Renaming file extensions using find-xargs-sed

Rohit_4739 · 01-29-2013, 05:01 AM

Hi All,

I am trying some experiments with find xargs sed and just wondering if this can be even achieved. I want to find all the ".txt" files and then change their name to ".bak"(lets say) and i want to do this using sed with xargs and not conventional "rename" command. Can anyone please help me on this. Below are the few variants i have tried but without any success.

Code:

find -type f -name "a.txt" | xargs bash -c "sed 's/dat/txt/' "
find -type f -name "a.txt" | xargs -I {} mv {} `echo {} | sed 's/txt/bak/'`

Any idea ?

divyashree · 01-29-2013, 06:20 AM

This simple oneliner will do your work.

Code:

for i in `find /tmp/tes -type f -name "*.txt"`; do mv  $i `echo $i | perl -pe 's/(.*)\..*/$1.bak/'` ; done

Replace /tmp/tes with your directory name.

tronayne · 01-29-2013, 06:33 AM

Maybe take a simpler way (there are many ways to skin this cat):

Code:

#!/bin/bash
#
#	Rename files in current tree
#
for file in $(find . -type f -name '*.txt')
do
	new_file=$(sed 's/\.txt/\.bak/' ${file})
	mv ${file} ${new_file}
done

The basic thing is that you need both the old name and the new name to mv.

Hope this helps some.

TobiSGD · 01-29-2013, 06:43 AM

If you use Bash anyways there is no need for Perl or sed, Bash can handle the substitution on its own (http://www.tldp.org/LDP/abs/html/par...stitution.html):

Code:

#!/bin/bash
#
#	Rename files in current tree
#
for file in $(find . -type f -name '*.txt')
do
	mv ${file} ${file%.txt}.bak
done

allend · 01-29-2013, 09:36 AM

If you use bash there is no need for 'find'.

Code:

#!/bin/bash
#
#	Rename files in current tree
#
shopt -s globstar

for file in **/*.txt; do
  mv "$file" "${file%.txt}.bak";
done

Rohit_4739 · 01-29-2013, 08:51 PM

Hello Everyone,

Well first of all thanks for all of your replies. However i am aware of all the methods that are mentioned above, my original question is IF I CAN ACHIEVE THE SAME USING a combination of FIND, XARGS and SED.

Can anyone answer that ?

allend · 01-30-2013, 05:53 AM

Something horrible without xargs and requiring a shell supporting process substitution such as bash.

Code:

while read aline; do mv "$aline" "$(echo $aline | sed 's/txt/bak/')"; done < <(find . -name "*.txt" -print)

The problem with xargs is it only supports one command, yet you want to do a 'mv' command including an argument that requires a second command. You could put the processing in a script that becomes the xargs command.

Apart from the intellectual exercise, is there a reason for this?

colucix · 01-30-2013, 06:21 AM

Code:

find . -name \*.txt -print0 | xargs -0 bash -c 'while [[ $# -gt 0 ]]; do mv "$1" "${1/.txt/.bak}" ; shift; done' dummy

and if you really want to use sed:

Code:

find . -name \*.txt -print0 | xargs -0 bash -c 'while [[ $# -gt 0 ]]; do mv "$1" "$(echo $1|sed s/.txt/.bak/)" ; shift; done' dummy

allend · 01-30-2013, 06:30 AM

Thanks colucix! Very instructive.

colucix · 01-30-2013, 07:04 AM

Quote:

Originally Posted by allend

Thanks colucix! Very instructive.

You're welcome! I prefer simple solutions like yours in post #5, anyway. Mine is like to drive a Ferrari to the local store near home!

tronayne · 01-30-2013, 08:39 AM

Something to keep in mind as you go along is the basic purpose of xargs (which is so you don't overflow the input buffer of a following command) when your working with a large batch of input (such as what you typically get with find).

The sed editor, the streaming editor, works on entire lines either in a file or an input stream (such as files names as in your example).

With sed, you can "in-line" directives or you can create a file that contains any number of directives (commands to be executed) by sed in multiple lines to work line by line on the content of a file.

For example, this is a "sed" file that does clean-up of an all upper case and abbreviated file:

Code:

cat 401.sed
s/LOAN BALANCE/Loan Balance/g
s/PIMCO REAL RETURN FUND INST/PIMCO Real Return Fund Institutional/g
s/PIMCO TTL RETURN INSTITUTIONAL/PIMCO Total Return Instutional Fund/g
s/VANGUARD DEV MRKTS INDEX FUND/Vanguard Developed Markets Index Fund/g
s/VANGUARD EXTENDED MKT INX ADMR/Vanguard Extended Market Index Fund Admiral Shares/g
s/VANGUARD INSTITUTIONAL INDEX/Vanguard Institutional Index Fund Institutional Shares/g
s/EMPLOYEE PRE-TAX/CONTRIBUTIONS - EMPLOYEE PRE-TAX/g
s/EMPLOYER MATCH/CONTRIBUTIONS - EMPLOYER MATCH/g
s/ROLLOVER/CONTRIBUTIONS - ROLLOVER/g
s/CONVERSION LOAN FUND/Conversion Loan Fund/g
s/VANGUARD DVLPD MKTS INDEX INST/Vanguard Institutional Developed Markets Index/g
s/VANGUARD EXTENDED MKT IND I CL/Vanguard Extended Market Index Fund Institutional Shares/g

It's used by

Code:

sed -f 401.sed file_name > new_file_name

Another might be

Code:

cat ./src/maps/gnis.sed
/\<SHORT_FORM\>/d;/\<ADM[0-2]\>/p;/\<PPL\>/p;/\<PPLA\>/p;/\<PPLC\>/p;/\<PPLL\>/p

for extracting certain lines from a (really large) file containing geographic names, latitudes and longitudes for placement on a map.

All of the methods described in the above posts are, of course, valid (and some quite inventive, too). But, like many things, simpler can be better -- not always, but frequently. I've pretty much always found it useful to think in terms of making a tool that will be used more than once so I try to keep things as simple as possible with a nod toward efficiency while I'm at it. I'm not being critical here, I'm just sayin'.

Complicated stuff will come back an bite you more often than not and I've tried to avoid those -- plus, half the time I can't remember why in heck I did what I did so I try not to do stuff like that.

For your purpose, renaming files, xargs and sed probably are, well, overkill. The best of the bunch up above is the shell substring extraction (${file%.txt}.bak) -- simple, elegant and efficient.

You know, it's sort of like when all you need is a Phillips screwdriver it doesn't make a heck of lot of sense to drag out the air gun?

Hope this helps some.

David the H. · 01-31-2013, 03:23 PM

Re posts #2, #3, & #4:

Never Read Lines With For!! Particularly when working with filenames like this. Always use a while+read loop, along with null separators whenever possible, such as with find's -print0 option.

Notice also Colucix's use of null separators with xargs too, for the same reason.

Actually though, you rarely need to use xargs in combination with find, since it has it's own built-in -exec option that can generally do the job just as well.

Here are a couple of good links about using find:
http://mywiki.wooledge.org/UsingFind
http://www.grymoire.com/Unix/Find.html