How do I tar files and put them in a list with gitbash

Question

I am using Git bash and wanted to archive my files with the "tar" function in git Bash.

As you can see, these files have dates in their name and I want to tar all dates in separate tar files. The result would be in this case that I have 2 tar files with 2 different dates.

My idea was to "find" the 6 digits in these files with find iname "jo* | \d\d\d\d\d\d | sort | tar -czvf testarchive.zip ~/test/targetfolder and then tar them. But then, I have to put the first found dates in a list, tar them until the date changes and then put the second datefiles in a list and so on...

Since I have no experience with bash and scripting, I dont know how to solve this problem. I would be very happy for help. I was not able to find a solution in the internet yet..

PS I am not sure if git bash is Linux/Unix or a different script language or if it is just Git bash, so sorry if i didnt meet up all the requirements.

Answer 1

At first glance, I'd build an associative array of unique dates, then tar each set that matched.

$: echo jo* # I made some similar local files
jo02042018ab jo02042018cd jo10112018ab jo10112018cd jo10112018ef

Filenames aren't exact, but close enough to make the point.

$: declare -A dateLst # create an *associative* array, keys are dates

An associative array uses strings as indexes (even if they are numbers).

$: for f in jo[0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9]?? # for each file
>  do dateLst["${f//[^0-9]/}"]=1                         # assert date as a key
>  done

dateLst[somekey]=1 creates a unique somekey in dateLst .
${f//[^0-9]/} removes all non-numerals from the filename, leaving only the date. (This will probably work fine even if they actually have .txt filename extensions, though you'll need to edit the glob in the for statement.)
So, dateLst["${f//[^0-9]/}"]=1 sets the date from the file as a key in the lookup table. If it already exists, it just sets it again.

We now have a table of unique dates we can use to create tarballs.

$: for t in "${!dateLst[@]}"; do echo tar -cvzf $t.tgz jo$t*; done
tar -cvzf 10112018.tgz jo10112018ab jo10112018cd jo10112018ef
tar -cvzf 02042018.tgz jo02042018ab jo02042018cd

${!dateLst[@]} is the keys from the table, so this loops over the distinct dates and echo s the tar command to create each tarball. jo$t* lists all matching files to be included in that archive. Remove the echo to execute the commands.

Answer 2

Perhaps you are looking for something like this?

for pat in jo[0-9][0-9][0-9][0-9][0-9][0-9]*; do
    tail=${pat#jo[0-9][0-9][0-9][0-9][0-9][0-9]}
    head=${pat%"$tail"}
    date=${head#jo}
    # If this tarball exists, skip
    if [ -e "$date.tar.gz" ]; then
        continue
    fi
    # Else, tar this date
    tar czvf "$date.tar.gz" "jo$date"*
done

It's unclear where or how you hoped ~/test/targetfolder would be used. If that's where the files are, perhaps just cd there before running this. If that's where you want the tar files to be created, put it before "$date.tar.gz" everywhere in the above.

The parameter expansions ${variable#pattern} and ${variable%pattern} produce the value of $variable with, repectively, any prefix or suffix match on pattern removed.