Replacing tabs with spaces

I have a file with lots of tabs and I want to replace all the tabs with five spaces. The tabs are scattered around the text file and in no particular order. Here's what I've tried:

>>> sed 's/\t/ /g' oldfile > newfile

On this code I used \t because it's the escape character in C++ to print tabs, plus the ascii table says that what tabs are. This code literally replaced all the lower case t's with 5 spaces.

***************************************

>>> sed 's/\011/ /g' oldfile > newfile

On this code I used \011 because it's the octal notation for tabs. It didn't do anything as far as I can tell because as far as I know there aren't any strings containg the chars 011. So that didn't work either. Also, since the octal notation didn't work I don't see why the decimal or hex notation would work either.

***************************************

>>> sed 's/\^I/ /g' oldfile > newfile

On this code I tried ^I because it's the hidden character for tabs in vi, I didn't think it would work and it didn't.

I'm sure there's a simple solution to this, but right now I'm drawing a blank. Any help would be appreciated. thanks.

Hmm.. strange that \t doesn't work. As you can see it works for me.





What version of sed are you using?

The remote server I log into is using GNU sed version 3.02. Maybe, for some reason the admin doesn't allow this??? Also, just to be thorough I also tried (but also didn't work) >>>

cat oldfile | sed 's/\t/ /g' oldfile > newfile

I still want to try and find a solution. Thanks.

That's a fairly old version of sed... rather than trying to find a workaround for sed, might I suggest a perl solution?

That perl code works! Thank you.

At work, I'm told we use the latest version of Sun, tomorrow I'm going to try out the sed code. I'm willing to bet it works just fine.

did you forgot to turn off the \ in the regex?

I just tested it out on Solaris 10 update 10 with no luck. However I did discover that the `expand` command will do exactly what you're looking for.

YES! That works nicely. calling up >>>

$ info expand

Says, using the general form of the command >>>

$ expand oldfile > newfile

says that the default is to replace all the tabs with eight spaces. However, on the remote antiquated system I tried this on it was seven spaces. No big deal because you can specify the number of spaces for tab replacement, for example, if you want five spaces then >>>

$ expand -t5 oldfile > newfile


* * * * * * * * *

Someone asked if I turned off the \ in the regex, for the original sed code I tried >>>

$ sed 's/\t/ /g' oldfile > newfile

I didn't since I didn't know I would need to. How is that done? The perl code works great and expand command works great, but I still want to see if I can get it to work with sed. I'm just stubborn like that.

Use the tab key, not the "tab" character.

As you will notice you cannot "see" the tab character but sed clearly replaced tabs with spaces because number of characters in the file list.c grew.

Try this:
cat oldfile | sed -e 's/[EnterTabSpace]//g' > NewFile

I do this all the time. For some reason, neither typing "\t" nor hitting the 'tab' key work for me. Instead, I type:

ctrl + v + i

This is the only way I can enter a tab in my terminal... not sure why, but I've become so used to doing it this way, I don't even notice it anymore.

That behaviour matches mine. I use the Korn shell (ksh) and due to far too many years to admit using vi, the first thing I set up is a set -o vi in my .profile.

Your syntax is fine, the issue is that the version of sed installed with Solaris is a part of the core pkg.. It hasn't changed in a long time and prob will not be updated.

As I'm a Solaris admin and regular user of sed, I almost always dump a freshly compiled copy of it in /usr/local.