Download the PHP package flownative/google-cloudstorage without Composer

Google Cloud Storage Adaptor for Neos and Flow

This Flow package allows you to store assets (resources) in Google Cloud Storage and publish resources to GCS. Because Neos CMS is using Flow's resource management under the hood, this adaptor also works nicely for all kinds of assets in Neos.

Key Features

• store all assets or only a specific collection in a private GCS bucket
• publish assets to a private or public GCS bucket
• supports GZIP compression for selected media types
• command line interface for tasks like connection check, emptying an GCS bucket or consistency check with clean up

Using this connector, you can run a Neos website which does not store any asset (images, PDFs etc.) on your local webserver.

Installation

The Flownative Google Cloud Storage connector is installed as a regular Flow package via Composer. Simply include flownative/google-cloudstorage into the dependencies of your Flow or Neos distribution:

Configuration

In order to communicate with the Google API, you need to provide the credentials of an account which has access to GCS (see next section for instructions for setting up the service user). Add the following configuration to the Settings.yaml for your desired Flow context (for example in Configuration/Production/Settings.yaml) and make sure to replace the private key with your own data:

Instead of using a file, the private key can also be specified directly, as a base64-encoded string. This allows for providing the private key via an environment variable:

You can test your settings by executing the connect command with a bucket of your choice.

Right now, you can only define one connection profile, namely the "default" profile. Additional profiles may be supported in future versions.

Bucket permissions, service users and keys

Public access

For the bucket to be readable by everyone on the Internet (needed to deliver assets), follow these steps:

1. In the Google Cloud console, go to the Cloud Storage Browser page.
2. In the list of buckets, click the name of the bucket that you want to make public.
3. Select the Permissions tab near the top of the page.
4. Click the + Add button.
5. The Add principals dialog box appears.
6. In the New principals field, enter allUsers.
7. In the Select a role drop down, select the Cloud Storage sub-menu, and click the Storage Object Viewer option. Note: That includes permission to list the objects in the bucket. If you don't want to grant listing publicly, use Storage Legacy Object Reader instead.
8. Click Save.
9. Click Allow public access.

This is takes care of the "read side" of affairs.

Service account setup

For Flow to be able to manage assets, you need to set up a service account with the needed rights.

1. In the Google Cloud console, go to the Service accounts page.
2. In the list of service accounts, click + Create Service Account
3. Fill in a name, adjust the generated Service Account ID as you like, add a description
4. Click Create And Continue
5. Now grant the Storage Object Admin role to the user
6. Click Continue
7. Click Done

Key generation

In order to authenticate for a service account, you need a service account key.

1. Go to the service account details page
2. Select the Keys tab near the top of the page.
3. Click the Add Key button, select Create new key
4. Select JSON as key type
5. The private key file will be downloaded - keep it safe! When lost, a new key must be generated.

The key file can now be used with the privateKeyJsonPathAndFilename option or as a base64-encoded string for privateKeyJsonBase64Encoded, as described above.

Publish Assets to Google Cloud Storage

Once the connector package is in place, you add a new publishing target which uses that connect and assign this target to your collection.

Since the new publishing target will be empty initially, you need to publish your assets to the new target by using the resource:publish command:

This command will upload your files to the target and use the calculated remote URL for all your assets from now on.

Switching the Storage of a Collection

If you want to migrate from your default local filesystem storage to a remote storage, you need to copy all your existing persistent resources to that new storage and use that storage afterwards by default.

You start by adding a new storage with the GCS connector to your configuration. As you might want also want to serve your assets by the remote storage system, you also add a target that contains your published resources.

Some words regarding the configuration options:

The keyPrefix option allows you to share one bucket across multiple websites or applications. All object keys will be prefixed by the given string.

The baseUri option defines the root of the publicly accessible address pointing to your published resources. In the example above, baseUri points to a subdomain which needs to be set up separately. If baseUri is empty, the Google Cloud Storage Publishing Target will determine a public URL automatically.

In order to copy the resources to the new storage we need a temporary collection that uses the storage and the new publication target.

Now you can use the resource:copy command (available in Flow 3.1 or Neos 2.1 and higher):

This will copy all your files from your current storage (local filesystem) to the new remote storage. The --publish flag means that this command also publishes all the resources to the new target, and you have the same state on your current storage and publication target as on the new one.

