work on pbs-absrepo-convert

3 files changed, 75 insertions, 78 deletions

diff --git a/config b/config
index e97e3e6..85c8003 100644
--- a/config
+++ b/config
@@ -1,6 +1,7 @@
 [core]
 	cachedir = /var/cache/pbs
 	sourcedir = /var/pbs-sources
+	rewritedir = /var/pbs-rewrite
 	treedir = /var/pbs
 
 [source "arch-packages"]
diff --git a/pbs-absrepo-convert b/pbs-absrepo-convert
index 97ef936..c03d58e 100755
--- a/pbs-absrepo-convert
+++ b/pbs-absrepo-convert
@@ -1,4 +1,4 @@
-#!/bin/bash -eu
+#!/bin/bash -euE
 
 export TMPDIR="`mktemp -d --tmpdir pbs-absrepo-convert.XXXXXXXXXX`"
 cleanup() {
@@ -16,19 +16,36 @@ abort() {
 }
 
 ##
+# Usage: trailing-newline FILE
+#
+# Adds a trailing newline to ${FILE} if there isn't one.
+#
+# Useful because different versions of git are inconsistent about
+# when there is one on output.
+###
+trailing-newline() {
+	local file=$1
+	if [[ "`sed -n '$p' "$file"|wc -l`" = 0 ]]; then
+		echo >> "$file"
+	fi
+}
+
+##
 # Usage: collect-data MODE
 # MODE is either 'correct' or 'fast'
 #
+# 'fast' may omit some packages that have been deleted.
+#
 # Assumptions:
-#  - currently in a working copy of the repo
-#  - git branch "master" exists and is untouched
-#  - TMPDIR is set
-#  - file(s) exist in "${TMPDIR}/missing-packages.in/"
+#  - currently in the working repo
+# Effected by:
+#  - file ../missing-packages
 # Effects:
+#  - creates file "../find"
+#  - creates file "../packages"
+#  - creates file "../architectures"
+# Side effects:
 #  - creates file "${TMPDIR}/commits"
-#  - creates file "${TMPDIR}/find"
-#  - creates file "${TMPDIR}/packages"
-#  - creates file "${TMPDIR}/architectures"
 ##
 collect-data() {
 	msg "$(gettext "Collecting package data...")"
@@ -36,45 +53,44 @@ collect-data() {
 	local mode=$1
 
 	case "$mode" in
-		correct)
-			git log --pretty=format:'%H' master > "${TMPDIR}/commits";;
-		fast)
-			echo master > "${TMPDIR}/commits";;
-		*)
-			usage; return 1;;
+		correct) git log     --pretty=format:'%H' master > "${TMPDIR}/commits";;
+		fast)    git log -n1 --pretty=format:'%H' master > "${TMPDIR}/commits";;
+		*)       usage; return 1;;
 	esac
+	trailing-newline "${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
+		printf "\\r$(gettext "Scanning commit %s (%d/%d)")" "$commit" "$n" "$count" >> /dev/tty
 		git ls-tree -rd --name-only "$commit"
-	done | fgrep /repos/ | sort -u > "${TMPDIR}/find"
+	done | fgrep /repos/ | sort -u > ../find
 	echo # newline
 
 	# extract some things ##################################################
 	# packages
 	{
 		if [[ $mode = fast ]]; then
-			cat "${TMPDIR}"/{packages,missing-packages}.in
+			cat ../{packages,missing-packages} 2>/dev/null || true
 		fi
-		< "${TMPDIR}/find" sed -e 's|^\.||' -e 's|/.*||'
-	} | sort -u > "${TMPDIR}/packages"
+		< ../find sed -e 's|^\.||' -e 's|/.*||'
+	} | sort -u > ../packages.tmp
+	mv ../packages{.tmp,}
 	# architectures
 	{
 	    echo master
-	    < "${TMPDIR}/find" sed -e 's/.*-//' -e '/^any$/d'
-	} | sort -u > "${TMPDIR}/architectures"
+	    < ../find sed -e 's/.*-//' -e '/^any$/d'
+	} | sort -u > ../architectures
 }
 
 ##
 # Usage: convert-package SOURCE PACKAGE
+#
 # Assumptions:
-#  - currently in a working copy of the repo
-#  - TMPDIR is set
+#  - currently in the working repo
 # Effects:
 #  - creates git branch "packages/${SOURCE}/${PACKAGE}"
-#  - creates file "${TMPDIR}/packages-${PACKAGE}.commits"
+#  - creates file "../packages-${PACKAGE}.commits"
 # Side effects:
 #  - changes git working tree
 #  - prints output of git commands
@@ -84,32 +100,37 @@ convert-package() {
 	local source=$1
 	local package=$2
 
-	local branch="packages/${source}/${package}"
+	local obranch="packages/${source}/${package}"
+	local ibranch
 	local dir
-
 	if git checkout "packages/${package}" &>/dev/null; then
 		# special case (common; optimization)
-		git checkout -b "$branch"
-		git branch -D "packages/${package}"
+		ibranch="packages/${package}"
 		dir=trunk
 	else
 		# general case (uncommon)
-		git checkout -b "$branch" orig
+		ibranch=master
 		dir="${package}/trunk"
 	fi
-	git filter-branch -f --prune-empty --subdirectory-filter "$dir" "$branch"
 
-	git log --pretty=format:'%T %H' > "${TMPDIR}/packages-${package}.commits"
+	git rewrite-branch --svn "$ibranch" "$obranch" \
+		--prune-empty --subdirectory-filter "$dir"
+
+	if [[ $ibranch != master ]]; then
+		git branch -D "$ibranch"
+	fi
+
+	git log "$obranch" --pretty=format:'%T %H' \
+		> "../packages-${package}.commits"
 }
 
