Cut from right to left?

LocoMojo · 02-24-2007, 09:59 AM

Is there a way I could use the cut command going from right to left? For example I want to cut from the last "/" in a url:

/home/me/documents/resume

I want "resume" in the output.

In this instance I know I could use:

cut -d "/" -f 4

But that would only work for directories 4 deep. I want this to work for directories 2, 3, 5, 6 deep.

Is there a way I could cut the first "/" going from right to left?

Another question:

Using a bash script, how do I grab Konqueror's "$PWD". In other words, if I navigate to /home/me/documents/resume in konqueror, how can I grab that url in bash? I'm not very familiar with DCOP, is there a way I could get it via DCOP or something?

Thanks in advance.

LocoMojo

LocoMojo · 02-24-2007, 10:04 AM

Quote:

Originally Posted by LocoMojo

Is there a way I could use the cut command going from right to left? For example I want to cut from the last "/" in a url:

/home/me/documents/resume

I want "resume" in the output.

In this instance I know I could use:

cut -d "/" -f 4

But that would only work for directories 4 deep. I want this to work for directories 2, 3, 5, 6 deep.

Is there a way I could cut the first "/" going from right to left?

Another question:

Using a bash script, how do I grab Konqueror's "$PWD". In other words, if I navigate to /home/me/documents/resume in konqueror, how can I grab that url in bash? I'm not very familiar with DCOP, is there a way I could get it via DCOP or something?

Thanks in advance.

LocoMojo

Ooops, I got it.

I wasn't aware of the "rev" command.

Still looking for the Konqueror's url deal though.

LocoMojo

firstfire · 02-26-2007, 03:52 AM

Hello!

Take a look at `man basename'.

bigearsbilly · 02-26-2007, 04:10 AM

you should ask what you need to be done, not how you think it should be done

kshkid · 02-26-2007, 06:37 AM

Code:

echo "/home/me/documents/resume" | sed 's/^.*\///g'

LocoMojo · 02-26-2007, 07:48 AM

Quote:

Originally Posted by firstfire

Hello!

Take a look at `man basename'.

Hi Firstfire,

Thanks for that, I'll look into basename.

LocoMojo

LocoMojo · 02-26-2007, 07:54 AM

Quote:

Originally Posted by bigearsbilly

you should ask what you need to be done, not how you think it should be done

Hello Bigbearsbilly,

I need for a bash script to get the current working directory of Konqueror.

In other words, when I call a bash script from within Konqueror via a service menu I need for the bash script to know which directory Konqueror is currently in.

Thanks,
LocoMojo

LocoMojo · 02-26-2007, 07:58 AM

Quote:

Originally Posted by kshkid

Code:

echo "/home/me/documents/resume" | sed 's/^.*\///g'

Hi Kshkid,

That seems to be far more elegant than what I wound up doing:

Code:

echo "/home/me/documents/resume" | rev | cut -d "/" -f 2- | rev

I really need to break down and learn about sed and awk.

Thanks,

LocoMojo

alienDog · 02-26-2007, 10:15 AM

Sed isn't too bad, but my humble opinion is that rev|cut|rev is more elegant. It might also be faster to execute than sed (not sure though, would need testing). On a single "cut" the difference might be marginal, but if you do massive amounts of cutting it may make a big difference.

I guess it's also a matter of taste...

schneidz · 02-26-2007, 11:30 AM

by and by, since idont use konqueror can someone explain if nautilus scripts are portable in konqueror ?

they are usually bash based.

kshkid · 02-26-2007, 10:01 PM

Quote:

Originally Posted by alienDog

Sed isn't too bad, but my humble opinion is that rev|cut|rev is more elegant. It might also be faster to execute than sed (not sure though, would need testing). On a single "cut" the difference might be marginal, but if you do massive amounts of cutting it may make a big difference.

I guess it's also a matter of taste...

But with the other option, there is a need to use too many processes and kernel data structures where the task could be done with just 1 process and also without any kernel data structures ( assuming input to be in a file and piped through the output of other process )

So, I think the solution with sed should be economical and faster !

bigearsbilly · 02-27-2007, 02:25 AM

why you need to know what dir konqueror is in?
don't you just need to know the cwd of the script.
i should think konq will just be in the directory it lives in.

$PPID is the parent of the shell script.
do ptree $PPID see what happens

oops, ptree is solaris, do you have on linux?

alienDog · 02-27-2007, 12:16 PM

Quote:

Originally Posted by kshkid

But with the other option, there is a need to use too many processes and kernel data structures where the task could be done with just 1 process and also without any kernel data structures ( assuming input to be in a file and piped through the output of other process )

So, I think the solution with sed should be economical and faster !

Could, could be... I was just guessing

Sed usually isn't blazingly fast...

cfaj · 02-27-2007, 05:01 PM

Quote:

Originally Posted by LocoMojo

Is there a way I could use the cut command going from right to left? For example I want to cut from the last "/" in a url:

/home/me/documents/resume

I want "resume" in the output.

In this instance I know I could use:

cut -d "/" -f 4

But that would only work for directories 4 deep. I want this to work for directories 2, 3, 5, 6 deep.

Is there a way I could cut the first "/" going from right to left?

var=/home/me/documents/resume
basename=${var##*/}

Quote:

Another question:

Using a bash script, how do I grab Konqueror's "$PWD". In other words, if I navigate to /home/me/documents/resume in konqueror, how can I grab that url in bash? I'm not very familiar with DCOP, is there a way I could get it via DCOP or something?

Where do you want to "grab" it? What do you want to do with it?

alienDog · 02-28-2007, 05:51 AM

Quote:

Originally Posted by kshkid

But with the other option, there is a need to use too many processes and kernel data structures where the task could be done with just 1 process and also without any kernel data structures ( assuming input to be in a file and piped through the output of other process )

So, I think the solution with sed should be economical and faster !

Well... curiosity got the better of me and I just had to test it

Actually rev|cut|rev seems to be significantly faster than cat|sed, here's how I tested it:

Code:

#!/bin/bash

# Make testfile with 500000 entries if it doesn't exist already:
if [ ! -e "testfile" ]; then
    START=$(date +%s)
    for i in  $(seq 500000); do
        echo "/foo/bar/bax" >> testfile
    done
    echo "Creating a testfile took $(expr $(date +%s) - $START) seconds"
fi

# With rev|cut|rev:

START=$(date +%s)
rev testfile | cut -f 1 -d "/" | rev > /dev/null
echo "rev|cut|rev took: $(expr $(date +%s) - $START) seconds"

# With cat|sed:

START=$(date +%s)
cat testfile | sed 's/^.*\///g' > /dev/null
echo "cat|sed took: $(expr $(date +%s) - $START) seconds"

With rev|cut|rev I get an average of 3 seconds and with cat|sed >10 seconds... My machine is real old, though, and with newer machines the differences might be much smaller, if even existant.

Basename works fine as long as you're dealing with files and their paths, but it doesn't work with other types of data.