You must log in or # to comment.
My five thousand line bash script can do things that one hundred thousand lines of code could not do.
On the brightside, at least script monkeys can now look down on vibe coders.
I think the most I’ve done with awk is write a battery monitor applet with it, it involved parsing data from /sys and making choices based on it so I decided it was a decent choice
Nushell enjoyers represent
Hey I throw a
/^regexp.*/ {print $NF}in there sometimes!…but yes, it’s mostly
print $1—but only because I mix up the parameters whenever I try to usecut!cut -d ' ' -f1master racejoke so dark I had to turn up my screen brightness to enjoy it.
You can even do sum with awk, you don’t need excel
num-utilshasnumsum.
Why spend 30 seconds manually editing some text when you can spend 30 minutes clobbering together a pipeline involving awk, sed and jq
Or 3 minutes clobbering a pipeline in nushell. Yes, I’m a nushell fanboy. I still over-automate everything, but with nushell it’s actually fun.
Or 60 minutes making it all work just with jq functions.
TIL I am an OP wizard.
to be fair, out of those three, jq invokes the least existential dread in me
The important part is to learn the limits of any tool. Nowadays I no longer use jq for any long or complicated tasking. Filter and view data? jq is fine. Anything more and I just cook up a python script.
i’m more of a bash fan tbh. Ever since i started using linux, python started to irritate me
How do you get complex data structures to work? I was alienated from scripting on zsh because I wanted something like a dict and realised I would have to write my own implementation. Is there a work around for that?
I mean, there’s a point in data structure complexity where it’s useful to use Python.
But as to dicts, sure. You’re looking for zsh’s “associative array”. Bash has it too.
zsh
$ typeset -A mydict $ mydict[foo]=bar $ echo $mydict[foo] bar $bash
$ typeset -A mydict $ mydict[foo]=bar $ echo ${mydict[foo]} bar $This will do nicely - I had several workflows where I’d hit an API and get a massive super nested JSON as output; I’d use jq to get the specific data from the whole thing and do a bunch of stuff on this filtered data. I pretty much resigned to using python because I’d have successively complicated requirements and looking up how to do each new thing was slowing me down massively.
All my homies use dubious regex
Honestly I think 90% of people would never use awk if there was a simple preinstalled command for “print the nth column”
This is definitely somewhere that PowerShell shines, all of that is built in and really easy to use
People are hating on Powershell way too much. I don’t like its syntax really but it has a messy better approach to handling data in the terminal. We have nu and elvish nowadays but MS was really early with the concept and I think they learned from the shortcomings of POSIX compatible shells.
I really can’t stress enough how much power and flexibility comes with an object oriented shell, especially with the dotnet type system behind it.
I think most people who hate it just do so either because it came from Microsoft (which… Yeah, that’s understandable), or because it’s a different way of thinking about it (and/or they spent a lot of effort learning how to parse data from strings effectively and hate that it’s made easier?). But love or hate it, it is effective and powerful, and I find myself missing that when working with bash.
cut?
To be fair, a lot of the programs don’t use a single character, have multiple spaces between fields, and
cutdoesn’t collapse whitespace characters, so you probably want something more liketr -s " "|cut -d" " -f3if you want behavior likeawk’s field-splitting.$ iostat |grep ^nvme0n1 nvme0n1 29.03 131.52 535.59 730.72 2760247 11240665 15336056 $ iostat |grep nvme0n1|awk '{print $3}' 131.38 $ iostat |grep nvme0n1|tr -s " "|cut -d" " -f3 131.14 $This is awesome! Looks like an LPI1 textbook. Never got the certification but I’ve seen a couple books about it and remember seeing examples like this one.
I never understood why so many bash scripts pipe grep to awk when regex is one of its main strengths.
Like… Why
grep ^nvme0n1 | awk '{print $3}'over just
awk '/^nvme0n1/ {print $3}'Because by the time I use
awkagain, I’ve completely forgotten that it supports this stuff, and the discoverability is horrendous.
my favorite awk snippet is
!x[$0]++which is likeuniqbut doesn’t care about order. basically, it’s equivalent toprint_this_line = line_cache[$current_line] == 0; line_cache[$current_line] += 1; if $print_this_line then print $current_line end.really useful for those long spammy logs.
Oh that’s very interesting. I usually do
sort --uniqueorsort [...] | uniqif I need specific sorting logic (like by size on disk, etc).Looking at the above awk snippet, it’ll retain order, though. So, sort will normally change the order. The awk snippet won’t, just skip occurrences of a given line after the first. Depending upon the use case, that order retention could be pretty desireable.
same, that statement saved me so much effort
10 PRINT BUTTS
20 GOTO 10
Everything you do with
awk, you can do withpython, and it will also be readable.Hmm, but you have to install and run the Python environment for that. AWK is typically present on *NIX systems already. Python seem like overkill for basic text processing tasks.
On Debian the
pythonis preinstalled.
and
perl, if you want it less readableOr you are old and crazy
Or PowerShell if you want it extra verbose
In all my years I’ve only used more than that a handful of times. Just don’t need it really
Now jq on the other hand…
– [x] I’m in this picture and don’t like it.
