Build a static site from a Rails app using Wget and AWS S3

deploy_to_staging.sh

# precomile the static assets the HTML pages link to such as the .js, .css and .jpg files
RAILS_ENV=production bundle exec rake assets:precompile

# circleci has RAILS_ENV & RACK_ENV env variables set to test need to override. -d runs
# the server as a daemon.
RAILS_ENV=production RACK_ENV=production bundle exec rails s -d

# wait for server to load
sleep 10

# regex means it only grabs the clean URL HTML pages e.g /contact-us.
# everything else eg .js or .jpg is in the public folder now we have done the precompile
wget --recursive --reject-regex '.*\..*' localhost:3000

# misc pages that the crawl won't hit (--force-directories forces the creation of
# directory ./localhost:3000)
wget --force-directories \
  localhost:3000/unsupported-browser \
  localhost:3000/404

# to kill the server
kill `cat ./tmp/pids/server.pid`

# gzip up all the HTML files. -9 flag means it will do it at the highest compression
# level. gzipping is important for Google Page Speed ranking
gzip --recursive -9 localhost:3000
# remove .gz extension (% strips the .gz)
for f in `find localhost:3000 -iname '*.gz'`; do
  mv $f ${f%.gz}
done

# delete everything and start with a clean bucket
aws s3 rm s3://staging.mybucket.com --recursive --region eu-west-1

# sync the HTML pages. Note S3 can't infer content type because no extension
# on the clean URL's.
aws s3 sync ./localhost:3000 s3://staging.mybucket.com \
  --content-type text/html \
  --content-encoding gzip \
  --cache-control 'public, max-age=600' \
  --region eu-west-1

# gzip up the text files from the precompile. namely the application .js and .css files.
# note image files are binary and don't benefit from being gziped. They should be compressed
# by other means for example Photoshop.
cd ./public/assets
for f in `find . -name "*.css" -o -name "*.js"`; do
  gzip -9 $f
  mv $f.gz $f
  # cache control header can be at max age because asset pipeline uses cache busting
  # fingerprinting in the file names
  aws s3 cp $f s3://staging.mybucket.com/assets/ \
    --content-encoding gzip \
    --cache-control 'public, max-age=31536000' \
    --region eu-west-1
done
cd ..

# sync the rest of the asset folder such as .jpg files
aws s3 sync ./assets s3://staging.mybucket.com/assets \
  --region eu-west-1 \
  --cache-control 'public, max-age=31536000'

# this file is an exception and lives in the /public folder in a Rails application.
# It can not be in the assets pipeline because it can't have a fingerprint file name
aws s3 cp robots.txt s3://staging.mybucket.com/ --region eu-west-1

exit 0