Rename files by reversing number order

Here's a bash snippet to do the job assuming files are indeed named the way you've depicted them (data<one-digit>_<digits>.txt).

shopt -s extglob

#gather files into array
files=( data[[:digit:]]_+([[:digit:]]).txt )

#zip original files with their target file names and feed to mv
paste <(printf '%s\n' "${files[@]}" | sort -k1.5,1n -k2n -t'_') \
    <(printf '%s.ren\n' "${files[@]}" | sort -k1.5,1n -k2nr -t'_') | 
    xargs -n 2 mv --

#strip the temporary .ren suffix
for f in data*.ren; do mv -- "$f" "${f%.ren}"; done

First, rename all files to prefix "old-":

for i in *
do
    mv "$i" "old-$i"
done

Then run this command and eyeball the output to make sure it looks good:

ls -v | tac | sort -s -t _ -k1,1 | sed -e 's/^old-//' | paste <(ls -v) - | sed -e 's/^/mv /'

If it does, pipe the output to sh.

Here's what's happening.

• ls -v produces them in sorted order (the -v says to sort 11 after 9, for example)
• tac reverses the whole input (the entire file; bear with me!)
• the sort says do a stable sort on only the characters before the first _. The -k1,1 and -s are both important to make sure you get the right output. Without the -k1,1 the rest of the line is used to resolve duplicates, which we don't want, and without -s duplicates are ordered arbitrarily.

The rest is easy enough.

for a start you may try like this:

mkdir x
for base in $(ls data* | sed 's/_[0-9][0-9]*.txt//' | sort -u ) ;do
  ls ${base}_*| tac | for i in data1*; do
    read x;cp $i x/$x
  done
done

It works well only for special cases, you may have unexpected results with unpadded numbers like 1 2 11 111 or filenames like data1_somethingelse1.txt, not to mentions space or special characters in filenames.

Here is a simple solution and I believe it would work for your situation very well. What we are doing here is, first checking how many file types we got by prefix. Here by file type prefix I meant data1_, data2_ and so on. Then for each prefix type, we are then getting the total number of available files and storing them into an array called totalFilesForEachPrefix.

Then, in step 1 we are renaming the file to a temporary file. The reason for moving to otemp extension is to avoid name collision and overwrite the existing file. Here, I assumed that you have files like 1_1 1_2 1_3 1_4 and so forth.

Then, in step 2, we are just getting rid of .otemp extension.

#!/usr/bin/env bash
totalFileTypeByPrefix=$( for file in data*_1.txt; do echo $file; done | wc -l )
totalFilesForEachPrefix=()
for (( prefix = 1; prefix <= totalFileTypeByPrefix; prefix++ )); do
  totalFilesForEachPrefix+=( $( for file in data${prefix}_*.txt; do echo $file; done | wc -l) )
done
### Step 1
prefix=1; type=0;
while (( prefix <= totalFileTypeByPrefix )); do
  suffix=${totalFilesForEachPrefix[$type]}
  for file in data${prefix}_*.txt; do
    mv $file data${prefix}_${suffix}.txt.otemp; echo "$file renamed temporary --> data${prefix}_${suffix}.txt.otemp"
    suffix=$((suffix -1))
  done
  type=$((type+1))
  prefix=$((prefix+1))
done
### Step 2
echo "....Finally changing the temporary files"....
for tempfile in *.otemp; do
  file=${tempfile::-6};
  mv $tempfile $file; echo "$tempfile renamed final --> $file";
done

Here is the output with an explanation of whats happening:

data1_1.txt renamed temporary --> data1_3.txt.otemp
data1_2.txt renamed temporary --> data1_2.txt.otemp
data1_3.txt renamed temporary --> data1_1.txt.otemp
data2_1.txt renamed temporary --> data2_2.txt.otemp
data2_2.txt renamed temporary --> data2_1.txt.otemp
....Finally changing the temporary files....
data1_1.txt.otemp renamed final --> data1_1.txt
data1_2.txt.otemp renamed final --> data1_2.txt
data1_3.txt.otemp renamed final --> data1_3.txt
data2_1.txt.otemp renamed final --> data2_1.txt
data2_2.txt.otemp renamed final --> data2_2.txt

A stupid, but always functional method (following The KISS principle) is this:

/bin/ls data*.txt >rename.sh
# edit the file so that instead of
# data1_1.txt
# each line reads
# mv data1_1.txt data1_3.txt
# (you can use column selection in your editor)
sh rename.sh

Of course, it is obvious, but usually doing this is a way faster than fiddling with complicated (and one-of anyway) scripts.

(edited later) Comments are right, that in this case using this technique can lead to the loss of data (data1_3.txt can be wiped out). Therefore it is safe to make move in two steps: first change names, but move the file to some safe location (either add arbitrary false extension, or create a temporary subdirectory, where all files are moved), and then in the second step move all files to their final destination.