Shell Programming and Scripting

Greetings!

I'm trying to take a directory list (such as nn.list in dpkg/info) and sort out the line content which furnishes only dirs from that which provides paths to actual files with a bash "one liner". From there, I want to pass the culled result to stdout for more pipework down the line...

So, here's the pseudo (broken) code which I've been playing around with to no avail. From a terminal in dpkg/info:

Code:

read < xx.list | find -type f  '*.*' | sort

(I know, it's incomplete and rather "Monty Python" -- but it was the best I could figure )

At any rate, is there a better builtin or a straightforward way of doing something like this in Bash?

Thanks a bunch!

Show the input you have and the output you want please. Broken code which doesn't do what you want sadly isn't very instructive!

Try:

Code:

sort xx.list | while read -r file
do      if [ -f "$file" ]
        then    printf "%s\n" "$file"
        fi
done

Oops. That -f should have been a -r. It is fixed now.

Thanks for dropping by, folks!

@Corona688:

Sorry 'bout that. Here's a quick rundown of the functionality I was looking at.

To start, we have a hypothetical file, xx.list, which consists of nothing but these three lines:

Code:

/foo
/foo/bar.txt
/foo/bar/yum.txt

Now, from there, I was attempting to process this file, xx.list, through 'read' and 'find' to simply pass

Code:

/foo/bar.txt
/foo/bar/yum.txt

along to 'sort' and off to the next pipe for more processing (via xargs, etc).

For simplicity's sake, just wanted a quick "one liner" of code to do the work


@Don Cragun:

Thanks for the snippet!

As an adaptation, would it be possible to pipe

Code:

sort xx.list | xargs

to a quick inline test of some sort to cull out the true file paths for even more downstream processing? Still getting a handle on the "quirks and handshakes" of Bash, and quite amazed at the hidden functionality contained therein...

Thanks again, all

I'm lost. I thought you said you wanted a sorted list of regular files to be created from the unsorted list of all sorts of files found in a file named xx.list. The script I gave you does that and provides the output with one filename per line. The output of that script can be piped into anything you want.

If you're saying it has to be a one-liner, change it to:

Code:

sort xx.list|while read -r file;do if [ -f "$file" ];then printf "%s\n" "$file";fi;done

but I'll take readable over one-liner every time.

Code:

sort xx.list | xargs

produces a sorted list of all of the filenames (without checking for the file type) and uses a combination of spaces and newlines (depending on the number of files in the list and the lenghts of the filenames) as filename terminators. Why do you believe that xargs will weed out the names of files that are not regular files? In cases where one or more filenames spaces, feeding a list of those filenames through xargs will create a list that is much harder to process in anything it in a pipeline.

@Don Cragun:

I do apologize if my additional question presented a difficulty. The call to xargs was not for filtering, but formatting for the next stage, if necessary.

The command will be entered at the terminal directly; hence the need for a brief, one-line segment. Needed something easy enough to remember and keypunch without creating too many typos.

Thank you again for the post and your assistance...

Ignore this -- I should have Don's post more carefully....

You should consider testing each entry in the to see if it is a regular file, in case some directories exist below /foo - ex: /foo/path/anotherdir

Code:

[ -d $variable ]

or

Code:

[ ! -f $variable ]

if there no other file types at all except regular files and directories.

does that for you. I know that "right now" there may be none, but you seem to imply you want to keep this code around.

Code:

while read name
do
  [  -d "$name" ] && echo "$name"
done < inputfile |  ... rest of pipe stream