 ##
-# Usage: convert-packages SOURCE
+# Usage: convert-packages SOURCENAME
 # Assumptions:
-#  - meets assumptions of `convert-package`
-#  - TMPDIR is set
-#  - file "${TMPDIR}/packages" contains a newline-separated list of packages
+#  - currently in the working repo
+#  - file "../packages" contains a newline-separated list of packages
 # Effects:
-#  - runs `convert-package` for every package listed in "${TMPDIR}/packages"
+#  - runs `convert-package` for every package listed in "../packages"
 #  - prints status updates about what it is doing
 ##
 convert-packages() {
@@ -117,8 +138,8 @@ convert-packages() {
 	[[ $# = 1 ]] || { usage; return 1; }
 	local source=$1
 
-	local count="$(wc -l < "${TMPDIR}/packages")"
-	cat -n "${TMPDIR}/packages" | while read n package; do
+	local count="$(wc -l < ../packages)"
+	cat -n ../packages | while read n package; do
 		msg2 "$(gettext "(%d/%d) %s")" "$n" "$count" "$package"
 		convert-package "$source" "$package"
 	done
@@ -127,10 +148,10 @@ convert-packages() {
 ##
 # Usage: convert-arch ARCH
 # Assumptions:
-#  - currently in a working copy of the repo
+#  - currently in the working repo
+#  - file "../packages-${PACKAGE}.commits" exists for every package
 # Effects:
 #  - creates git branch "${ARCH}"
-#  - TMPDIR is set
 #  - creates file "${TMPDIR}/missing-packages/${ARCH}"
 # Side effects:
 #  - changes git working tree
@@ -143,10 +164,8 @@ convert-arch() {
 	mkdir -p "${TMPDIR}/missing-packages"
 	touch "${TMPDIR}/missing-packages/${arch}.tmp"
 
-	git checkout orig
-	git checkout -b "$arch"
-	git filter-branch -f --tree-filter \
-		"pbs-absrepo-convert--filterarch $arch" "$arch"
+	git rewrite-branch master "$arch" \
+		--tree-filter "pbs-absrepo-convert--filterarch $arch" "$arch"
 
 	sort -u \
 	    < "${TMPDIR}/missing-packages/${arch}.tmp" \
@@ -157,9 +176,8 @@ convert-arch() {
 ##
 # Usage: convert-arches
 # Assumptions:
-#  - meets assumptions of `convert-arch`
+#  - currently in the working repo
 #  - file "${TMPDIR}/architectures" contains a newline-separated list of arches
-#  - file "${TMPDIR}/packages-${PACKAGE}.commits" exists for every package
 # Effects:
 #  - runs `convert-arch` for every arch listed in "${TMPDIR}/architectures"
 #  - prints status updates about what it is doing
@@ -175,47 +193,25 @@ convert-arches() {
 	done
 }
 
-##
-# Usage: copy-metatada SOURCEDIR SOURCE FILE
-# Assumptions:
-#  - TMPDIR is set
-# Effects:
-#  - creates "${TMPDIR}/${FILE}.in"
-##
-copy-metadata() {
-	[[ $# = 3 ]] || { usage; return 1; }
-	local sourcedir=$1
-	local source=$2
-	local file=$3
-
-	if [[ -f "${sourcedir}/${source}.${file}" ]]; then
-		cp "${sourcedir}/${source}.${file}" "${TMPDIR}/${file}.in"
-	else
-		touch "${TMPDIR}/${file}.in"
-	fi
-}
-
 main() {
 	trap abort EXIT
 	[[ $# = 1 ]] || { usage; return 1; }
 	local source=$1
 
-	local cachedir="$(pbs-plumb-config get core.cachedir)"
+	local cache="$(pbs-plumb-config get core.cachedir)/${source}.git"
+	local workdir="$(pbs-plumb-config get core.rewritedir)/${source}"
 	local sourcedir="$(pbs-plumb-config get core.sourcedir)"
 
-	# init #################################################################
-	copy-metadata "$sourcedir" "$source" missing-packages
-	copy-metadata "$sourcedir" "$source" packages
-
-	msg "$(gettext "Creating working copy...")"
-	git clone "${cachedir}/${source}.git" "${TMPDIR}/repo"
-	cd "${TMPDIR}/repo"
-	git checkout -b orig
-	collect-data fast
-	git branch -D master
+	# init ###################################d##############################
+	if [[ ! -d "${workdir}" ]]; then
+		msg "$(gettext "Creating working copy...")"
+		git clone "$cache" "${workdir}/repo"
+	fi
+	cd "${workdir}/repo"
 
 	# main #################################################################
 
+	collect-data     fast
 	convert-packages "$source"
 	convert-arches
 
diff --git a/pbs-absrepo-convert--filterarch b/pbs-absrepo-convert--filterarch
index 4288014..5940fa0 100755
--- a/pbs-absrepo-convert--filterarch
+++ b/pbs-absrepo-convert--filterarch
@@ -20,7 +20,7 @@ rm -rf *
 cat "$tree" | while read mode type treehash path; do
 	package="${path%%/*}"
 	repo="$(sed -n "s/^${package} //p" "$repos")"
-	commithash="$(sed -n "s/^${treehash} //p" "${TMPDIR}/packages-${package}.commits")"
+	commithash="$(sed -n "s/^${treehash} //p" "../packages-${package}.commits")"
 	if [[ -n "$commithash" ]]; then
 		mkdir -p "$repo"
 		git submodule add ./ "${repo}/${package}"