Rewrite ftpdir-cleanup

* runtime reduced to < 1 second
* uses file lists and the comm command
* searches for missing packages
* removes empty legacy directories like extra/os/any

1 files changed, 92 insertions, 24 deletions

diff --git a/cron-jobs/ftpdir-cleanup b/cron-jobs/ftpdir-cleanup
index 3456e9c..0a43bd1 100755
--- a/cron-jobs/ftpdir-cleanup
+++ b/cron-jobs/ftpdir-cleanup
@@ -3,37 +3,105 @@
 . "$(dirname $0)/../db-functions"
 . "$(dirname $0)/../config"
 
-repos="$(get_repos_for_host)"
+clean_pkg() {
+	local pkg
+	local target
+
+	if ! ${CLEANUP_DRYRUN}; then
+		for pkg in "$@"; do
+			if [ -h "$pkg" ]; then
+				rm -f "$pkg"
+			else
+				mv -f "$pkg" "$CLEANUP_DESTDIR"
+			fi
+		done
+	fi
+}
+
+repos=($(get_repos_for_host))
 
 script_lock
 
-#adjust the nice level to run at a lower priority
-/usr/bin/renice +10 -p $$ > /dev/null
+for repo in ${repos[@]}; do
+	for arch in ${ARCHES[@]}; do
+		repo_lock ${repo} ${arch} || exit 1
+	done
+done
+
+${CLEANUP_DRYRUN} && warning 'dry run mode is active'
 
-repopaths=''
-for repo in $repos; do
-    $(dirname $0)/../misc-scripts/ftpdir-cleanup-repo $repo
-    repopaths="${repopaths} ${FTP_BASE}/${repo}/os/"
+for repo in ${repos[@]}; do
+	for arch in ${ARCHES[@]}; do
+		if [ ! -f "${FTP_BASE}/${repo}/os/${arch}/${repo}${DBEXT}" ]; then
+			warning "${FTP_BASE}/${repo}/os/${arch}/${repo}${DBEXT} not found, skipping"
+			continue
+		fi
+		# get a list of actual available package files
+		find "${FTP_BASE}/${repo}/os/${arch}" -xtype f -name "*${PKGEXT}" -printf '%f\n' | sort > "${WORKDIR}/repo-${repo}-${arch}"
+		# get a list of package files defined in the repo db
+		bsdtar -xOf "${FTP_BASE}/${repo}/os/${arch}/${repo}${DBEXT}" | awk '/^%FILENAME%/{getline;print}' | sort > "${WORKDIR}/db-${repo}-${arch}"
+
+		missing_pkgs=($(comm -13 "${WORKDIR}/repo-${repo}-${arch}" "${WORKDIR}/db-${repo}-${arch}"))
+		if [ ${#missing_pkgs[@]} -ge 1 ]; then
+			error "Missing packages in [${repo}] (${arch})..."
+			for missing_pkg in ${missing_pkgs[@]}; do
+				msg2 "${missing_pkg}"
+			done
+		fi
+
+		old_pkgs=($(comm -23 "${WORKDIR}/repo-${repo}-${arch}" "${WORKDIR}/db-${repo}-${arch}"))
+		if [ ${#old_pkgs[@]} -ge 1 ]; then
+			msg "Removing old packages from [${repo}] (${arch})..."
+			for old_pkg in ${old_pkgs[@]}; do
+				msg2 "${old_pkg}"
+				clean_pkg "${FTP_BASE}/${repo}/os/${arch}/${old_pkg}"
+			done
+		fi
+	done
 done
 
-to_cleanup=""
-poolpath="$FTP_BASE/$(get_pkgpool_for_host)/"
-pushd $poolpath >/dev/null
-for pkg in *$PKGEXT; do
-[ -f "$pkg" ] || continue # in case we get a file named "*.pkg.tar.gz"
-LINKS="$(/usr/bin/find $repopaths -type l -name "$pkg" 2>/dev/null)"
-if [ -z "$LINKS" ]; then
-	to_cleanup="$to_cleanup $poolpath/$pkg"
+# get a list of all available packages in the pacakge pool
+find "$FTP_BASE/$(get_pkgpool_for_host)" -name "*${PKGEXT}" -printf '%f\n' | sort > "${WORKDIR}/pool"
+# create a list of packages in our db
+cat "${WORKDIR}/db-"* | sort -u > "${WORKDIR}/db"
+
+old_pkgs=($(comm -23 "${WORKDIR}/pool" "${WORKDIR}/db"))
+if [ ${#old_pkgs[@]} -ge 1 ]; then
+	msg "Removing old packages from package pool..."
+	for old_pkg in ${old_pkgs[@]}; do
+		msg2 "${old_pkg}"
+		clean_pkg "$FTP_BASE/$(get_pkgpool_for_host)/${old_pkg}"
+	done
 fi
+
+# cleanup of legacy $repo/os/any directories
+for repo in ${repos[@]}; do
+	if [ ! -d "${FTP_BASE}/${repo}/os/any" ]; then
+		continue
+	fi
+	if [ -n "$(find "${FTP_BASE}/${repo}/os/any" -type d -empty)" ]; then
+		msg "Removing empty legacy directory ${repo}/os/any"
+		${CLEANUP_DRYRUN} && rmdir "${FTP_BASE}/${repo}/os/any"
+		continue
+	fi
+	find "${FTP_BASE}/${repo}/os/any" -name "*${PKGEXT}" -printf '%f\n' | sort > "${WORKDIR}/any-${repo}"
+	cat "${WORKDIR}/db-${repo}-"* | sort -u > "${WORKDIR}/all-${repo}"
+
+	old_pkgs=($(comm -23 "${WORKDIR}/any-${repo}" "${WORKDIR}/all-${repo}"))
+	if [ ${#old_pkgs[@]} -ge 1 ]; then
+		msg "Removing old packages from [${repo}] (any)..."
+		for old_pkg in ${old_pkgs[@]}; do
+			msg2 "${old_pkg}"
+			clean_pkg "${FTP_BASE}/${repo}/os/any/${old_pkg}"
+		done
+	fi
+done
+
+
+for repo in ${repos[@]}; do
+	for arch in ${ARCHES[@]}; do
+		repo_unlock ${repo} ${arch}
+	done
 done
-popd >/dev/null
-
-if [ -n "$to_cleanup" ]; then
-    msg "The following packages are no longer in any repo"
-    for f in $to_cleanup; do
-        msg2 "$(basename "$f")"
-    done
-    ${CLEANUP_DRYRUN} || mv $to_cleanup "$CLEANUP_DESTDIR"
-fi
 
 script_unlock