Renaming a set of files in specific order without losing them (on Linux)

Question

I have a set of files (the list is larger, 4x43 files with extensions I.sto, Q.sto, U.sto and V.sto):

probni45069Q.sto probni45080I.sto probni45080V.sto probni45083U.sto 
probni45069U.sto probni45080Q.sto probni45083I.sto probni45083V.sto
probni45069I.sto probni45069V.sto probni45080U.sto probni45083Q.sto

My goal is to rename them in the sorted order starting with number 1:

1. probni45069I.sto probni1I.sto
2. probni45080I.sto probni2I.sto
3. probni45083I.sto probni3I.sto
4. probni45069Q.sto probni1Q.sto
5. probni45080Q.sto probni2Q.sto
6. probni45083Q.sto probni3Q.sto
7. probni45069U.sto probni1U.sto
8. probni45080U.sto probni2U.sto
9. probni45083U.sto probni3U.sto
10. probni45069V.sto probni1V.sto
11. probni45080V.sto probni2V.sto
12. probni45083V.sto probni3V.sto

I have followed the instructions and guides from Renaming files in a folder to sequential numbers and created the following bash script:

model='probni'

for sto in I Q U V;
do
    i=1
    for j in $model*$sto.sto;
    do
        echo "$j" `printf $model%1d$sto.sto $i`
        mv "$j" `printf $model%1d$sto.sto $i` 2>/dev/null || true
        i=$((i + 1))
    done
done

This script works great when used only once. The problem is when I use it multiple times on already sorted set of files or when I have a set of sorted files along with additional unsorted files (e.g. probni1I.sto + probni1346I.sto), I lose a certain number of files. When used repeatedly on my original 4x43 set of files, I finally end with the set of 4x7 files.

My questions is how to make this script to be idempotent for sorted files or how just to add new unsorted files to the sorted list without losing any files.

Answer 1

The problem is that file names are sorted lexicographically, which causes problems with numbers because 1 2 1346 will be sorted as 1 1346 2 since the character 1 comes before 2. You need a smarter sort that handles numbers more intuitively.

sort -g is just what the doctor ordered. Filter your file list through sort -g and 1346 won't cause you any more troubles.

for j in $(ls $model*$sto.sto | sort -g); do
    ...
done

Normally parsing the output of ls is a bad idea. If you want to do things really right you'll need to restructure things a bit.

for sto in I Q U V; do
    find -maxdepth 1 -name "$model*$sto.sto" -print0 |
        sort --files0-from - -z |
    {
        i=1
        while IFS= read -rd $'\0' old; do
            new=$model$i$sto.sto
            [[ $old == $new ]] && continue
            echo "$old" "$new"
            mv -- "$old" "$new" 2>/dev/null || true
            ((i++))
        done
    }
done

It's a heck of a lot of work, but this version handles weird file names much better. It uses NUL separators at all stages so file names with spaces and newlines won't gum up the works.

I don't expect that you have weird files like that, but hey, better safe than sorry. It's always good to try to do the right thing.

Answer 2

What about incrementing i in a loop as long as the destination file already exists?

while [ ! -f FILENAME ]
do
    INCREMENT
done

Edit: I missed that your input and output files look the same. This would require you to either 1) change the naming scheme of either, 2) put them in different directories, or 3) start renaming no sooner than you encounter a gap in counting up from 1

Answer 3

Normally parsing the output of ls is considered bad. However, having that the folder only contains files following the shown pattern, you can use the following shell script:

ls | sort -k1.12 -k2.6n | awk '{
    n=substr($0,7,5);
    l=substr($0,12,1);
    c[l]++;
    system("mv "$0" probni"c[l]""l".sto")
}'

sort -k1.12 -k2.6n will sort the files by the letter at the end and the number in the middle. This works only the file names have all the same length and the number is always 5 characters long and the prefix always probni (6 characters long).

The awk script at the end of the pipe disassembles the file name using substr() and reassembles it following the desired pattern. system() is used to issue the mv command.