bash replace part of string based on location

Question

I've looked around for a one line solution ( as bash offers ) to replace part of filename.

Given that a folder has image sequence like

ve2_sq021_sc001_v000.0101.jpg
ve2_sq021_sc001_v000.0102.jpg
ve2_sq021_sc001_v000.0103.jpg
ve2_sq021_sc001_v000.0104.jpg

Need to replace only v000 with v09 ( say ). How is it possible (throughout directory).

while if i know that it is v000 in fourth position then i can do as suggested. My question is - How is it possible to go to fourth location after _ and replace that with "v09". Can regular expression come useful here. — nish, Feb 11 '14 at 10:21

slm · Accepted Answer · 2014-02-11T13:19:45.010

3

If you have the unix command rename installed you can do this trivially like this:

$ rename v000 v09 *.jpg

$ ls -1
ve2_sq021_sc001_v09.0101.jpg
ve2_sq021_sc001_v09.0102.jpg
ve2_sq021_sc001_v09.0103.jpg
ve2_sq021_sc001_v09.0104.jpg

NOTE: This is using the rename implementation that's included with the package util-linux.

edited Feb 11 '14 at 13:19

answered Feb 11 '14 at 08:39

slm

This is a useful command. You could merge it into @terdon's answer above. – Faheem Mitha Feb 11 '14 at 08:55
ok. This may be silly to ask. how will replace in a path in script. pth="/tmp/images" rename v000 v09 $pth/*.jpg this does not work. Error is "No such file or directory" – nish Feb 11 '14 at 12:06
@nish - there are 2 different implementations of rename. Which distro is this? – slm Feb 11 '14 at 12:55
CentOS release 6.5 (Final) – nish Feb 13 '14 at 07:14
@nish - OK so the example I showed you will work then, that's the CentOS/Fedora version as well. – slm Feb 13 '14 at 07:16

score 1 · Answer 2 · answered Feb 11 '14 at 08:44

1

for f in $(ls ve2*); do mv $f $(echo $f | sed s/v000/v09/g ); done

If you want to make it recursive, you could use find instead of ls ve2*

answered Feb 11 '14 at 08:44

rralcala

1

Do not ever parse the output of ls. You'll be sorry for it the instant you encounter a filename or path containing whitespace. Instead, use the globbing of you shell (for f in ve2*; do) or find's -exec parameter. – n.st Feb 11 '14 at 13:51
1

Also, there's no need to spawn a sed instance for each file to be renamed. Either use rename or at least let bash handle the strong replacement: ${f//v000/v09} evaluates to the value of f with all occurrences of 'v000' replaced by 'v09'. – n.st Feb 11 '14 at 14:00

2 Answers2