellScripting quandary:
(solution below)
I'm uncertain how shells delimit filename variables internally.
I have a shell (bash) function that loops through a set of PDF files, opening them in zathura in a "suckless" tabbed window (so that the PDFs always open within the same window, so I can just hit q to zip through them one-at-a-time.
Here is my variable declaration:
local files=${*-*.[pP][dD][fF]}
But for simplicity, you could just imagine it to be:
files=*
Then I'm doing a
for f in $files; do
I'm wanting to print a status line for each item viewed, so I have an idea how many more there are to go, so I'm using this:
echo "[[$f]] ($count/$tot)"
And of course, there's a ((count++)) at the beginning of the loop.
But how to get the total ($tot)??
echo $files |wc -l will always result in 1, as does echo "$files" |wc -l
What I ended up doing was just creating a
for f in $files; do ((tot++)); done
One-liner loop before the actual loop, just to get a total.
Is there a better way?
Am I missing something really obvious?
Solution, courtesy of @khm
Use an array!
local files=(${*-*.[pP][dD][fF]})
echo "[[$f]] ($count/${#files[@]})"
Should've made this a long time ago:
function ciglob {
#case-insensitive glob generator
echo "$*" |while read -N1 c; do
case "$c" in
[a-zA-Z]) echo -n "[${c^^}${c,,}]";;
*) echo -n "$c"
esac
done
}
~ $ ciglob "Hello, world!"
[Hh][Ee][Ll][Ll][Oo], [Ww][Oo][Rr][Ll][Dd]!
~ $ ls -ld $(ciglob documents)
drwxr-xr-x 52 ~~~ ~~~ 20,480 Apr 10 11:45 Documents
(Not the most useful example, but I did have a use case in mind when I wrote it ;)
P.S. (This is a valid way to close a parenthesis. Fight me ;)
#Example directory: 27a81c6a_06f2_4a0b_82a3_5eeee7ef91e8
h="[0-9a-f]" #hex value glob
for dir in $h$h$h$h$h$h$h${h}_$h$h$h${h}_$h$h$h${h}_$h$h$h${h}_$h$h$h$h$h$h$h$h$h$h$h$h; do
if [[ -d $dir ]]; then
...
Gotta say, I wish #bash had local / lexically-scoped functions, rather than having to rely on hacks like:
function widget {
function ___widget_parse {
...
}
...
unset ___widget_parse
}
Something like this (which does not currently work) would be lovely:
function widget {
local function parse {
...
}
}
OSS CLI tools appreciation post.
Sure, you use AI tools to generate CLI commands. I use CLI tools to generate AI prompts. We are not the same.
I've probably looked this up in `man` a few hundred times. But it's not an example in man... Gotta actually read to figure it out. Why did they choose `+`? That always throws me.
`date +"%Y%m%d"`
Here a function to download a YouTube thumbnail to file. Written in dnyScript for AquaShell.
today I learned:
#linux #shell
#ShellScripting #wetter #weather
füge das zu Deiner ~/.bashrc hinzu:
```
wetter(){
curl http://wttr.in/$1
}
```
danach:
(found at https://github.com/chubin/wttr.in)
Automate your tasks on BashCore! 
Tired of typing long commands over and over? Prepare your .sh, .py, .rb or .pl scripts on another PC, save them on a USB stick, and run them directly on BashCore. No need to rewrite everything—just execute and go!
Faster workflow Fewer errors Maximum efficiency
I want to like ShellSpec, but it's custom It/When/The DSL language feels so disconnected from actual code. It's also not in many package repos, where as shunit2 already is.
https://shellspec.info/
I figured out how to "stub" specific commands in shell script tests, but it's not pretty.
# stub the "command" command as a function
function command() {
if [[ "$1" == "-v" ]] && [[ "$2" == "java" ]]; then
# stub that java is not installed
return 1
else
/usr/bin/command $*
fi
}
# and to remove the stub
unset command
