Equivalent to tar's "--strip-components=1" in unzip?

Question

I have a script that extracts a tar.gz-file to a specified subdirectory mysubfolder:

mkdir mysubfolder; tar --extract --file=sourcefile.tar.gz --strip-components=1 --directory=mysubfolder;

Is there any equivalent way of doing this with a zip-file?

Answer 1

As Mathias said, unzip has no such option. But a one-liner bash script can do the job.

Problem is: the best approach depends on your archive layout. A solution that assumes a single top-level dir will fail miserably if the content is directly in the archive root: think about an archive containing /a/foo, /b/foo, /foo, and the chaos of stripping /a and /b.

And the same fail happens with tar --strip-component. There is no one-size-fits-all solution.

So, to strip the root dir, assuming there is one (and only one):

unzip -d "$dest" "$zip" && f=("$dest"/*) && mv "$dest"/*/* "$dest" && rmdir "${f[@]}"

Just make sure that second-level files/dirs do not have the same name of the top-level parent (for example, /foo/foo). But /foo/bar/foo and /foo/bar/bar are ok. If they do, or you just want to be safe, you can use a temp dir for extraction:

temp=$(mktemp -d) && unzip -d "$temp" "$zip" && mkdir -p "$dest" &&
mv "$temp"/*/* "$dest" && rmdir "$temp"/* "$temp"

If you're using Bash, you can test if top level is a single dir or not using:

f=("$temp"/*); (( ${#f[@]} == 1 )) && [[ -d "${f[0]}" ]] && echo "Single dir!"

Speaking of Bash, you should turn on dotglob to include hidden files, and you can wrap everything in a single, handy function:

# unzip featuring an enhanced version of tar's --strip-components=1
# Usage: unzip-strip ARCHIVE [DESTDIR] [EXTRA_cp_OPTIONS]
# Derive DESTDIR to current dir and archive filename or toplevel dir
unzip-strip() (
    set -eu
    local archive=$1
    local destdir=${2:-}
    shift; shift || :
    local tmpdir=$(mktemp -d)
    trap 'rm -rf -- "$tmpdir"' EXIT
    unzip -qd "$tmpdir" -- "$archive"
    shopt -s dotglob
    local files=("$tmpdir"/*) name i=1
    if (( ${#files[@]} == 1 )) && [[ -d "${files[0]}" ]]; then
        name=$(basename "${files[0]}")
        files=("$tmpdir"/*/*)
    else
        name=$(basename "$archive"); name=${archive%.*}
        files=("$tmpdir"/*)
    fi
    if [[ -z "$destdir" ]]; then
        destdir=./"$name"
    fi
    while [[ -f "$destdir" ]]; do destdir=${destdir}-$((i++)); done
    mkdir -p "$destdir"
    cp -ar "$@" -t "$destdir" -- "${files[@]}"
)

Now put that in your ~/.bashrc and you'll never have to worry about it again. Simply use as:

unzip-strip sourcefile.zip [mysubfolder] [OPTIONS]

This little beast will:

• Create mysubfolder for you if it does not exist
• Automatically detect if if your zip archive contains a single top-level directory and handle the extraction accordingly.
• mysubfolder is optional. If blank it will extract to a subdir of the current directory (not necessarily the archive directory!), named after:
  • The single top-level directory in the archive, if there is one
  • or archive file name, without the (presumably .zip) extension
• If destination path, given or derived, already exists as a file, increment the name until a suitable one is found (new path or existing directory).
• By default this will:
  • be silent
  • overwrite any existing files
  • preserve links and attributes (mode, timestamps, etc)
• You can pass extra OPTIONS to cp. Useful options are:
  • -v|--verbose: output each copied file, just like unzip does
  • -n|--no-clobber: do not overwrite existing files
  • -u|--update: only overwrite files that are newer than the destination
• Such extra OPTIONS are the 3rd argument onward. Craft a proper argument parser if needed.
• It will use double the extracted disk space during the operation, due to "extract to temp dir and copy" approach. No way around this without losing some of its flexibility/features.

Answer 2

You can use -j to junk paths (do not make directories). This is only recommended for somewhat common single-level archives. Archives with multi level directory structures will be flattened - this might even lead to name clashes for the files to extract.

From the man page of unzip:

   -j     junk  paths.   The  archive's directory structure is not recreated; all files are deposited in the
          extraction directory (by default, the current one).

Answer 3

As others have noted, unzip does not support this. However, bsdtar can extract zip files as well.

bsdtar xvf app.zip --strip-components=1 -C /opt/some-app

bsdtar can also extract streams (which unzip does not support).

curl -sSL https://... | bsdtar xvf - --strip-components=1 -C /opt/some-app

Answer 4

I couldn’t find such an option in the manual pages for unzip, so I’m afraid this is impossible. :(

However, (depending on the situation) you could work around it. For example, if you’re sure the only top-level directory in the zip file is named foo- followed by a version number, you could do something like this:

cd /tmp
unzip /path/to/file.zip
cd foo-*
cp -r . /path/to/destination/folder