Bash: How to replace 1 with 01 in filename without using rename function

Question

I have a a set of file wat follow the form L2_error_with_G_at_#_#.csv where the # symbol could be any number between 1 and 16. I want to rename the files using bash and sed so that any single digit number in the filename is buffered with a single 0, e.g. the filenames L2_error_with_G_at_2_9.csv and L2_error_with_G_at_1_16.csv would be replaced with filenames L2_error_with_G_at_02_09.csv and L2_error_with_G_at_01_16.csv. I'm seen solutions to this sort of problem that use the rename function from pearl. I don't have access to that function as I'm working on a system where I don't have installation privileges. Ideally the solution will use more of the basic features of bash, which is why I suggested a solution using sed.

EDIT: I do not need to use sed, I can use anything so long as it's a standard part of bash

Here's what I've tried

for file in L2_error_with_G_at_*.csv
do
    new=$(echo "$file" | sed 's/_\([0-9]\)/_0/g; s/_0\([0-9][0-9]\)/_\1/g;');
    mv "$file" "$new";
done;

Answer 1

Assumptions:

• all filenames have the same format: a_b_c_d_e_n1_n2.ext where ...
• a..e and ext do not contain white space
• n1/n2 are numbers

One bash idea:

for fname in L2_error_with_G_at_2_9.csv L2_error_with_G_at_1_16.csv
do
    printf -v new "%s_%s_%s_%s_%s_%02d_%02d.%s" ${fname//[._]/ }
    echo mv "${fname}" "${new}"
done

This generates:

mv L2_error_with_G_at_2_9.csv L2_error_with_G_at_02_09.csv
mv L2_error_with_G_at_1_16.csv L2_error_with_G_at_01_16.csv

Once the output is verified the echo can be removed from the code so that on the next run the mv occurs.

---

If the list of filenames is in a file:

while read -r fname
do
    printf -v new "%s_%s_%s_%s_%s_%02d_%02d.%s" ${fname//[._]/ }
    echo mv "${fname}" "${new}"
done < file.list

If the list of filenames is coming from a command (eg, find or ls):

while read -r fname
do
    printf -v new "%s_%s_%s_%s_%s_%02d_%02d.%s" ${fname//[._]/ }
    echo mv "${fname}" "${new}"
done < <(command that generates list of filenames)

Answer 2

$ cat tst.sh
#!/usr/bin/env bash

olds=( L2_error_with_G_at_2_9.csv and L2_error_with_G_at_1_16.csv )

for old in "${olds[@]}"; do
    if [[ "$old" =~ ^(.*_)([0-9]+)(_)([0-9]+)([^_]*)$ ]]; then
        printf -v new '%s%02d%s%02d%s' "${BASH_REMATCH[@]:1:5}"
        echo mv "$old" "$new"
    fi
done

$ ./tst.sh
mv L2_error_with_G_at_2_9.csv L2_error_with_G_at_02_09.csv
mv L2_error_with_G_at_1_16.csv L2_error_with_G_at_01_16.csv

populate olds as you see fit or make the loop for old in L2_error_with_G_at_*.csv or whatever you need to get a list of old file names to iterate on. Obviously remove the echo when you're happy that it'll do what you want.

Answer 3

Try this Shellcheck-clean pure Bash (except for the mv) code:

#! /bin/bash -p

shopt -s nullglob

filename_rx='^(.+)_([[:digit:]]{1,2})_([[:digit:]]{1,2}).csv$'

for file in L2_error_with_G_at_*_*.csv; do
    if [[ $file =~ $filename_rx ]]; then
        printf -v newfile '%s_%02d_%02d.csv' "${BASH_REMATCH[@]:1:3}"

        [[ $newfile == "$file" ]] || echo mv -v -- "$file" "$newfile"
    fi
done

• shopt -s nullglob makes globs expand to nothing when nothing matches (otherwise they expand to the glob pattern itself, which is almost never useful in programs). That means the loop below will do nothing if nothing matches the L2_..._*_*.csv pattern.
• See mkelement0's excellent answer to How do I use a regex in a shell script? for an explanation of [[ ... =~ ... ]].
• Remove the echo from echo mv -v ... if you are happy that the code will do what you want.
• The code in the loop should work for filenames containing any characters (including spaces, newlines, or glob characters).

Answer 4

Using parameter expansion - substring removal, requiring at_ as part of the string and printf

Only using these 2 constructs

• %pattern* remove everything after and including "pattern"
• #*pattern remove everything before and including "pattern"

for i in L2_error_with_G_at_2_9.csv L2_error_with_G_at_1_16.csv
do 
   base="${i%at_*}at_"
   num="${i#*at_}"
   num_2="${num#*_}"
   echo "$i" "${base}$(printf %02d "${num%_*}")_$(printf %02d "${num_2%.*}").csv"
done
L2_error_with_G_at_2_9.csv L2_error_with_G_at_02_09.csv
L2_error_with_G_at_1_16.csv L2_error_with_G_at_01_16.csv

Answer 5

Using GNU AWK

awk -F'[._]'  -v OFS='_' '
    {
        srcfile = $0
        for (i=NF-1; i>=NF-2; i--){    
             $i = sprintf("%02d", $i)
        }
        trgfile = gensub(/^(.*)_([^_]*)$/,"\\1.\\2",1)
        if(srcfile != trgfile)
            print |"mv \"" srcfile "\" \"" trgfile "\""
    }
' <(ls L2_error_with_G_at_*.csv)