Unzip all gz files in all subdirectories in the terminal

Question

Is there a way to unzip all gz files in the folder containing the zipfiles. When zip files are in subdirectories. A query for

find -type f -name "*.gz"

Gives results like this:

./datasets/auto/auto.csv.gz
./datasets/prnn_synth/prnn_synth.csv.gz
./datasets/sleep/sleep.csv.gz
./datasets/mfeat-zernike/mfeat-zernike.csv.gz
./datasets/sonar/sonar.csv.gz
./datasets/wine-quality-white/wine-quality-white.csv.gz
./datasets/ring/ring.csv.gz
./datasets/diabetes/diabetes.csv.g

Answer 1

If you want, for each of those, to launch "gzip -d" on them:

cd theparentdir && gzip -d $(find ./ -type f -name '*.gz')

and then, to gzip them back:

cd theparentdir && gzip $(find ./ -type f -name '*.csv')

This will however choke in many cases

• if filenames have some special characters (spaces, tabs, newline, etc) in them
• other similar cases
• or if there are TOO MANY files to be put after the gzip command!

A solution would be instead, if you have GNU find, to do :

find ... -print0 | xarsg -0 gzip -d # for the gunzip one, but still choke on files with "newline" in them

Another (arguably better?) solution, if you have GNU find at your disposal:

cd theparentdir && find ./ -type f -name '*.gz' -exec gzip -d '{}' '+'

and to re-zip all csv in that parentdir & all subdirs:

cd theparentdir && find ./ -type f -name '*.csv' -exec gzip '{}' '+'

"+" tells GNU find to try to put as many found files as it can on each gzip invocation (instead of doing 1 gzip incocation per file, very very ressource intensive and very innefficient and slow), similar to xargs, but with some benefits (1 command only, no pipe needed)

Answer 2

There is an option for recursivity (-r).

gzip -dr ./datasets

All archive will be decompressed in their own directory.

Example: gzip -dr ./a

a/b/c/test1.gz
a/b/d/test2.gz
a/e/test3.gz

After execution:

a/b/c/test1
a/b/d/test2
a/e/test3