On Tue, Oct 25, 2011 at 4:50 PM, Eric Bélanger <[email protected]> wrote:
> No. I was waiting for feedback for the new version that I appended at
> the end of my last email. Allan was supposed to give it a look but I
> guess he was too busy with the pacman 4 release like everyone else. If
> that version is OK, let me know and I'll resubmit an updated git
> patch.
Looked at this tonight. A few observations:
* We use 'KiB' instead of 'K' now, so the magic matching was busted
* running one `pacman -Qi` call and one awk invocation per package
really stinks and is very slow
A revised version is below that pipelines and reduces calls to all
programs involved. It takes 1.8 seconds now instead of a lot longer.
(you can make it go faster, 0.5 seconds total, if you add a single
grep call in the pipeline, but not too concerned at this point).
Dave, I hope you cry at my bash skillz.
-Dan
$ time bash pacsysclean.sh >/dev/null
real 0m1.862s
user 0m1.803s
sys 0m0.070s
#!/bin/bash
PACMAN_OPTS=
usage() {
echo "pacsysclean - Sort installed packages by decreasing installed
size."
echo
echo "Usage: pacsysclean [options]"
echo
echo "Options:"
echo " -o <options> Specify custom pacman query options (e.g.,
-dt)"
echo " -h, --help Show this help message and exit"
}
if [ -n "$1" ]; then
case "$1" in
-o) PACMAN_OPTS="${2}" ;;
-h|--help) usage; exit 0 ;;
*) usage; exit 1 ;;
esac
fi
IFS=$'\n'
name="^Name.*: (.*)$"
size="^Installed Size.*: (.*) KiB$"
for line in $(LANG=C pacman -Qi $PACMAN_OPTS); do
if [[ $line =~ $name ]]; then
printf "%s\t" ${BASH_REMATCH[1]}
elif [[ $line =~ $size ]]; then
printf "%s\n" ${BASH_REMATCH[1]}
fi
done | sort -g -k2 | awk '
BEGIN {
split("KiB MiB GiB TiB PiB EiB ZiB YiB", suffix)
}
function format_size(size) {
count = 1
while (size + 0 > 1024) {
size /= 1024
count++
}
sizestr = sprintf("%.2f %s", size, suffix[count])
return sizestr
}
{
printf("%s\t%s\n", format_size($2), $1);
}'
