publish.sh

# Upload the compiled website to S3.
#
# Since Jekyll doesn't fully support incremental builds (where only
# modified files get recompiled), lots of files in _site/ end up getting
# re-compiled even if they don't change; to avoid uploading these to S3,
# rsync _site/ against a persistent _s3_site/ directory using file
# checksums rather than timestamps, and then sync *that* with S3. This
# ensures that only files whose contents have changed get uploaded.

print_msg(){
	echo "$(tput bold)$1$(tput sgr0)"
}

print_msg "rsyncing..."
rsync -ivac --no-t --delete _site/ _s3_site/
echo ""

print_msg "The following files will be synced:"
aws s3 sync --dryrun _s3_site s3://sevko.io
echo ""

while true; do
	read -p "$(tput bold)Does that look right? (y/n) $(tput sgr0)" yn
	case $yn in
		[y]* ) aws s3 sync _s3_site s3://sevko.io; break;;
		[n]* ) print_msg "Cancelling"; break;;
		* ) print_msg "Enter y or n";;
	esac
done