Shell Programming and Scripting

Petro_Project/Regional Work/Charge Access/Studies_PG2014/P_Tables A_394857_figure.pdf

First I apologize for my novice question. I'm still learning sed & awk.

I have a file with the following data format:

Code:

Petro_Project/Regional Work/Charge Access/Studies_PG2014/P_Tables/A_394857_figure.pdf

This file contains over 8000 data lines in which the path may be more or less then what is shown. I've been requested to do a simple format (or so I thought) on the file to look like the following:

Code:

Petro_Project/Regional Work/Charge Access/Studies_PG2014/P_Tables	A_394857_figure.pdf

There is a tab that replaces the last slash character which will seperate the last field.

To start, The first awk line I've used to do this is:
awk -F"/" '{$NF=""}1'

I know this kills the last $NF, but again, I'm just learning this (getting what I can from the Sed & Awk, O'Reilly book), and am trying to separate the filename from the path of a single line.

I've searched through numerous threads, and even though I know there is a solution available, I couldn't find one.

Your help and explanations would be greatly appreciated!

Hi, this can be done with sed:

Code:

sed 's|\(.*\)/|&      |' file

The TAB is a hard-tab, entered with CTRL-V TAB

--
Some sed's can use a \t code, but this is a non-standard extension, for example GNU sed:

Code:

sed 's|\(.*\)/|&\t|' file

--
With awk, you could do this:

Code:

awk -F/ '{sub("/[^/]*$", "\t" $NF)}1' file

How about

Code:

awk -F"/" '{sub (FS $NF "$", "\t" $NF)}1' OFS="/"  file
Petro_Project/Regional Work/Charge Access/Studies_PG2014/P_Tables    A_394857_figure.pdf

WOW, that did it. I never thought of using sed.

Many Many Thanks for your help! Much appreciated!!

Small correction to Scrutiizer's proposal (which does NOT replace the last slash):

Code:

sed 's|\(.*\)/|\1	|' file
Petro_Project/Regional Work/Charge Access/Studies_PG2014/P_Tables    A_394857_figure.pdf

I checked yours too RudiC. It worked and I'm checking on some of your usage in the book...

Thank You for your help!

Yes it should be \1 rather than & of course :roll eyes:

--
Just one note about the use of awk. RudiC and my approach will give the same result in most cases.
However, if $NF is used in the matching part (the first field) of the sub() function and it contains special regular expression characters then that may give undesired results.
The first field is interpreted as ERE, an extended regular expression (unless it consists of a single character in double quotes)

There is no problem with using $NF in the replacement part of the sub() function (the 2nd field), since that is a string.