Clean up a messy folder

For 3, sed can strip everything up to the last dot (s/.*\.//), then sort | uniq -c | sort -rn counts and ranks, just like the page counts in lesson 8.

$ find . -type f | wc -l
113
$ du -sh .
32K     .
$ ls | sed 's/.*\.//' | sort | uniq -c | sort -rn
     40 jpg
     21 pdf
     15 png
     12 txt
     12 JPG
      6 tmp
      3 log
      3 csv
      1 conf

Notice JPG and jpg are counted separately. Linux is case-sensitive, so you'll need -iname when you sort the photos.

$ mkdir -p sorted/{images,documents,data,logs,config,other}
$ tree sorted
sorted
├── config
├── data
├── documents
├── images
├── logs
└── other

7 directories, 0 files

find has -empty. Run it without -delete first. Careful: the photos and invoices you made with touch are empty too, so limit the search to the .tmp files.

$ find . -type f -name "*.tmp" -empty
./empty-1.tmp
./empty-2.tmp
./empty-3.tmp
./empty-4.tmp
./empty-5.tmp
./empty-6.tmp
$ find . -type f -name "*.tmp" -empty -delete
$ ls *.tmp
ls: cannot access '*.tmp': No such file or directory

This step is why you always look before you delete. find . -empty -delete on its own would have wiped every photo and invoice in this practice folder, because touch made them all empty.

find with \( -iname "*.jpg" -o -iname "*.png" \) matches all the photos. GNU mv has -t dir, which puts the destination first, so -exec mv -t ../sorted/images {} + moves them in one go. Names with spaces are safe this way, because find passes each name as one argument.

$ find . -type f \( -iname "*.jpg" -o -iname "*.png" \) -exec mv -t ../sorted/images {} +
$ find . -type f \( -name "*.pdf" -o -name "*.txt" \) -exec mv -t ../sorted/documents {} +
$ mv *.csv ../sorted/data/
$ mv *.log ../sorted/logs/
$ mv *.conf ../sorted/config/
$ ls | wc -l
0

downloads is empty now. Check where everything went:

$ for d in sorted/*; do echo "$d $(ls "$d" | wc -l)"; done
sorted/config 1
sorted/data 3
sorted/documents 33
sorted/images 67
sorted/logs 3
sorted/other 0

That for loop counts each directory. Lesson 28 explains loops; for now, copy it as it is.

find with -name "* *". To rename one, quote the old name: mv "meeting notes.txt" meeting-notes.txt.

$ find . -name "* *"
./meeting notes.txt
./final final v2.txt
$ mv "meeting notes.txt" meeting-notes.txt
$ mv "final final v2.txt" final-v2.txt
$ find . -name "* *" | wc -l
0

grep -c counts per file. For 2, grep -h leaves out the file names, sed can strip the ERROR prefix, and sort | uniq -c | sort -rn ranks them. tee saves and shows.

$ grep -c ERROR *.log
server-1.log:3
server-2.log:3
server-3.log:2
$ grep -h ERROR *.log | sed 's/^ERROR //' | sort | uniq -c | sort -rn | tee ~/project1/error-report.txt
      5 db timeout
      2 disk full
      1 auth failed

db timeout is the problem: five times across three servers.

find sorted -type f -mtime +90. GNU tar reads names from a file with -T old-files.txt, and --remove-files deletes each file after it's safely in the archive.

$ find sorted -type f -mtime +90 > old-files.txt
$ wc -l < old-files.txt
38
$ tar -czf old-$(date +%F).tar.gz --remove-files -T old-files.txt
$ tar -tzf old-*.tar.gz | wc -l
38
$ find sorted -type f | wc -l
69

ls -t lists newest first, so ls -t sorted/data/*.csv | head -n 1 is the newest. Put that inside $(...) in an ln -sfn command.

$ ls -lt sorted/data
total 8
-rw-rw-r-- 1 ubuntu ubuntu 2492 Oct 21 11:37 sales-2026-10.csv
-rw-rw-r-- 1 ubuntu ubuntu 3092 Oct 18 11:37 sales-2026-09.csv
$ ln -sfn "$(ls -t sorted/data/*.csv | head -n 1)" latest.csv
$ ls -l latest.csv
lrwxrwxrwx 1 ubuntu ubuntu 29 Oct 21 11:37 latest.csv -> sorted/data/sales-2026-10.csv
$ wc -l latest.csv
650 latest.csv

Careful: the newest modified file isn't always the one with the latest date in its name. Here sales-2026-10.csv was created last, so it wins.

Paths are full of slashes, so use another separator: s|old|new|g. Two changes, one sed with -e twice.

$ sed -i.bak -e 's|/home/olduser|/home/ubuntu|g' -e 's/^port = .*/port = 9000/' app.conf
$ cat app.conf
app_name = shop
data_dir = /home/ubuntu/data
log_dir = /home/ubuntu/logs
port = 9000
$ ls
app.conf  app.conf.bak

$ rmdir downloads
$ tree -L 3 --filelimit 15
.
├── error-report.txt
├── latest.csv -> sorted/data/sales-2026-10.csv
├── old-2026-10-21.tar.gz
├── old-files.txt
└── sorted
    ├── config
    │   ├── app.conf
    │   └── app.conf.bak
    ├── data
    │   ├── sales-2026-09.csv
    │   └── sales-2026-10.csv
    ├── documents  [16 entries exceeds filelimit, not opening dir]
    ├── images  [47 entries exceeds filelimit, not opening dir]
    ├── logs
    │   ├── server-1.log
    │   ├── server-2.log
    │   └── server-3.log
    └── other

8 directories, 11 files

rmdir would have refused if anything were left. --filelimit 15 stops tree from listing directories with more than 15 entries and shows how many there are instead, which keeps the big images and documents directories from flooding the screen.