Middleman::S3Sync

This gem determines which files need to be added, updated and optionally deleted and only transfer these files up. This reduces the impact of an update on a web site hosted on S3.

Why not Middleman Sync?

Middleman Sync does a great job to push Middleman generated websites to S3. The only issue I have with it is that it pushes every files under build to S3 and doesn't seem to properly delete files that are no longer needed.

Version Support

• Use middleman-s3_sync version 4.x for Middleman 4.x
• Use middleman-s3_sync version 3.x for Middleman 3.x

Installation

Add this line to your application's Gemfile:

gem 'middleman-s3_sync'

And then execute:

$ bundle

Or install it yourself as:

$ gem install middleman-s3_sync

Usage

You need to add the following code to your config.rb file:

activate :s3_sync do |s3_sync|
  s3_sync.bucket                     = 'my.bucket.com' # The name of the S3 bucket you are targeting. This is globally unique.
  s3_sync.region                     = 'us-west-1'     # The AWS region for your bucket.
  s3_sync.aws_access_key_id          = 'AWS KEY ID'
  s3_sync.aws_secret_access_key      = 'AWS SECRET KEY'
  s3_sync.delete                     = false # We delete stray files by default.
  s3_sync.after_build                = false # We do not chain after the build step by default.
  s3_sync.prefer_gzip                = true
  s3_sync.path_style                 = true
  s3_sync.reduced_redundancy_storage = false
  s3_sync.acl                        = 'public-read'
  s3_sync.encryption                 = false
  s3_sync.prefix                     = ''
  s3_sync.version_bucket             = false
  s3_sync.index_document             = 'index.html'
  s3_sync.error_document             = '404.html'
end

You can then start synchronizing files with S3 through middleman s3_sync.

Configuration Defaults

The following defaults apply to the configuration items:

Setting AWS Credentials

There are several ways to provide the AWS credentials for s3_sync:

Through config.rb

You can set the aws_access_key_id and aws_secret_access_key in the block that is passed to the activate method.

I strongly discourage using this method. This will lead you to add and commit these changes to your SCM and potentially expose sensitive information to the world.

Through .s3_sync File

You can create a .s3_sync at the root of your middleman project. The credentials are passed in the YAML format. The keys match the options keys.

The .s3_sync file takes precedence to the configuration passed in the activate method.

A sample .s3_sync file is included at the root of this repo.

Make sure to add .s3_sync to your ignore list if you choose this approach. Not doing so may expose credentials to the world.

Through the Command Line

The aws credentials can also be passed via a command line options --aws_access_key_id (-k) and --aws_secret_access_key (-s). They should override any other settings if specified.

Through Environment

You can also pass the credentials through environment variables. They map to the following values:

The environment is used when the credentials are not set in the activate method or passed through the .s3_sync configuration file.

Through IAM role

Alternatively, if you are running builds on EC2 instance which has approrpiate IAM role, then you don't need to think about specifying credentials at all – they will be pulled from AWS metadata service.

IAM Policy

Here's a sample IAM policy that will allow a user to update the site contained in a bucket named "mysite.com":

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": "s3:*",
      "Resource": "arn:aws:s3:::mysite.com"
    },
    {
      "Effect": "Allow",
      "Action": "s3:*",
      "Resource": "arn:aws:s3:::mysite.com/*"
    }
  ]
}

This will give full access to both the bucket and it's contents.

Command Line Usage

Push All Content to S3

There are situations where you might need to push the files to S3. In such case, you can pass the --force (-f) option:

$ middleman s3_sync --force

Overriding the destination bucket

You can override the destination bucket using the --bucket (-b) switch. The command is:

$ middleman s3_sync --bucket=my.new.bucket

Overriding the destination prefix

You can override the destination prefix using the --prefix (-p) switch. The command is:

$ middleman s3_sync --prefix=my/new/prefix

Specify a Middleman environment

You can specify which environment to run Middleman under using the --environment (-e) option:

$ middleman s3_sync --environment=production

You can set up separate sync environments in config.rb like this:

	configure :staging do
		activate :s3_sync do |s3_sync|
			s3_sync.bucket = '<bucket'
			... 
    	end
    end

See the Usage section above for all the s3_sync. options to include. Currently, the .s3_sync file does not allow separate environments.

