nixpkgs/pkgs/build-support/fetchzip/default.nix
Eelco Dolstra ea36f3b868 fetchFromGitHub: Use .tar.gz instead of .zip
Also clean up the name attribute of fetchzip derivations a bit.
2014-05-09 15:53:44 +02:00

45 lines
1.3 KiB
Nix

# This function downloads and unpacks an archive file, such as a zip
# or tar file. This is primarily useful for dynamically generated
# archives, such as GitHub's /archive URLs, where the unpacked content
# of the zip file doesn't change, but the zip file itself may
# (e.g. due to minor changes in the compression algorithm, or changes
# in timestamps).
{ lib, fetchurl, unzip }:
{ # Optionally move the contents of the unpacked tree up one level.
stripRoot ? true
, url
, ... } @ args:
fetchurl ({
# Remove the extension, because otherwise unpackPhase will get
# confused. FIXME: fix unpackPhase.
name = args.name or lib.removeSuffix ".zip" (lib.removeSuffix ".tar.gz" (baseNameOf url));
recursiveHash = true;
downloadToTemp = true;
postFetch =
''
export PATH=${unzip}/bin:$PATH
mkdir $out
cd $out
renamed="$TMPDIR/${baseNameOf url}"
mv "$downloadedFile" "$renamed"
unpackFile "$renamed"
''
# FIXME: handle zip files that contain a single regular file.
+ lib.optionalString stripRoot ''
shopt -s dotglob
if [ "$(ls -d $out/* | wc -l)" != 1 ]; then
echo "error: zip file must contain a single directory."
exit 1
fi
fn=$(cd "$out" && echo *)
mv $out/$fn/* "$out/"
rmdir "$out/$fn"
'';
} // args)