Just started with linux mint, need help with a command - eviltoast

As the title says, I just started with linux mint and am falling in love with bash scripts 😍 Actually I’m not sure if it’s considered a script, but I want to delete the last 2 files in all subfolders in a folder. So far I’ve (after great effort) got the terminal to list the files, but I want to delete them. Here is how I get them listed:

for f in *; do ls $f | tail -n 2; done

All their names come satisfyingly up in the terminal. Now what? I tried adding | xargs rm but that didn’t delete them. I also tried something with find command but that didn’t work either. Some folders have 3 items, so I want to delete #2 and 3. Some folders have 15 items so I want to delete #14 and 15. Folders are arranged by name, so it’s always the last 2 that I want to delete.

It’s frustrating to be sooooo clooooose, but also very fun. Any help is appreciated!

EDIT: Thanks for the awesome help guys! The next part of this is to move all the .html files into one folder (named “done”), prepending their name with an integer. So far I got:

n=1; for f in *; do find ./"$f" -type f | sort | xargs mv done/"$n$f"; n=$((n+1)); done

but that is… not really doing anything. The closest I have gotten so far is some error like

mv: Missing destination file operand

Any help is again appreciated!

  • skaarl@feddit.nlOPEnglish
    1·
    2 days ago

    Thank you for the tips, but now I’m getting “Cannot remove: No such file or directory” all the way down! The files are there, I see them, they come up in the terminal, but for some reason xargs rm does not want to delete them. When I put the -f flag, rm doesn’t give an error but the files are still there! wtf

    • harsh3466@lemmy.ml
      2·
      2 days ago

      When you run the command without the xargs bit, like this:

      for f in \*; do ls $f | tail -n 2; done\,

      Does the output give you the full file path, or just the file names?

      The full file path will look something like:

      /dir1/dir2/actual-file

      And of course the file name would just be:

      some-file

      If you’re getting just the file name, that’s the problem. Unless you’re in the directory with the file you wish to delete, rm needs the full path.

      Edit: grammar

      • skaarl@feddit.nlOPEnglish
        1·
        2 days ago

        Yea that must be it! It’s spitting out just the file name and not the whole path. There is only 1 level of depth, so I want to remove

        • ./folder1/file 3
        • ./folder1/file4
        • ./folder2/file11
        • ./folder2/file12

        so how do I get the whole path into xargs? I tried xargs "$f"/ but fortunately that didn’t work because it was trying to delete all the directories lmao XD

        • harsh3466@lemmy.ml
          3·
          2 days ago

          Here’s the command to delete the files:

          for f in *; do find ./"$f" -type f | sort | tail -n 2 | xargs -n 1 rm; done

          If you want to insure it will target the correct files, first run this command (I HIGHLY recommend you do this first. Verify BEFORE you delete so you don’t lose data):

          for f in *; do find ./"$f" -type f | sort | tail -n 2; done

          I’ll be adding another comment reply with a breakdown of the command shortly (just need to write it up)

          • harsh3466@lemmy.ml
            4·
            2 days ago

            Here’s what’s happening in the command;

            for f in *; do

            You already know this for loop, which is using the * glob to iterate over each directory in the current directory.

            find ./"$f" -type f

            Instead of your original ls command, which gives the file names, and not their full paths, we’re using GNU find, which outputs the full path of what it finds. The arguments are:

            ./"$f" - This tells find where to start its search. I double qouted the $f variable to properly expand the directory name even if it has nonstandard characters in it like spaces.

            -type f - This tells find what kind of file object to look for. So it’s two parts. -type to tell find there will be a specific type to look for, and the f flag, which means file. Meaning, it will only find files

            The output of find is not sorted alaphabetically, so before piping the output to tail, we first pipe it to sort, which by default will sort alphanumerically, which we then pipe to tail to grab just the last two files, and finally we get to the xargs bit.

