improve performance of pbs-absrepo-convert's collect-data

2 files changed, 20 insertions, 21 deletions

diff --git a/pbs-absrepo-convert b/pbs-absrepo-convert
index 7b0bdaf..29f749f 100755
--- a/pbs-absrepo-convert
+++ b/pbs-absrepo-convert
@@ -15,23 +15,17 @@ abort() {
 	cleanup
 }
 
-_collect-data-gc() {
-	local base=$1
-	local files=`echo $base.new*`
-	cat -- $files 2>>/dev/null | sort -u > $base.tmp
-	rm -f -- $files
-	mv $base.tmp $base.new
-}
-
 collect-data-gc() {(
 	cd "$TMPDIR"
+	local files
 	while [[ ! -f collect-data-filter-branch.done ]]; do
-		_collect-data-gc packages
-		_collect-data-gc repo-arch
+		files=`echo find.new*`
+		cat -- $files 2>>/dev/null | sort -u > find.tmp
+		rm -f -- $files
+		mv find.tmp find.new
 		sleep 10
 	done
-	mv packages.new packages
-	mv repo-arch.new repo-arch
+	mv find.new find
 	rm -f collect-data-filter-branch.done
 )}
 
@@ -42,25 +36,33 @@ collect-data-gc() {(
 #  - git branch "master" exists and is untouched
 #  - TMPDIR is set and exists
 # Effects:
+#  - creates file "${TMPDIR}/find"
 #  - creates file "${TMPDIR}/packages"
-#  - creates file "${TMPDIR}/repo-arch"
 #  - creates file "${TMPDIR}/architectures"
 ##
 collect-data() {
 	msg "$(gettext "Collecting package data...")"
 	[[ $# = 0 ]] || { usage; return 1; }
+
+	# main data collection #################################################
 	collect-data-gc &
 	git filter-branch --tree-filter pbs-absrepo-convert--filterlist master
 	# notify collect-data-gc that filter-branch is done
 	touch "${TMPDIR}/collect-data-filter-branch.done"
 	# wait for collect-data-gc to finish
-	while [[ ! -f "${TMPDIR}/packages" ]]; do
+	while [[ ! -f "${TMPDIR}/find" ]]; do
 		sleep 10;
 	done
-	# extract some data from those files
+
+	# extract some things ##################################################
+	# packages
+	{
+		< "${TMPDIR}/find" sed -r 's|^\./([^/]*)/.*|\1|'
+	} | sort -u > "${TMPDIR}/packages"
+	# architectures
 	{
 	    echo master
-	    < "${TMPDIR}/repo-arch" sed -e 's/.*/-/' -e '/^any$/d'
+	    < "${TMPDIR}/find" sed -e 's/.*/-/' -e '/^any$/d'
 	} | sort -u > "${TMPDIR}/architectures"
 }
 
diff --git a/pbs-absrepo-convert--filterlist b/pbs-absrepo-convert--filterlist
index 645c351..df24066 100755
--- a/pbs-absrepo-convert--filterlist
+++ b/pbs-absrepo-convert--filterlist
@@ -1,7 +1,4 @@
 #!/bin/bash
 
-printf '%s\n' * > "${TMPDIR}/packages.tmp.${GIT_COMMIT}"
-find */repos/* -type d -printf '%f\n' >> "${TMPDIR}/repo-arch.tmp.${GIT_COMMIT}"
-
-mv "${TMPDIR}/packages.tmp.${GIT_COMMIT}" "${TMPDIR}/packages.new.${GIT_COMMIT}"
-mv "${TMPDIR}/repo-arch.tmp.${GIT_COMMIT}" "${TMPDIR}/repo-arch.new.${GIT_COMMIT}"
+find . -mindepth 3 -maxdepth 3| fgrep /repos/ > "${TMPDIR}/find.tmp.${GIT_COMMIT}"
+mv "${TMPDIR}/find.tmp.${GIT_COMMIT}" "${TMPDIR}/find.new.${GIT_COMMIT}"