cvstogit

#!/bin/bash

# Convert RPM Fusion packages from CVS to Git.

# This script will convert the packages in the "free" and "nonfree" CVS
# repositories to Git. It downloads the owners.list file, and then cycles
# through each package in the list.

# The script converts each package to Git using the "git cvsimport" command.
# After "cvsimport", the branches are not usable, so we have to transform each
# subdirectory in CVS into a proper Git branch.

# The script also does some miscellaneous cleanups like moving .cvsignore to
# .gitignore and removing stray CVS files.

# Lastly the script automatically merges any branches that are identical to
# master. This allows git's fast-forward merges to work between branches.

# To skip a git-cvsimport bug triggered by a tag in the RPM Fusion dosemu
# package CVS repo, you will need the following patch to git-cvsimport:
# https://bugzilla.redhat.com/850640


# This setting is used for the conversion commit.
#AUTHOR="RPM Fusion Sysadmins <rpmfusion-sysadmin@lists.rpmfusion.org>"
AUTHOR="Nicolas Chauvet <kwizart@gmail.com>"

RF_CVSROOT=":pserver:anonymous@cvs.rpmfusion.org:/cvs"


# CVS-to-Git dist name translations.
declare -A BRANCHES
BRANCHES=(
['devel']='master'
['F-21']='f21'
['F-20']='f20'
['F-19']='f19'
['F-18']='f18'
['F-17']='f17'
['F-16']='f16'
['F-15']='f15'
['F-14']='f14'
['F-13']='f13'
['F-12']='f12'
['F-11']='f11'
['F-10']='f10'
['F-9']='f9'
['F-8']='f8'
['F-7']='f7'
['F-6']='f6'
['F-5']='f5'
['F-4']='f4'
['EL-6']='el6'
['EL-5']='el5'
['EL-4']='el4'
)

# Import a remote RPM Fusion CVS repo into a local git repo, using the
# git-cvsimport command, and writing into a "$package.cvsimport" directory. If
# the directory already exists, cvsimport will pull in any updates from the
# remote repo into the local Git version.
# Arguments:
#    - RPM Fusion repo (free or nonfree)
#    - RPM Fusion package (eg ffmpeg or openafs)
git_cvsimport() {
	if [ -z "$1" ]; then
		echo "Error: specify a repository (free or nonfree)"
		exit 1
	fi
	repo=$1
	if [ -z "$2" ]; then
		echo "Error: specify a package (eg. ffmpeg)"
		exit 1
	fi
	package=$2

	# Import from CVS into a Git repository.
	# TODO: use "-A <author-conv-file>" to convert authors' Unix account
	# names to Git author emails.

	git cvsimport -a \
		-d "$RF_CVSROOT/$repo" \
    -A ../rpmfusion-cvs-users.txt \
		-S 'branch' \
		$package \
		-C "$package.cvsimport" || exit 1
}


# This function takes a raw "git-cvsimport" git repository with subdirectories,
# such as:
#    openafs/
#         F-12/
#         F-13/
#         F-14/
# and moves each subdirectory into unique Git branches, following the branch
# naming mappings in the $BRANCHES array.
# Arguments:
#    - location to a git directory on disk.
convert_branches() {
	if [ -z "$1" ]; then
		echo "Error: specify a git repository directory"
		exit 1
	fi
	pushd $1
	# http://stackoverflow.com/questions/359424/detach-subdirectory-into-separate-git-repository
	for cvsbranch in "${!BRANCHES[@]}"; do
		# cvs branch name is $cvsbranch
		# git branch name is ${BRANCHES["$branch"]}
		gitbranch=${BRANCHES[$cvsbranch]}

		# Only operate on branches that exist
		if [[ ! -d $cvsbranch ]]; then
			# This branch does not exist in CVS; skip it.
			continue;
		fi

		# Debugging
		echo "Converting $cvsbranch to $gitbranch"

		# git-cvsimport creates an "origin" branch, and we'll use that
		# as our "base" as we create our other branches.

		# Create a new Git branch.
		# (We can skip creating a "master" branch, because cvsimport
		# already has already created one for us.
		if [[ $gitbranch != 'master' ]]; then
			git branch $gitbranch origin
		fi
		# Switch to our new branch.
		git checkout $gitbranch
		# Remove Makefile and branch along with moving .cvsignore to .gitignore
		# Rewrite this branch, isolating the changes from this one
		# directory in CVS.
		git filter-branch \
			--tag-name-filter cat --prune-empty --subdirectory-filter $cvsbranch \
			--tree-filter 'rm -f branch Makefile ; if [ -f .cvsignore ] ; then  git mv .cvsignore .gitignore ; fi' \
       HEAD
		# Cleanup after filter-branch.
		git reset --hard
		git for-each-ref --format="%(refname)" refs/original/ | xargs -n 1 git update-ref -d
		# Maybe we could do these two at the end of the loop, rather than
		# running after each branch?
		git reflog expire --expire=now --all
		git gc --aggressive --prune=now

		git checkout origin
	done
	popd
}

# This function takes a git repository with branches, and checks each branch
# against master, and attempts to merge any identical branches into master.
# Arguments:
#    - location to a git directory on disk.
merge_branches() {
	if [ -z "$1" ]; then
		echo "Error: specify a git repository directory"
		exit 1
	fi
	pushd $1