Now you can overwrite your old collection configuration and remove the temporary one:

Clear caches and you're done.

One- or Two-Bucket Setup

You can either create separate buckets for storage and target respectively or use the same bucket as storage and target.

One Bucket

In a one-bucket setup, the same bucket will be used as storage and target. All resources are publicly accessible, so Flow can render a URL pointing to a resource right after it was uploaded.

This setup is fast and saves storage space, because resources do not have to be copied and are only stored once. On the backside, the URLs are kind of ugly, because they only consist of a domain and the resource's SHA1:

Two Buckets

In a two-bucket setup, resources will be duplicated: the original is stored in the "storage" bucket and then copied to the "target" bucket. Each time a new resource is created or imported, it will be stored in the storage bucket and then automatically published (i.e. copied) into the target bucket.

You may choose this setup in order to have human- and SEO-friendly URLs pointing to your resources, because objects copied into the target bucket can have a more telling name which includes the original filename of the resource (see for the publicPersistentResourceUris options further below).

Customizing the Public URLs

The Google Cloud Storage Target supports a way to customize the URLs which are presented to the user. Even though the paths and filenames used for objects in the buckets is rather fixed (see above for the baseUri and keyPrefix options), you may want to use a reverse proxy or content delivery network to deliver resources stored in your target bucket. In that case, you can tell the Target to render URLs according to your own rules. It is your responsibility then to make sure that these URLs actually work.

Let's assume that we have set up a webserver acting as a reverse proxy. Requests to assets.flownative.com are re-written so that using a URI like https://assets.flownative.com/a817…cb1/logo.svg will actually deliver a file stored in the Storage bucket using the given SHA1.

You can tell the Target to render URIs like these by defining a pattern with placeholders:

The possible placeholders are:

• {baseUri} The base URI as defined in the target options
• {bucketName} The target's bucket name
• {keyPrefix} The target's configured key prefix
• {sha1} The resource's SHA1
• {filename} The resource's full filename, for example "logo.svg"
• {fileExtension} The resource's file extension, for example "svg"

For legacy and convenience reasons, the default pattern depends on the setup being used:

• no pattern and no baseUri set: https://storage.googleapis.com/{bucketName}/{keyPrefix}{sha1}
• no pattern set: {baseUri}/{keyPrefix}{sha1}/{filename}

The respective setup is auto-detected by the Target and the patterns set accordingly. You may, of course, override the patterns, by specifying the pattern setting as explained above.

Dynamic Custom Base Uri

Your application may take the responsibility to render a base URI by registering a custom method. After the options were set, the Target will call your method and use the returned string as a base URI.

This mechanism allows you to tweak the domain, or other parts of the base URI, depending on the current request. In the following example, we replace the domain "example.com" by "replaced.com", using a customer base URI method.

The following options are passed to your render method:

• targetClass
• bucketName
• keyPrefix
• baseUri
• persistentResourceUriEnableSigning

Publish Uris with Limited Lifetime

You can protect access to your resources by creating a private Google Cloud Storage bucket. For example, you can declare a bucket policy which grants access only to a service key owned by your application.

Let's say you generate invoices as PDF files and want to store them securely in a private bucket. At some point you will want to allow authorized customers downloading an invoice. The easiest way to implement that, is to generate a special signed link, which allows access to a given resource for a limited time.

The Google Cloud Storage Target can take care of signing links to persistent resources. Just enable signing and specify a signature lifetime (in seconds) like in the following example. Be aware though, that anyone with such a generated link can download the given protected resource wile the link is valid.

With this configuration, generated links will look like the following:

GZIP Compression

Google Cloud Storage supports GZIP compression for delivering files to the user, however, these files need to be compressed outside Google Cloud Storage and then uploaded as GZIP compressed data. This plugin supports transcoding resources on the fly, while they are being published. Data in the Google Cloud Storage storage is always stored uncompressed, as-is. Files which is of one of the media types configured for GZIP compression are automatically converted to GZIP while they are being published to the Google Cloud Storage target.

You can configure the compression level and the media types which should be compressed as such:

Note that adding media types for data which is already compressed – for example images or movies – will likely rather increase the data size and thus should be avoided.

Full Example Configuration for GCS