How do you transpose parts of filename in Linux?

Question

Take a variable length filename 'abcdef - ghijkl.pdf' and rename it to 'ghijkl - abcdef.pdf'

How would I achieve that in a command line script for all files in a single folder?

Answer 1

Since you are using Ubuntu, you should have at least one of the perl-based rename or prename

$ rename -n -- 's/(\w*) - (\w*)/$2 - $1/' *.pdf
rename(abcdef - ghijkl.pdf, ghijkl - abcdef.pdf)

$ prename -n -- 's/(\w*) - (\w*)/$2 - $1/' *.pdf
abcdef - ghijkl.pdf renamed as ghijkl - abcdef.pdf

Alternatively, you could install and use mmv

$ mmv -n '* - *.pdf' '#2 - #1.pdf'
abcdef - ghijkl.pdf -> ghijkl - abcdef.pdf

(In all cases, remove the -n to actually complete the operation).

Answer 2

Using bash:

[[ $f =~ ^([^-]+)\ -\ ([^\.]+)\.(.*) ]] && 
  mv "$f" "${BASH_REMATCH[2]} - ${BASH_REMATCH[1]}.${BASH_REMATCH[3]}"

This uses bash's regex test operator =~ to see if the given filename (in $f) matches the pattern: "anything except dash, followed by space dash space, followed by anything except a period, followed by a period, followed by anything-or-nothing". If the filename matches that pattern, the matching bits of the parenthesized subsections of the regex get assigned to elements of the BASH_REMATCH array.

To do that for every file in a particular folder:

cd /to/that/folder
for f in *
do
  [ -f "$f" ] || continue
  [[ $f =~ ^([^-]+)\ -\ ([^\.]+)\.(.*) ]] && 
    mv "$f" "${BASH_REMATCH[2]} - ${BASH_REMATCH[1]}.${BASH_REMATCH[3]}"
done

Using only POSIX-specified variable expansions:

pre=${f%%.*}
ext=${f##*.}
mv "$f" "${pre##*- } - ${pre%% -*}.$ext"

Answer 3

You can create the script with that command:

\ls *-*.pdf | sed 's/\(.*\)-\(.*\).pdf/mv \1-\2.pdf \2-\1.pdf/' > script_mv.bash

Then you check that the script is correct, then you can execute it:

. ./script_mv.bash

I have written \ls, because it cancels options on ls.

The sed pattern means:

• find anything1, a "-", anything2, ".pdf" .
• Both anythings are enclosed in ( ) so that they get names \1 and \2 .
• then I copy the mv line, for that I can recreate filenames thanks to \1 and \2 .

Answer 4

bashy answer, specific to the "xxx - yyyy.ddd" name pattern..

find -type f | while read flnm ; do basename "$flnm" | awk -F'[\ .]' '{print $3 " " $2 " " $1 "."$4}'; done

Find a file and hand the name off to a while read loop. In there, use basename to extract the name and hand it off to Awk. Awk, using multiple field separators, space and period, then re-prints the fields, in the desired order.

abcdef - ghijkl.pdf

becomes

ghijkl - abcdef.pdf

practical application... rename files to this new pattern name

find -type f | while read flnm ; do mv "$(basename "$flnm")"  "$(basename "$flnm" | awk -F'[\ .]' '{print $3 " " $2 " " $1 "."$4}')"; done

The only difference is the inclusion of the mv command and the wrapping for the names as strings.