	# If the contents of a branch are identical to master, merge them
	# together. For example:
	# git checkout master
	# git merge f17
	# git checkout f17
	# git merge master

	# First part: merge any eligible $gitbranch into master.
	declare -a gitbranches
	declare -a mergedbranches
	gitbranches=$(git branch | cut -c 3-)
	for gitbranch in $gitbranches; do
		# Skip these two $gitbranch's:
		case $gitbranch in
		'master')
			# We won't be merging master with itself. Skip this one.
			continue
			;;
		'origin')
			# Don't bother the origin branch
			continue
			;;
		esac

		# Compare this $gitbranch to master.
		diffs=$(git diff master..${gitbranch})
		if [[ $diffs == '' ]]; then
			# This $gitbranch is identical to master; merge them
			# together.
			git checkout master
			git merge $gitbranch -m "Merge branch '$gitbranch'"
			# Record this $gitbranch for later, so we can go back
			# and merge master into it.
			mergedbranches+=( $gitbranch )
		fi
	done

	# Debugging
	echo
	echo "Found the following total branches in Git:"
	echo ${gitbranches[@]}
	echo "I have merged the following Git branches with master:"
	echo ${mergedbranches[@]}
	echo

	# Second part: Now ff-merge master into all our $mergedbranches.
	for gitbranch in ${mergedbranches[@]}; do
		git checkout $gitbranch
		# This should be a fast-forward merge.
		git merge master $gitbranch
	done

	# We're done using the "origin" branch, so remove it.
	git checkout master
	git branch -D origin

	popd
}



# Convert an entire RPM Fusion CVS repo tree to git modules.
# This function will write everything into a directory "$repo".
# Arguments:
#    - RPM Fusion repo (free or nonfree)
do_repotree() {
	if [ -z "$1" ]; then
		echo "Error: specify a repository (free or nonfree)"
		exit 1
	fi
	repo=$1

	if [[ ! -d $repo ]]; then
		mkdir $repo || exit 1
	fi
	pushd $repo

	# Grab the owners.list for this repo
	if [[ -d 'owners' ]]; then
		# We already downloaded an owners.list. Just update it.
		pushd owners
		cvs up
		popd
	else
		# Do a fresh owners.list checkout.
		cvs -d "$RF_CVSROOT/$repo" co owners
	fi

	# Cycle through each package.
	packages=$(grep -v "^#"  owners/owners.list | cut -d \| -f 2)
	for package in $packages; do

		# Skip these packages.
		case $package in
    'rpmfusion-package-config-smart')
      #fatal: refs/heads/origin: not a valid SHA1
      continue
      ;;
		'spek')
			# Present in CVS, but cannot do a checkout.
			# https://bugzilla.rpmfusion.org/2946
			continue
			;;
		'vlc-phonon-backend')
			# Invalid CVS module. "phonon-backend-vlc" is the correct one.
			# https://bugzilla.rpmfusion.org/1309#c20
			# https://bugzilla.rpmfusion.org/2949
			continue
			;;
		'xmms-aac')
			# Listed in free owners.list, but does not exist in CVS.
			# I emailed Orcan, and he said he cannot remember this one.
			# Perhaps it was orphaned and retired long ago.
			# https://bugzilla.rpmfusion.org/2947
			continue
			;;
		'slmodem')
			# Present in nonfree CVS, but cannot do a checkout.
			# (Similar symptoms to "spek".)
			# https://bugzilla.rpmfusion.org/2946
			continue
			;;
		'slmodem-kmod')
			# Present in nonfree CVS, but cannot do a checkout.
			# (Similar symptoms to "spek".)
			# https://bugzilla.rpmfusion.org/2946
			continue
			;;
		'vboxgtk')
			# Listed in nonfree owners.list, but does not exist in nonfree CVS.
			# Need to remove this from nonfree owners.list, and add to free list.
			# https://bugzilla.rpmfusion.org/2948
			continue
			;;
    'bsnes')
      continue
     ;;
    'xmris')
     # Cannot write branch origin for update
     continue
     ;;
     *)
       # Test if the directory already exists, in which case we do not convert again
      if [ -d $package ] ; then
        continue
      fi
      ;;
		esac

		echo "Importing $repo $package"
		# TODO: if git-cvsimport says "Already up-to-date.", then we don't need to
		# rewrite the whole git repo again, and we can skip over the package at
		# that point to save time.
		git_cvsimport $repo $package
		rm -rf $package
		cp -r "$package.cvsimport" $package
		# Do the rest of the conversion steps.
		convert_branches $package
		merge_branches $package
	done
	popd
}

do_repotree 'free'
do_repotree 'nonfree'

# vim: tabstop=2 shiftwidth=2