ansaurus

Question

Answer 1

A:

su[0-9] will match a single digit.

Ignacio Vazquez-Abrams 2010-01-18 22:57:40

Answer 2

+1 A:

$ sed -e 's|/su[^/]*|/newfolder|' /tmp/files\
dcdd62defb908e37ad037820f7  /sp/dir/newfolder/89/asga.gz
7d59319afca23b02f572a4034b  /sp/dir/newfolder/89/sfdh.gz
...

If you want to get rid of the checksums as well:

$ sed -r -e 's|/su[^/]*|/newfolder|' -e 's/^[^ ]+ +//' /tmp/files\
/sp/dir/newfolder/89/asga.gz
/sp/dir/newfolder/89/sfdh.gz
...

Wayne Conrad 2010-01-18 22:58:41

As I said in my other comment, you don't need `/` to be the delimiter for `s`, so you can write: `sed -e 's#/su[^/]*#/newfolder#' /tmp/files` for more readability.

Alok 2010-01-18 23:05:27

what if one of the directory looks like /sp/su4dir/su1/89/24.gz ?

ghostdog74 2010-01-18 23:08:52

@ghostdog74: the problem is not very well-defined, so we can only guess about what pattern the OP really needs.

Alok 2010-01-18 23:10:25

@Alok, Thanks .

Wayne Conrad 2010-01-18 23:13:12

Answer 3

+2 A:

I think you want

sed 's/su./newfolder/g'

If you actually want to keep the number in su1...su7 as a part of newfolder (for example newfolder1...newfolder7), you can do:

sed 's/su\(.\)/newfolder\1/g'

It also depends upon how "strict" do you want your patterns to be. The above will match su followed by any character and do the replacement. On the other hand, a command like s#/su\([0-9]\)/#/newfolder\1/#g will only match /su followed by a digit, followed by /. So you may need to adjust your pattern accordingly.

Alok 2010-01-18 22:59:25

Answer 4

A:

sed requires a dirty amount of metacharacter escaping, some of it may be slightly off.

sed -i -e 's/\/su[^\/]+\//\/newFolder\//g' myfile

Paul Creasey 2010-01-18 22:59:47

You don't need `/` to be the character after `s`, so you can pick any character not in your pattern: so, `'sxsu[^/]+/x/newFolder/xg'` does the same (if I didn't make a typo).

Alok 2010-01-18 23:03:45

@Alok: Thanks for the tip, i always knew that was possible in perl, didn't reaise you should do it in SED, very useful, since escapes make sed regex near unreadable!

Paul Creasey 2010-01-18 23:07:44

it works when i changed + to .*. also, it doesn't take care of paths like : /sp/su4dir/su1/89/24.gz

ghostdog74 2010-01-18 23:19:02

Answer 5

A:

I vote for Wayne Conrad's answer as the most likely to be what the OP wants, but I'd suggest using an alternate character for the sed expression separator, thus:

sed 's|/su[^/]*|/newfolder|'  /tmp/files

That makes it a bit cleaner.

Note also that the trailing 'g' is probably not wanted.

BobS 2010-01-18 23:05:59

Answer 6

A:

use awk. since there is a delimiter you can use , '/'. after that, column 4 is what you want to change. So if you have paths like /sp/su3dir/su2/89/sfdh.gz , su3dir will not be affected.

awk -F"/" '{$4="newfolder";}1' OFS="/" file

ghostdog74 2010-01-18 23:06:24

ansaurus

tags:

views:

answers:

select part of filename using regex

related questions