            Here I added the -n 1 argument to xargs to get it to work on the files one at a time. This isn’t actually necessary. You could just run it as xargs rm. I didn’t realize that before I posted the command. (I’m still learning too! The learning never ends. :D )

            • skaarl@feddit.nlOPEnglish
              1·
              2 days ago

              Thanks so much harsh!!! I will study this and hit Enter after I understand it.

              Thanks again, that’s epic.

              • harsh3466@lemmy.ml
                2·
                2 days ago

                You’re welcome! Happy I could help.

                One other quick note, do the filenames or directories have spaces in them? If they do, that will cause a problem with the command as it is and need some additional modification. I accounted for the possible spaces in the directory names with the find command, but not with xargs. I just realized that as I was looking it over again.

                • skaarl@feddit.nlOPEnglish
                  2·
                  2 days ago

                  That was it! Thank you. I got rid of over 150 files in 127 directories with a lot less clicks than through the file explorer.

                  Luckily this time there were no spaces in the names. Spaces in names are a PITA at my stage of learning, and I’m never sure if I should use ’ or ".

                  Btw, new challenge in the edited original post, if you haven’t yet exhausted your thinking quota for the day lol.

                  • harsh3466@lemmy.ml
                    1·
                    2 days ago

                    I’d tackle this as two different commands rather than trying to come up with a oneliner to handle moving and renaming with incrementing numbers all in one go. It could be done all in one go, but to accomplish that, I’d probably write a bash script over a one liner.

                    So we’ll start with moving them, which is the easy part using find:

                    find html-dir -name '*.html' -exec mv {} /path/to/new/dir/ \;

                    Let’s look at each argument:

                    find - calls find

                    html-dir - is the path of the top level directory that contains all of the html files and/or other directories that contain the html files you want to move. This command will go recursively through all directories it finds inside of this top level directory.

                    -name - this flag tells find to use the following expression to match filenames

                    '*.html' - This is the expression being used to find filenames. It’s using globbing to match any file that has the .html extension

                    -exec this is kind of like xargs, but for find. It allows you to take what has been found and act on it.

                    mv - the move command to move the files to the new directory

                    {} - this is the placeholder to call the find results

                    /path/to/new/dir/ - the directory you want to move the .html files to

                    \; - this terminates the exec action. The backslash escapes the termination character. You can use two different termination characters, ; and +. The semicolon tells exec to pass the results through one at a time, while the plus tells exec to pass the results through all at once.

                    Once all of the files are in the new folder, cd into that folder and use this to enumerate the files:

                    n=1; for file in *; do pad="$(printf '%02d' $n)"; mv $file $pad$file; ((n++)); done

                    This is a one liner similar to what you were trying to do.

                    n=1; - sets our first number for the enumeration with a semicolon to terminate and set up the next command

                    for file in *; do - Sets up our for loop. file will be the variable in the loop representing each file as it loops, and * will bring in all the .html files in the directory you are currently in, and ; do terminates and sets up the loop.

                    pad="$(printf '%02d"'' $n)"; - You may or may not want to use this. I like to pad my numbers so they list properly. This creates a new variable inside the loop called pad. Using command substitution, the printf command will take our $n variable and pad it. The padding is dictated by the '%02d', which will turn 1 into 01, and so on. If there will be more than 99 .html files, and you want padding to three digits, then change '%02d' to '%03d'. Again the semicolon terminates and sets up the next command

                    mv $file $pad$file; - this is pretty straightforward. It’s taking the file being acted on in the loop as represented by the variable $file and moving it inside the same directory with the mv command to give it a new enumerated name by concatenating the $pad and $file variables into a single filename. so a.html would become 01a.html. This command is also terminated with a semicolon

                    ((n++)); - this is an easy way in bash to increment a numeric variable, again terminated by a semicolon,

                    done ends the loop, and the command.

                    Let me know how it works for you!

                  • harsh3466@lemmy.ml
                    1·
                    2 days ago

                    Okay, for the HTML files, do the integers need to be apied to the files in a certain order, like, alphanumerically from the folders they’re coming from?