How can I move all folders in a directory to subdirectories based on their value? - regex

Question

A legacy web application I've inherited control over seems to have hit the maximum number of subdirectories in a folder on my web server. Whenever an article is created in the system, it's static content is placed in a subdirectory of the document root matching the pattern /uploads/story/{STORY_ID}/. But now the system is unable to create any new directories in the /uploads/story/ folder.
I'd like to address this in 2 steps, but I'm not sure how to run the necessary linux commands to achieve this. Would you be able to help?
As a temporary fix to buy me more time to implement a better directory structure, I'd like to archive the static content of all stories with a STORY_ID of less than 1000. These should be moved from /uploads/story/ to /uploads/story_archive/.
I'll change the upload path to be /uploads/story/{THOUSANDS}/{STORY_ID}/ in the code, but will need to be able to move all folders within /uploads/story/ into this format. e.g. /uploads/story/65312/ would become /uploads/story/65/65312/. How can I do this?
Edit
Fixing (1) was as simple as running:
$ cd /path/to/uploads/
$ mkdir story_archive
$ for i in {1..999}; do mv story/$i story_archive/; done

Answer 1

Given that you are sure /uploads/story/* will give you a number in the * part, you can do the following (Note: backup the whole thing just in case):
# update this based on your actual directory
path_to_fix=/uploads/stories
# move the directories out of the way so they don't get mixed up
mv $path_to_fix $path_to_fix/../temp
mkdir $path_to_fix > /dev/null 2>&1
# get all directories to be moved
dirs=$(ls $path_to_fix/../temp)
# for each of them
for d in $dirs; do
# get the basename, which is store_id
id=$(basename $d)
# divide by 1000
sub=$((id / 1000))
# create the appropriate directory
mkdir $path_to_fix/$sub > /dev/null 2>&1
# move the original directory to that sub-directory
mv $path_to_fix/../temp/$d $path_to_fix/$sub
done
# cleanup
rm -Rf $path_to_fix/../temp

Answer 2

for the temporary fix you can use for loop and a glob
for path in /uploads/story/*; do
storyid=${path##*/}
if [[ ${storyid} -lt 1000 ]]; then
mv "${path}" /uploads/story_archive/${storyid}
fi
done
that will iterate over all directories in /uploads/story/ directory with the full path in $path var.
the bash construct ${variable##pattern} will remove the longest substring matching pattern from the left hand side of variable, in this case all the leading directories leaving just the storyid to get stored in the var.
we then check to see if storey id is less than 1000 and move it to the story archive.
The next bit.
for path in /uploads/story/*; do
storyid=${path##*/}
if [[ $storyid -ge 1000 ]]; then
thous=${storyid%%???}
[[ -d/uploads/story/$thous/ ]] || mkdir /uploads/story/$thous/
mv $path /uploads/story/$thous/
fi
done
ok here again we iterate over all the directories and pluck off the story id. this time though we make sure that the storyid is greater than or equal to 1000 and use the %%??? to remove the last three characters from storyid ( the same as the ## trick but from the right hand side of the variable. )
Then we see if the thousads dir exists and make it if it doesn't and move the dir over.
you could even do one sweep and do both tasks at once
for path in /uploads/story/*; do
storyid=${path##*/}
if [[ $storyid -lt 1000 ]]; then
mv "${path}" /uploads/story_archive/${storyid}
else
thous=${storyid%%???}
[[ -d/uploads/story/$thous/ ]] || mkdir /uploads/story/$thous/
mv $path /uploads/story/$thous/
fi
done