Dry Run

You can perform a dry run to see what would be the result of a sync operation using the --dry_run (-d) option:

$ middleman s3_sync --dry_run

Print instrument messages

The --instrument (-i) option will output more information about Middleman and s3_sync.

Run build before synchronizing

The --build (-B) option will ensure that Middleman build is run before the synchronization with the S3 bucket.

AWS Configuration

Pushing to a folder within a bucket

You can push to a folder within an S3 bucket by adding using the prefix option in the config block:

activate :s3_sync do |s3_sync|
  # ...
  s3_sync.prefix = 'prefix'
end

Bucket Versioning

You can enable bucket versioning by setting the version_bucket setting to true within the bucket configuration.

Versioning is enabled at the bucket level, not at the object level.

You can find out more about versioning here.

HTTP Caching

By default, middleman-s3_sync does not set caching headers. In general, the default settings are sufficient. However, there are situations where you might want to set a different HTTP caching policy. This may be very helpful if you are using the asset_hash extension.

Setting a policy based on the mime-type of a file

You can set a caching policy for every files that match a certain mime-type. For example, setting max-age to 0 and kindly asking the browser to revalidate the content for HTML files would take the following form:

caching_policy 'text/html', max_age: 0, must_revalidate: true

As a result, the following Cache-Control header would be set to max-age:0, must-revalidate

Setting a Default Policy

You can set the default policy by passing an options hash to default_caching_policy in your config.rb file after the activate :s3_sync ... end block:

default_caching_policy max_age:(60 * 60 * 24 * 365)

This will apply the policy to any file that do not have a mime-type specific policy.

Caching Policies

The Caching Tutorial is a great introduction to HTTP caching. The caching policy code in this gem is based on it.

The following keys can be set:

Setting Expires Header

You can pass the expires key to the caching_policy and default_caching_policy methods if you insist on setting the expires header on a results. You will need to pass it a Time object indicating when the resource is set to expire.

Note that the Cache-Control header will take precedence over the Expires header if both are present.

A Note About Browser Caching

Browser caching is well specified. It hasn't always been the case. Still, even modern browsers have different behaviors if it suits it's developers or their employers. Specs are meant to be ignored and so they are (I'm looking at you Chrome!). Setting the Cache-Control or Expires headers are not a guarrantie that the browsers and the proxies that stand between them and your content will behave the way you want them to. YMMV.

Path Specific Content Type

You can now set the content type of a path through the s3_sync.content_types hash. This hasi will take precendence over the content type discovered by the mime_types gem. The associated pull request has a few good examples on how to use this feature.

ACLs

middleman-s3_sync will set the resources's ACL to public-read by default. You can specificy a different ACL via the acl configuration option. The valid values are:

• private
• public-read
• public-read-write
• authenticated-read
• bucket-owner-read
• bucket-owner-full-control

The full values and their semantics are documented on AWS's documentation site.

Encryption

You can ask Amazon to encrypt your files at rest by setting the encryption option to true. Server side encryption is documented on the AWS documentation site .

GZipped Content Encoding

You can set the prefer_gzip option to look for a gzipped version of a resource. The gzipped version of the resource will be pushed to S3 instead of the original and the Content-Encoding and Content-Type headers will be set correctly. This will cause Amazon to serve the compressed version of the resource. In order for this to work, you need to have the :gzip extension activated in your config.rb.

Custom S3 Index and Error Documents

You can enable a custom index document and error document settings. The index_document option tells which file name gets used as the index document of a directory (typically, index.html), while error_document specifies the document to display for 4xx errors (ie, the 404 page).

A Debt of Gratitude

I used Middleman Sync as a template for building a Middleman extension. The code is well structured and easy to understand and it was easy to extend it to add my synchronization code. My gratitude goes to @karlfreeman and his work on Middleman sync.

Many thanks to Gnip and dojo4 for supporting and sponsoring work on middleman-s3_sync.

Contributing

1. Fork it
2. Create your feature branch (git checkout -b my-new-feature)
3. Commit your changes (git commit -am 'Add some feature')
4. Push to the branch (git push origin my-new-feature)
5. Create new Pull Request