Zipping files is appending by default - Watch out!

This is not a bug in the age-old zip Linux program. It's maybe a bug in its intuitiveness.

I have a piece of automation that downloads a zip file from a file storage cache (GitHub Actions actions/cache in this case). Then, it unpacks it, and plucks some of the files from it into another fresh new directory. Lastly, it creates a new .zip file with the same name. The same name because that way, when the process is done, it uploads the new .zip file into the file storage cache. But be careful; does it really create a new .zip file?

To demonstrate the surprise:

$ cd /tmp/
$ mkdir somefiles
$ touch somefiles/file1.txt
$ touch somefiles/file2.txt
$ zip -r somefiles.zip somefiles
  adding: somefiles/ (stored 0%)
  adding: somefiles/file1.txt (stored 0%)
  adding: somefiles/file2.txt (stored 0%)

Now we have a somefiles.zip to work with. It has 2 files in it.

Next session. Let's say it's another day and a fresh new /tmp directory and the previous somefiles.txt has been downloaded from the first session. This time we want to create a new somefile directory but in it, only have file2.txt from before and a new file file3.txt.

$ rm -fr somefiles
$ unzip somefiles.zip
Archive:  somefiles.zip
   creating: somefiles/
 extracting: somefiles/file1.txt
 extracting: somefiles/file2.txt
$ rm somefiles/file1.txt
$ touch somefiles/file3.txt
$ zip -r somefiles.zip somefiles
updating: somefiles/ (stored 0%)
updating: somefiles/file2.txt (stored 0%)
  adding: somefiles/file3.txt (stored 0%)

And here comes the surprise, let's peek into the newly zipped up somefiles.txt (which was made from the somefiles/ directory which only contained file2.txt and file3.txt):

$ rm -fr somefiles
$ unzip -l somefiles.zip
Archive:  somefiles.zip
  Length      Date    Time    Name
---------  ---------- -----   ----
        0  2023-10-04 16:06   somefiles/
        0  2023-10-04 16:05   somefiles/file1.txt
        0  2023-10-04 16:06   somefiles/file2.txt
        0  2023-10-04 16:06   somefiles/file3.txt
---------                     -------
        0                     4 files

I did not see that coming! The command zip -r somefiles.zip somefiles/ doesn't create a fresh new .zip file based on recursively walking the somefiles directory. It does an append by default!

The solution is easy. Right before the zip -r somefiles.zip somefiles command, do a rm somefiles.zip.