A script to automate converting a SourceForge.net CVS repo into a git repo.

sfcvs2git.sh

#!/usr/bin/env bash
#

#---------------------------------------------------------------------
usage ()
{
	cat <<EOT

${0##*/}
    Wraps up the commands required to dump a SourceForge CVS repo
    (that you have SSH access to) and convert it into a git repo.
    Will create a working directory in whatever path you run this
    from.

    Requires the cvs, git and cvs2svn commands. If these are not
    already present and in your PATH, this script will attempt to
    install them for you using MacPorts.

Usage:
    bin/${0##*/}


EOT

	exit ${1:-0}  # Exit with code 0 unless an arg is passed to the method.
}
if [ "$1" = '-h' ]; then
	usage
fi

#---------------------------------------------------------------------
echoerr() { echo "$@" 1>&2; }


GIT_USER=$(git config --get user.name)
GIT_EMAIL=$(git config --get user.email)
WORKING_DIR="sfcvs2git"
REQUIRED_COMMANDS=("cvs" "cvs2svn" "git" "zip")


# Input validation.
if [ -z "$1" ]; then
	echoerr "!! You must provide the SourceForge project name as the first arg."
	exit 1
fi
PROJECT_NAME="$1"


# Environment verification.
for CMD in "${REQUIRED_COMMANDS[@]}"; do
	if ! command -v "$CMD" >/dev/null 2>&1; then
		echoerr "!! You must have the \`${CMD}\` command installed."

		if command -v 'port' >/dev/null 2>&1; then
			echoerr "!! Trying \`sudo port install ${CMD}\`."

			if ! sudo port install $CMD; then
				exit 1
			fi
		else
			echoerr "!! Please install it in your PATH before continuing."
			exit 2
		fi
	fi
done


# Create the working dir.
if [ ! -d "$WORKING_DIR" ]; then
	mkdir -p "$WORKING_DIR/tmp"
fi
cd "$WORKING_DIR" >/dev/null 2>&1


# Download a full dump of the CVS repo.
echo "## Downloading CVS repo."
rsync -aq rsync://$PROJECT_NAME.cvs.sourceforge.net/cvsroot/$PROJECT_NAME/\* "${PROJECT_NAME}"


# Loop over all module folders (other than CVSROOT) in the downloaded
# folder, as long as there isn't already an output git folder (that we
# don't want to overwrite.)
IFS="$(printf '\n\t')"
for MODULE in "$PROJECT_NAME/"* ; do
	if [ -d "$MODULE" ] && [[ ! -d "output/${MODULE}" ]] && [[ ! "$MODULE" =~ CVSROOT$ ]]; then
		echo "## Found module \`${MODULE}\`"
		cvs2git -qq --blobfile=tmp/blob.dat --dumpfile=tmp/dump.dat --username=$USER "${MODULE}"

		# Post-process the dump file to set better committer info.
		sed -i '' "s/^committer $USER \<\>/committer $GIT_USER \<$GIT_EMAIL\>/" tmp/dump.dat

		# Create the git repo and import blob/dump data.
		(
			echo "## Creating git repo."
			git init --quiet --bare "output/${MODULE}"

			echo "## Importing blob and dump data."
			cd "output/${MODULE}" >/dev/null 2>&1
			cat ../../../tmp/{blob,dump}.dat | git fast-import --quiet
		)
		echo "## Repo done: \`${WORKING_DIR}/output/${MODULE}\`"
	fi
done