I have a number of directories that I know I want to run a bash script agenst.

I can find the file that I want with find / -iname "file"

/namedir1/unknown-name1/../../../../../file
/namedir2/unknown-name2/../../../../../file
/namedir3/unknown-name3/../../../../../file
/namedir4/unknown-name4/../../../../../file

I want to work out how to take

/namedir4/unknown-name*/ and remove the rest of the folder information while keeping the first part of the directory information.

I used to have a solution for such a issue but my mind has gone blank on what would be the best way to do the following.

Regards
Hyperdaz

will return /namedir1/unknown-name1/../../../../../

then you can use a script or awk or perl or xargs or find -exec or whatever method works for you to use the returned path

conversely basename will return the filename from the string, optionally dropping suffix

Thanks for you reply _BSD,

Yes it was really the sed or awk section I am looking for..

ideally it should give the full unknown-name1, while possibily dealing with spaces ((I imagine it needs to search for the second accurance of "/" and remove the namedir1 and any other "/"

Cheers
Daz

There are lots of various web pages with sed one liners.
Search for a regexp that matches the pattern you describe, it simple for a greedy pattern matcher
try 's#.*/\(.*\)#\1#'

Hi.

or

Thanks Firtfire

I guess I need another sed ((combo)) command or two to remove

/namedir1/
and all
"/"

So only the "unknown-name1" part is left...

Regards
Hyperdaz

No.
I thought you need first two parts..
Try this:
I will try to explain how it works. Your input string consists of a series of '/'-characters and "not '/'-characters" (that is almost any character except slash) between them. You need to extract second block of "not '/'-characters", right? Such a block (directory name) can be matched by regular expression '[^/]+', which reads "one or more consecutive not slashes". '\1' is a reference to text matched by expression in first parentheses. So we should skip first block and print only the second. The `s' (substitute) command uses the character after `s' to delimit its arguments: "s/.../.../" is the same as "s|...|...|"; this way we do not have to escape slashes in the regular expression.

See `info sed', `info grep' for more details.

Hope this helps.

Did you not try the pattern that I gave you?
It works with sed

@_bsd - OP is not looking for filename but rather 'unknown-name1' from your example.

How about we keep it simple:

1 members found this post helpful.

_BSD

Yes I did try the example that you gave me, but I am not after the "filename" from the string but after "unknown-name1"

Sorry I thought that was clear from my post(s)

Many thanks grail, my thought was to use tr and then awk but -F is a little more faster

Thanks for all help

Hyperdaz

no, I'm sorry, I misread the initial post, didn't see what you were after, my bad.