confused with sed

paperplane · 12-29-2007, 06:09 PM

hi,

i want sed to send only the first line of a text file to another text file, but it sends them all. this is the commandline ive tried

'cat list1 | sed -e '1p' > list2'

ive had a look at some documentation on sed on the net but i cant figure it out. can anyone help with this?

jozyba · 12-29-2007, 06:25 PM

If you look closely at the output of your command you will see that it prints the first line and then prints the whole file; so the first line gets printed twice. To suppress this behaviour you should use the -n switch:

Code:

sed -n '1p' list1 > list2

Alternatively you could just delete all lines from line two to the end:

Code:

sed '2,$d' list1 > list2

syg00 · 12-29-2007, 07:15 PM

And, as implied by jozyba, the "cat" is redundant.

And if it's a big file,you might want to quit after that first line rather than spin through the entire file. Try this

Code:

sed -n '1p;1q' list1 > list2

ghostdog74 · 12-29-2007, 10:24 PM

alternative:

Code:

head -1 file > newfile

paperplane · 12-30-2007, 09:39 AM

thanks guys, in the end i went with 'sed q', which emulates "head -1".

by the way, can the sed command rename files, or just work on the text within them?

i need to rename alot of .gif files which contain whitespace, and substitute it with a dot (.)

if sed cant do this, does anyone know any other way of doing this?

gerbenny · 12-30-2007, 11:31 AM

That's really easy one, here's a quick bash loop to ls over a whole directory, and output a file with a full list of rename commands:

Code:

#!/bin/bash
for i in $( ls ); do
  echo $i |sed "s/[[:space:]]/./g" |sed "s/^/ren $1 /g" >> script_to_rename.sh
done
chmod 777 script_to_rename.sh

Then just run the newly created script_to_rename.sh file.

ghostdog74 · 12-30-2007, 12:10 PM

Quote:

Originally Posted by gerbenny

Code:

#!/bin/bash
for i in $( ls ); do
  echo $i |sed "s/[[:space:]]/./g" |sed "s/^/ren $1 /g" >> script_to_rename.sh
done
chmod 777 script_to_rename.sh

that's a redundant for loop.
Essentially you can just do this:

Code:

ls * | sed .......| >> file

paperplane · 12-30-2007, 12:30 PM

Quote:

Originally Posted by ghostdog74

that's a redundant for loop.
Essentially you can just do this:

Code:

ls * | sed .......| >> file

ghostdog74, could you explain please, what should i put in the ... bit?

pixellany · 12-30-2007, 01:50 PM

Quote:

Originally Posted by paperplane

ghostdog74, could you explain please, what should i put in the ... bit?

He's just showing you an overall structure---you would put it whatever SED syntax is required for your task. This thread already includes several ways of solving the same problem---with SED or otherwise.

Here is my favorite SED tutorial: http://www.grymoire.com/Unix/Sed.html

pixellany · 12-30-2007, 01:55 PM

Quote:

Originally Posted by ghostdog74

that's a redundant for loop.
Essentially you can just do this:

Code:

ls * | sed .......| >> file

I'm not sure that you would use a pipe followed by a redirection operator????

i.e.: ls *|sed ...... >>file

gerbenny · 12-30-2007, 03:58 PM

Actually, in this case the for loop is a compliment for the sed command, if I was to use sed to pull the original filename to rename, instead of the $i varibale I would have to write the sed command diferently. So in this case the for loop is not redundant.

gerbenny · 12-30-2007, 04:01 PM

Quote:

Originally Posted by pixellany

He's just showing you an overall structure---you would put it whatever SED syntax is required for your task. This thread already includes several ways of solving the same problem---with SED or otherwise.

I agree, there are many ways to skin a cat, and no one way is correct when is comes to linux.

paperplane · 12-30-2007, 04:18 PM

thanks all

and thanks for the info on sed, ive just discovered it this weekend and it looks to be a very useful command.

jozyba · 12-30-2007, 04:36 PM

Quote:

Originally Posted by gerbenny

Actually, in this case the for loop is a compliment for the sed command, if I was to use sed to pull the original filename to rename, instead of the $i varibale I would have to write the sed command diferently. So in this case the for loop is not redundant.

Firstly your script won't work because you forgot to quote "$(ls)" and "$i".

Code:

#!/bin/bash
for i in "$(ls)"; do
  echo "$i" |sed "s/[[:space:]]/./g" |sed "s/^/ren $1 /g" >> script_to_rename.sh
done

Secondly the only place you use the variable "$i" is to echo it to sed. As ghostdog74 points out if you use 'ls *' you don't have to echo "$i" so the loop is redundant:

Code:

ls * | sed -e "s/ /./g" -e "s/^/ren $1/g" >> script_to_rename.sh

gerbenny · 12-30-2007, 05:54 PM

Quote:

Originally Posted by jozyba

Firstly your script won't work because you forgot to quote "$(ls)" and "$i".

Code:

#!/bin/bash
for i in "$(ls)"; do
  echo "$i" |sed "s/[[:space:]]/./g" |sed "s/^/ren $1 /g" >> script_to_rename.sh
done

You are correct, newbie mistake to forget those quotes....

Quote:

Originally Posted by jozyba

Secondly the only place you use the variable "$i" is to echo it to sed. As ghostdog74 points out if you use 'ls *' you don't have to echo "$i" so the loop is redundant:

Code:

ls * | sed -e "s/ /./g" -e "s/^/ren $1/g" >> script_to_rename.sh

That's a new one for me, I didn't realise the $1, $2, $3 etc, variables could be passed to sed that way. Thanks.

When you stop learning, it's time for a funeral.