I have a lot of files that have a shared pattern in their name that I would like to remove. For example I have the files, "a_file000.tga" and "another_file000.tga". I would like to do an operation on those files that would remove the pattern "000" from their names resulting in the new names, "a_file.tga" and "another_file.tga".
Add a comment
|
5 Answers
Bash can do sed-like substitutions:
for file in *; do mv "${file}" "${file/000/}"; done
answered Sep 30, 2009 at 19:36
Dennis Williamson
364k95 gold badges386 silver badges446 bronze badges
8 Comments
efrem
Nice, can you explain the function of "{" ?
Pavel Zubkou
For documentation on this functionality please see section 3.5.3 Shell Parameter Expansion from Bash Reference Manual, look for
${parameter/pattern/string}. More detailed examples of parameter substitution could be found in section 10.2 Parameter Substitution from Advanced Bash-Scripting Guide.
Dennis Williamson
@irnc: Another: Parameter Expansion
John Manko
Best answer, IMO.
Dennis Williamson
@hrdom: You will need to use special quoting in order to have the newline escape interpreted:
for file in *; do mv "${file}" "${file/$'\n/'}"; done . And if your filenames might have more than one newline and you want to remove all of them then double the first slash: for file in *; do mv "${file}" "${file//$'\n/'}"; done |
Try this (this works in plain old Bourne sh as well):
for i in *000.tga
do
mv "$i" "`echo $i | sed 's/000//'`"
done
Both arguments are wrapped in quotes to support spaces in the filenames.
3 Comments
Chris Lutz
+1 for most portable solution (and because I'm running too low on upvotes to keep upvoting all the good answers).
Charles Duffy
Portable, but slow and buggy (
echo $i is itself buggy for the reasons described in I just assigned a variable, but echo $variable shows something else, and running a separate copy of sed for each name is very slow)Charles Duffy
...also, backticks are bad form --
$( ) has been standardized as part of the POSIX sh standard (which supersedes 1970s Bourne) since the early 1990s; there's no good reason to keep using the old form that's hard to nest and changes the meaning of backslashes within them.A non-bash solution, since I know two speedy posters have already covered that:
There's an excellent short perl program called rename which is installed by default on some systems (others have a less useful rename program). It lets you use perl regex for your renaming, e.g:
rename 's/000//' *000*.tga
Comments
Use rename, maybe you need to install it on linux, Its python script
rename (option) 's/oldname/newname' ...
so you can use it like
rename -v 's/000//' *.tga
that means we are instructing to replace all files with .tga extension in that folder to replace 000 with empty space. Hope that works
Comments
Piping the output of ls is widely considered to be very dangerous because it's not predictable or portable, but if you are willing to take the risk:
#!/bin/bash
ls | while read name; do
echo mv $name ${name/$1//}
done
4 Comments
Javier
@Laurence Gonsalves: using shell wildcards expand the command line, and there's a not-so big limit there. "while read" avoids that
Cascabel
@Laurence: This is also much more extensible. It's commonly used with
find on the front.Charles Duffy
while read is an improvement (if it were used correctly, with an empty IFS and -r, and -d '' and a NUL-delimited input stream), but ls is the opposite. See Why you shouldn't parse the output of ls
whiskeychief
To be clear, you shouldn't parse the output of
ls.