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

1 files changed, 10 insertions, 23 deletions

diff --git a/pbs-absrepo-convert b/pbs-absrepo-convert
index 29f749f..901b924 100755
--- a/pbs-absrepo-convert
+++ b/pbs-absrepo-convert
@@ -15,20 +15,6 @@ abort() {
 	cleanup
 }
 
-collect-data-gc() {(
-	cd "$TMPDIR"
-	local files
-	while [[ ! -f collect-data-filter-branch.done ]]; do
-		files=`echo find.new*`
-		cat -- $files 2>>/dev/null | sort -u > find.tmp
-		rm -f -- $files
-		mv find.tmp find.new
-		sleep 10
-	done
-	mv find.new find
-	rm -f collect-data-filter-branch.done
-)}
-
 ##
 # Usage: collect-data
 # Assumptions:
@@ -36,6 +22,7 @@ collect-data-gc() {(
 #  - git branch "master" exists and is untouched
 #  - TMPDIR is set and exists
 # Effects:
+#  - creates file "${TMPDIR}/commits"
 #  - creates file "${TMPDIR}/find"
 #  - creates file "${TMPDIR}/packages"
 #  - creates file "${TMPDIR}/architectures"
@@ -44,15 +31,15 @@ collect-data() {
 	msg "$(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}/find" ]]; do
-		sleep 10;
-	done
+	git log --pretty=format:'%H' master > "${TMPDIR}/commits"
+
+	# actual data collection ###############################################
+	local count="$(wc -l < "${TMPDIR}/commits")"
+	cat -n "${TMPDIR}/commits" | while read n commit; do
+		printf '\rscanning commit %s (%d/%d)' "$commit" "$n" "$count" >> /dev/tty
+		git ls-tree -rd --name-only "$commit"
+	done | fgrep /repos/ | sort -u > "${TMPDIR}/find"
+	echo # newline
 
 	# extract some things ##################################################
 	# packages