删除Linux中除最新文件以外的所有.csv文件

Question

I'm trying to remove the files in a directory called Date_Sources . This directory has 5 subfolders in it, called Test1 Test2 ... Test5 .

I know how to remove the files every 90 days :

find /home/deployer/Data_Sources/.../..   -mtime +90 -type f -exec rm -r '{}' \; 

How can I make sure my script will remove all the files after 90 days EXCEPT the newest file (Per each subdirectory)?

So the script has to go to Date_Sources/Test1 , Data_sources/Test2 , ..., etc. and make sure all the files are removed after 90 days except the newest one.

Answer 1

Something like:

find /home/deployer/Data_Sources -maxdepth 1 -mindepth 1 -type d > afile
while IFS= read -r subdir
do
  find "$subdir" -mtime +90 -type f -printf "%T+\t%p\n" \
      | sort \
      | head -n-1 \
      | sed 's|[[:blank:]]\+| |' \
      | cut -f 2 -d " " > bfile
  while IFS= read -r each
  do
    rm -vf "$each"
  done < bfile
done < afile
rm -f bfile afile

Explanation:

find all directories in /home/deployer/Data_Sources within the immediate directory ( -maxdepth 1 ) and excluding itself ( -mindepth 1 ), and dump this to a file called afile

find /home/deployer/Data_Sources -maxdepth 1 -mindepth 1 -type d > afile

Read each line from afile and store in environment variable subdir (See the done line of this loop to see that afile is being piped to the while loop)

while IFS= read -r subdir
do

Run a find command on each subdir , listing only files older than 90 days (and print the timestamp with the file)

  find "$subdir" -mtime +90 -type f -printf "%T+\t%p\n" \

sort the output of the find

      | sort \

grab all but the last line of the sort output (the last line is the newest file)

      | head -n-1 \

replace all multiple whitespaces to a single space

      | sed 's|[[:blank:]]\+| |' \

grab the second column from the output, defining space as delimiter; all dumped to a file called bfile . (Could probably use a tab as a delimiter somehow and not use the sed above, but I wasn't sure how to specify it so it would work; -d "\\t" certainly didn't do the trick)

      | cut -f 2 -d " " > bfile

Now, read bfile line-by-line and store in each environment variable (Again: See the done line to show bfile being piped in)

  while IFS= read -r each
  do

Actually do the removal of the old files (being verbose and suppressing prompts)

    rm -vf "$each"
  done < bfile
done < afile

Remove the temporary files bfile and afile

rm -f bfile afile

Note:

-Updated to remove use of for loops to properly handle special characters and spaces in paths; also double quotes on variables (as suggested by @mklement0)